Sources of Human Exposure

Aluminium and its compounds comprise about 8% of the Earth’s surface; aluminium occurs naturally in silicates, cryolite, and bauxite rock. Natural processes account for most of the redistribution of aluminium in the environment. Acidic precipitation mobilizes aluminium from natural sources, and direct anthropogenic releases of aluminium compounds associated with industrial processes occur mainly to air. Certain uses lead to the presence of aluminium in drinking water and foodstuffs.

Bauxite is the most important raw material used in the production of aluminium. Bauxite is refined to produce alumina from which aluminium metal is recovered by electrolytic reduction; aluminium is also recycled from scrap. Aluminium hydroxide is produced from bauxite. In 2004, primary aluminium was being produced in 41 countries, the largest producers being China, Russia, Canada and the United States. In that year, worldwide production of primary aluminium, alumina and aluminium hydroxide reached about 30, 63, and 5 million tonnes per annum, respectively. More than 7 million tonnes of aluminium is recovered annually from recycled old scrap.

The largest markets for aluminium metal and its alloys are in transportation, building and construction, packaging and in electrical equipment. Transportation uses are one of the fastest growing areas for aluminium use. Aluminium powders are used in pigments and paints, fuel additives, explosives and propellants. Aluminium oxides are used as food additives and in the manufacture of, for example, abrasives, refractories, ceramics, electrical insulators, catalysts, paper, spark plugs, light bulbs, artificial gems, alloys, glass and heat resistant fibres. Aluminium hydroxide is used widely in pharmaceutical and personal care products. Food related uses of aluminium compounds include preservatives, fillers, colouring agents, anti-caking agents, emulsifiers and baking powders; soy-based infant formula can contain aluminium. Natural aluminium minerals especially bentonite and zeolite are used in water purification, sugar refining, brewing and paper industries.

Aluminium has not been classified with respect to carcinogenicity; however, “aluminium production” has been classified as carcinogenic to humans by the International Agency for Research on Cancer (IARC) (for further explanation, please see Effects on Humans, Effects from Occupational Exposure, Cancer). Occupational limits exist in several countries for exposures to aluminium dust and aluminium oxide. For non-occupational environments, limits have been set for intake in foods and drinking water; the latter are based on aesthetic or practical, rather than health, considerations.

Environmental Levels and Human Exposure

Aluminium may be designated as crustal in origin, and thus surface soils at uncontaminated sites constitute a source of soluble aluminium species in surface water and aluminium-containing particulates in sediments and ambient-air aerosols. Not surprisingly, the latter are present extensively in air samples in agricultural communities and when road dust is extensive. Environmental acidification is known to mobilize aluminium from land to aquatic environments. Interestingly, aluminium levels and its various forms (species) are often similar in source water and after its treatment with potassium alum as a flocculent during drinking water purification.

Workers in the aluminium production and user industries, as well as aluminium welders, experience considerable exposures to the metal and/or its compounds. In absence of occupational exposures and chronic use of aluminium-containing antacids and buffered aspirin, food is the major intake source of aluminium, followed by drinking water. When considering bioavailability, namely the fraction that is actually taken up into the blood stream, food is again the primary uptake source for individuals not occupationally exposed. However, chronic use of antacids, buffered aspirins and other medical preparations would likely constitute the major uptake source, even when exposed at work.

Kinetics and Metabolism

Effects on Laboratory Mammals and In Vitro Test Systems

Regardless of the duration of exposure, the toxicity attributed to aluminium is dependent upon the physiochemical properties (solubility, pH, bioavailability, etc.), type of aluminium preparation, route of administration, and physiological status (presence of renal dysfunction). Following oral exposure, aluminium distributes throughout the organism with accumulation in bone, kidneys and brain being of concern to humans with evidence of renal dysfunction, anemia or neurobehavioural alterations reported after excessive doses. The presence of aluminium in vaccines was found to be associated with macrophagic myofasciitis (MMF) at the site of i.m. injection. The toxicity of aluminium is affected by chelating agents and ligands although the mechanisms underlying toxicity remain unknown. However, it should be noted that only at excessive concentrations of aluminium are toxic manifestations seen and, hence aluminium is considered to possess a “low” potential for producing adverse effects.

Oral administration of aluminium did not affect reproductive capacity in males or females. Exposure to aluminium during gestation did not affect maternal health or development of the foetuses and neonates. Further, there was no evidence of teratogenic alterations in the foetuses of mothers fed dietary aluminium. Maternal dietary exposure to excessive amounts of aluminium during gestation and lactation resulted in neurobehavioural abnormalities in mouse offspring. At physiological concentrations the reproductive system does not appear to be a target for aluminium-induced effects; and if there is exposure during pregnancy, the growth and development of offspring of metal-treated mothers is not adversely affected.

The form of aluminium most often presented to tissues outside of the blood stream is expected to be bound to Tf. In brain, aluminium is prone to dissociate from Tf as a soluble citrate salt. Most cells of the central nervous system (CNS) express Tf receptor, and thus receptor-mediated uptake would be one mechanism by which aluminium could enter cells of the brain. Free flow endocytosis of aluminium citrate could be an alternative route of uptake. As outlined in Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, In Vivo Models, Neuropathology, there is at least one example of human pathology which is consistent with this mode of tissue exposure. Choroid plexus epithelia, cortical glia, and cortical neurons of patients exhibiting dialysis associated encephalopathy (DAE) develop intracellular argentophylllic granules that are lysosome-derived and intracytoplasmic. Uptake of aluminium-Tf complexes via receptor-mediated endocytosis would be expected to produce just such pathology. Whether aluminium, of any amount or speciation, escapes these compartments to impact on intracellular processes in humans is unknown. If relatively high doses produce pathology of such a distinctive nature, then it is reasonable to presume that lower doses of aluminium would follow similar pathways into the nervous system of humans.

In the studies of animals, it is important to note that a few reports have documented a pathologic accumulation of aluminium in intracellular lysosome-derived structures. Aluminium accumulation in lysosome-like cytoplasmic granules of retinal neurons in rats exposed to very high doses of aluminium was reported (see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, In Vivo Models, Rodent Models of Aluminium Toxicity by Direct Injection). Severe atrophy of the retina and loss of photoreceptors was also noted. Similarly, another study noted intracellular accumulations of aluminium in the brain of rats feed diets high in aluminium. For CNS it seems likely that the mode of delivery to the tissue is through Tf-mediated uptake. From animal studies and the clear association of aluminium exposure and DAE, it is clear that high levels of aluminium in CNS can lead to neurotoxicity. From the current literature it remains difficult to assess what a concentration of aluminium in serum (chronic levels) correlates with neurotoxicity. The effects of aluminium on the developing nervous system have also not been thoroughly addressed.

In regards to mechanisms by which aluminium could play a role in AD, there are both direct and indirect modes of potential action. In a direct mode, aluminium could potentiate the aggregation of molecules known to form pathologic lesions in AD. There is evidence that aluminium can promote the aggregation of β-amyloid peptide in vitro. However, whether aluminium would dissociate from Tf at an appreciable rate and bind β-amyloid peptide in vivo is unclear. One study found no association between AD-like pathology and long-term ingestion of aluminium. Indeed in this study of older patients, the incidence of AD-associated pathology in patients with DAE was no different from controls. Although these studies would suggest that there is little direct evidence for an association between AD and aluminium, a study of transgenic mice that produce Alzheimer-type amyloid pathology noted that mice feed diets high in aluminium showed increased levels of amyloid (see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, Alzheimer’s Disease). Moreover, it is well established in the rabbit that exposure to aluminium induces the formation of filamentous structures containing cytoplasmic neurofilament protein (see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, Motor Neuron Disease). Therefore, it is difficult to determine how a life-time of exposure to aluminium might influence the development of Alzheimer-type pathology by affecting the folding or clearance of “at-risk” proteins such as β-amyloid, tau, and α-synuclein.

Apart from the potential that aluminium might interact directly with molecules implicated in AD and related neurodegenerative disorders, studies in animals have revealed potential mechanisms by which aluminium might indirectly impact on the function of the nervous system. In Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, Alzheimer’s Disease, studies are described that reported aluminium may affect levels of cholesterol, which has been suggested in numerous studies as a potential modulator or Alzheimer-type amyloid formation. Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, In Vivo Models, Rodent Models of Aluminium Toxicity by Direct Injection describes several studies that have reported elevated levels of markers of oxidative stress in animals exposed to aluminium. These studies suggest potential mechanisms by which long-term exposure to aluminium could be deleterious and could synergistically worsen cognitive abilities in individuals that have pathologic abnormalities associated with AD.

However, there has not been strong evidence from animal studies that aluminium directly modulates cognitive function. As described in Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity,Behavioural Studies of Laboratory Animals Exposed to Aluminium, there have been several studies that have examined the cognitive abilities of mice and rats exposed to aluminium. For the most part, these studies did not report profound cognitive impairment even when exposed to very high levels of aluminium. Therefore, it seems unlikely that aluminium might lower the threshold for AD by blunting cognitive ability of adults.

Outside of the nervous system, the data regarding the potential for alumimium to cause abnormalities is mixed. There is clear evidence that sustained exposure to high levels of aluminium can cause bone abnormalities. Aluminium is clearly deposited in bone at sites of new growth. Bones in animals exposed to aluminium may show increased weakness and increased brittleness. Deficiencies in calcium or magnesium may exacerbate the effects of aluminium. Aluminium overload leads to PTH suppression and with regards to the bone, may be associated with altered calcium homeostasis.

Aluminium may also have negative effects on hematopoiesis. However, these effects are relatively mild unless animals are deficient in iron. In this latter setting, there will be increased levels of free Tf, which can then bind aluminium and compete for Tf receptor; further limiting the amount of iron available for erythrogenesis. Aluminium may also interfere with the metabolism of other metals. On this latter point, the strongest data, meaning most reproducible, suggest that aluminium exposure can lead to increased excretion of phosphorous.

From the present data, however, it is difficult to determine what level of exposure poses a risk for human health or which systems are most vulnerable. Based on projections from studies in dogs, individuals with sustained aluminium levels in serum that are 10-fold higher than the average range, or 1-2 μg/L, may be at increased risk for bone abnormalities. The exposure levels at which other systems might be affected are more difficult to project, particularly when trying to assess risk for late-onset illnesses.

Although not reported in every study, the majority of studies that utilized high doses of aluminium reported significant reductions in weight gain, particularly in studies initiated in young animals. The physiologic basis for this outcome is unclear, but it was reported that animals exposed to high doses of aluminium in drinking water consumed less food. Whether general effects of aluminium on metabolic processes depress metabolism or reduce nutritional efficiency remains to be resolved.

Experimental aluminium inhalation has been shown to produce effects interpreted as alveolar proteinosis and lipid pneumonia. Inhalation of aluminium had some protective effect against quartz dust-induced fibrosis in some, but not all, studies. Intratracheal aluminium instillation produced nodular fibrosis. Aluminium is used as an adjuvant in vaccines and hyposensitization treatments to precipitate toxins and toxoids, enhance their antigenic properties and reduce their rate of absorption and elimination. Aluminium can produce aluminium-species-dependent dermal irritation.

Experimental animal studies have failed to demonstrate carcinogenicity attributed solely to aluminium compounds. Often the response reported is associated with a tissue response to a foreign body rather than a direct effect of aluminium exposure. This appeared to be consistent across various routes of exposure from inhalation to intraperitoneal (i.p.) injection.

In agreement with their non-carcinogenic activity, aluminium compounds failed to show positive results in most short-term mutagenic assays and animal experiments to determine genotoxic potential of aluminium compounds lead to contradictory results with suggestions of an anti-genotoxic potential.

There is little reported for aluminium compounds in the way of immunotoxicity. There may be an altered immune response to challenge following excess aluminium exposure and this may be influenced by the health and hormonal status of the dam with increased susceptibility to bacterial infection seen in pregnancy.

Effects on Humans

Conclusions

This report synthesizes data from relevant studies on potential health effects of exposure to aluminium to quantify risk using the four-step process specified by the National Research Council: 1) hazard identification, 2) exposure assessment, 3) dose-response assessment, and 4) risk characterization.

Hazard identification qualitatively identifies adverse effects by route of exposure, and determines whether those effects are likely in humans at some level of exposure, perhaps much greater than exposure levels experienced in the population of interest. It is important to note that the identification of effects that can be caused by aluminium says nothing about how likely those effects are at exposure levels in human populations. That probability depends on the level of exposure and the dose-response relationship. This report classified the weight of evidence for each exposure pathway and health effect as strong, modest, limited, or having no clear evidence (see Table 25). We concluded that there is strong evidence that aluminium can cause irritation following exposure via either inhalation or injection. Modest evidence of an effect exists for reproductive toxicity following oral exposure, for neurological toxicity following either oral or injection exposure, and for bone toxicity following injection exposure. All other effects were judged to be supported by either limited evidence or no clear evidence at all. Exposure assessment, dose-response assessment, and risk characterization were conducted for those effects for which the evidence was judged to be either strong or modest. The remainder of this section describes our findings for the general population, subpopulations at special risk, and occupationally-exposed populations.

Table 25

Strength of evidence for health effects.

General population

Exposure assessment quantified aluminium intake and uptake (i.e., absorption of aluminium into systemic circulation) for a variety of pathways (see Table 26). For the general population, intake of aluminium from food (7.2 mg/day for females and 8.6 mg/day for males) dominated that from drinking water (0.16 mg/day) and inhalation exposure (0.06 mg/day). Antacids and buffered aspirin can contribute on the order of thousands of mg/day to aluminium intake. Relative contributions to uptake are ranked similarly to these intake contributions. However, because inhaled aluminium is approximately seven times more bioavailable than aluminium in drinking water, the contribution of inhaled aluminium to uptake (1.7 × 10^-5 mg/kg b.w./day) exceeds the corresponding contribution from drinking water (6.9 × 10^-6 mg/kg b.w./day). Uptake of aluminium in food is approximately 1 × 10^-4 mg/kg b.w./day. Aluminium uptakes from antacids and buffered aspirin amount to 3.1 × 10^-1 and 4.3 × 10^-2 mg/kg b.w./day, respectively.

Table 26

Intake and uptake of aluminium.

Relevant exposure levels of concern for the general population identified as part of dose response assessment included: irritation following inhalation (50 mg/m³), neurological effects due to drinking water exposure (100 μg aluminium/L water), reproductive toxicity due to oral intake (400 mg/kg-b.w./day), and irritation following injection (1 injection). We characterized risk (see Table 27) by calculating a margin of exposure, or MOE (the exposure level of concern divided by actual exposure), for each of these pathway-endpoint combinations. The MOE values were large for local irritation following inhalation (7000) and reproductive toxicity associated with oral intake (2900). For irritation following injection, the MOE is less than unity, although the severity of this endpoint is limited. For neurological effects associated with drinking water exposure, the MOE may be as small as unity. The evidence supporting this effect, however, comes from studies that have a number of methodological limitations, a finding that suggests the causal nature of the association is uncertain.

Table 27

Comparison between effects at which aluminium induces adverse health effects and levels of human exposure.

Subpopulations at special risk

Individuals with impaired renal function do not clear aluminium as effectively as healthy individuals. This population can also be exposed to extremely high levels of aluminium that are administered inadvertently via their intravenous feeds. This route of exposure may be particularly significant because it bypasses the barrier imposed by GI absorption characteristics. Infants, especially those born pre-term, are also vulnerable to aluminium exposure due to immaturity of the GI wall, the BBB, and the renal system. In addition to their added susceptibility due to compromised renal function, patients on dialysis may be subject to higher aluminium exposure levels if dialysis or intravenous fluid becomes contaminated, a problem that was more common in the past. Although not explicitly quantified, the susceptibility of these populations suggests that the exposure level of concern is less than it is for the general population. At the same time, some sensitive populations may have been exposed to very high aluminium exposures in the past. Because of the substantial quantities of injected fluids received by dialysis patients and their increased susceptibility, the MOE for this pathway for this population may be less than unity.

Occupationally-exposed populations

Occupational populations can be exposed to airborne concentrations of aluminium exceeding concentrations to which the general population is exposed by approximately three orders of magnitude (see Table 26). Aluminium intake resulting from these exposures is estimated to be 21 mg/day, compared to 0.06 mg/day for the general population, with uptake for occupationally exposed individuals amounting to 6 × 10^-3 mg/kg b.w./day, compared to 1.7 × 10^-5 for the general population (Table 26). The resulting margin of exposure for occupationally exposed populations is approximately 8, compared to 7000 for general population exposure to airborne aluminium (see Table 27).

Purity/Impurities, Additives

Most of the substances listed in Tables 1 and ​and22 are generally available in high purity and thus impurities are not an issue from a risk assessment perspective. However, it is clear that for many of the aluminium compounds, the degree of hydration can vary. Recently, the presence of a thin surface coating of ultrafine particles of sodium fluoride on aluminium oxide particulates has been demonstrated for aerosols collected in an aluminium refinery (Höflich et al., 2005; L’vov et al., 2005).

Physical and Chemical Properties

Properties of aluminium compounds

Table 3 summarizes the available physico-chemical properties of the compounds. Most of the aluminium compounds are solids exhibiting high melting points; some are liquids. No gaseous substances were identified. Only a few of the compounds sublimate, namely anhydrous aluminium chloride and fluoride, aluminium nitride and sulphide, as well as the complex with 8-hydroxyquinoline. Most of the substances are white or colourless.

Table 3

Available physical-chemical properties of aluminium and its compounds^a,^b

The water solubility of aluminium compounds is limited except for its salts, namely the chloride, nitrate, sulphate and chlorate (often as a corresponding hydrate). Salts of low molecular organic acids also have some water solubility (e.g., acetate, benzoate and lactate), as do salts containing aluminium anion complexes (e.g., ammonium hexafluoroaluminate and tetrachloroaluminate; sodium and potassium aluminate.) As explained in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation, pH is often a factor that can limit solubility in water. Solubility of inorganic aluminium compounds in organic solvents is limited to those which are anhydrous such as the bromides, chlorides, and iodides. Aluminium alkyls, alkyl halides, alkoxides and complexes of long-chain FAs and of high molecular mass organic ligands exhibit solubility in organic solvents.

Aluminium metal, aluminium oxide and aluminium hydroxide are nearly insoluble in water and organic solvents, while freshly prepared aluminium metal surfaces do react with water to form an inert protective coating. By contrast, powdered aluminium can react with water to yield hydrogen gas (see below Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation).

In terms of chemical reactivity, the following compounds are notable for their reactions with water: aluminium alkyls, alkyl halides, hydrides; the anhydrous halides (namely bromide, chloride and iodide); and the carbide, chlorate, nitride and phosphide. Explosive gases are released on contact with water, specifically hydrogen (H₂) from the hydrides and methane (CH₄) from the carbide. Release of toxic gases on hydration can also occur, that is chlorine dioxide (ClO₂) from the chlorate; ammonia (NH₃) from the nitride; phosphine (PH₃) from the phosphide; and hydrogen sulphide (H₂S) from the sulphide.

As described in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation, the Al³⁺ ion has a very high affinity for the hydroxide ion, even at relatively low pH values. This is consistent with the Class A (Hard) cation reactivity classification of Al³⁺, that is, it strongly prefers oxygen-containing organic ligands over those with nitrogen or sulphur as the donor atom. Its affinity for the halide anions increases in the order I⁻<Br⁻<Cl⁻≪F⁻ (Nieboer & Fletcher, 1996; Nieboer et al., 1999; Nieboer & Richardson; 1980). This reactivity classification is consistent with the stability or instability patterns towards water outlined for the aluminium compounds listed in Tables 1-​-3.3. As a Class A (Hard) cation, the chemistry of Al³⁺ resembles that of Mg²⁺, Ca²⁺, Na⁺ and K⁺. In fact, it may be viewed as a super Ca²⁺ or Mg²⁺ ion (Nieboer et al., 1999; Nieboer & Richardson, 1980), thereby often inhibiting the biological roles of these essential divalent cations, for example on the surface tissues of fish gills (Reid et al., 1991; Wilkinson et al., 1993).

In biological systems, Al³⁺, like Mg²⁺ and Ca²⁺, seeks out carboxylate and phosphate groups linked to macromolecules (i.e., proteins, RNA and DNA) or as constituents of low-molecular-mass ligands such as amino acids, nucleotides, citrate, phytates, lactate, carbonate, phosphate and sulphate (Harris, 1992). Because of the small size of the unhydrated Al³⁺, it can also bond to the phenolic group of the amino acid tyrosine in proteins. Most of the Al³⁺ in human serum is bound to the protein Tf (see Toxicokinetics, Distribution (Including Compartmentalization), Human Studies, Transport in Blood), which is a recognized carrier of trivalent metal ions, especially Fe³⁺ (Barker et al., 1990; Harris, 1992; Harris et al., 1996). Involvement of tyrosine phenolate groups in the Fe³⁺-Tf complex is well established (DaSilva & Williams, 1991). Under certain instances, such as in its citrate complex, Al³⁺ can also bind to a deprotonated alcohol group (Feng et al., 1990).

Chemical and morphological speciation

Formally, the definition of elemental speciation is limited to a chemical perspective; thus a chemical species is defined as: “a specific form of a chemical element, such as a molecular or complex structure or oxidation state” (Caruso et al., 2003; Templeton et al., 2000). However, Nieboer et al. (1999; 2005) subscribe to a broader working definition of “speciation”, that is: “an interdisciplinary field of activity concerned with all dimensions of the occurrence and measurement of an element in separately identifiable forms (i.e., chemical, physical or morphological)”. The former and more restrictive definition is employed below for the solution chemistry of Al³⁺, while the latter is helpful when considering the reactivity of aluminium oxide and aluminium hydroxide solids.

The strong dependence of Al³⁺ speciation on pH is illustrated by typical distribution curves in Figures 1 and ​and22.

Figure 1

Distribution of soluble, mononuclear aluminium ion species in aqueous solutions. Ordinate: mole fraction of aluminium ion occurring as each designated species. At any pH the individual mole fractions sum to unity. Reprinted with permission from Martin ...

Figure 2

Mole fraction of Al(III) v.s. pH distribution curves (solids) for a solution containing 0.1 mM citrate, 3 mM Ca²⁺, and 1 pM total Al(III). The distribution is only very weakly dependent on the concentrations of the three components. The dashed line labelled ...

From Figure 1, it is clear that, at physiological pH of 7.4, little or no free (hydrated) Al³⁺ exists in aqueous solution; the anion Al(OH)₄⁻ predominates. The distribution curve in Figure 2 illustrates the competition between hydroxide and citrate as ligand molecules. Under the conditions indicated in the legend to Figure 2, the competition with the OH⁻ is suppressed by the citrate tetra-anion. At physiologic pH, the AlLH_-1(OH)^2- complex dominates.

The formation of aluminium fluoride complexes in fluoridated drinking water has been debated extensively. Fluoridation of municipal drinking water supplies is a common practice for the prevention of dental caries; fluoride is added at a concentration of around 50 μmol/L (1 mg/L), corresponding to a pF (that is a −log[F]) of 4.3. Further, the pH of municipal water supplies is typically 8.0 ± 0.4 (e.g., Nieboer et al., 1995). Consequently, and with reference to Figure 3, the distribution curves depicted for pH 7.5 are the most relevant. Thus, again, Al(OH)₄⁻ is the dominant form in which Al³⁺ occurs, with little evidence for complexation with the fluoride ion.

Figure 3

Mole fraction of total Al(II1) v.s. pF = -log [F^-], where [F^-] is the ambient fluoride molar concentration; for fluoride complexes of aluminium at two pH values, dashed curves for pH 4 and solid curves for pH 7.5. Symbols on curves designate number of ...

As reviewed elsewhere (see, for example, Baes & Mesmer, 1976; Cotton & Wilkinson, 1980; Nieboer et al., 1995; Stumm & Morgan, 1996; Teagarden et al., 1981), at relatively high concentrations of Al³⁺ (≥ 100 μg/L) in the pH range 5.3 to 6.5, polymerization occurs and results in the formation of polynuclear species such as Al₁₃O₄(OH)₂₄⁷⁺.

In Human Exposure, Environmental Levels, Water, and Human Exposure, General Population Exposures, Drinking Water, the various operationally defined forms of aluminium in surface and drinking water are discussed, including complexes of natural organic ligands.

The reactivity of aluminium powders depends on their morphology (size, shape and surface area), bulk density and aluminium content. For example, Ilyin et al. (2002), have demonstrated that nanoparticles of fine aluminium powders exhibit maximal values of oxidation (combustion) rates compared to microparticles, and this occurred at lower temperatures. Not surprisingly, because of their thinness and corresponding high surface area, aluminium flake powders (see Sources of Human Exposure, Anthropogenic Sources, Uses, Aluminium Powders) also are relatively reactive. Trunov et al. (2005) and Meda et al. (2004) have reported similar findings. Consequently fine and ultrafine aluminium powders show better promise as propellant additives than do the more conventional-sized (of the order of 10μm) aluminium powders. Interestingly, the combustion products of nano-sized aluminium powders are also different, such as a higher proportion of low-temperature aluminium oxide polymorphs (Meda et al., 2004) (see below) and product morphology. Ilyin et al. (2002) demonstrated that combustion of spherical micro-sized aluminium powders resulted in spherical products, while spherical fine powders produced submicron needles. As explained in the next paragraph, this is important for human exposure characterization.

The size of aerosols is important in terms of where they are deposited in the respiratory system and where they exert their toxic effects. In the workplace it is now common to consider three health-related aerosol fractions (Nieboer et al., 2005; Oller & Bates, 2005; Vincent, 1995). The inhalable aerosol fraction corresponds to the total amount of airborne particulates that enters the body through the nose and/or mouth during breathing (aerodynamic diameters (d_ae) of ≤ 100μm). The thoracic aerosol fraction penetrates the tracheoalveolar region of the lung (d_ae < 28μm), while the respirable aerosol fraction (d_ae < 10μm) penetrates the alveolar region of the lung (includes the respiratory bronchioles, the alveolar ducts and sacs). For outdoor aeorosols, it has been customary to measure the particulate matter (PM)₁₀ fraction, which corresponds to the thoracic fraction but with particles larger than d_ae=10μm excluded (Nieboer et al., 2005; Vincent, 2005). For a description of these differences, the reader is referred to the review by Vincent (2005) in which the various aerosol fractions criteria are depicted graphically as a function of aerodynamic diameter. More recently, exposure to fine particles (d_ae<2.5μm) has become the focus in relation to increased rates of cardiovascular and respiratory diseases (Dockery et al., 1993; Englert, 2004; Pope et al., 2002). Ultrafine particles (d_ae < 100 nm) are also gaining in importance. Fine and ultrafine particles are included in the respirable and PM₁₀ fractions. It is clear from the material presented in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Properties of Aluminium Metal and the previous paragraph that, depending on the exact industrial process involved, aluminium powder workers have the potential of being exposed to some or all of the aerosol fractions discussed. This also appears to be the situation in aluminium refinery workers (Höflich et al., 2005; Skaugset et al., 2005; Thomassen et al., 2006).

Aluminium oxide (Al₂O₃) occurs in two major forms. α-Al₂O₃ [corundum; CAS No 1302-74-5] constitutes a high temperature form and is formed on heating aluminium hydroxide, Al(OH)₃, at a temperature of 1000 °C or above. It is very hard and resistant to hydration and attack by acids; it occurs in nature as corundum (Cotton & Wilkinson 1980). γ-Al₂O₃ is generated at 500°C and readily takes up water and dissolves in acids. Other minor forms generated when heating Al(OH)₃ include: χ-, κ-, δ-, θ-Al₂O₃ (Pearson, 1992; Trunov et al., 2005). Recent studies suggest that the structure of the surface layers of α-Al₂O₃ depends on hydration: on the surface of a highly polished single crystal, the aluminium atom (more correctly Al³⁺ ions) are exposed, while in the ideal (unrelaxed) structure the oxygen atoms (actually oxide anions, O^2-) are at the surface. On hydration, the latter are overlaid by a semi-ordered absorbed water layer with the presence of extensive hydrogen bonding and hydroxyl groups resulting in a relaxed structure. In fact, the hydrated surface structure appears to be in between that of the ideal α-Al₂O₃ surface and that of γ-Al(OH)₃. The Al³⁺ ions in the Al-terminated surface are strong Lewis acids and react strongly with water, while surface hydroxyl groups are Lewis bases that interact with metal ions (Eng et al., 2000). Surface reactivity constitutes the basis for the use of aluminium oxide and hydroxide as industrial catalysts, absorbents and chromatography packing materials (i.e., as stationary phases). Not surprisingly, the reactivity of solid Al₂O₃ depends on its specific crystal structure and hydrophilic/hydrophobic surface properties, and thus the degree of surface hydration (Pearson, 1992).

Gibbsite, γ-Al(OH)₃ [CAS No 14762-49-3] is one of the three minerals that make up bauxite ore; the others being Al₂O₃ and the oxyhydroxides boehmite [α-Al₂O₃ • H₂O or α-AlO(OH)] and diaspore [β- Al₂O₃ • H₂O or β-AlO(OH)]. Of the latter, diaspore is the high temperature/high pressure form (Pearson, 1992). Gibsite has three structural polymorphs, namely bayerite [α-Al(OH₃); CAS No 20257-20-9], nordstandite [β-Al(OH₃); CAS No 13840-05-6] and doyleite which is rather rare (Mineralogy Database, 2006). On heating the Al(OH)₃, polymorphs follow a different Al₂O₃ transition sequence in reaching the high temperature α-Al₂O₃ form (Pearson, 1992).

In simple terms, the surface layer of a metal binary compound has exposed metal ions (and anions) with reduced coordination number, which can behave as Lewis acids (or bases). At the solid-solution interface, proton association and dissociation can lead to pH-dependent surface charges and complexation. This is further complicated by specific adsorption of cations or anions right at the solid/solution interface, with some ordering of counter-ions in a more diffuse layer at points further into the solution (Stumm & Morgan, 1996). A zero point of charge occurs, which is pH dependent. It is also influenced by the extent of specific ion adsorption. In addition to surface electrostatic and complexation reactions, hydrogen-bonding and London-van der Waals forces can be involved in the adsorption of surfactants, non-polar organic solutes, polymers and polyelectrolytes (Stumm & Morgan, 1996).

Adsorption capacity is central to some of the major uses of Al₂O₃, Al(OH)₃ and other aluminium compounds (e.g., aluminium phosphate, AlPO₄). Adsorption of antigens onto Al(OH)₃ and AlPO₄ constitutes the basis for their use as vaccine adjuvents (Gupta, 1998). At neutral pH, gels of these compounds have different charges, the phosphate being negative and the hydroxide positive. This is important, relative to the charge borne by the antigen at physiological pH. The various forces described are optimized by adjustment of the pH and ionic strength of the medium, temperature, particle size of the adsorbent, and the surface area of the latter (Gupta, 1998). Similarly, the natural hydrophilic surface characteristic of Al₂O₃ is central to its use as solid-phases in chromatography. More recently, stable surface coatings have been developed which render the Al₂O₃ surface hydrophobic, which makes it even more versatile in chromatographic applications. Al₂O₃ and the aluminium oxyhydroxides are used to remove moisture from gases such as argon, alkanes, and sulphur dioxide. They are also used to remove hydrogen fluoride (HF) from air by adsorption; fluoride ions can also be removed effectively by them from water. Fluoride is adsorbed on to alumina at low pH values and can be desorbed on increasing the pH. Clearly, the inhalation of particulates of Al₂O₃ and related oxyhydroxides in aluminium smelting operations may constitute a delivery vehicle for adsorbed HF (Höflich et al., 2005; L’vov et al., 2005). Finally, the use of finely divided aluminium metal, Al(OH)₃, aluminium potassium silicate and other aluminium compounds to generate dye (colour) lakes (see Table 2) stems from the surface adsorptive forces/capacities described.

Classification

Of the three substances reviewed in detail in this report, only aluminium powder is classified in Annex 1 of the European Economic Union Council Directive 67/548. Indeed, very few of the compounds listed in Tables 1 and ​and33 are classified; those that are listed are recognized as hazardous and are widely used, such as cryolite and aluminium phosphide (see Identity, Physical and Chemical Properties, Analytical Methods, Identity).

Although aluminium (EINECS # 013-001-00-6) is not included in a priority list, it is considered hazardous in powdered form. Aluminium powder is classified as: F; R 10,15 and the following risk (R) and safety (S) phrases have been assigned:

Cryolite is on the EEC Priority List #3 and has the risk classification: T; R20/22-N; R48/23/25-Xn; R51-53. Assigned risk (R) and safety (S) phrases are:

Aluminium phosphide is not listed on a priority list and has the classification of: F; R15/29-T; R28,32-N; R50. The risk and safety phrases are:

Anthropogenic Sources

Aluminium is released and dispersed in the environment by natural processes and from human activity. Natural processes account for most of the redistribution of aluminium in the environment (ATSDR, 1999; IPCS, 1997; Wagner, 1999) as a result of the weathering of rocks and minerals in which it is present. Mobilization from natural sources can, however, also result from the deposition of acidic precipitation (IPCS, 1997; Wagner, 1999). Direct anthropogenic releases of aluminium compounds occur primarily to air and these are associated with industrial processes. Thus, the mining and processing of aluminium ores and the production of aluminium metal, alloys and compounds can lead to the release of aluminium compounds into the environment. The use of aluminium and its compounds in processing, packaging and storage of food products, and as flocculants in the treatment of drinking-water may contribute to its presence in drinking-water and food stuffs (ATSDR, 1999).

Production levels and processes

Trends in production

Bauxite

In 1992, worldwide production of bauxite was 106 million tonnes; based on a comparison of this quantity with the quarterly average values for 1993 and 1994, production in major producing countries appeared to be fairly constant (IPCS, 1997; World Bureau of Metal Statistics, 1994). However, by 2003 world mine production of bauxite had reached approximately 146 million tons (150 million tonnes) and rose to about 156 million tons (160 million tonnes) in 2004 (USGS, 2005a)¹.

Alumina

In 1992, total alumina production worldwide was 33.8 million metric tonnes; this total production figure included 30 million tonnes for metallurgical uses and 3 million tonnes for non-metallurgical uses (IPAI, 1993; IPCS, 1997). By 1998, alumina production reached 47.5 million tonnes (USGS, 2002) and by 2004 was close to 63 million tonnes; Table 4 shows the production figures for different global regions for the period 2000 to 2005 (IAI, 2006a; USGS, 2005b; USGS, 2006a).

Table 4

World production of alumina 2000-2005^a.

Aluminium hydroxide

Worldwide production of aluminium hydroxide in 2004 was estimated to be 5 million tonnes per annum (E. Nordheim, personal communication, 2006).

Aluminium

It is noted by the International Programme on Chemical Safety (IPCS) (IPAI, 1993; IPCS, 1997) that annual worldwide production of primary aluminium was 14.8 million tonnes in 1992; by 2004 production essentially doubled to 29.6 million tonnes (Table 5). Between 2001 and 2004 production increased by between 6 and 8% per annum. In 2004, primary aluminium was being produced in 41 countries, the largest producers being China (22% of world production) followed by Russia (12%), Canada (9%), the United States (8.5%), Australia (3.8%), Brazil (5%), and Norway (4.5%) (USGS, 2004). China and Russia surpassed the United States as the largest producers of primary aluminium in 2001 (Figure 4).

Figure 4

Trends in aluminium production by country. Data from USGS (2004).

Table 5

Primary aluminium production worldwide for 2000-2005^a.

In addition to primary aluminium production, more than 7 million tonnes is produced per year from post consumer (old) recycled scrap. Almost 100 per cent of all production scrap and over 60 per cent of all old scrap are recycled; it has been noted that the proportion of aluminium produced from scrap (“recycled aluminium”) is rising rapidly (IAI, 2005).

The recycling of aluminium requires much less energy than that used to recover the metal from its ores. The annual amount of aluminium recovered from purchased and tolled (new and old) scrap was about 15 million tonnes in 2004 (IAI, 2005). The automotive industry is the largest consumer of recycled cast aluminium accounting for about 70% of production. It was expected that demand for secondary aluminium would increase significantly as the automotive industry addressed the growing need for lighter vehicles (Wagner, 1999).

In 2004, the total shipments worldwide of aluminium ingots including primary and recycled metal were estimated at 62 million tonnes; shipments of aluminium powder were estimated to be 0.2% (75 000 tonnes per year) of this total (E. Nordheim, personal communication, 2006).

Uses

Industrial and non-industrial uses of aluminium and its compounds are summarized in Tables 6 and ​and77.

Table 6

Summary of industrial uses of aluminium and its compounds – industrial compounds^a.

Table 7

Summary of uses of aluminium and its compounds - pharmaceuticals, food additives, cosmetics and other household products^a.

Aluminium

Aluminium metal is light in weight and is durable because surfaces of products made from it are oxidized to form a thin protective coating of aluminium oxide (alumina). However pure aluminium is extremely soft and therefore is often mixed with other metals and elements (e.g., copper, magnesium, manganese, silicon, lithium and zinc) to form alloys which are stronger and harder and hence of increased versatility (ATSDR, 1999; Wagner, 1999). The tensile strength of some copper-aluminium alloys can exceed that of mild steel by as much as 50% (Wagner, 1999). Reference to the uses of aluminium thus normally relates to those for aluminium as an alloy.

Aluminium metal and its alloys are used extensively in building construction (e.g., siding, roofing, doors, windows), in transportation (in the manufacture of automobiles and aircraft), in packaging (e.g., for beverage cans), and in electrical equipment. Other uses include die-cast motor parts, cooking utensils, decorations, road signs, fencing, beverage cans, coloured kitchenware, food packaging, foil, corrosion-resistant chemical equipment, solid fuel rocket propellants and explosives, dental crowns, jewellery and denture materials. Aluminium is also used for power lines, electrical conductors, insulated cables and wiring (ATSDR, 1999; IAI, 2006b). The largest markets for aluminium are transportation (27%), building and construction (23%), packaging (16%) and electrical equipment (10%). Transportation uses are one of the fastest growing areas for aluminium use showing a growth rate of about 4% per annum in 1999 (Wagner, 1999). Table 8 lists the approximate distribution of the different product segments in which aluminium is used on a global basis (E. Nordheim, personal communication, 2005)

Table 8

Global markets for aluminium products^a.

The major markets for aluminium products in the United States in 1997 are summarized in Table 9 (ATSDR, 1999; USGS, 1997).

Table 9

Major (U.S.) markets for aluminium products in 1997^a.

Over 95% of beer and carbonated drinks are packaged in aluminium cans. Other food-related applications of the metal are as sheet and foil for pie plates, frozen food trays and other food packaging, cooking utensils. Consumer product and medical uses include toys, jewellery, paints and protective coatings, and dental alloys for crowns and dentures (ATSDR, 1999; USGS, 1997).

Environmental Levels

Air

Prior to 1987, it was the practice to measure particulate air pollution levels as total suspended particles (TSP) (Samet et al., 2000). TSP constitutes air-borne particles with d_ae <30 μm (Cyrys et al., 2005). In 1987, annual and 24-hr standards were promulgated by the U.S. EPA for the PM₁₀ aerosol fraction (see Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation) of 50 and 150 μg/m³, respectively. In 1997, this agency added standards for PM_2.5, namely 15 (annual) and 65 μg/m³ (24-hr) (EPA, 1997). Both the PM₁₀ and PM_2.5 aerosol fractions have been associated with adverse health effects as reported in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation. Reported ambient aluminium concentrations in these fractions are summarized in Table 13. A perusal of these data indicates mean PM_2.5 aluminium levels were in the range 0.035 to 1.82 μg/m³ (excluding the arctic and Antarctica sites) and 0.58-6.97 μg/m³ for PM₁₀. By comparison, the mean total particulate mass reported in the PM_2.5 and PM₁₀ fractions for non-remote sites were near or often exceeded the annual standards indicated. Detailed analyses of these data outlined in the original publications have identified the trends and interpretations summarized in the next paragraph.

Table 13

Concentrations of aluminium in ambient aerosol fractions.

Aluminium, like silica, may be designated as crustal in origin (Laden et al., 2000; Lee et al., 1994). Calculations of the enrichment factor (the ratio of the concentration of an element of interest and that of a signature element in the air sample divided by the same concentration ratio in the Earth’s crust) support this, namely values below 7.0 (Lee et al., 1994). Further corroboration comes from principal component analysis and the general characteristic that anthropogenic elements (e.g., sulphur and lead) accumulate predominantly in the PM_2.5 fraction, while crustal elements (e.g., silicon, titanium, iron, and aluminium) occur frequently in coarser fractions such as such as PM₁₀ (Orlic et al., 1999). The latter was observed for the central California data in Table 13, for which aluminium concentrations for the two agricultural sampling stations were considerably higher in the PM₁₀ fraction (Chow et al., 1996). Road dust, industry and urbanization are also recognized factors in generating higher ambient aluminium levels (Chow et al., 1996; Lee et al., 1994; 2003; Pinto et al., 1998). Finally, remote sites well away from towns and cities exhibit the lowest aluminium air concentrations. Examples are (see Table 13) the Sequoia and Yosemite National Parks in California (Chow et al., 1996), the high arctic (Maenhaut et al., 1996) and Antarctica (Mazzera et al., 2001).

Occupational Exposure

Aluminium production

Exposures experienced by workers employed in aluminium production are influenced by the industrial processes involved; an overview of the primary refining steps and a brief introduction to secondary refining of this metal are presented in Sources of Human Exposure, Anthropogenic Sources, Production Levels and Processes, Aluminium.

Aluminium refinery workers can be exposed to aluminium hydroxide in the Bayer process, and “to airborne particles of variable composition (e.g., aluminium oxides, cryolite, soot), a number of gases (e.g., carbon oxides, sulphur dioxide (SO₂), hydrogen fluoride (HF)) and PAHs are generated” (Höflich et al., 2005) in the Hall-Héroult process (also Benke et al., 1998). The emission of dust, including PAHs is more extensive for the Søderberg pots and anode plants. As already pointed out (see Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation), adsorption of HF and SO₂, and presumably also PAHs, onto aluminium oxide and oxyhydroxides is an issue of concern; beryllium emissions pose an additional health risk (Benke et al., 1998). Ultrafine particles in the 10-300 nm diameter range have also been identified in potrooms (Höflich et al., 2005; L’vov et al., 2005; Thomassen et al., 2006). By comparison, for secondary aluminium smelting and relative to accepted occupational exposure limits, inhalable dust, particulate fluorides, HF, lead and aluminium appear to constitute a health risk (Healy et al., 2001).

Although personal measurements of exposure to dust containing aluminium oxide and other aluminium compounds have been made in the aluminium production industry, they are not extensively documented in the published literature. The emphasis has rather been on PAHs, fluorides and HF measurements (Benke et al., 1998). However, the reports by Pierre et al. (1995; 1998) provide a helpful overview. They surveyed 234 workers employed in primary aluminium refining, 88 in secondary aluminium refining and 13 in aluminium powder production. A summary of their findings is given in Table 14. Aluminium air concentrations were measured as the “total” aerosol fraction. Relative to the inhalable aerosol fraction defined in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation, “total” personal samplers undersample particles larger than 15 μm (Vincent, 1994). Unfortunately, for most of the worker groups surveyed, only mean values are available.

Table 14

Individual exposures to aluminium and its compounds: average daily concentrations in mg/m³ (adapted from Pierre et al., 1995; 1998).

Several noteworthy trends are evident from the data in Table 14. Exposure to bauxite, aluminium and the metal (as powder or sheets) is associated with low concentrations of water-soluble aluminium, by contrast to exposures in the potrooms, secondary smelting, and AlF₃ production. Consequently, one might expect to observe higher urinary aluminium concentrations among workers in the latter group, which was indeed the case. Surprisingly, the aluminium powder workers exhibited substantially higher before-shift concentrations in urine than other workers. This observation suggests that particulates that deposited and accumulated in the respiratory tract serve as sinks. Further, Gitelman (1995) and Gitelman et al. (1995) conducted a survey involving 40 control subjects and 235 workers employed in 15 plants in the USA engaged in primary and secondary aluminium refining and in the manufacture of products (e.g., powder technologies, cables, rolling mills). Personal exposures ranged from 0.01-1.20 mg/m³, with a median of 0.025 mg/m³ for respirable fractions (<10 μm), and 0.001-3.0 mg/m³ with a median of 0.10 mg/m³ for the “total” aerosol fractions. The maxima in their survey likely pertained to primary refining and work with powders.

The reports by Röllin et al. (1996; 2001) suggest that newer potroom technologies can substantially reduce exposure to aluminium. In a new primary smelter, the median “total” aluminium levels were 0.03 to 0.084 mg/m³, while in a plant using the more standard potroom smelting approach exposures were: mean (range) in mg/m³ of 1.47 (1.25-1.66), potroom 1; 0.35 (0.20-0.57), potroom 2; and other operator, 0.036 (0.02-0.13).

An extensive survey of 7 secondary aluminium smelters in the UK found inhalable aluminium air concentrations of 0.04 to 0.90 mg/m³, with a mean value of 0.31 mg/m³. These measurements are comparable in magnitude to the concentrations reported in Table 14 for jobs involving slag treatment. Healy et al. (2001) did indeed identify the slagging out of rotary furnaces as a dusty operation. Comparable concentrations have been reported for two foundries in South Africa: 0.17 mg/m³ for smelters; 0.027 mg/m³ for operators; and 0.58 mg/m³ for fettlers (i.e., those lining furnaces) (Rollin et al., 1991a).

General Population Exposures

Food and beverages

It is clear that, in the absence of aluminium containing additives, frequently eaten foods contain relatively low concentrations of aluminium. Mean fresh-weight concentrations (in μg/g) typically increase in the order: beverages (1.5); fruit (2.7); fish (fresh or tinned, 3.2); milk, dairy products (4.5); meat, sausage, offal (5.4); vegetables (5.7); sugar, sugar-rich products (6.7); bread, cake and pastries (7.4); edible seeds (beans, peas, etc., 9.3); and meal, flour (9.5) (Müller et al., 1998). Items with higher mean concentrations (in μg/g fresh weight) of which smaller amounts are ingested include: herbs (19); cocoa and cocoa products (33); spices (145) and tea (900) (Müller et al., 1998). Steeped teas often contain 2-5 mg/L of aluminium (Baxter et al., 1989; Jackson & Huang, 1983; Müller et al., 1998; Pennington & Jones, 1989). Rajwanshi et al. (1997) reported concentrations double these values. Pennington & Jones (1989) and Pennington & Schoen (1995) clearly illustrated that food items containing aluminium additives (see Table 7) can appreciatively add to the dietary aluminium intake. They identified the following items as contributing significantly to the estimated daily intake of 11.5 mg/day calculated for a 14-16 year old male: cornbread (36.6% of total intake); American processed cheese (17.2%); pancakes (9.0%); yellow cake with icing (8%); taco/tostada (3.5%); cheeseburger (2.7%); tea (2.0%); hamburger (1.8%); fish sticks (1.5%). It was suspected that aluminium additives in processed cheese and baking ingredients were responsible. Saiyed & Yokel (2005) have confirmed that, due to the presence of aluminium additives (e.g., baking soda; leavening, anticaking, or emulsifying agents (Nieboer et al., 1995); also see Table 7), certain food ingredients and items constitute major sources of dietary aluminium. Examples are: processed cheese; pizzas; flour mixes for cakes and pancakes, and thus many baked goods and crusts; table salt; hot cocoa mix; pickle relish; and non-dairy creamers. Although the details are addressed in Human Exposure, General Population Exposures, Medical, aluminium intake from antacids, buffered aspirins, and antidiarrheal agents would increase by many fold the daily intake of this metal.

In the IPCS (1997) monograph, estimates of the average adult dietary daily intake of aluminium are tabulated for 8 countries. These data and other reported values for adults are depicted in Figure 5. It is clear that, since the mid 1980s, the estimates fall below 15 mg/day. As suggested in the IPCS (1997) review, the lower values in this group likely reflect reduced use of aluminium food additives in some of the countries. The intakes in Figure 5 estimated for or reported in 1985 or before of 20-25 mg/day may be due to the inclusion of larger portions of food items with additives, although analytical limitations cannot be excluded.

Figure 5

Estimates of aluminium consumption (by year)

Intakes for infants appear considerably lower; for example, a 4-month old infant consuming cow’s milk-based formula was estimated to take in 0.03-0.05 mg/day, compared to 0.27-0.53 mg/day when fed soya-based formula (MAFF, 1993; also see Dabeka & McKenzie, 1992; and Pennington & Schoen, 1995). Navarro-Blasco & Alvarez-Galindo (2003) have reported comparable findings (based on the previous PTWI of 7 mg/kg b.w., rather than the current value of 1 mg/kg b.w.): “Standard formulae gave lower intakes amounting to about 4% PTWI; specialized and preterm formulae resulted in moderate intake (11 to 12 and 8 to 10% PTWI, respectively); and soya formulae contributed the highest intake (15% PTWI)”.

The daily intakes in Figure 5 derived from the Pennington & Jones (1989) and Pennington & Schoen (1995) studies include foods that have been in contact with aluminium foil or aluminium containers or that have been cooked in aluminium utensils. It is clear that these storage and preparation steps contribute aluminium to daily intake, especially for acidic foods (e.g., ATSDR, 1999; Fimreite et al., 1997; IPCS, 1997; Neelam et al., 2000; Ranau & Oehlenchälger, 2001). However, relative to the aluminium food additives, such contributions appear not to be appreciable, representing only a small fraction of the total dietary intake.

Drinking water

A perusal of recent reports of total and total dissolved aluminium (for definitions see Human Exposure, Environmental Levels, Water and Table 15) in drinking water after municipal treatment indicated that levels are generally below 7.4 μmol/L (200 μg/L). Recently reported concentrations of dissolved aluminium and the year collected are: 1.36±1.28 μmol/L (38±35 μg/L; 1995/6; (Gauthier et al., 2000); and 2.0-3.6 μmol/L (53-96 μg/L; 1995/6; see Table 15). Total aluminium levels are of comparable magnitude: 0.85-3.9 μmol/L (23-105 μg/L; 1993/7; (Zhao et al., 2001); 0.2-6.1 μmol/L (4-165 μg/L; c.2000; López et al., 2002); and a mean of 2.2 μmol/L (60 μg/L; 1999/2000; Gillette-Guyonnet et al., 2005). Compared to the compilation for the 1980s reported by Nieboer et al. (1995) there appears to be some moderate downward trend in the reported total aluminium concentrations.

Table 15

Mean concentrations of total aluminium (μg/L) and proportion of constituent fractions % in drinking water (after purification treatment).

The speciation summary provided in Table 15 illustrates that, after purification treatment, dissolved aluminium levels are not very different from those observed for surface (raw) water (see Human Exposure, Environmental Levels, Water). As in raw water, both inorganic and organic forms are present. However, operational filtering processes in treatment plants using aluminium coagulation can remove or add to the residual dissolved forms in drinking water (Berubé, 2004; Cech & Montera, 2000; Flaten, 2001). Other operational parameters are also known or suspected to influence the amount of dissolved aluminium in drinking water. As the pH decreases or increases from 6, the water-solubility of aluminium increases (Driscoll & Letterman, 1995; Driscoll & Schecher, 1989; Nieboer et al., 1995). Mean pH values of treated drinking water typically are around 8.0, with minimum values as low as 5.3 and maxima just above 9.0 (e.g., Nieboer et al., 1995). Further, addition of filter aids such as sulphate and non-ionic polymers appear effective in repressing residual dissolved aluminium; and solubility increases with water temperature (Driscoll & Letterman, 1995; Kvech & Edwards, 2002). Interestingly, the water quality data reported by Gauthier et al. (2000) suggest that the source of the raw water (e.g., lake, river or ground) may have some bearing. And finally, the type of water purification technology employed is an important determinant. Consequently, and as observed, dissolved aluminium concentrations in household water may be expected to be quite variable.

Many of the organic and inorganic water-soluble species discussed in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation and those shown to exist in surface waters and soil extracts (see Human Exposure, Environmental Levels, Water) may be expected to be stable at the pH values of drinking water.

Aluminium concentrations in bottled commercial drinking water have been reported in a small number of instances. Rosborg et al. (2005) report a median level of total aluminium of 1.33 μmol/L (36 μg/L). Interestingly, when they compared carbonated samples from a single brand, one contained in a plastic bottle and the other in an aluminium can, the respective concentrations were 0.63 μmol/L (17 μg/L) and 2.7 μmol/L (72 μg/L) respectively. Gillette-Guyonnet et al. (2005) reported that aluminium concentrations in 3 of 8 commercial mineral waters obtained in 1992 were below the detection limit of 0.1 μmol/L (3 μg/L), while the remainder had values between 0.2 μmol/L (5 μg/L) and 1.2 μmol/L (32 μg/L). By comparison, levels for 6 city water supplies sampled in the same time period ranged from 0.4 μmol/L (10 μg/L) to 2.3 μmol/L (63 μg/L), and another was 2.2 μmol/L (60 μg/L) in 1999-2000. Finally, López et al. (2002) found a mean ± SD of recoverable aluminium (that found on digestion at 120°C for 90 minutes) of 2.1±1.6 μmol/L (58±43 μg/L) in both 15 regional samples of tap water and 11 samples of glass bottled water bought in supermarkets. By contrast, mineral water purchased in plastic bottles had considerable higher concentrations, namely 4.5±1.2 μmol/L (121±32 μg/L). Leaching from the plastic storage bottles was suspected in this case.

Miscellaneous exposures

Because of the multiple uses of aluminium and its compounds (see Tables 6 and ​and7),7), employment in workplaces where they are manufactured or used can potentially lead to exposures not systematically documented in the scientific literature.

Physico-chemical properties (explosivity, flammability, oxidizing potential)

As described in Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Properties of Aluminium Compounds / Chemical and Morphological Speciation, and Identity, Physical and Chemical Properties, Analytical Methods, Classification, explosivity, flammability, oxidizing potential and related reactivities are not considered to be of primary interest with respect to assessing the health impacts of exposure to aluminium and its compounds.

Absorption

Animal studies

Inhalation exposure

Inhalation exposure to aluminium is known to result from cosmetic use and from occupational and environmental sources.

The size of the inhaled particles is expected to have a profound effect on the deposition and absorption of aluminium in the lung. Dust comprises particles from < 1 to > 100 μm in diameter; inhalable particles are those with diameters up to 10 μm. Particles having diameters up to 10 and 2.5 μm are now classified as PM₁₀ and PM_2.5 respectively (for more details on these aerosol fractions see Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Chemical and Morphological Speciation). Ultrafine particles have diameters < 0.1 μm. Particles can be removed from the respiratory tract by mucociliary clearance, the movement of mucous that covers the respiratory epithelium by cilia projecting from cells lining the respiratory tract. The mucous is moved up and out of the respiratory tract into the upper GI tract. This process has the potential to contribute to the oral route of exposure for substances initially deposited in the respiratory tract. Experimental studies have not isolated the pulmonary from other absorption sites (Rollin et al., 1991b).

The only data from which one can estimate the percentage of aluminium absorbed from inhalation exposure is from exposures in the occupational environment (see 5.1.2.1 below). As the percentage of aluminium estimated to be absorbed during inhalation exposure is greater than from oral aluminium intake (see 5.1.2.2 below), it seems unlikely that absorption from the GI tract accounts for the absorption of all inhaled aluminium. There has been no estimate of the percentage of aluminium absorbed via the intra-nasal route, so its role in relation to pulmonary absorption cannot be delineated.

Aluminium chlorohydrate is present in many aerosol anti-perspirants. Rats and guinea pigs were exposed to aerosolized aluminium chlorohydrate, 0.25, 2.5, or 25 mg/m³, 6 hr/day, 5 days/week, for up to 21 months (guinea pigs) or 24 months (rats). Neither animal species showed appreciable aluminium accumulation in the brain, heart, spleen, kidney, liver or serum, whereas significant increases in aluminium concentrations were seen in the lung of both species, adrenal glands of rats, and peri-bronchial lymph nodes of the guinea pigs (Stone et al., 1979).

Deposition of ~ 2 to 12% of fly ash into the lungs of rats was observed in three studies. Lung and pulmonary deposition of aluminium were 9.8 and 7.9% for spherical monodisperse aluminosilicate particles having a diameter of 2.2 μm (Raabe et al., 1977). Aluminium deposition into rat lung after 7 days of exposure to power plant fly ash with a mass median aerodynamic diameter (MMAD) of ~ 2 μm was 11.8% (Raabe et al., 1982). In another study, the percentage of aluminium in fly ash that deposited in the lung was calculated from the amount of aluminium in the lung of rats after exposure to 73 mg fly ash/m³ for 23 hr/day, 5 days/week for 1 month (Tanaka et al., 1983). The deposition fraction was 1.8%. Similar exposure of rats to a fly ash that was 9.7% aluminium and had a MMAD of 3 μm at 10.4 mg/m³ for 7 hr/day, 5 days/week for 1 month resulted in a deposition fraction of 5.1% (Matsuno et al., 1986). The authors attributed the differences to particle diameter, noting a smaller apparent deposition fraction with larger particles. Rabbits exposed to a mean concentration of 0.56 mg aluminium oxide/m³ for 8 hr/day, 5 days/week for 5 months showed significant increases in aluminium concentrations in the brain, lung, and bone that were, respectively, 247, 15,800 and 122% of the values for the controls, whereas aluminium levels in the heart were significantly lower (70% those of controls) (Rollin et al., 1991b). Serum aluminium levels increased during this study, although not consistently over time.

Oral administration

There has been considerable research on aluminium pharmacokinetics, including its oral absorption. However, most of this research, including all of that carried out prior to 1990, has been conducted using ²⁷Al. To determine oral aluminium bioavailability, which is very low, and to see a significant increase of aluminium in blood, urine or tissue above the endogenous aluminium concentration, it was necessary to give very large doses of ²⁷Al. More recently, studies have been conducted to estimate oral aluminium bioavailability using ²⁶Al.

Bioavailability (fractional absorption) is the amount of a substance absorbed compared to the amount administered. With respect to the toxicokinetics of aluminium, systemic bioavailability, the fraction that ultimately reaches systemic circulation from where it has access to the brain and bone, the target organs for its toxicity, is most relevant. Oral aluminium bioavailability has been determined using several methods. Each of the methods has strengths and weaknesses. One of the first used was the balance study in which absorption was estimated based on the difference between intake and faecal, or urinary-plus- faecal, excretion. Estimating aluminium absorption based on the difference between intake and faecal excretion is not accurate for aluminium, for which oral bioavailability is very low. This approach assumes the difference between aluminium intake and faecal excretion is that which is absorbed and retained or excreted in the urine, which is the major route of elimination of absorbed aluminium. Small errors in determination of aluminium in faeces can significantly influence the estimate of bioavailability. As oral aluminium absorption is 1% or less under most conditions, with the balance passing through the GI tract unabsorbed, it would be very difficult to accurately determine the 1% loss of aluminium due to absorption based on the difference between oral intake and faecal excretion. This was acknowledged by Cam et al. (1976). Therefore, the results of studies that utilized this method (Allen & Fontenot, 1984; Cam et al., 1976; Clarkson et al., 1972) are not considered to be reliable estimates of oral aluminium absorption. Balance studies that estimate retention based on the difference between intake and urinary-plus-faecal excretion tend to overestimate the bioavailability of aluminium because aluminium can be retained on the gut wall and then be eventually excreted in faeces, but not absorbed. Therefore, the results of a study that utilized this method (Gorsky et al., 1979) are not considered a reliable estimate of oral aluminium absorption.

Estimating aluminium bioavailability based on urinary excretion compared to intake has been the method most commonly used to determine aluminium bioavailability. This method has many advantages. Collection of urine is less invasive than the sample collection requirements for most other methods that estimate oral bioavailability, which typically include blood and/or tissue. However, in the single dose, non-steady-state study, collection of all urine, or at least urine collection for a sufficient duration to ensure that nearly all of the urinary aluminium output from the dose has been obtained, is required. This is difficult when ²⁷Al has been used as the aluminium dose because it is necessary to distinguish urinary excretion of the aluminium administered in the test dose from that ingested prior to the study or from other sources during the study. The requirement for total urine collection is also a compliance issue for the human subject. In the steady-state study, a sample collection period representing normal urinary aluminium output should be sufficient to reduce the compliance issue. Calculation of bioavailability assumes that all absorbed aluminium is excreted in the urine. This method may underestimate bioavailability due to the aluminium eliminated in bile, although this is only ~ 1%, (see 5.3.1.2 below), and the aluminium retained during the duration of the study. In a human study, Stauber et al. (1999) utilized two correction factors to estimate oral aluminium bioavailability. They collected urine for 24 hr and multiplied the amount of aluminium excreted by 2.2 to correct for the percentage of i.v. injected aluminium found in the urine after 7 days (72%), and the percentage of total aluminium excreted in 7 days that was excreted in the first day (62%).

Estimation of absorption from a single serum sample and the calculated volume of distribution would be expected to underestimate bioavailability because this approach does not account for aluminium that has not yet been absorbed, has distributed out of the vascular compartment, or has been excreted. It does not assure that peak serum aluminium was sampled unless independently determined. Underestimation of bioavailability by this method was shown by Hohl et al. (1994) who found that peak serum ²⁶Al suggested 0.01% bioavailability, whereas cumulative urine ²⁶Al estimated it to be 0.1%. Similarly, an approximate 10-fold greater estimate of bioavailability was obtained based on urinary aluminium excretion compared with that derived from a single (1, 4 or 24 hr) blood sample (Priest et al., 1996; 1998). The authors concluded: “…bioavailability cannot be accurately determined from blood ²⁶Al or ²⁷Al levels at a single time after administration” (Priest et al., 1996). Furthermore, an accurate estimate of volume of distribution assumes an accurate estimate of the t_½ of elimination.

The use of the product of “tissue aluminium concentrations × tissue weights” to determine aluminium bioavailability assumes no aluminium elimination from the sampled tissues. If all tissues are not sampled, this method results in an underestimate of bioavailability. This method was employed by Wilhelm et al. (1992) who used bone and Zafar et al. (1997) who used liver, kidney, spleen, femur, brain, pancreas and blood. Some other studies estimated oral bioavailability from the sum of urinary aluminium excretion and levels of aluminium in bone (and liver and brain) tissue, thereby partially overcoming one of the limitations of using urinary aluminium excretion only to estimate bioavailability (Drüeke et al., 1997; Jouhanneau et al., 1993; 1997a).

Comparison of areas under the “plasma aluminium concentration (AUC) × time curve” after oral vs. i.v. dosing is the method generally accepted for determining the oral bioavailability of most substances (Rowland & Tozer, 1995). This method requires repeated blood sampling, which is a disadvantage. The use of ²⁶Al and this method were employed by Zafar et al. (1997) who compared the AUC after oral, to i.p., not i.v., systemic injection, and by Yokel (2001) who compared the AUCs, or their equivalent, after oral administration of ²⁶Al and i.v. administration of ²⁷Al.

It is clear that aluminium can be orally absorbed. This has been shown by studies in which neurobehavioural changes and elevations of serum, urine and tissue aluminium following oral aluminium dosing of animals and humans have been reported. This corrects misinformation in 1965 stating: “… Large doses of soluble [Al] compounds taken orally will produce … no systemic effects” and “No entity ‘chronic aluminium poisoning’ has been identified in human beings” (Maynard, 1965).

Numerous studies have shown increased serum aluminium and/or urinary aluminium excretion after oral administration of various ²⁷Al-containing products; the increases have often been seen to be influenced by the chemical form of the aluminium. These results suggest aluminium absorption from the GI tract (Beynon & Cassidy, 1990; Gorsky et al., 1979; Haram et al., 1987; Ittel et al., 1991a; 1991b; Kaehny et al., 1977; Knoll et al., 1984; Rauch et al., 1989; Recker et al., 1977; Robertson et al., 1989; Tsou et al., 1991; Winterberg et al., 1987a). However, most of the studies do not permit estimation of oral aluminium bioavailability.

In addition to elevations of serum or urine aluminium levels, numerous studies have shown increased levels of aluminium in the brain after oral ²⁷Al exposure, demonstrating oral aluminium absorption. Some examples follow. Rats and dogs were fed 300 or 3000 mg aluminium hydroxide daily in their food for 5 months. These animals exhibited a significant elevation of brain aluminium concentrations, compared to those not receiving the aluminium hydroxide (Arieff et al., 1979). A significant elevation of brain aluminium was claimed after a single administration of oral aluminium hydroxide to mice (Cutrufo et al., 1984), although these authors did not report their actual results. Rat bone, but not brain, aluminium was significantly elevated after consumption for 8 weeks of a diet containing 570 mg sucralfate/kg. Sucralfate is a sucrose aluminium sulphate complex. The consumption of aluminium was approximately 4 mg/kg/day (Burnatowska-Hledin & Mayor, 1984). Consumption, by rabbits, of distilled water containing 0, 100 or 500 mg Al/L, introduced as aluminium chloride, for 12 weeks produced a positive correlation between aluminium exposure and aluminium in bone, stomach, intestine and kidney, but not in brain (Fulton & Jeffery, 1990). Brain aluminium was significantly increased in rats after 90 daily oral doses of 30 or 100 mg/kg of aluminium chloride (Bilkei-Gorzó, 1993). Consumption by mice of drinking water containing 400 mg Al/L (as aluminium lactate) for 6 months increased aluminium in brain and other tissues (Anghileri et al., 1994). Mice consuming ~20 μg Al/day as aluminium hydroxide gel in their drinking water for 105 days were reported to have 30, 60 and 340% increases in kidney, liver and brain aluminium concentrations, respectively (Sahin et al., 1994).

On the other hand, in many studies, increased blood or tissue aluminium levels were not found after oral ²⁷Al administration. For example, rats consuming approximately 0, 0.01, 0.2 or 5.5 mg Al/kg/day in their drinking water, as aluminium chloride, in the absence or presence of acetate or citrate, failed to show an increase of aluminium in bone or brain (Fulton et al., 1989). Consumption of 0.112 mg Al/day for 20 days by one baboon did not result in increased serum aluminium compared to a 10-day period of 0.06 mg Al/day consumption in the same animal (Turnquest & Hallenbeck, 1991). This exposure level and duration were much smaller than in most other studies, so the negative result from this one animal is not informative. Prior to their report of 0.03% oral aluminium absorption from Zeolite A, Cefali et al. (1995; 1996) did not find significant aluminium absorption from Zeolite A, sodium aluminosilicate or aluminium hydroxide, containing 3.36, 0.90 and 27.8 mg Al/kg, respectively. Rats consuming drinking water containing 3.8 mg Al/L (as the chloride) for 10 weeks, and a diet containing 4 to 5 mg Al/kg, had no elevation of brain, bone or liver aluminium compared to those drinking water containing 0.05 mg Al/L (Glynn et al., 1995). Tissue aluminium concentrations significantly decreased from 0 to 10 weeks after consumption of this diet, perhaps due to consumption of a diet containing more aluminium prior to the study, which might have masked any changes due to the consumption of aluminium in drinking water during the study. At the pHs of these drinking waters, 4.3 to 4.6, speciation calculations suggested 99% of the aluminium should be labile species, and therefore available for absorption (Glynn et al., 1995).

In light of the evidence for oral absorption of aluminium presented in the paragraph immediately above, it must be assumed that there was insufficient aluminium absorbed in the studies described in the previous paragraph. Contributing to the lack of a significant increase of aluminium could be endogenous aluminium in the organism, contamination, and variability in the aluminium assay. The negative studies cannot be taken as proof of lack of oral absorption or as support for the null hypothesis that aluminium is not absorbed after oral administration.

Application of accelerator mass spectrometry to quantify ²⁶Al has enabled study of aluminium toxicokinetics under physiological conditions. This method has been used since ~ 1990. For example, in one study, 2 rats were given a single gastric administration of ²⁶Al in the absence of ligand during access to a normal diet; 2 rats did not receive ²⁶Al. Seven days later the ²⁶Al-treated rats had higher brain ²⁶Al levels than the controls (Fink et al., 1994). In a second study by this group, using conditions that simulate drinking water, 8 rats were given a single gastric administration of ²⁶Al in the absence of ligand after 30 hr of no access to food; 2 rats did not receive ²⁶Al. Two weeks later, brain ²⁶Al concentrations of 2 of the ²⁶Al-dosed rats were comparable to those of controls (3 × 10^-9 of the ²⁶Al dose), whereas brain ²⁶Al in the other 6 ²⁶Al-dosed rats ranged up to 100 times higher (Walton et al., 1995). The large variability is disconcerting. These results were frequently cited as evidence supporting the hypothesis that aluminium in drinking water contributes to brain aluminium accumulation. Assuming that a comparable fraction of orally consumed aluminium reaches and is permanently retained by the human brain, Walton et al. (1995) concluded that humans drinking alum-treated water (ATW) over seven to eight decades would have ~ 1 mg Al/kg wet brain. This is greater than normal levels of aluminium in the brain. On the other hand, these results could support the hypothesis that aluminium in drinking water does not contribute to brain aluminium accumulation if one uses the lowest brain aluminium level obtained after ²⁶Al dosing, or if aluminium is not permanently retained by the brain. Subsequent studies utilizing ²⁶Al, reviewed below, have clarified this issue.

Studies in animals utilizing ²⁶Al, also reviewed below, have shown the ability of aluminium to enter bone after oral administration.

No published information was found on buccal aluminium absorption.

In summary, there are sufficient studies in which ²⁷Al was utilized by many different research groups to investigate neurobehavioural endpoints or in which blood, urine and/or tissue aluminium levels were studied following controlled administration of ²⁷Al or ²⁶Al, to show that aluminium can be absorbed after oral administration. These studies also show that this route of exposure has the potential to produce toxicity. However, many of the studies with ²⁷Al were conducted utilizing supra-physiological exposures or doses of aluminium. This has led to the controversy of whether or not aluminium exposure, under normal conditions, has the potential to produce toxicity.

Drinking water

The average daily intake of aluminium from drinking water for a 70 kg person is approximately 160 μg, or about 2.3 μg/kg b.w./day. Studies that best model oral aluminium bioavailability from drinking water meet two criteria. They use aluminium doses ~ 2 μg Al/kg b.w./day, which reasonably compares with daily oral aluminium intake from water by adult humans of ~ 2.3 μg Al/kg b.w./day, (see Human Exposure, Total Human Uptake from All Environmental Pathways (Combined Exposure) and Evaluation of Human Health Risks, Health Effects, Exposure Characterization) and they introduce the aluminium either as a chemical species that might be found in drinking water, that is, as the chloride, sulphate or hydroxide, which have the ability to release the free aluminium ion in solution, or they use water from municipal water supplies. As noted above, by necessity, all studies with ²⁷Al use aluminium doses much greater than 2 μg Al/kg. The use of ²⁶Al enables determination of oral aluminium bioavailability after administration of ≤ 2 μg Al/kg.

Estimates of the oral bioavailability of aluminium, based on administration of ²⁷Al chloride in the rat, were 0.06 to 0.2% (Ittel et al., 1987) and 0.04% (Froment et al., 1989a). Administration of ²⁷Al as the chloride, nitrate, lactate or citrate resulted in absorption of 0.57, 1.16, 0.7 to 1.9, and 2.18% in the rabbit (Yokel & McNamara, 1985; 1988). The difference in results obtained in these studies may relate to the difference in physiology of the stomach of these two animal species and/or gastric pH. The rat has a single compartment stomach whereas the rabbit’s stomach has more than one compartment. The pH of suckling rabbit gastric juice is ~ 4 to 6.5 dropping to ~ 1.5 to 2.5 post-weaning (Beauville et al., 1966; Kaplan & Timmons, 1979; Yu & Tsen, 1993) whereas the pH of rat gastric juice is ~ 3.2 (Cunha-Melo et al., 1983; Lysiak, 1963). The human stomach pH is typically 1 to 3. The lower pH may result in conversion of more aluminium to monomeric species. However, as aluminium is not significantly absorbed from the stomach (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, The Site and Mechanisms of Oral Aluminium Absorption), its chemical species would be expected to change as it enters the upper GI tract. Estimates based on studies utilizing ²⁷Al suggest the oral bioavailability of the aluminium ion is < 1%.

Several studies utilized the tracer ²⁶Al and total aluminium doses of 0.1 to 2 μg Al/kg. However, many of these studies had very small sample sizes and some were not well controlled. The amount of ²⁶Al, after administration in the absence of ligand in distilled water that was estimated to be in the plasma of two rats 8, 24 and 48 hr after dosing was 0.0001 to 0.001% (Jouhanneau et al., 1993). These rats had free access to food and water. When plasma aluminium concentrations obtained by Jouhanneau et al. (1993) are compared to those obtained at similar times from one rat after i.v. ²⁶Al injection, in pH 5 saline solution (Meirav et al., 1991), oral ²⁶Al bioavailability is estimated to be 0.033% (8 hrs), 0.087% (24 hrs) or 0.2% (48 hrs). In a study in which ²⁶Al was given to 9 rats/group, the fraction of the dose that was in plasma 0.5 to 5 hr later was estimated (Schönholzer et al., 1997). They pooled the plasma samples at each time point for a single ²⁶Al analysis. The plasma ²⁶Al area under the curve suggested greater ²⁶Al bioavailability in the presence of ²⁷Al citrate + Na citrate than ²⁷Al citrate or ²⁷Al hydroxide. Estimates of bioavailability based on urinary aluminium excretion were 5.1, 0.7 and 0.1%, respectively (Schönholzer et al., 1997). Studies with aluminium hydroxide and aluminium chloride, which may release the free aluminium ion at the low total aluminium concentrations tested, suggested an oral aluminium bioavailability of about 0.06 to 0.1% (Drüeke et al., 1997; Schönholzer et al., 1997). Schönholzer et al. (1997) may have underestimated aluminium bioavailability because they only collected the eliminated urine for 300 minutes after dosing and the urine in the bladder at 300 minutes. Although Jouhanneau et al. (1997a) studied only 2 rats at each euthanasia time point, there were 8 rats/group in the study by Drüeke et al. (1997). Jouhanneau et al. (1997a) collected total urine output for up to 720 hr. Of the total urinary aluminium output, 85% was in the first day, about 4% in the second, an additional 5% in days 3 to 5, and 6% in days 6 to 29. They based their estimate of oral aluminium bioavailability on the sum of the aluminium excreted in urine and that present in bone, liver and brain. Drüeke et al. (1997) based their estimate on cumulative 48 hr urinary aluminium excretion and on aluminium found in the skeleton at 48 hr. These studies were conducted under conditions in which it was likely that the subjects had food in their stomachs (Drüeke et al., 1997; Jouhanneau et al., 1997a) or, possibly, had some stomach contents. Food and/or faeces have been found in the stomachs of rats after a 24-hr (Walton et al., 1995) and 36-hr fast (Yokel et al., unpublished results). This is partly because rats recycle faeces (the practice of coprophagia). Therefore, it is probable that there were stomach contents in the rats studied by Schönholzer et al. (1997) after the 16 hr overnight fast. Zafar et al. (1997) gave ²⁶Al (in the presence of ²⁷Al chloride) by the oral route to 3 subjects and by i.p. injection to 3 others and concluded that oral aluminium bioavailability was 1 to 1.6%. Based on ²⁶Al recovered in urine, liver, spleen and bone after an oral administration of 20 ng ²⁶Al and 200 μg ²⁷Al, as the chloride, oral bioavailability was estimated to be 0.133 % in control rats (Ittel et al., 1997). Oral aluminium bioavailability was determined by concurrent administration of oral aluminium as one isotope, ²⁶Al (in the absence of a ligand), and i.v. aluminium (as Al potassium sulphate) as another, ²⁷Al (Yokel et al., 2001a). Oral aluminium bioavailability was ~ 0.28% in rats that had no stomach contents. Using the rat in situ intestinal perfusion technique, absorption of aluminium from aluminium chloride and equimolar trisodium citrate, based on aluminium in blood and a number of tissues, was estimated to be 0.2% (Arnich et al., 2004).

There is some evidence that fractional absorption of aluminium is dose dependent in the fasted animal. Oral aluminium bioavailability in the rabbit dosed with 108 or 540 mg Al/kg, as aluminium lactate, was 0.7 and 1.9% (Yokel & McNamara, 1985). Although not significantly different, these results suggest a positive correlation between aluminium dose and bioavailability. On the other hand, absorption of aluminium into rat blood and tissues after perfusion of the gut with 48 or 64 mM aluminium chloride at pH 3 was not concentration dependent (Arnich et al., 2004).

Overall, these results suggest that oral aluminium bioavailability from water in the rat and rabbit is in the range of 0.05 to 0.4% and most likely ~ 0.3%.

Beverages and foods

Although food comprises the primary source (> 90%) of aluminium for the typical human (see Human Exposure, General Population Exposures, Food and Beverages), there are very few data on oral aluminium bioavailability from foods, or beverages other than water. The difficulties in estimating aluminium bioavailability from water using ²⁷Al, discussed above, apply to food. It has generally been assumed that oral aluminium bioavailability from food is less than that from water due to the aluminium being incorporated in high molecular weight, relatively insoluble, complexes (Glynn et al., 1995).

It has been suggested that the aluminium in tea leaves has low oral bioavailability (Glynn, 1995). In tea, 91 to 100% of aluminium is in organic complexes with a M_r > 20,000. Even at pH 2, ~ 83% remained bound to the organic matter (French et al., 1989; Gardner & Gunn, 1991). For example, a much lower percentage (15%) of the aluminium in tea was found to be in chemically labile species, compared with that in drinking water (61 to 75%) (Stauber et al., 1999). Approximately 50% of aluminium in tea infusions was as soluble species, ~ 50% as non-labile monomeric aluminium species and a small fraction as labile monomeric aluminium, whereas > 90% of aluminium in tap waters was labile monomeric aluminium (Chen et al., 2004). A large fraction of the aluminium in tea infusions was very strongly bound to unidentified ligands (Alberti et al., 2003). Kralj et al. (2005) found that ~ 10 to 35% of the aluminium in tea was negatively charged aluminium citrate. They thought the remainder was bound to phenolic compounds. Addition of citrate increased the negatively charged aluminium citrate species by up to 40% of total aluminium, whereas milk complexed most of the aluminium that was not associated with citrate to protein, mainly casein. The authors suggested that addition of citrate and milk protein would enhance aluminium absorption. Reiber et al. (1995) suggested that a substantial portion of aluminium, regardless of the form consumed, will be solubilized to monomeric aluminium in the stomach and subsequently converted to poorly soluble aluminium species in the near neutral pH of the upper intestine. As the stomach is not an important site of aluminium absorption, this implies that oral aluminium bioavailability should be aluminium species independent. Citrate and other ligands influence aluminium absorption, suggesting that this hypothesis is an oversimplification.

Weanling rats whose liquid source for three weeks was a diet soft drink from aluminium cans containing 0.47 mg Al/L, had significantly higher liver and bone aluminium concentrations than rats which drank a soft drink from glass bottles containing 0.38 mg Al/L (Kandiah & Kies, 1994). Bone aluminium was 69% higher in rats that drank the soft drink from the aluminium cans than in rats that drank distilled water containing 0.023 mg Al/L. However, liver aluminium concentration was significantly lower, and the concentration of aluminium in bone was non-significantly lower, in rats that drank bottled soft drink than in those that consumed distilled drinking water containing 0.23 mg Al/L, thereby casting doubt on the results of this study. It is surprising that a 24% increase in aluminium intake from beverage would result in a 62, 127 and 84% increase in brain, bone and liver aluminium levels. The authors reported that the rat ration contained 110 mg Al/kg, which is much greater than the beverages. In other studies, rat diet has been found to contain 100, 5, and 51 mg Al/kg (Glynn et al., 1995; Gupta et al., 1986; Yokel et al., unpublished results), mouse diet to contain 131 and 64.5 mg Al/kg (Dlugaszek et al., 2000; Fosmire et al., 1993), guinea pig diet to contain 47 and 60 mg Al/kg (Golub et al., 1996a; Owen et al., 1994), and rabbit diet to contain 297, 1215 and 335 mg Al/kg (Fulton & Jeffery, 1990; Yokel & McNamara, 1985; Yokel et al., unpublished results).

No increases in blood or liver aluminium concentrations were seen in rats which consumed tea as the only source of fluid for 28 days (Fairweather-Tait et al., 1991). In one study, it was reported that increased tissue aluminium concentrations were attributed to the intake of aluminium in food. Guinea pigs that ate a test diet of sponge cake three times weekly for 3 weeks providing a total of 40 mg of aluminium, as acidic sodium aluminium phosphate (SALP) showed a significant elevation in bone aluminium concentrations compared with those that ate only guinea pig chow, which provided a total of 3 mg aluminium (Owen et al., 1994).

In an ambitious but not definitive study, Walton et al. (1994) orally administered various beverages and foods to anaesthetized rats, in the absence of aluminium-treated water. Tail blood and urine, withdrawn by needle aspiration of the bladder, were obtained prior to, and 1, 2, 3 and 4 hr after, dosing. Considering the short period of observation (4 hr) serum aluminium is probably a better indicator of absorbed aluminium than is urinary excretion. Margarine and 8 mg of aluminium, as aluminium sulphate, increased serum aluminium levels. The aluminium content of the margarine was not determined. The other beverages and foods tested, beer, Coca Cola®, coffee, orange juice, tea, wine, apple, broccoli, butter, meat and Vita Wheat®, did not appreciably increase serum aluminium levels when administered alone. This is not surprising as the aluminium dose provided by these beverages and foods, which ranged from 0.005 to 8.6 μg, was ≤ 0.1% of the 8 mg of aluminium in the 1 mL of water which produced an elevation of serum aluminium of 4 to 5-fold. Determination of relative oral bioavailability of aluminium from food and water was not the objective of this study. Little can be learned from these results concerning the oral bioavailability of aluminium from foods.

Although rigorous pharmacokinetic determination of oral aluminium bioavailability from milk could not be determined, it was estimated to be < 1% in rabbits (Yokel & McNamara, 1985).

Weanling rats fed 1 to 2.7 gm of Al/kg diet, added as aluminium hydroxide in the absence and presence of added sodium citrate dehydrate, were estimated to absorb 0.01 to 0.04% of the aluminium (Greger & Powers, 1992). The percentage of aluminium absorbed was lower with the higher dietary concentration of aluminium as the citrate, in contrast to above studies, thereby showing a positive correlation between aluminium dose and fraction absorbed.

There are sufficient studies with reasonably similar results to estimate the oral bioavailability of aluminium from water. Those who suggest water may be a significant contributor of the aluminium body burden assume that aluminium in water is more bioavailable than the aluminium in food. Some recent data support this assumption, as presented in the next paragraph (Yokel & Florence, 2006). Diet provides most of the human’s daily aluminium intake. Those who wish to allay any concern about drinking water as a source of aluminium suggest that the oral bioavailability of aluminium from food and water are similar (Reiber et al., 1995) and have provided some data to support this conclusion (Stauber et al., 1999).

The bioavailability of aluminium from selected foods has been estimated in the rat. ²⁶Al was incorporated into the synthesis of acidic SALP, used as a leavening agent in baked goods, and then incorporated into a biscuit. Basic SALP, used as an emulsifier in cheese, was incorporated into a processed cheese (Yokel et al., 2005). When rats, that had no stomach contents, ate the biscuit containing acidic SALP, it was estimated that oral aluminium bioavailability was ~ 0.1%, significantly less than from water (Yokel & Florence, 2006). Oral Al bioavailability from the cheese containing basic SALP was ~0.1 to 0.3% (Yokel et al., unpublished results). Comparison of oral Al bioavailability from these representative foods to that obtained from water (~0.3%) times the contributions of food and drinking water to the typical human’s daily Al intake (~5 to 10 mg from food and 0.1 mg from water, respectively) suggests food provides ~25-fold more Al to systemic circulation, and potential Al body burden, than does drinking water.

The very limited data available, above, suggest that oral aluminium bioavailability from food is less than from water.

Drugs

Relevant published data on the oral bioavailability of aluminium species from drugs include the use of aluminium hydroxide as an antacid and phosphate binder; sucralfate as an antacid; aluminium lactate as a component in dental products for sensitive teeth; an aluminium silicate product, Zeolite A, which is an inducer of osteoblast proliferation; and aluminium in the presence of citrate. Some of the earlier work, conducted with ²⁷Al, was reviewed by Wilhelm & Ohnesorge (1990). Because of its poor absorption, use of large doses of aluminium was necessitated in studies of bolus ²⁷Al dosing to determine oral aluminium bioavailability. The chemical species of the aluminium, when given in large doses, particularly when the aluminium species has good buffering capacity, such as aluminium hydroxide, would not be expected to be totally affected by the normal chemical reactions in the gut. It is less likely that the series of chemical reactions in the gut that might produce comparable aluminium species irrespective of aluminium species ingested, as proposed by Reiber et al. (1995), would be relevant. Therefore, these studies might more closely reflect the absorption of the aluminium species introduced, albeit after very high doses, assuming there is no saturation of uptake processes.

The bioavailability of aluminium from ingested aluminium hydroxide appears to be less than from the aluminium ion, based on studies using ²⁷Al. In the rabbit, 0.45% of aluminium as the hydroxide was absorbed compared with 0.57 to 1.16% from aluminium chloride and nitrate (Yokel & McNamara, 1988). In the rat, 0.01% of aluminium from the hydroxide was absorbed compared with 0.037% from the chloride (Froment et al., 1989a). Wilhelm et al. (1992) could not detect aluminium absorption after an oral dose of 1 mg Al/kg, as aluminium lactate, illustrating the necessity of using large doses in bolus ²⁷Al dosing studies. Their estimate of 1% oral bioavailability, based on bone aluminium 17 days after oral aluminium dosing, would be an underestimate if any bone aluminium were eliminated within that time, although there may not have been significant bone aluminium elimination (see Toxicokinetics, Elimination and Excretion, Animal Studies, Elimination Rate).

The absorption of aluminium from aluminium citrate is greater than from aluminium hydroxide: 2.18 vs. 0.45% in the rabbit (Yokel & McNamara, 1988), 1.49 vs. 0.015% and 1 vs. 0.1% in the rat (Froment et al., 1989a; Schönholzer et al., 1997). Sucralfate, which, like aluminium hydroxide, is insoluble in water but soluble in acid and base, exhibits oral bioavailability comparable to that of aluminium hydroxide and lower than that of soluble aluminium species. Aluminium bioavailability from sucralfate in the rabbit was 0.6%, and was 0.63% from aluminium lactate, compared with 0.57 to 1.16% from aluminium chloride and nitrate (Yokel & McNamara, 1988). In the rat, oral aluminium bioavailability was 0.015% from aluminium hydroxide and sucralfate, 0.037% from aluminium lactate and aluminium chloride, and 1.49% from aluminium citrate (Froment et al., 1989a).

Cefali et al. (1996) studied aluminium absorption from Zeolite A^®, an aluminium silicate product, in dogs; they observed highly variable plasma aluminium levels during the control phase and a small increase of plasma aluminium as a result of the treatments. They did not use baseline-corrected data in their determination of aluminium pharmacokinetics. Therefore, one cannot have much confidence in these results.

Factors influencing oral aluminium absorption

Oral aluminium absorption is dependent on many factors. A summary of the reported factors affecting oral aluminium absorption in animals is shown in Table 17. The following sections describe these effects in much more detail.

Table 17

Summary of the influence of factors that affect oral aluminium absorption from studies in animals.

Solubility

There is evidence of greater aluminium absorption from more soluble aluminium species. Aluminium borate, glycinate, hydroxide and sucralfate are much less soluble in water than aluminium chloride, lactate, nitrate and citrate and were generally less well absorbed (0.27, 0.39, 0.45 and 0.60 vs. 0.57, 0.63, 1.16 and 2.18%, respectively (Yokel & McNamara, 1988). Similarly, aluminium hydroxide and sucralfate are less soluble at pH 3, 6 and 7 than aluminium lactate and chloride and were also less well absorbed (0.015 vs. 0.037%), based on urinary aluminium excretion (Froment et al., 1989a).

pH

The chemical speciation of aluminium is known to be greatly influenced by pH (Harris et al., 1996). The bioavailability of aluminium, to some extent, is influenced by the aluminium species. However, Reiber et al. (1995) suggested that the human GI tract acts as a series of reactors. They predicted that, in the stomach, all aluminium would be converted to small molecular weight soluble species due to the predominance of the hydrated aluminium ion at pH 1 to 2. They suggested that, when the stomach contents enter the duodenum/jejunum where the pH increases to 4.5 to 6.8, all aluminium would be converted to hydroxides. They predicted uniform oral bioavailability of aluminium, independent of the chemical species of the aluminium consumed. However, this ignores the presence of ligands to compete with hydroxide to associate with aluminium at the site of absorption. Using the in situ rat intestine preparation with systemic and portal blood sampling, Van der Voet & de Wolff (1986-1987) found aluminium absorption to be higher at pH 4 than at pH 7, presumably due to generation of more soluble aluminium species. However, the predicted aluminium species at acidic pHs, in the absence of binding ligands, are Al(H₂O)₆³⁺ and Al(OH)²⁺ (Harris et al., 1996), which would not be expected to permeate the gut wall by diffusion. This is consistent with the assumed primary site of GI aluminium absorption, the upper intestine, discussed below (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, The Site and Mechanisms of Oral Aluminium Absorption).

Carboxylic acids

Citrate and other carboxylic acids form coordination complexes with aluminium. The co-administration of citrate with aluminium-containing drugs is relevant to iatrogenic (medical) aluminium exposure.

The chemistry of aluminium interaction with citrate has been quite extensively studied (Harris et al., 1996) (see also Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Properties of Aluminium Compounds). Citrate can complex aluminium, forming an electrically neutral 1:1 citrate:aluminium complex at pH 2 to 5 in the absence of great excess of citrate. At higher pHs, this complex deprotonates, with a pKa of ~ 4, to a complex with one negative charge held by a carboxylate group that is not involved in the aluminium citrate:complex (Lakatos et al., 2001). More complex chemical species slowly form at higher concentrations (Harris et al., 1996).

Citrate may enhance oral aluminium absorption if there is sufficient citrate present to compete with other binding ligands for aluminium in the GI tract. Speciation calculations indicated that citrate would solubilize ~ 97% of aluminium in the stomach (Glynn et al., 2001). Citrate is the major small molecular weight ligand for aluminium in plasma. Formation of aluminium citrate also enables the distribution of aluminium out of plasma, and may enhance aluminium elimination by the kidney in the presence of renal function. However, in the absence of renal function, citrate can significantly increase aluminium-induced toxicity, presumably by enhancing aluminium distribution out of the blood and the resultant tissue aluminium accumulation. However, the enhanced solubility of aluminium by citrate did not completely account for enhanced aluminium absorption (Froment et al., 1989b). It has been suggested that aluminium citrate is sufficiently lipid soluble from pH 2.5 to 8 to be absorbed by diffusion (Partridge et al., 1992). Taylor et al. (1998) argued that the alkaline environment of the upper intestine could produce insoluble aluminium species, as the aluminium citrate species would not predominate at neutral and higher pHs. However, the typical pH of the human duodenum/jejunum ranges from 4.5 to 6.8, which is not alkaline. Furthermore, in the presence of equimolar aluminium and citrate, aluminium citrate-hydroxide complexes would predominate (W.R. Harris, personal communication, 2004). Alternatively, enhanced aluminium absorption may require sufficient citrate to activate another mechanism of aluminium absorption. Other hypotheses to explain the enhancement of aluminium absorption by citrate are the transport of aluminium citrate into mucosal cells, and citrate opening the tight junctions of the intestinal cells, as discussed in Toxicokinetics, Absorption, Animal Studies, Oral Administration, The Site and Mechanisms of Oral Aluminium Absorption.

Numerous animal studies have shown that aluminium is more bioavailable when administered as the citrate rather than as other chemical species (Cunat et al., 2000; Deng et al., 1998; 2000; Drüeke et al., 1997; Froment et al., 1989b; Partridge et al., 1989; 1992; Schönholzer et al., 1997; Sutherland & Greger, 1998; Van der Voet et al.,1989; Yokel & McNamara, 1988). It was estimated that 3.4 to 4.2% of aluminium was absorbed by rats, from aluminium citrate; however, no direct comparison was made with aluminium dosing in the absence of citrate (Sutherland & Greger, 1998). Numerous studies in humans have also shown enhanced aluminium absorption in the presence of citric and other carboxylic acids (see Toxicokinetics, Absorption, Studies in Humans, Oral Administration, Factors Influencing Oral Aluminium Absorption, Carboxylic Acids). However, a few studies failed to find an increase of aluminium absorption in the presence of citrate. These were conducted in subjects who received aluminium during free food access (Jouhanneau et al., 1993; 1997a).

An increase in the citrate:²⁶Al ratio increased oral aluminium bioavailability in one study (Schönholzer et al., 1997). However, the increase of citrate was accompanied by an increase of the ²⁶Al dose and pH of the delivered solution, confounding the interpretation of the variable that increased oral aluminium bioavailability 5-fold in this study. A citrate-induced increase of brain and bone aluminium has been reported in many studies (Bilkei-Gorzó, 1993; Deng et al., 1998; Domingo et al., 1991a; 1993; Ecelbarger et al., 1994a; Ecelbarger & Greger, 1991; Fulton et al., 1989; Fulton & Jeffery, 1990; Greger & Powers, 1992; Owen et al., 1994; Radunovic et al., 1998; Slanina et al., 1984; 1985; Testolin et al., 1996) though not in others (Jouhanneau et al., 1997a; Van Ginkel et al., 1993). Moreover, Maitani et al. (1994) found that citrate addition decreased tissue aluminium concentrations. It was concluded that the chronic consumption of aluminium and citrate by gavage resulted in greater aluminium retention than when aluminium was delivered in the diet which, in turn, was greater than when delivered in drinking water (Greger & Sutherland, 1997). The authors suggested that the presence of other substances in the GI tract from the diet, which would not be present when aluminium is administered by gavage, may have modified the effect of citrate on aluminium absorption.

Concurrent acute administration of 2 mmole aluminium with 2 mmole citric, gallic, chlorogenic, caffeic or protocatechuic acids resulted in a significant increase in rat blood aluminium 2 hr later only with citric acid. There was no effect on liver aluminium from any ligand, an increase of kidney aluminium with citric and chlorogenic acid, and an increase in tibial aluminium with all but chlorogenic acid (Deng et al., 2000). Consumption by rats of a diet containing 16 mmole/kg of aluminium chloride and carboxylic acid resulted in no significant differences of aluminium in blood, liver, kidney or tibia; but an increase of aluminium in brain after citric and chlorogenic acids (Deng et al., 2000). The very high aluminium concentrations reported in blood and tissues in the control subjects reduce confidence in these results.

Citrate may enhance oral absorption and may also enhance distribution into and out of tissues as well as from the organism by renal elimination, in the presence of renal function, as suggested by (Maitani et al., 1994). Citrate inhibited or produced only a small increase of aluminium uptake into neuroblastoma and human erythroleukemia cells (Guy et al., 1990; McGregor et al., 1991; Shi & Haug, 1990).

Ultrafilterable species may redistribute within the organism, particularly in the presence of reduced renal function. Quartley et al. (1993) gave rats a single oral dose of aluminium citrate; 2, 4 and 24 hr later they found elevated aluminium in all tissues but the brain. The aluminium concentration decreased in most tissues from 2 to 24 hr, except for the bone, in which it increased. Citrate may have enabled redistribution of the aluminium into bone.

Other short chain carboxylic acids, including acetate, oxalate, lactate, malate, tartrate, gluconate, ascorbate, and carbonate have been similarly shown to increase aluminium absorption or tissue aluminium accumulation in some, but not all, animal studies. However, these substances were generally less effective than citrate (Colomina et al., 1994; Domingo et al., 1991a; 1993; 1994; Gómez et al., 1994; Nestel et al., 1994). This may be due to the formation of a more stable complex between aluminium and citrate than these other ligands.

Aluminium can form a complex with the dietary long chain mono-unsaturated FA, oleic acid, but the stability of this complex has apparently not been described (Krasnukhina et al., 1971; Lesnikovich et al., 1993). It appears aluminium does not complex with the saturated long chain FA stearate (Ross & Takacs, 1983). No information was found on aluminium complexation with the dietary saturated long chain FA, palmitic acid, the long chain poly-unsaturated FAs, α-linolenic and arachidonic acids, and the long chain branched chlorophyll degradation products, phytonic and pristonic acids. Aluminium can form a 1:1 complex with phytic acid but the reaction is slower and less favoured thermodynamically than phytic acid complexation with divalent cations (Evans & Martin, 1988). Insoluble aluminium-phytate complexes form at higher aluminium:phytate mole ratios (Evans & Martin,1988).

Silicon-containing compounds

There is evidence suggesting that increased dietary intake of silicon (Si)-containing compounds may reduce aluminium absorption and facilitate aluminium excretion. There are also studies that did not find an effect of silicon-containing compounds on aluminium absorption. Silicon is absorbed in the animal GI tract as monomeric silicic acid. It can react with aluminium to form hydroxyaluminosilicate species and slowly, but eventually, amorphous solids (Birchall et al., 1996). This was thought to reduce aluminium bioavailability, as shown by reduced systemic aluminium absorption in fish (Birchall & Chappell, 1989) and reduced oral aluminium absorption in humans (see Toxicokinetics, Absorption, Studies in Humans, Oral Administration, Factors Influencing Oral Aluminium Absorption, Silicon-Containing Compounds). Addition of sodium metasilicate to the diet was reported to reduce brain hippocampal aluminium accumulation in 28, but not 23, month-old rats (Carlisle & Curran, 1987). The basis for the age difference is unknown. These results have not been independently replicated and are not considered reliable. Addition of 0.5 mM silicic acid to water from 5 days before to 4 hr after aluminium citrate dosing by oral gavage reduced aluminium in most tissues of rats (Quartley et al., 1993). Rats drinking water containing 59 or 118 mg Si/L (as sodium silicate with 27% SiO₂) and receiving 450 mg aluminium nitrate nonahydrate injections 5 days weekly had less brain, liver, bone, spleen and kidney aluminium than rats not consuming silicon, suggesting that silicon-containing compounds may have reduced aluminium absorption and/or enhanced its elimination (Bellés et al., 1998). However, as the molar dose of silicon consumed was considerably less than the dose of aluminium administered, the mechanism does not seem to be simply aluminium complexation by silicon. Drüeke et al. (1997) did not find an effect of co-administered silicon dioxide and ²⁶Al on absorption of ²⁶Al given with citrate to rats after eating. It is possible that there was insufficient silicon in the presence of the food to compete with the citrate and interact with the aluminium to influence its bioavailability.

It has been suggested, based on the very low concentrations of silicon in biological fluids, that it is very doubtful that a monomeric aluminium silicate species plays any significant role in the biological chemistry of aluminium, after the aluminium has been absorbed (Harris et al., 1996).

Fluoride

Fluoride appears to be able to increase aluminium absorption. Fluoride forms numerous complexes with aluminium (see Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Properties of Aluminium Compounds). Speciation calculations indicated fluoride would solubilize > 60% of aluminium in the stomach (Glynn et al., 2001). Aluminium reduced fluoride absorption (Allain et al., 1996; Hobbs et al., 1954; Li et al., 1991; Xiao et al., 1992). The influence of fluoride on aluminium absorption is less clear. Rats consuming a diet containing fluoride and aluminium (chemical species not reported) had decreased levels of aluminium in liver, brain, heart, testes and femur compared to rats consuming aluminium alone, suggesting that fluoride decreased aluminium absorption or enhanced its clearance (Ondreicka et al., 1971). Increasing the aluminium (as the chloride) concentration from 0 to 100 to 500 ppm decreased plasma, bone, liver and urine fluoride concentrations in rabbits concurrently consuming fluoride in their drinking water (Ahn et al., 1995). Increasing the drinking water fluoride concentration from 0 to 1 to 4 to 50 ppm had no significant effect on aluminium levels in plasma, urine, or liver, although the rabbits that drank water containing 50 ppm fluoride and no added aluminium had increased bone aluminium concentrations (Ahn et al., 1995). Sterna, obtained from another study in which rats drank water containing 79 ppm fluoride for 2 years with no added aluminium, showed a significant increase of aluminium (Ahn et al., 1995). Co-administration of aluminium and fluoride to mice increased plasma aluminium 3 hr later when compared to when aluminium chloride or aluminium phosphate was dosed. The increase was similar to that seen with aluminium citrate. Administration of aluminium fluoride to rats increased plasma aluminium more than did administration of aluminium chloride. The rats were given 1.87 mmole Al/kg orally by cannula as aluminium fluoride or aluminium chloride. The time course (sampling at 20, 40, 60 and 90 min) showed aluminium from the aluminium fluoride peaked at ~ 8 μM at ~ 40 min; aluminium from aluminium chloride peaked at ~ 3 μM at ~ 40 min (Allain et al., 1996). In rats fed 130 or 300 mg F/kg, 300 or 1200 mg Al/kg, 130 mg Al + 130 mg F/kg, 300 mg Al + 300 mg F/kg or 1200 mg Al + 300 mg F/kg in their diet for 12 weeks, the lower fluoride concentrations reduced, and the higher concentrations increased, aluminium absorption (Xiao et al., 1992).

Overall, these results suggest that the presence of fluoride can enhance aluminium absorption.

Iron

Iron (Fe) status impacts on the absorption of aluminium and its accumulation in the brain; thus aluminium absorption was generally found to increase in the presence of low amounts of iron. Rats maintained on an iron-deficient diet had greater (0.0065%), and rats maintained on an iron-supplemented diet had lower (0.0028%) oral aluminium bioavailability than controls (0.0040%) (Winklhofer et al., 2000). Oral exposure to aluminium hydroxide produced a greater increase in aluminium excretion and brain aluminium levels in iron-deficient than in normal and iron-overloaded rats, whereas serum aluminium did not show consistent changes (Cannata et al., 1991; Fernández et al., 1989). Similarly, oral exposure to aluminium chloride produced non-significant increases in serum, spleen and liver aluminium concentrations in iron-deficient rats (Brown & Schwartz, 1992). Iron deficiency significantly increased tissue aluminium levels in rats co-administered aluminium and citrate (Brown & Schwartz, 1992). Iron depletion increased aluminium uptake, introduced as the hydroxide, by an intestinal cell line (Cannata et al., 1991; Fernandez Menendez et al., 1991). This is perhaps due to a similar mechanism of GI aluminium and iron uptake, which was suggested to involve an active process mediated by Tf, because aluminium transport was 3-fold greater as the Tf than the citrate (Cannata et al., 1991; Fernandez Menendez et al.,1991). Divalent iron, Fe(II), but not trivalent iron, Fe(III), increased the disappearance of aluminium, introduced as the hydroxide form, from the intestinal lumen in the in situ rat gut preparation and decreased the appearance of aluminium in portal and systemic blood (Van der Voet & de Wolff, 1987b). The authors suggested that Fe(II) may enhance Tf-mediated aluminium uptake, followed by the binding of aluminium in mucosal cells by ferritin. However, although there was a negative correlation between the iron status of Caco-2 cells and their aluminium uptake, both iron-depleted and iron-overloaded Caco-2 cells showed less transport of aluminium citrate, lactate and nitrilotriacetate across the cells than did cells with normal iron content (Alvarez-Hernandez et al., 1994), which is in contrast to the above results. Addition of Tf failed to facilitate aluminium uptake into Caco-2 cells when the aluminium was introduced as aluminium citrate (Alvarez-Hernandez et al., 1994). On the other hand, concurrent administration of 48 or 480 mg/kg Ferromia (sodium ferrous citrate, tetrasodium biscitrato iron (II), E0708) with 50 mg Al/kg as aluminium hydroxide daily 5 times weekly for 16 weeks to 5/6 nephrectomized rats did not affect serum or tissue aluminium concentrations compared to administration of aluminium hydroxide or aluminium citrate alone (Yamazaki et al., 1995). Similarly, iron deficient and iron overloaded rats did not show significantly different urinary aluminium excretion following i.v. aluminium chloride injection than controls, with the exception of less urinary aluminium excretion within the 5 days after aluminium treatment in iron-overloaded rats (Ittel et al. 1996). Brain, bone, liver and muscle aluminium concentrations were not different in iron-depleted vs. iron overloaded rats after 41 days of aluminium supplementation in the drinking water and diet, whereas spleen aluminium concentration was higher in the iron-supplemented rats (Ittel et al., 1996). Surprisingly, urinary aluminium output was greater when aluminium chloride was given orally with iron than saline (Ittel et al., 1996).

Nearly all of the interactions between iron and aluminium are consistent with an enhanced aluminium uptake and retention in the presence of iron deficiency (anaemia). In contrast to iron, zinc (Zn) deficiency did not produce measurable increases of tissue aluminium in rats after 28 days (McNall & Fosmire, 1996).

Calcium and sodium

Like iron, calcium (Ca) status impacts on aluminium absorption and accumulation. Dietary calcium deficiency increased the rate and extent of aluminium absorption, when introduced as the chloride, tissue aluminium accumulation, and aluminium-induced neuropathology in rats (Provan & Yokel, 1990; Taneda, 1984). Increased calcium decreased aluminium uptake and its appearance in plasma in studies that used the rat everted gut sac and in situ rat gut technique, suggesting a common uptake mechanism for aluminium, introduced as the chloride, and calcium (Cunat et al., 2000; Feinroth et al., 1982). Based on ionic radii, it is more likely that aluminium would compete with magnesium than with calcium. Although there is some evidence for aluminium-magnesium competition in vivo, this has not been well investigated.

A reduction of sodium (Na) has been reported to increase aluminium uptake. Replacement of sodium in the perfusate of the in situ gut perfusion study with equimolar choline chloride at pH 3 significantly (~ 50-fold) increased blood aluminium levels, suggesting a negative interaction between sodium and aluminium on aluminium absorption, introduced as the chloride, from the GI tract (Van der Voet & de Wolff, 1987a).

Ethanol

The results of one of two studies suggest that ethanol can affect the toxicokinetics of aluminium, but the mechanism is not known. Rats were given saline, 10% ethanol in drinking water, 25 mg Al/kg, as aluminium nitrate, by gastric intubation (i.g.) or 10% ethanol in drinking water and 25 mg Al/kg i.g. for 6 days/week for 6 weeks (Flora et al., 1991). After the 6 weeks, aluminium concentrations were significantly higher in blood and liver (24 and 76%, respectively), and non-significantly higher in kidney and brain (28 and 31%), in the ethanol-and-aluminium-treated group compared to the group treated only with aluminium. The amount of ethanol consumed was not reported. If fluid consumption was typical for the rat, it would have been 10 to 12 mL/100 g/day. The ethanol-and- aluminium-treated group also showed significant differences from the aluminium-only-treated group with respect to δ-aminolevulinic acid dehydratase, zinc protoporphyrin, and glutamic oxaloacetic transaminase in blood; glutamic pyruvic transaminase in liver; δ-aminolevulinic acid in urine; and dopamine and homovanillic acid in brain. These effects were generally an accentuation of the aluminium effects. However, it is not known if the elevation of blood and liver aluminium levels was due to ethanol alteration of aluminium absorption, distribution or excretion. It is also not known if ethanol increased aluminium toxicity or if these effects were the result of the combined toxicity of these two agents. It has been noted that chronic ethanol ingestion compromises GI tract integrity (Davis, 1993). The author speculated that it may increase aluminium absorption. Results of the combined administration of ethanol and aluminium in rats suggested that ethanol enhances the effects of aluminium, introduced as the chloride, but no information was provided to indicate if ethanol affected aluminium toxicokinetics (Rajasekaran, 2000). However, daily gavage with a 30% v/v ethanol solution delivering 3 g/kg and 91.8 mg aluminium lactate/kg for 90 days did not significantly affect serum aluminium, compared to aluminium alone (Kohila et al., 2004). There are no reports that assess directly the influence of ethanol on aluminium absorption, distribution or elimination.

Uraemia

There is evidence that uraemia enhances aluminium absorption. The primary documented problems with aluminium as a toxicant to bone and the brain have occurred in persons with uraemia (see Toxicokinetics, Absorption, Studies in Humans, Oral Administration, Factors Influencing Oral Aluminium Absorption, Uraemia). Aluminium levels were higher in the liver, but not in other organs, of chronically uraemic rats given oral aluminium hydroxide supplementation than in pair-fed controls (Drüeke et al., 1985). However, this could be due to a difference in absorption, distribution and/or elimination. Blood aluminium and/or urinary aluminium excretion were higher in uraemic, compared to normal, subjects after oral aluminium chloride, hydroxide or lactate administration (Ittel et al., 1987; 1988; 1991a; 1991b; 1996; 1997; Olaizola et al., 1989). The lack of increase of urinary aluminium after i.v. aluminium chloride or lactate administration to uraemic rats suggests the increased urinary aluminium in uraemic rats is due to enhanced absorption (Ittel et al., 1991b). The oral bioavailability of aluminium introduced as ²⁶Al plus ²⁷Al chloride was estimated to be 0.133% in controls and 0.175% in uraemic (5/6 nephrectomized) rats (Ittel et al., 1997). Reduction of renal function in rabbits, to ~ 23% of controls, increased the percentage of aluminium absorbed from aluminium chloride, citrate and lactate by ~ 50 to 100% (Yokel & McNamara, 1988). Systemic clearance was reduced to ~ 74% of that of controls. Ultrafilterable serum aluminium increased, perhaps accounting for the smaller reduction in systemic aluminium clearance than might be expected with a renal function that is 23% that of controls. Serum aluminium concentration was lower in 5/6 nephrectomized rats but urinary aluminium clearance was not significantly different from that of controls, suggesting a greater free fraction of aluminium in uraemia (Ittel et al., 1997). Uraemia may increase aluminium uptake through the paracellular pathway, indicated by increased serum and urinary aluminium in the presence of chemically-induced intestinal mucosal atrophy and by reduced serum aluminium when kinetin, a paracellular pathway blocker, was added to oral but not i.v. aluminium chloride administration (Ittel et al., 1992a; 1996).

Age

It is unclear from animal studies whether age influences aluminium absorption. Comparison of blood and urine aluminium in weanling and growing rats after oral aluminium hydroxide administration led Olaizola et al. (1989) to conclude that there is an inverse relationship between aluminium absorption and age. Rat brain aluminium concentration inversely correlated with age in control rats that were 21 days, 8 months and 16 months old (Domingo et al., 1996), in contrast to most observations in the human (see Toxicokinetics, Distribution (Including Compartmentalization), Human Studies, Tissue Aluminium Concentrations, Brain). After consuming aluminium and citrate in drinking water for 6.5 months, elderly rats generally had higher concentrations of aluminium in liver, kidney, spleen, bone and testis than the young rats (Gómez et al., 1997). In contrast, the brain aluminium concentrations were significantly lower in old than young rats (Domingo et al., 1996). The results of these studies are not consistent. Furthermore, none of these studies directly addressed absorption. The differences seen could be due to distribution or elimination.

Foods and dietary components

It has been assumed that the presence of food in the stomach inhibits aluminium absorption, due to aluminium association with organic ligands in food. Only a few studies have directly addressed this hypothesis. The results do not consistently show an influence of food on aluminium absorption. Some of the studies of oral aluminium bioavailability that model aluminium exposure from drinking water restricted food access, whereas others did not. Rodents and rabbits recycle their faeces to increase essential nutrient absorption and usually have stomach contents 24 to 36 hr after food removal. Therefore, simply depriving the animal of food does not guarantee that there will be no stomach contents. In some studies, rats were dosed after a 16 or 24 hr fast (Drüeke et al., 1997; Schönholzer et al., 1997), whereas in others free access to food was allowed (Drüeke et al., 1997). Oral aluminium bioavailability in these studies ranged from 0.06 to 0.36%. The oral bioavailability of ²⁶Al, given in the absence of ligand, in rats that were deprived of food for 24 hr was 0.94% (Drüeke et al., 1997). As stomach contents were found in rats deprived of food for ≥ 24 hr, as noted above, it cannot be assured that there were no stomach contents in the rats in this study. In contrast, Yokel et al. (2001b) did not find a difference in oral ²⁶Al bioavailability, when the Al was given in the absence of ligands, in rats exposed to a procedure that resulted in no stomach contents compared to rats that did have chow in their stomachs. Similarly, the addition of calcium and magnesium carbonates, designed to simulate hard water, did not affect oral aluminium bioavailability (Yokel et al., 2001b). Walton et al. (1994) conducted an ambitious study to assess the influence of beverages and foods on the oral absorption of aluminium given as the sulphate, to produce alum in solution. Their results show increased peak serum aluminium concentrations and urinary aluminium excretion after co-administration of orange juice and, to a much smaller extent, coffee and wine. Meat and carbohydrate/cereal products decreased aluminium absorption. However, neither the blood nor urine samples obtained hourly for 4 hr after dosing enable determination of oral aluminium bioavailability.

Dietary components such as phytate and polyphenols, that chemically associate with aluminium and reduce oral absorption of other minerals, may affect aluminium absorption (Powell & Thompson, 1993; Powell et al., 1993). However, no published studies were found that directly assessed this.

It was speculated that the species of aluminium in the GI tract, which may be present as insoluble or soluble forms and associated with various ligands and, perhaps more importantly, the form of mucus and interaction of aluminium with mucus, vary as a function of time since food consumption and, perhaps, composition of the diet. This may result in variable amounts and composition of absorbable aluminium species, contributing to the variable results in relation to the effect of stomach contents on oral aluminium absorption.

Overall, results suggesting that food composition of the presence of food in the stomach significantly affects oral aluminium bioavailability have been obtained in a few studies (Walton et al., 1994; Drüeke et al., 1997; Yokel & Florence, 2006) whereas another study did not find a significant effect of water quality or the presence of food in the stomach on oral aluminium bioavailability (Yokel et al., 2001b).

Chemical speciation

The chemical form in which aluminium is administered can affect its absorption (Cunat et al., 2000; Deng et al., 2000; Froment et al., 1989b; Schönholzer et al., 1997; Yokel & McNamara, 1988). As noted above in Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Carboxylic Acids, addition of citrate, which is believed to solubilize aluminium and perhaps open the paracellular pathway, increased oral aluminium bioavailability in many studies. Consumption by rats of diets for 18 days containing ~ 200 to 260 mg Al/kg diet as aluminium hydroxide, palmatate, lactate or phosphate, or aluminium hydroxide as a reagent grade chemical, or 2 different sources of desiccated gel, resulted in some significant differences in brain, bone and kidney aluminium concentrations (Greger, 1985). For example, aluminium hydroxide raised brain aluminium more than aluminium palmitate and one form of aluminium hydroxide gel raised kidney aluminium levels more than another form but resulted in lower aluminium levels in the tibia. Using the in situ rat gut technique, significantly more aluminium appeared in the plasma after perfusion with the citrate, lactate, tartrate and gluconate forms of aluminium, whereas plasma aluminium was not elevated after perfusion with the chloride, nitrate, sulphate and glutamate forms (Cunat et al., 2000).

It has been suggested that the chemical species of aluminium in drinking water in the absence of fluoride would be primarily labile, monomolecular Al(H₂O)₆ below pH 5 and Al(OH)_{1 to 4}^{+2 to -1} at higher pHs (Lazerte et al., 1997). In the presence of fluoride (1 ppm; 53 μM), AlF₂⁺ and AlF₃ would be expected at pH < 6.5. At higher, pHs mixed Al(OH)_xF_y complexes or Al(OH)₄^- would be expected (Nieboer et al., 1995). It is thought that these species would favour aluminium absorption from drinking water compared with that from food (Martyn et al., 1989) in which aluminium is presumed to be bound to phosphorus-rich compounds such as phytates and casein, a phosphoprotein (Glynn et al., 1995). Glynn et al. (1995) determined soluble and quickly reacting aluminium species (presumably species that might be absorbed) in the presence of 4 mg Al/L, the chloride, added to simulated gastric contents containing 40% rat feed. Less than 50% of the added aluminium was in the soluble fraction. As nearly 100% would be expected to be labile species (Al⁺³, monomeric hydroxo and sulphato complexes) under these conditions, the authors concluded that the labile aluminium rapidly complexed to components of food. They hypothesized that foods may alter aluminium absorption. The presence of food may change the pH. It has been suggested that food may bind aluminium, hence reducing its bioavailability. Therefore, it has been suggested that enhanced aluminium absorption may occur in the presence of an empty stomach, especially during periods of fasting. This has been discussed in Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Foods and Dietary Components. There is no strong evidence to support this notion.

In a subsequent study, rats drank water containing 0, 10, 50 or 500 mg Al/L, introduced as the chloride, for 9 weeks while consuming a diet containing 5 mg Al/kg (Glynn et al., 1999). Bone aluminium increased above control concentrations in the rats drinking the waters containing the higher two aluminium concentrations. In vivo fractionation of the stomach contents of these rats showed an increase of dissolved aluminium as water aluminium concentration increased, up to 50 mg Al/L, and the appearance of measurable quickly reacting aluminium species in those exposed to 500 mg Al/L. The authors interpreted their results as supportive of their hypothesis that aluminium absorption will increase when the aluminium-binding capacity of food in the stomach has been saturated. When compared to typical drinking water aluminium concentrations of 50 to 100 μg Al/L, these results suggest that aluminium would not be absorbed when stomach contents are present, as the aluminium binding capacity would not be saturated. This is not supported by several studies utilizing ²⁶Al in much lower aluminium doses than employed by Glynn et al. (1999). These studies found oral aluminium absorption in the presence of stomach contents (Drüeke et al., 1997; Jouhanneau et al., 1997a; Yokel et al., 2001b).

In contrast to the above hypothesis, Reiber et al. (1995) suggested that a substantial portion of aluminium, regardless of the form consumed, will be solubilized to monomeric aluminium in the stomach. They based this on studies showing solubilization of aluminium from particulate and colloidal sources within 2 hr at pH 1 to 2 and the average residence time in the human stomach, which has a pH of 1.5 to 2, of 1 to 4 hr. They concluded: “…the often raised issue of the relative bioavailability of the aluminium source would seem to be a moot point”, because “Regardless of the consumptive form, the bulk of aluminium will be converted to monomolecular species in the stomach” As they claimed that the stomach is impervious to charged species, they concluded: “In a healthy stomach, however, it is unlikely that any free or complexed aluminium can be absorbed across the stomach wall.” They claimed that the pH of stomach contents rapidly rises to 6.2 when in the duodenum, and then to pH 7.3 after leaving the duodenum. Because the nadir of aluminium solubility, due to aluminium hydroxide formation, is at pH 6.2 (Harris et al., 1996), they suggested: “…more than 99.9 percent of the consumed aluminium should ultimately be excreted in the stool as an aluminium hydroxide precipitate, leaving less than 0.1 percent available for uptake”, explaining why only a small percentage of oral aluminium is absorbed (Reiber et al., 1995). They recognized that this scenario is simplistic. However, at the pH of the human intestine, ~ 6.8, the major aluminium hydroxide species which would be formed in the absence of other ligands to associate with aluminium would be the soluble Al(OH)₄^- (Harris et al., 1996), suggesting a lower percentage of precipitated aluminium than stated by Reiber et al. (1995). If studies utilizing large doses of ²⁷Al are not considered, there is very little data relevant to comparative oral aluminium bioavailability for different aluminium forms to test the suggestion of Reiber et al. (1995).

The most ambitious study of the influence of beverages and foods on oral aluminium absorption was conducted by Walton et al. (1994). This is a non-peer reviewed report of a study in which large sample sizes were not used. However, an issue that has not been well investigated was addressed. They exposed anesthetized rats to 2 mL of water containing no aluminium, or 8 mg aluminium as aluminium sulphate and/or various beverages and foods by gastric administration. Blood was obtained from the tail, prior to and 1, 2, 3 and 4 hr after, dosing. Urine was obtained, at the same times after dosing, by needle aspiration from the bladder. Considering the short period of observation (4 hr), serum aluminium is probably a better indicator of absorbed aluminium than is aluminium excreted in the urine. Administration of 8 mg aluminium in water increased peak serum aluminium about 4.5-fold and produced a urinary aluminium concentration of 2.6 μg/L. Co-administration of orange juice, coffee and wine with the aluminium significantly increased peak serum aluminium levels, compared to administration of aluminium alone, approximately 17-, 2.5-, and 1.9-fold, respectively. Serum aluminium peaked at 1 hr when given alone. The time of peak serum aluminium in the presence of co-administered foods was not reported. Therefore, it is not known if the rates of aluminium absorption are similar and, as a result, if the 4 blood samples obtained represent similar time profiles of the aluminium absorbed. Urinary aluminium concentration increased approximately 10.5-, 1.3- and 1.8-fold when these three beverages were co-administered with aluminium. Meat and Vita Wheat® biscuits (a carbohydrate/cereal product) significantly attenuated the elevation in serum aluminium caused by aluminium dosing to 62 and 65%, when compared to administration of aluminium alone. Urinary aluminium was attenuated by these two foods to 45 and 57%. These foods were available for 36 hr prior to their co-administration with the aluminium, to “load the gastrointestinal tract with the single food and its breakdown products.” In contrast, no food was available for 36 hr prior to aluminium alone or beverage (plus aluminium) dosing. In preliminary studies, the investigators found food in the stomach 24 hr after removal of food access, so they went to 36-hr food deprivation. They did not verify the lack of stomach contents after 36-hr food deprivation. As noted above, this does not assure the lack of stomach contents. Co-administration of beer, Coca-Cola®, tea, apple, broccoli, butter and margarine produced non-significant effects. Co-administration of both orange juice and Vita Wheat® with aluminium resulted in an increase of serum aluminium levels of 9.1-fold, an attenuation of the result seen with orange juice alone, but urine aluminium concentration was not attenuated by the addition of Vita Wheat® to the aluminium plus orange juice administration. No definitive conclusions can be made from this study. The results from co-administration of coffee and wine, which slightly increased serum and urine aluminium levels, and meat and the cereal product, which decreased serum and urine aluminium levels, suggest that co-administration of certain foods with drugs or foods that contain significant amounts of aluminium warrants further study.

The site and mechanisms of oral aluminium absorption

Aluminium, like most substances, is better absorbed from the upper intestine than from the stomach. The stomach is lined by a thick, mucus-covered membrane. It has a much smaller surface area than the intestine. The primary function of the stomach is digestive, whereas that of the intestine in absorptive. Ionized molecules are usually unable to penetrate the lipid bilayer of cell membranes due to their low lipid solubility (Benet et al., 1996). As aluminium would be expected to be present primarily as the free ion, with associated waters of hydration, at the low pH of the stomach, non-carrier-mediated absorption would not be predicted from the stomach. Studies with in vivo isolated duodenal segments suggested aluminium uptake at pH 2 was due to both an aluminium-concentration-dependent non-saturable process and a saturable process that is at least partially vitamin D dependent, suggesting that aluminium may compete with calcium (Adler & Berlyne, 1985). Uptake of aluminium from the citrate into stomach sacs was much less than into small bowel and colon (Whitehead et al., 1997). Plasma aluminium concentrations peaked 45 minutes after oral citrate intake by rats (Froment et al., 1989a; 1989b). They observed a simultaneous peak of plasma aluminium and glucose and concluded that the site of aluminium absorption is probably the proximal small intestine (Froment et al., 1989b). The lower pH of the proximal duodenum would be expected to result in greater aluminium absorption than more distal intestinal segments (Greger & Sutherland, 1997).

GI aluminium absorption appears to be a two-step process, an initial mucosal cell uptake, followed by much slower release into the blood. This was suggested by Feinroth et al. (1982), based on results obtained with everted rat jejunal sacs and aluminium chloride. Others reached a similar conclusion, based on results with in vivo isolated gut segments and the in situ intestinal perfusion technique with portal and systemic blood sampling after aluminium chloride (Adler & Berlyne, 1985; Van der Voet & de Wolff, 1986-1987) and aluminium chloride, lactate and nitrate addition (Jäger et al., 1991), as did Cochran et al. (1990), after aluminium chloride addition, using a duodenal perfusion preparation. It was shown that aluminium binds to mucus glycoproteins in the GI tract (Crowther & Marriott, 1984). A much greater loss of aluminium from the intestine than appeared in the blood was typically found in the above studies. This loss was attributed to mucosal cell or mucous uptake of aluminium. Support for this suggestion was provided by the finding that nearly all of the aluminium not recovered in the perfusate effluent from the rat gut could be recovered in intestinal mucus (Powell et al., 1994; 1999). This group found tissue aluminium uptake to be up to 1000-fold that of aluminium transport across the gut tissue (Whitehead et al., 1997). Powell et al. (1999) found that the association of aluminium with gelatinous mucus prevents formation of aluminium-hydroxy precipitates. They interpreted their results to show that aluminium is primarily associated with insoluble mucus, which, due to its slow reaction kinetics, reduces aluminium absorption, enabling its faecal elimination (Powell et al., 1999). The finding that ~ 60% of aluminium, introduced as the chloride, taken up from the in situ perfused rat intestine was in the small intestine is consistent with this interpretation (Arnich et al., 2004).

The mechanisms mediating GI aluminium absorption have been suggested to include passive (diffusion) and active (carrier- and vesicular-mediated) transport across intestinal cells as well as paracellular diffusion between these cells. Support has been provided for an energy-dependent uptake process. Dinitrophenol (DNP), cyanide, vanadate, the absence of glucose and low temperature inhibited the uptake of aluminium, introduced as the chloride, in studies utilizing the rat everted gut sac, isolated jejunal slice, and a duodenal perfusion technique (Cochran et al., 1990; Feinroth et al., 1982; Provan & Yokel, 1988b). Reduced glucose in the in situ rat gut technique perfusate decreased aluminium in plasma (Cunat et al., 2000). In contrast, Farrar et al. (1987) and Provan & Yokel (1988a) did not find an effect of DNP using rat duodenal, jejunal or ileum preparations and the in situ rat gut preparation. Whitehead et al. (1997) found no effect of ouabain, suggesting the lack of dependence on Na/K-ATPase. A plateau of aluminium uptake, introduced as the chloride, was observed by Feinroth et al. (1982) and Provan & Yokel (1988b), suggesting a carrier-mediated mechanism. Deleting phosphorus from the in situ rat gut technique perfusate reduced aluminium uptake into plasma which, the authors suggested, could be due to the role of phosphorus in cellular respiration (Cunat et al., 2000). However, the presence of phosphorus, which causes aluminium phosphate precipitation, has been associated with decreased aluminium absorption (Kaehny et al., 1977).

Evidence for an interaction with calcium uptake has been obtained. Increasing calcium perfusate concentration decreased aluminium, introduced as the chloride, uptake (Feinroth et al., 1982) and addition of aluminium chloride to the in situ perfused duodenum decreased calcium uptake (Adler & Berlyne, 1985). Calcium channel blockers decreased, and calcium channel activators increased, uptake into rat jejunal slices (Provan & Yokel, 1988b) and duodenum (Cochran et al., 1990) when introduced as aluminium chloride; however, Provan & Yokel (1988a) did not find an effect of calcium channel blockers or a facilitator in the in situ rat gut preparation.

It has been suggested that citrate facilitates aluminium absorption by opening the tight junctions between GI mucosal cells, by chelating calcium, which is required for tight junction integrity, and that the aluminium is absorbed as aluminium citrate (Froment et al., 1989b). This mechanism of enhanced aluminium absorption was favoured by Taylor et al. (1998). However, they found peak serum citrate concentrations 31 to 32 minutes after aluminium citrate consumption, whereas aluminium concentrations peaked at 77 to 108 minutes, suggesting that aluminium was not absorbed as aluminium citrate due to the faster absorption of citrate and was not released into blood as aluminium citrate (Taylor et al., 1998).

Some evidence suggests a role for sodium transport processes. Low sodium and amiloride, a sodium uptake blocker, increased the uptake t_½ of aluminium (Provan & Yokel, 1988a). In contrast, Van der Voet & de Wolff (1987a) found a negative interaction between sodium and aluminium uptake. Further study by Van der Voet & De Wolff (1998) showed that the presence of calcium, but not sodium, reduced aluminium uptake. The presence of both calcium and sodium attenuated the calcium effect, leading the authors to speculate: “… aluminium attempts to mimic calcium in its Na-related intestinal passage”.

Paracellular pathway blockers increased the t_½ of aluminium uptake in the in situ gut preparation (Provan & Yokel, 1988a). More support for the hypothesis that the paracellular pathway is involved in aluminium absorption was provided by further studies which used the in situ rat gut preparation and aluminium chloride or citrate (Froment et al., 1989a; Partridge et al., 1992; Provan & Yokel, 1988a), inverted intestinal segments and aluminium citrate (Farrar et al., 1987; Whitehead et al., 1997), and Caco-2 cells exposed to aluminium in the absence of ligand or as the citrate, fluoride, hydroxide or maltolate (Zhou & Yokel, 2005). However, paracellular pathway blockers did not affect aluminium uptake by the jejunal slice (Provan & Yokel, 1988b), which might be expected when the endpoint is uptake into, rather than flux across, tissue.

Tf addition to the medium vascularly perfusing an intestinal preparation increased aluminium uptake, suggesting it facilitated entry of aluminium, introduced as the chloride, into blood, perhaps from intestinal cells (Jäger et al., 1991). It has been suggested that Tf mediates aluminium release from mucosal cells into blood (Greger & Sutherland, 1997). There is no single unifying explanation for these results. It is probable that there is more than one mechanism of aluminium uptake, which might be aluminium species-, pH- and intestinal region-dependent. Multiple mechanisms would account for the lack of ability of manipulations to totally block aluminium uptake, and perhaps the significant results obtained by some, but not other, investigators.

Lower serum aluminium concentrations resulted from portal vein than from i.v. injections of the same aluminium dose, as the sulfate, to rats, indicating significant first-pass pre-systemic clearance of aluminium by the liver (Xu et al., 1992a). These results suggest that determination of aluminium absorption based on area-under-the-curve calculations from multiple blood/serum aluminium determinations over time might underestimate absorbed aluminium, although this method estimates the aluminium that reaches systemic circulation and is available for distribution to organs such as the brain.

The chemical speciation of aluminium, consumed in beer, as it passes through the upper part of the GI tract was modelled by (Sharpe et al., 1995). The citrate concentration in beer exceeded the aluminium concentration, resulting in the predominant aluminium species in lager, the mouth and stomach being aluminium citrate. In the duodenum and jejunum, where the pH is ~ 6.5, the predicted predominant species was aluminium phosphate. This is the primary site of aluminium absorption. At pH 7, neutral aluminium species exceeded charged and “solid” (insoluble) aluminium species at total aluminium concentrations below 7600 μg Al/L, which is well above the median aluminium concentration in the beers studied (100 μg Al/L).

Studies in humans

Oral administration

Drinking water

The first human studies attempting to estimate oral aluminium bioavailability utilized ²⁷Al. These were balance studies, in which aluminium absorption was estimated from the difference between intake and faecal excretion. Aluminium retention was estimated from the difference between intake and faecal-plus-urinary excretion. However, it has become apparent that the percentage of aluminium that is orally absorbed is quite low (as noted in the animal studies above and more recent human studies, below). Therefore, studies using these methods are not considered reliable, as discussed above.

One of the few human studies of oral aluminium absorption that models drinking water is that reported by Stauber et al. (1999). The 21 subjects consumed a diet that provided a total intake of about 3 mg Al/day. They drank either 1.6 L daily of an ATW from a municipal treatment plant that they found contained 140 μg Al/L or reconstituted soft water (RSW) that had < 1 μg Al/L, with and without sodium citrate. The ATW provided 208 to 233 (in the absence of added sodium citrate) or 253 μg Al/day (when sodium citrate was added). The RSW provided < 1 μg Al/day (when no sodium citrate was added) or 46 μg Al/day (when sodium citrate was added). Oral bioavailability was estimated from the 24 hr urinary aluminium output times 2.2. The value 2.2 was used to correct for the estimated fraction of total urinary aluminium excretion that occurs within the first 24 hr and the fraction of absorbed aluminium that is retained (e.g., does not appear in the urine). This is based on the fraction of aluminium excreted in the urine over 24 hr compared to that excreted over 7 days, which was estimated to be 0.62 by (Priest, 1993). It is also based on the fraction of aluminium administered intravenously that was found in the urine within 7 days by Talbot et al. (1995), 0.72, suggesting the balance of the aluminium was retained. They concluded that the oral bioavailability of aluminium from water was 0.39% in the absence, and 0.36% in the presence, of citrate. This was based on the increase in 24 hr urinary aluminium excretion when the subjects consumed ATW and a controlled diet compared to RSW and the same diet, and the increased aluminium in ATW compared to RSW. The controlled diet was given to all subjects in the same amounts for three meals, three snacks, tea and a banana daily, delivering 1777 calories and 2.9 mg aluminium. It contained standard amounts of specific food components, including meat, cereal, milk, tea, and cookies, each of which was analyzed to determine its aluminium concentration. Aluminium in ATW represented 6 to 7% of total aluminium intake. Its consumption raised urinary aluminium output approximately 9%. Therefore, the response seen was < 10% of background urinary aluminium excretion. A concern about this study is that one might expect the variability in the excretion of aluminium in the urine from food to mask any ability to see an increase in urinary aluminium output from the aluminium in water. The range of 24 hr urinary aluminium output in these subjects was 5-fold (1.8 to 9.3 μg). However, within-subject variability must have been considerably less for the results to produce a statistically significant difference in urinary aluminium output after consuming the ATW compared to RSW. This is indicated by the 95% confidence intervals (CI) that suggest considerably less variability than the 5-fold range of their subjects. The design of this study favoured the study objectives, to measure the amount of aluminium absorbed from water, above that contributed by food. Their study included within-subject comparisons and a defined diet containing a low amount of aluminium for the day before and the day of the study (3 mg vs. normal intake of 3 to 7 mg/day in Australia, according to the authors). Glynn et al. (1995) stated that the concentration of aluminium in drinking water has to be extremely high in relation to food to ensure that aluminium from the water is a significant part of the total oral aluminium intake. Similar concern expressed was: “…biokinetic studies of injected or ingested aluminium in man and animals is complicated by the investigator’s inability to distinguish between aluminium from a test dose and that already present in the body. Most studies of stable aluminium are, thus, of poor sensitivity.” (Priest, 1993). Apparently Stauber et al. (1999) were able to overcome this reservation. The correction factors they used to estimate urinary aluminium output were based on two human studies conducted by other investigators in which ²⁶Al was used. These estimates may not accurately predict the fraction of aluminium excreted in urine in 1 day compared to that excreted to time infinity, and may not accurately predict the percentage retained, when that estimate is also based on 1 vs. 7 days of observations. Although these estimates may influence the absolute estimate of oral bioavailability, they are less likely to influence the relative estimates of aluminium bioavailability from water compared to that from food; this is discussed below.

A second study relevant to consumption of aluminium from drinking water involved 4 subjects who consumed a soft mineral water containing < 1 μg Al/L, the same water with 300 μg Al/L as Al₂(SO₄)₃, and deionized water containing 300 μg Al/L as Al₂(SO₄)₃ (Gardner & Gunn, 1995). Unfortunately, urine was collected only 7 hr after aluminium administration; urinary aluminium excretion rate would not have yet returned to the pre-aluminium-treatment rate in many cases. Therefore, the estimate of oral aluminium bioavailability, which was < 0.1%, is below the true value.

A study of 2 humans who consumed ²⁶Al as aluminium chloride, Hohl et al. (1994) produced an estimate of oral aluminium bioavailability of 0.1%. The authors collected urine for 4 days (although missing the collection of part of the sample from one of the two subjects on day 1). Their estimate of 0.1% oral aluminium bioavailability was not corrected by the aluminium that was retained by the subjects or excreted after the period of sample collection. The authors suggest this would not result in a large error.

Another study that modelled aluminium consumption from drinking water employed two subjects who consumed ²⁶Al added to water from a public supply (Priest et al., 1998). Faeces and urine were collected for 7 days after ²⁶Al administration; these contained 97.6 and 100.4% of the ²⁶Al administered to the two subjects. These results suggest that little aluminium was absorbed. The results also illustrate the inadequacy of the balance method to determine oral aluminium bioavailability. The ²⁶Al in blood obtained 1, 4 and 24 hr after oral ²⁶Al dosing was multiplied by the estimated blood volume of each subject. The results suggested oral aluminium bioavailability of 0.027, 0.034 and 0.012%, based on these three sampling times. Cumulative urine excreted suggested oral aluminium bioavailability was 0.20 and 0.14%, for these two subjects. The authors corrected these values for the percentage of aluminium voided in the urine in the same time period after i.v. injection (72%) (Talbot et al., 1995), as conducted by Stauber et al. (1999). This correction resulted in an estimate of 0.22% oral aluminium bioavailability. This is an order of magnitude greater than the results they obtained by estimating absorption from a single serum sample and the calculated volume of distribution, suggesting this latter method underestimates oral aluminium bioavailability.

Overall, the above results suggest that oral aluminium bioavailability from drinking water is in the range of 0.1 to 0.5%; and most likely approximately 0.3%.

An error resulted in the introduction of 20 tons of 8% aluminium sulphate into a municipal water supply near Camelford, England. Some of the consumers reported that the water had an unpleasant metallic taste; others reported various symptoms (Eastwood et al., 1990). Excessive aluminium, copper, lead and zinc were found in the water. Two of the affected individuals showed elevated urinary aluminium output. Bone biopsies showed the presence of elevated aluminium (Eastwood et al., 1990) that was still elevated 6 and 7, but not 18, months later (McMillan et al., 1993). These results suggest the possibility of elevated bone aluminium in humans with normal renal function after oral consumption of excessive amounts of aluminium. However, only two reports were found of elevated bone aluminium in humans with normal renal function after massive oral consumption of aluminium. One was a 49 year old male with a 25 year history of consumption of aluminium-containing antacids whose bone aluminium concentration, 24 mg/kg dry weight, was between that of three non-dialysis subjects at autopsy, which averaged 6.4 mg Al/kg, and that of 3 dialyzed subjects at autopsy, which averaged 125 mg Al/kg (Recker et al., 1977). The other was a 39 year old female who consumed antacids containing a total of ~ 18 kg of elemental aluminium over 8 years. She had stainable aluminium on 27.6% of the bone surface (Woodson, 1998).

Beverages and foods

Oral aluminium bioavailability from the diet was estimated to be 0.1 to 0.3% based on normal urinary aluminium excretion of 20 to 50 μg/day and a daily aluminium intake of 20 mg (Ganrot, 1986). Daily aluminium intake is now believed to be less. Priest (1993) estimated oral aluminium bioavailability from food to be ~ 0.1% based on a daily aluminium intake of 15 mg, a daily urinary excretion of 0.025 mg and 5% aluminium retention in the body. The same estimate was obtained by comparing an average daily urinary aluminium excretion of 0.004 to 0.012 mg to average daily aluminium intake from food of 5 to 10 mg (Nieboer et al., 1995). Based on daily dietary aluminium intake of 10 mg, a terminal t_½ of aluminium of 50 years, and aluminium body burdens of 5 and 60 mg, oral aluminium bioavailability was estimated to be 0.14 and 1.6%, respectively (Priest, 2004).

The bioavailability of aluminium from a low aluminium diet was estimated to be 0.78%, compared to 0.09% when aluminium, as aluminium lactate, was added to the diet, (Greger & Baier, 1983). These results are in the range of values obtained for aluminium from drinking water (see Toxicokinetics, Absorption, Studies in Humans, Oral Administration, Drinking Water). The results suggest an inverse relationship between aluminium dose and oral absorption. However different aluminium species were consumed under these two conditions.

Aluminium bioavailability was estimated based on 24 hr urinary aluminium excretion in subjects consuming a controlled diet that included tea and RSW water (which provided essentially no aluminium) (Stauber et al., 1999). The ~ 3 mg Al/day provided by this diet is below typical dietary intake. Absorption of aluminium from food consumed prior to the study, which likely provided > 3 mg Al/day, may have contributed to the urinary aluminium excretion during the study. This would produce an over-estimation of aluminium bioavailability from the controlled diet. The estimate of oral aluminium bioavailability from food-plus-tea in this study was ~ 0.53%, assuming that 10% of the aluminium in the tea was available for absorption. Based on the assumption that 100% of the aluminium in the tea was available for absorption, oral aluminium bioavailability from food-plus-tea was estimated to be 0.28%. Therefore, the authors concluded that oral aluminium bioavailability from water and food is comparable. They claimed “… recent research suggesting that aluminium in food and aluminium in water have similar bioavailabilities”, citing Priest (1993). However, Priest does not make this statement or provide data to support, or refute, it.

Some studies found increased aluminium in the urine after tea consumption, suggesting absorption of aluminium from the tea. Urinary aluminium concentration increased in the 12 hr after tea consumption (Koch et al., 1988). Equal volumes of coffee or water did not increase the urinary aluminium concentration. They did not report urine volume, or urinary aluminium output, so it is unknown if aluminium excretion increased. However, if urine volume was greater after consuming tea than water, as reported by Powell et al. (1993), urinary aluminium output would have been greater following tea consumption. In the 24 hr after consumption of 2 litres of tea, presumably containing a total of 218 μmoles of aluminium, by one subject, urinary aluminium output was 0.725 μmoles, compared to 0.14 μmoles after consumption of water by the same subject (Powell et al., 1993). This would suggest 0.3% aluminium bioavailability. In contrast, tea, with or without lemon juice or milk, or mineral water was consumed one day with a defined diet, in a cross-over study. The tea contained 2.3 to 2.8 mg aluminium and comprised about 31% of the total daily dietary aluminium intake. Blood was obtained 11 times, from immediately before to 0.5 to 24 hr after dosing. No elevation of serum aluminium, above the pre-treatment mean concentration of 4.2 μg/L, was seen (Drewitt et al., 1993). Four subjects who consumed 2 litres of tea containing ~ 4 mg Al/L eliminated an average of 0.003 mg aluminium within the subsequent 7 hr in their urine, or ~ 0.04% of the aluminium in the tea (Gardner & Gunn, 1995). However, this is clearly an underestimate of the oral aluminium bioavailability because urinary aluminium excretion had not yet returned to the pre-treatment rate in most subjects, suggesting insufficient time to follow aluminium absorption and/or incomplete excretion of the absorbed aluminium.

It was suggested that polyphenols in tea bind most of the aluminium, thus greatly reducing its oral bioavailability (Flaten & Odegard, 1988). Collection of ileostomy effluent from a subject who had not consumed food orally for two weeks but had consumed tea suggested there was no breakdown of the polyphenols from the tea (Powell et al., 1993). These results suggest that digestion does not change the major ligand that binds aluminium in tea.

Chewing gums contain a significant amount of aluminium (0.2 to 4 mg/stick for 3 brands) (Lione & Smith, 1982). Chewing released 2 to 21% of the aluminium, 0.01 to 0.44 mg. The authors concluded that moderate use of chewing gums would not contribute substantially to daily aluminium intake. However, chewing gum may exceed drinking water as a daily source of aluminium.

Drugs

Oral aluminium bioavailability was estimated to be 0.006 and 0.007% from two studies of a commercial aluminium hydroxide product (Linc®); estimates were based on the increase above baseline of urinary aluminium excretion after consumption of 1, 4 or 8 tablets (Haram et al., 1987; Weberg & Berstad, 1986). Weberg & Berstad (1986) observed an inverse relationship between dose and oral aluminium bioavailability. They suggested that this was due to the ability of more tablets to produce a greater increase in the pH of the intragastric milieu, resulting in decreased solubility. However, if aluminium is primarily absorbed from the upper intestine, where the pH is about 6 or 7, this may not be the explanation. It is unknown if diet influenced these results, as this was not controlled or documented. Oral aluminium bioavailability from aluminium hydroxide, and aluminium glycinate taken with aspirin, was estimated to be 0.003 and 0.22%, respectively (Meshitsuka & Inoue, 1998; Meshitsuka et al., 1999).

Oral aluminium bioavailability was estimated from 6-day urinary ²⁶Al output in 2 subjects (Priest et al., 1996). Both subjects received intragastric dosing of ²⁶Al incorporated into aluminium hydroxide, ²⁶Al hydroxide in the presence of citrate, and ²⁶Al citrate. Oral aluminium bioavailability was estimated from urinary aluminium output, corrected for the fraction of aluminium administered intravenously that was found in the urine within 7 days, 0.72, as demonstrated by Talbot et al. (1995), and as used by Stauber et al. (1999) above. Estimates of the percentage of ²⁶Al absorbed from ²⁶Al hydroxide in the absence and presence of citrate averaged 0.01 and 0.14%, respectively; aluminium bioavailability from ²⁶Al citrate was 0.52%. These limited results mirror many reports showing very low aluminium bioavailability from aluminium hydroxide (and sucralfate) and enhancement by citrate.

Oral aluminium bioavailability from aluminium hydroxide appears to be less than from food or water.

Factors influencing oral aluminium absorption

Studies in humans have also shown that oral aluminium absorption is dependent on many factors, although this has been less well studied than in animals. A summary of the reported factors affecting oral aluminium absorption in humans is shown in Table 18. The following sections describe these effects in much more detail.

Table 18

Summary of the influence of factors that affect oral aluminium absorption from studies in humans.

Solubility

Oral bioavailability of aluminium from quite insoluble forms, such as aluminium hydroxide and sucralfate, was generally reported to be quite low, e.g. ~ 0.001 to 0.007% (Haram et al., 1987; Weberg & Berstad, 1986), whereas from drinking it water was ~ 0.22 and 0.35% (Priest et al., 1998; Stauber et al., 1999). Higher plasma levels were observed after oral administration of sucralfate in suspension than in tablet form (Conway et al., 1994). This is not consistent with the notion that bioavailability is independent of the chemical species of the aluminium (Reiber et al., 1995), a notion which does not appear to be valid.

pH

Administration of ranitidine, which increased gastric pH, reduced urinary aluminium excretion (Rodger et al., 1991), consistent with an enhancement of aluminium absorption at lower gastric pH. Serum and urine aluminium levels increased in pre-surgery ulcer patients with normal renal function who had a mean gastric pH of 2.09 following administration of 30 mg/kg aluminium hydroxide, whereas the opposite was seen in post-surgery patients whose gastric pH averaged 5.78 (Olaizola Ottonello et al., 1991). In contrast, limited observations in patients with chronic renal failure failed to show a correlation between gastric acid secretion and elevation of serum aluminium after oral aluminium consumption, suggesting gastric acid secretion may not play a critical role in aluminium absorption (Beynon & Cassidy, 1990). There are insufficient reported results to draw a firm conclusion about any effect of gastric pH on oral aluminium absorption.

Carboxylic acids

Numerous studies in humans have shown enhanced aluminium absorption from aluminium hydroxide in the presence of citrate, other carboxylic acids and orange juice (Coburn et al., 1991; Fairweather-Tait et al., 1994; Lindberg et al., 1993; Mauro et al., 2001; Nestel et al., 1994; Nolan et al., 1990; Nordal et al., 1988a; Priest et al., 1996; Rudy et al., 1991; Slanina et al., 1986; Walker et al., 1990; Weberg & Berstad, 1986). For example, Weberg & Berstad (1986) reported that 0.004% of 1 g of aluminium, given as an antacid, was absorbed in the absence of citrate, compared with 0.03% when consumed with orange juice and 0.2% when consumed with citric acid solution that delivered 0.7:1 citrate:aluminium. Similar results have been observed in animal studies (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Carboxylic Acids). In contrast, co-administered sodium bicarbonate and calcium acetate were not found to increase aluminium absorption (Mauro et al., 2001; Walker et al., 1990).

The first study of oral aluminium absorption using ²⁶Al was conducted in one human subject who received aluminium citrate. The results suggested ≥ 1% oral absorption (Day et al., 1991), whereas two other studies involving 5 and 2 subjects led, respectively, to estimates of 0.015 (Edwardson et al., 1993) and 0.52% (Priest et al., 1996). However, there is a human study that failed to find an increase of aluminium absorption in the presence of citrate. This was conducted in subjects who received aluminium with their diet (Stauber et al., 1999). Generally, increasing the amount of citrate consumed with aluminium hydroxide increased blood aluminium concentrations (Taylor et al., 1992).

Although citrate can increase aluminium absorption, it has not always been reported to increase tissue or cellular aluminium retention. Some animal studies have reported a citrate-induced increase of brain and aluminium (as described above in Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Carboxylic Acids), whereas other animal studies and one human study have not (Sakhaee et al., 1993). Therefore, citrate may enhance oral aluminium absorption but may also enhance its distribution into and out of tissues as well as from the organism by renal elimination, in the presence of renal function, as suggested by Maitani et al. (1994), and discussed below (see Toxicokinetics, Elimination and Excretion, Animal Studies, Urinary Excretion).

The primary humic substances in water and soil are humic and fulvic acids. One healthy person who took 1.75 g of aluminium hydroxide, adjusted to pH 2, and 1.25 gm of concentrated humic substances, excreted more aluminium in the urine over the subsequent 50 hr than when the aluminium was taken alone (1.4 v.s. 0.6% of the aluminium dose), suggesting humic substances can increase aluminium absorption (Alexander et al., 1990).

Silicon-containing compounds

Sodium silicate (100 μM) reduced the GI absorption of ²⁶Al (consumed in orange juice, a source of citrate) by 85% in fasted humans (Edwardson et al., 1993). Taylor et al. (1995) cited previous studies showing an inverse relationship between aluminium and silicon concentrations in drinking water. An earlier study showed an inverse relationship between renal function and plasma silicon but no correlation between serum aluminium and silicon in patients with chronic renal failure on regular haemodialysis (Roberts & Williams, 1990). In a later study there appeared to be an inverse relationship between serum aluminium and silicon levels in a subset of haemodialysis patients (Parry et al., 1998). Studies of the aluminium and silicon dioxide concentrations in drinking water suggested that moderate silicon concentration has a slight protective effect against aluminium-associated impaired mental function (Forbes et al., 1995; Forbes & Gentleman, 1998). These results suggest that increasing dietary silicon may reduce aluminium absorption and facilitate excretion.

Fluoride

Fluoride forms numerous complexes with aluminium (see Identity, Physical and Chemical Properties, Analytical Methods, Physical and Chemical Properties, Properties of Aluminium Compounds). Aluminium appears to inhibit fluoride absorption (Brudevold et al., 1973; Hobbs et al., 1954; Li et al., 1991; Spencer et al., 1980a; 1980b; 1981; 1985; Spencer & Kramer, 1985; Spencer & Lender, 1979).

In fluoridated drinking water, decreased pH would favour formation of aluminium-fluoride complexes. In one survey, 19% of aluminium in treated water was complexed with fluoride, when 58 μM (~ 1 mg/L, 1 ppm) fluoride was added (Driscoll & Letterman, 1988). Moderate concentrations of fluoride in drinking water were found to have a slight protective effect against aluminium-associated impaired mental function (Forbes & Agwani, 1994a; 1994b; Forbes & Gentleman, 1998). However, there was no obvious benefit of 40 mg of fluoride given daily for 1 year to patients in the early stages of AD (D. Shore, personal communication, 1992). Although complexation of aluminium with fluoride can be important at the more acidic pHs found occasionally in water, there is not sufficient fluoride in plasma to form any significant amount of binary aluminium fluoride complexes (W.R. Harris, personal communication, 2000). It is not clear if fluoride significantly influences aluminium toxicokinetics.

Iron

Haemodialysis patients who had low or normal serum ferritin levels and were given aluminium hydroxide for 7 days had increased serum aluminium, whereas patients with high serum ferritin did not show an increase of serum aluminium concentration (Cannata et al., 1984). Serum aluminium negatively correlated with serum iron, serum ferritin and Tf saturation in chronic haemodialysis patients (Cannata et al., 1993; Huang et al., 1992). Similarly, there was a negative correlation between serum iron and aluminium absorption in dialysis patients (Cannata et al., 1993). These results are consistent with those of animal studies showing enhanced aluminium absorption in the presence of iron deficiency (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Iron).

Calcium

Addition of calcium to aluminium hydroxide did not affect aluminium absorption in humans with normal renal function who presumably had normal calcium status (Nolan et al., 1990). Based on ionic radii, it is more likely that aluminium would compete with magnesium than calcium. Although there is some evidence for aluminium-magnesium competition in vivo, this has not been well investigated.

Uraemia

The primary documented problems with aluminium as a toxicant to bone and the brain have occurred in persons with uraemia, who accumulate aluminium, due to their inability to excrete it, and may develop aluminium-induced toxicity (Alfrey, 1980). There is evidence that uraemia may also enhance aluminium absorption. Ittel et al. (1991a) found greater serum aluminium levels in acute and chronic renal failure patients than those with normal renal function receiving the same daily dose of aluminium (1.18 g) as hydroxy-magnesium aluminate, greater daily urinary aluminium excretion in the chronic renal failure patients, and a significant negative correlation between renal aluminium excretion and creatinine clearance. They concluded that there was an enhancement of GI aluminium absorption in the presence of chronic renal failure. However, calculations using the biokinetic model, described in Toxicokinetics, Pharmacokinetic Modelling, failed to find evidence for increased oral aluminium absorption by humans with chronic renal failure (Steinhausen et al., 2004).

Age

The information related to the affect of age on aluminium absorption is from adult and geriatric subjects. No published studies of children were found. A greater increase in blood aluminium was seen in subjects aged > 77 than in those aged < 77 (serum aluminium 101 vs. 38 μg/L at 1 hr), who consumed ~ 4.5 mg/kg aluminium hydroxide and 3.3 to 6.5 g citrate (citrate:aluminium, 1.6:1 to 3.2:1) after an overnight fast (Taylor et al., 1992). Comparing oral aluminium bioavailability in the subjects < 59 with those > 59 years of age failed to reveal a difference (Stauber et al., 1999). Taking these observations together with the results from animal studies, (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Urinary Excretion) there is a lack of consistent data to be able to conclude that there is a significant effect of age on aluminium absorption, distribution or retention.

Dementia

In a study of 20 AD subjects aged 65 to 76 (n=10) and 77 to 89 (n=10) blood aluminium levels were compared to those of 20 controls. Subjects and controls consumed ~ 4.5 mg/kg aluminium hydroxide and 3.3 to 6.5 g citrate (citrate:aluminium, 1.6:1 to 3.2:1) after an overnight fast (Taylor et al., 1992). In the younger AD subjects, the blood aluminium was significantly greater at 60 minutes than in control subjects (104 vs. 38 μg/L). In the older subjects, the increase in blood aluminium levels was greater, but not statistically different, in the controls than in patients, in contrast to the younger subjects. Aluminium absorption was studied in AD subjects and compared to age-matched controls after consumption of a fruit drink containing ²⁶Al (Moore et al., 2000). The percent of aluminium absorbed in subjects, estimated from single plasma samples obtained 1 hr after the oral aluminium consumption, was 164% of that seen in controls. The authors attributed these differences to absorption, although reduced renal aluminium clearance in the aged could also contribute to this difference. The lack of consistent overall differences as a function of age or dementia status makes it difficult to draw any general conclusion from this study. Zapatero et al. (1995) found significantly higher serum aluminium concentrations in 17 AD subjects, compared to age-matched controls, but no difference between 15 subjects with senile dementias and controls. Based on greater serum and urine aluminium levels in 8 patients with dementia including the Alzheimer’s type who were 65 to 86 years old compared to 144 controls who were 30 to 65 years old (18 and 6 μg/L and 77 and 26 μg/L, respectively), Roberts et al. (1998) claimed to have confirmed earlier findings that patients with dementia appear to absorb more aluminium from the diet than healthy subjects. However, the difference in the serum and urine aluminium levels could be due to factors other than dementia, such as the significant difference in the age of the subjects. AD appears to be associated with a higher serum aluminium concentration than seen in controls. Nothing can be concluded from these observations to elucidate the mechanism of this difference.

Utilizing both ²⁷Al and ²⁶Al in separate studies, greater aluminium absorption was seen in subjects with Down’s syndrome than in controls, 0.14 vs. 0.030% for ²⁷Al administered with citrate and 0.14 vs. 0.022% for ²⁶Al administered with orange juice (Moore et al., 1997). However, these results are based on a single blood sample drawn 60 minutes after aluminium administration, a method noted above in Toxicokinetics, Absorption, Animal Studies, Oral Administration that may not very reliably estimate oral bioavailability. Down’s syndrome is associated with changes in neurodegeneration, particularly increased β-amyloid deposition, that resemble AD.

Aluminium dose

Aluminium bioavailability following consumption of a commercial aluminium hydroxide product was found to be 0.007, 0.004 and 0.001% from aluminium doses of ~ 3.5, 14 and 28 mg/kg, respectively (Weberg & Berstad, 1986). This inverse relationship between dose and oral bioavailablity is not consistent with results from animal studies suggesting increased aluminium absorption with increased dose, as discussed in Toxicokinetics, Absorption, Animal Studies, Oral Administration, Drinking Water.

Site of oral absorption

A significant increase in serum aluminium concentration was seen 30 minutes after consumption of aluminium and citrate, suggesting aluminium absorption from an upper intestinal site (Nordal et al., 1988a). Peak serum aluminium was seen 30 minutes after consumption of an aluminium antacid (Nagy & Jobst, 1994). Oral dosing of human subjects with antacids and citrate produced a peak blood aluminium level ~ 4 hr later (Weberg & Berstad, 1986) that was interpreted as support for intestinal aluminium absorption (Powell & Thompson, 1993). Blood aluminium peaked in four subjects ~ 30 to 45 minutes after they had consumed aluminium citrate (Nordal et al., 1988a; Taylor et al., 1992) and at 30 minutes in one subject after oral ingestion of an aluminium-containing antacid (Nagy & Jobst, 1994). Serum glucose also peaked 30 minutes after oral administration of 50 g of glucose (Nagy & Jobst, 1994). Hohl et al. (1994) saw maximal serum aluminium approximately 2 hr after oral aluminium intake, although the peak was not much higher than seen at 30 minutes, suggesting quite rapid absorption. As concluded from studies in animals (see Toxicokinetics, Absorption, Animal Studies, Oral Administration, The Site and Mechanisms of Oral Aluminium Absorption), aluminium seems to be primarily absorbed from the upper intestine.

Distribution (Including Compartmentalization)

Human studies

The human whole body aluminium burden has been estimated to be 80 mg (Tipton & Shafer, 1964), probably 35 to 40 mg (Alfrey, 1989) and, more recently, 30 to 50 mg (ATSDR, 1999). Slight age-related increases in blood, bone, brain, and other soft tissues have been reported. Aluminium concentrations increased from ~ 160 mg/kg (in ash) in the lung of 0 to 3 month olds, to ~ 625 in 1 to 12 year olds and to > 2000 in 19 to 89 year old adults, in liver from ~ 100 in 0 to 3 month olds, to ~ 150 in 1 to 12 year olds and to ~ 550 in adults, and in kidney from ~ 150 in 0 to 3 month olds, to ~ 300 in 1 to 12 year olds and to ~ 350 in adults (Stitch, 1957). Similarly, median lung aluminium concentrations increased from ~ 150 mg/kg (wet tissue) in 0 to 12 year olds to ~ 2500 in ≥ 60 year olds (Tipton & Shafer, 1964), and from ~ 2 mg Al/kg (wet tissue) in 21 to 30 year olds to ~ 40 in > 81 year olds (Roider & Drasch, 1999). Brain and bone aluminium increases with age, as discussed in Toxicokinetics, Distribution (Including Compartmentalization), Human Studies, Tissue Aluminium Concentrations, Brain / Bone. Kidney, liver and spleen showed a similar trend to lung and brain, an increase with age up to ~ 40 years old, a plateau or slight decrease to age 70, then an increase later in life. Ten 32 to 46 year olds had mean hippocampal and frontal cortex aluminium concentrations of 0.014 and 0.020 mg/kg (wet tissue) whereas those of fifteen 75 to 101 year olds were 0.402 mg/kg and 0.373 mg/kg, respectively (Shimizu et al., 1994). Serum aluminium levels in 356 healthy 20 to 80 year olds increased with age with a significant linear regression of r² = 0.067 and a mean of 7.3 μg Al/L (Zapatero et al., 1995). In contrast, tissue aluminium levels were not found to increase with age in 36 subjects from Denver, Colorado, U.S. or in 21 from Brisbane, Australia. However, the ages of the subjects were not reported (Alfrey et al., 1980).

In some studies increased levels of aluminium have been found in the brain of persons who had suffered from AD; in other studies no such increases were observed (see Table 19). More details of some of these studies are presented in Speziali & Orvini (2003). It has been noted that aluminium was reported to be higher in the cortex and hippocampus than in other brain structures in normal and AD brains (Gupta et al., 2005). A review was conducted of reports of aluminium in the brain of Parkinson’s disease subjects compared to controls. It revealed some studies that found a significant increase in the former group. However, all of the studies were conducted by the same research team (Speziali & Orvini, 2003).

Table 19

Comparison of aluminium in the bulk brain samples from AD subjects with those of controls. Adopted and modified from Yokel (2000).

NFT bearing neurons from AD brain showed several-fold more aluminium in the tangle than in the nucleus which, in turn, had 1.5 to 2-fold more aluminium than cytoplasm and neuropil. Aluminium distribution in the nucleus, cytoplasm and neuropil of tangle-free neurons was similar (Good et al., 1992). Similarly, another group found more aluminium in the nuclei than cytoplasm of tangle- and non-tangle-bearing neurons of Alzheimer and non-Alzheimer brains (Lovell et al., 1993).

Iron status negatively correlates with tissue aluminium accumulation. Ferritin isolated from the brains of 2 subjects with AD had more aluminium than that from 2 normal human subjects (18.9 vs. 3.4 mole Al/mole ferritin) (Fleming & Joshi, 1987). After equilibrium dialysis of ferritin isolated from human brain and liver against 20 μM aluminium, ferritin from liver had more aluminium than from brain (Fleming & Joshi, 1987). In contrast, aluminium levels were not found to be different in the brain cortex from normal, AD and chronic renal dialysis patients (6.2, 8.9 and 7.2 aluminium atoms/ferritin molecule) (Dedman et al., 1992b).

Tissue aluminium concentrations

Aluminium is unequally distributed throughout the body in normal and aluminium-intoxicated humans (Alfrey et al., 1980; Di Paolo et al., 1997). Tissue aluminium concentrations in normal adults at steady state (in mg/kg wet weight unless otherwise stated), were 20 in lung, 1 to 3 in bone (based on dry weight), 1 in liver and spleen, 0.5 in the kidney, 0.45 in the heart, 0.4 in muscle, 0.35 in brain and ~ 0.002 in blood (Nieboer et al., 1995). Similar reference values, of 2.21 to 15.3 mg/kg in the lung, 1.0 to 2.45 mg/kg in the liver and 0.55 to 1.31 mg/kg in the kidneys were obtained from a review of literature published in the prior 30 years (Caroli et al., 1994). With increased age and accumulation of aluminium, late life aluminium body burden has been estimated to be 25 to 50 mg in bone, 20 mg in lung, and 9 to 24 mg in soft tissue (Keith et al., 2002). Based on typical organ weights for a 70 kg adult, ~ 58, 26, 11, 3, 0.95, 0.3, 0.25 and 0.2% of the body burden of aluminium would be in the bone, lung, muscle, liver, brain, heart, kidney and spleen, respectively. The higher concentration in lung of normal humans may reflect entrapment of airborne aluminium particles whereas the higher concentrations in bone, liver and spleen may reflect aluminium sequestration. Skin, taken from the back of 11 chronic haemodialysis patients, had greater aluminium concentration than from 9 controls (1.02 vs. 0.26 mg/kg) (Subra et al., 1991). Skin and serum aluminium concentrations were greater in the patients who had received haemodialysis for a longer time (averages of 156 vs. 49 months) (Subra et al., 1991). It has been suggested that up to 14% of the body burden of aluminium is in the skin, but the potential contribution of contamination to this value has been raised (Priest, 2004). Because of its very low bioavailability by most routes of exposure and the effective urinary clearance of aluminium from blood, human tissue and fluid, aluminium concentrations are low compared to aluminium concentrations in most exposure sources.

Victims of the dialysis encephalopathy syndrome showed elevated aluminium exposure in all tissues (Alfrey et al., 1980). The respective mean aluminium concentrations, in mg/kg dry defatted tissue, in controls, non-dialyzed uraemic patients, dialyzed uraemic patients and dialyzed uraemic patients with encephalopathy were 4.1, 25.5, 161 and 301 in liver; 3.0, 35.3, 243 and 493 in spleen; 4.5, 27.4, 116 and 281 in bone; 1.1, 6.9, 22.5 and 42.7 in heart; 1.2, 2.6, 9.1 and 14.9 in muscle; 55, 75, 90 and 215 in lung; and 2.4, 4.1, 8.5 and 24 in brain grey matter. In a more recent episode of aluminium intoxication in dialysis patients, 10 of 17 patients exposed to dialysate prepared from tap water containing 650 μg Al/L died. The serum aluminium concentrations of those that died and those that did not averaged 808 and 255 μg Al/L, respectively. Tissue aluminium concentrations in 4 of the victims were above reference values: liver 4.7 to 51.7 (reference < 2), bone 7.5 to 88.7 (reference < 2) and cerebral cortex 1.09 to 1.78 (reference 0.14 to 0.22) mg Al/kg.

Two male twins, 32 weeks at birth, received 2500 and 5350 mg total aluminium orally as antacids and 2.96 and 3.2 mg in total parenteral nutrition (TPN) solutions over 5 months. When they died at age 149 and 157 days, their bone, brain, liver, and lung, and in one of the two, kidney, aluminium concentrations were greater than the average values from infants that had not received i.v. fluids or TPN solution for more than 21 days (Bozynski et al., 1989). Considering oral aluminium bioavailability of 0.1%, ~ ½ to ⅔of the aluminium that reached systemic circulation was from the antacids. They received comparable amounts of aluminium from the TPN solution but the twin that received ~ 2-fold more from antacids had higher aluminium concentrations in all organs but the brain.

Aluminium, as the phosphate, was shown to be elevated ~ 10-fold in the synovial fluid, synovial membrane and articular cartilage of 28 chronic haemodialysis patients, some of whom took 5 to 15 g of aluminium hydroxide daily, compared to patients who had not taken aluminium (Netter et al., 1984; 1988). Speciation calculations indicate that a major aluminium species in synovial fluid is the citrate (Silwood & Grootveld, 2005).

Lung, liver and brain aluminium concentrations were greatly elevated (~ 20-, 120- and 20-fold) in a worker who developed fibrosis and encephalopathy after exposure to aluminium powder for 13.5 years (McLaughlin et al., 1962). Aluminium levels were elevated in the lung, liver, spleen and hilar lymph node of a stone mason (Teraoka, 1981).

Gastric mucosal membrane aluminium concentration was elevated in aluminium-consuming patients with normal renal function and those with renal insufficiency who were predialysis and those receiving dialysis, compared to patients with normal renal function who were not consuming aluminium-containing antacids (Zumkley et al., 1984a). The increases paralleled increases in plasma aluminium, and were attributed to consumption of aluminium antacids.

In a case of Wilson disease, increased urinary clearance of aluminium as well as copper and increased aluminium content of the liver was observed (Yasui et al., 1979).

The observations of increased serum aluminium during acute infection and concurrent neurological dysfunction and some deaths suggest infection in humans who have a significant body burden of aluminium causes aluminium release from its storage sites (Davenport et al., 1988; Fenwick et al., 2005).

Brain

A review of the literature on the reports of aluminium concentrations in various brain regions led the authors to conclude that aluminium is generally higher in grey than white matter (Speziali & Orvini, 2003). Human brain aluminium concentration correlated positively with age in several studies. Brain aluminium concentrations in 3 subjects aged 25, 43 and 65 averaged 1.59 mg/kg (dry tissue) and 2.74 mg/kg in 6 subjects aged from 75 to 99 years (McDermott et al., 1979). Brain aluminium levels in 7 subjects from premature to 6 months old were ~ 0.3 mg/kg (wet tissue), increasing as the age of subjects increased, to ~ 0.7 mg/kg in 4 subjects 80 to 99 years old (Markesbery et al., 1984). Similarly, brain aluminium levels increased from ~ 0.2 mg Al/kg (wet tissue) in 21 to 30 year olds to ~ 0.55 mg/kg in > 81 year olds (Roider & Drasch, 1999). Ten adults 32 to 46 years old were compared to fifteen adults 75 to 101 year old. The aluminium concentrations in the hippocampus and frontal lobe were 0.014 and 0.020 vs. 0.40 and 0.37 mg Al/kg, wet weight, respectively (Shimizu et al., 1994). The increase in brain aluminium concentrations with age could be due to increased exposure with age, a decreased ability to remove aluminium from the brain with age, or very slow, or no, elimination of aluminium from the brain. Patients who had elevated aluminium intake and who received haemodialysis before successful renal transplantation up to 8 years prior to death had elevated post-mortem aluminium levels in the brain (McDermott et al., 1978; Reusche et al., 1996), suggesting accumulation of aluminium during haemodialysis that was slowly, or not, cleared after establishment of renal function. Using morin, aluminium was visualized in the disintegrating NFT and senile plaque amyloid core of non-demented elderly subjects. In normal brain tissue, aluminium was seen in the wall of the capillary vessels of the BBB, perivascular glial supporting tissues, nuclei of astrocytes, and nuclei and nucleoli of neurons (Shimizu et al., 1994). However a study of only 4 subjects who died between 58 to 74 years of age and who had a mean brain aluminium concentration of 0.54 mg/kg wet weight failed to find an association between age and brain aluminium concentration (Jacobs et al., 1989). The aluminium concentration in 12 brain regions obtained from 8 neurologically normal subjects was reported to range from a mean of 58 mg Al/kg, wet weight, in the pons to 196 mg Al/kg in temporal cerebrum (Rajan et al., 1997). These very much higher values do not appear to be correct, perhaps due to contamination or incorrect reporting of the units.

Neuromelanin from human brain was reported to contain 220 mg Al/kg, dry weight (Zecca et al., 1994).

Patients who received brain surgery for a tumour after consuming an aluminium-rich antacid for 10 days had 2-fold higher brain aluminium concentration than patients consuming an antacid low in aluminium (Winterberg et al., 1987a), demonstrating aluminium absorption and distribution into the human brain in the presence of normal renal function.

A case of generalized amyloidosis showed greater brain aluminium accumulation than seen in the control or parkinsonism groups, 89.5 mg/kg vs. 17.7 mg/kg and 31.8 mg/kg, respectively (Yasui et al., 1980).

The CNS shows lower aluminium concentrations than many other tissues, even in the presence of overt neurotoxicity. Increased brain aluminium concentrations of ~ 4- to 6-fold in rabbits and somewhat higher increases in victims of dialysis encephalopathy syndrome were associated with neurotoxicity (Alfrey, 1980; Crapper et al., 1976; Yokel, 1983). Results of the many studies of aluminium concentration in bulk brain samples and sub-cellular sites of AD victims inconsistently show elevated aluminium (see Tables 19, ​,2020 and ​and21).21). The magnitude of elevation of bulk brain aluminium, when reported, is generally a few-fold or less in AD victims than in controls. The inconsistent findings of a small increase in brain aluminium in AD in relation to the small increase in bulk brain aluminium sufficient to produce neurotoxicity has hindered the resolution of the role of aluminium in AD. If aluminium is elevated in AD brain, it is not reflected in CSF aluminium, which has generally not been found to be elevated (Jagannatha Rao et al., 1999; Kapaki et al., 1993). CSF aluminium concentration was reported to be higher in patients with virus nephro infections, residual manifestations from brain and spinal cord trauma, and pain syndromes, than in controls. Blood serum aluminium concentrations were also reported to be elevated in those with residual effects of brain and spinal cord trauma (0.014 and 0.012% aluminium in the ash vs. 0.0088% in controls) (Del’va, 1962).

Table 20

Comparison of aluminium in NFTs from AD subjects with those in controls, determined by microprobe analysis or staining. Adopted and modified from Yokel (2000).

Table 21

Determination of aluminium in plaques by microprobe methods or staining. Adopted and modified from Yokel (2000).

The indigenous people of several western Pacific foci, the Chamorro on Guam, Japanese on the Kii peninsula of Honshu Island, Japan and the Auyu and Jakai of southern West New Guinea, developed two syndromes having features of ALS and a parkinsonism-dementia (PD). From the time this variant of ALS was first identified in 1945 until 1960, its incidence was 50 to 150 times higher than elsewhere in the world (Kihira et al., 2004). It was suggested that a high aluminium and low calcium and magnesium concentration in the environment contributed to these syndromes (Yase, 1972). Although the manganese concentrations in the soil, river and drinking water were greater in ALS than control areas, the manganese content of plants, crops, livestock and fish was not (Iwata et al., 1976). A specific localization of manganese, aluminium and calcium was observed in the spinal cord of ALS patients (Iwata et al., 1976). Food was not found to be high in aluminium and low in calcium, but the soil was high in aluminium (McLachlan et al., 1989). The aluminium, and calcium, concentration in the brain of victims of ALS has been reported to be greater than in controls, whereas magnesium was not elevated, using neutron activation analysis (Yoshimasu et al., 1976; 1980). Brain aluminium concentrations averaged 33.1 mg/kg in 6 ALS cases and 36.8 mg/kg in 4 PD cases compared to 17.7 mg/kg in controls, determined by neutron activation analysis. Aluminium in the ALS and PD groups was statistically greater than in the controls (Yase, 1980). Calcium was also elevated in the ALS and PD subjects. X-ray microanalysis showed similar calcium, aluminium and manganese distribution in spinal cord of ALS patients (Yase, 1980). Using EAAS, Traub et al. (1981) found elevated brain aluminium levels in two Guamian ALS cases to be 1.7 and 8.9 and in two Guamian PD cases to be 2.0 and 3.9 mg/kg, compared to an average of 1.38 mg/kg in 4 normal subjects. Aluminium, silicon, calcium, vanadium, iron and zinc were reported to be elevated in the frontal cortex of humans with ALS compared to normal subjects and patients with parkinsonism (Mizumoto et al., 1983).

Aluminium was significantly increased in 26 CNS regions in 2 of 6 patients with ALS compared to those in 5 patients without neurological abnormalities. Mean concentrations were 88 and 136 in the two cases vs. 26 and 23 mg Al/kg dry weight in the other 4 cases and controls, respectively (Yasui et al., 1991b; 1991c).

Using SEM with energy dispersive spectrometry, NFT-bearing neurons from ALS-PD and non-afflicted patients were found to have a high aluminium concentration (Perl et al., 1982). Aluminium and calcium were co-localized in the NFT-bearing neurons. Using wavelength dispersive spectrometry coupled with electron beam X-ray microprobe analysis, aluminium and calcium were found to be co-localized in the NFTs of two Guamian PD patients but not in the non-NFT-containing regions of either the PD patients or two control lifelong Guamian residents (Garruto et al., 1984). Semi-quantitative estimates of the highest concentrations were 7200 and 500 mg/kg calcium and aluminium, dry weight, respectively. Garruto & Yase (1986) compared silicon distribution in 5 Guamian Chamorros who had PD and in 2 who had ALS with 2 Guamian and 2 Caucasian normal controls and found a similar distribution, estimated to be up to 3000 mg/kg, with no detectable silicon in the controls. The average brain aluminium concentration was higher in 6 Guam PD cases than 7 Chamorro controls (179 vs. 57 mg/kg dry weight) (Yoshimasu et al., 1985). Aluminium and calcium were seen in the cytoplasm of hippocampal neurons bearing NFTs, using laser microprobe mass spectroscopy (Perl & Pendlebury, 1986). Aluminium and calcium were found to be associated with NFT-bearing hippocampal neurons of PD patients, using secondary ion mass spectrometry (Linton et al., 1987). Using histochemical staining, aluminium was visualized in the hippocampus, spinal cord and frontal cortex in most of 3 Guamian patients with ALS and 5 with PD who had NFTs but not in the 5 neurologically and neuropathologically normal Guamian or Caucasian patients (Piccardo et al., 1988). Staining was observed in the cytoplasm, nucleoli, neuropil, white matter and some endothelial cells and walls of cerebral vessels. X-ray microanalysis confirmed the presence of aluminium.

Using neutron activation analysis, aluminium levels were reported to be higher in 3 Guamian cases than in 4 non-demented controls. The Guamian cases also had high calcium levels in grey and white matter and low zinc levels in grey matter (Yoshida et al., 1993). Using PIXE, extremely high aluminium concentrations were reported in lumbar spinal cord and hippocampus of patients with ALS from Guam and the Kii peninsula of Japan, compared with those in cases with sporadic ALS and in controls (Yoshida et al., 1997; Yoshida, 1999). Aluminium concentrations negatively correlated with calcium and magnesium contents in the birthplace area’s rivers (Yoshida et al., 1997) and positively correlated with iron and copper, and negatively correlated with zinc, in the neural tissue (Yoshida, 1999).

Toenail aluminium concentrations, measured as an indicator of metal exposure, were not different between 22 patients with ALS and 40 controls. Median values were 34.5 and 37.5 mg Al/kg, respectively (Bergomi et al., 2002).

Aluminium levels in 4 cases of Parkinson’s disease were compared to those in the 5 patients without neurological abnormalities in the above cited studies and were found to be significantly higher in the hippocampal gyrus, caudate nucleus, globus pallidus and substantia nigra as well as in the liver, kidney and spleen (Yasui et al., 1992). Magnesium, but not calcium, levels were significantly decreased in these same brain regions, and others as well. However, aluminium was not significantly different in the frontal cortex, caudate nucleus, substantia nigra, and cerebellum of 9 Parkinson’s disease patients or 15 patients with other chronic neurological diseases compared to 12 controls (Uitti et al., 1989).

Bone

Aluminium concentrations, on a dry weight basis, in the bone of normal humans were a few-fold higher than those in the brain (Alfrey et al., 1980; Di Paolo et al., 1997), ~ 1 to 3 mg/kg (Nieboer et al., 1995). Human bone aluminium concentration significantly increased with age in a study of one hundred seventy-two 16-98 year old subjects. It was <0.4 μg/gm dry bone weight in quartile 1 compared to >1.7 in quartile 4 (Hellström et al., 2005). Aluminium contamination in parenteral nutrition solutions in patients with normal renal function led to stainable bone aluminium and osteomalacia (Ott et al., 1983). Oral consumption of aluminium in drinking water and aluminium carbonate by people with normal renal function led to detectible bone levels of aluminium (Eastwood et al., 1990; Recker et al., 1977). Bone aluminium concentrations averaged 6.4 mg/kg dry bone weight in 3 non-dialysis and 125 mg/kg in 3 dialysis patients (Recker et al., 1977) and 5.6 and 102 mg/kg fresh weight in 2 controls and 4 patients with dialysis osteomalacia (Zumkley et al., 1984b). Consumption of a total of ~ 18 kg of elemental aluminium in antacids by a 39 year old female with normal renal function over 8 years led to stainable aluminium on 27.6% of the bone surface, elevated serum and urinary aluminium, and phosphate-depletion-induced osteomalacia that was attributed to aluminium (Woodson, 1998).

Dialysis patients had an average increase of brain and bone aluminium concentrations of 5 and 10-fold, respectively (Alfrey et al., 1980; Di Paolo et al., 1997). Those with dialysis encephalopathy had brain and bone aluminium concentrations about 10- and 85-fold higher than those of controls, respectively (Alfrey et al., 1980). Considerably more aluminium enters, and resides in, the human skeletal system than in the CNS, due to the larger mass of the former. Aluminium was found to be localized at the mineralization front and osteoid of bone (Boyce et al., 1981; Cournot-Witmer et al., 1981; Ott et al., 1982; Schmidt et al., 1984). The aluminium was within 2μm of the calcified bone/osteoid interface at a concentration 20 to 40-fold greater than the whole bone aluminium concentration (Boyce et al., 1981).

Six patients who received aluminium-contaminated TPN solutions for 6 to 72 months had bone aluminium concentrations of 14 to 265 mg/kg, well above the normal range, as well as elevated plasma aluminium (98 to 214 μg/l) and urinary aluminium outputs (Klein et al., 1982). Bone aluminium concentration averaged 2 mg/kg dry weight in infants who received limited i.v. therapy compared to 20 in infants who received ≥ three weeks of i.v. therapy (Sedman et al., 1985).

As bone is a major site of aluminium storage, prolonged urinary aluminium excretion may reflect a prolonged t_½ of aluminium in bone. A t_½ of 7 years was estimated in one human who had received an i.v. injection of ²⁶Al citrate 3.2 years earlier (Priest et al., 1995). An updated estimate in this individual, based on whole-body monitoring collected up to 3000 days after the injection, suggests the t_½ is ~ 50 years (Priest, 2004). This prolonged whole-body t_½ may largely reflect the t_½ of aluminium in bone.

Clinical evidence suggested that PTH had a protective effect against aluminium-induced bone disease, and encephalopathy (Cannata et al., 1988). Bone aluminium did not correlate with PTH levels in uraemic patients (Alfrey et al., 1979). Subtotal PTX of 10 dialysis patients who had refractory secondary hyperparathyroidism resulted in a significant increase of bone aluminium in 6 of the 7 consuming aluminium (De Vernejoul et al., 1985). These results are not consistent with observations in animals (see Toxicokinetics, Distribution (Including Compartmentalization), Animal Studies).

Aluminium-containing implants have been implicated as a cause of encephalopathy, associated with elevated CSF, serum and urine aluminium concentrations (Hantson et al., 1994; Renard et a., 1994). Using PIXE, it was shown that aluminium uniformly leaked from the site of an implanted aluminium-containing alloy into surrounding bone (Passi et al., 2002).

It has been suggested that the accumulation of aluminium in bone and liver protects patients from the toxic effect of aluminium in other organs (Berend et al., 2001).

Hair

Hair has been used to estimate the body burden and to indicate excessive exposure to metals since the 1960s (Villain et al., 2004). For example, hair has been shown to reflect methylmercury exposure (Johnsson et al., 2005) and to have a 1:270 ratio with blood mercury (IPCS, 1990). The validity of hair to predict the aluminium body burden has not been well established. There can be additional problems with the analysis of hair for metals; procedures for collection have not been standardized. Although hair has been shown in some studies to correlate with other indicators of body burden, it is seldom the preferred tissue for this purpose; commercial panels that test hair for multiple metals have questionable validity (Kales & Goldman, 2002; Villain et al., 2004).

Aluminium was shown to be taken up into human hair from aqueous solution (Wilhelm et al., 1989a). Normal hair aluminium concentration was reported to be < 0.24 to 67 mg/kg in 194 people (Imahori et al., 1979); 6.5 mg/kg in English samples (Alder et al., 1976); 3.7 for US, 8.8 mg/kg for rural Japan, 11.9 mg/kg for Hong Kong and 13.6 mg/kg for samples from Tokyo (IAEA, 1978) and 15 to 18 mg/kg in samples obtained in Kentucky, US, depending on the method used to wash the samples to remove surface contamination (Yokel, 1982). A reference value for aluminium in hair of 0.1 to 36 mg/kg was derived from a literature review of studies published in the prior 30 years (Caroli et al., 1994).There are a few cases reporting elevated hair aluminium in children with emotional problems (Rees, 1979), but the results could be due to contamination. Unexpectedly high aluminium concentrations were observed in the hair of patients with neurological and other disorders, which was thought to be due to dolomite in many of the cases (Roberts, 1981).

The aluminium concentration in hair of 6 control subjects was reported to be 97 ± 25 mg/kg, 118 ± 46 mg/kg in 11 haemodialysis patients and 370 ± 266 in 8 non-dialyzed chronic renal failure patients (Tsukamoto et al., 1979). Hair aluminium concentration positively correlated with AUC and duration of dialysis. Similarly, hair aluminium concentrations were higher in non-dialyzed and dialyzed chronic renal failure patients than in controls, whereas haemofiltered chronic renal failure patients did not have higher concentrations of hair aluminium (Marumo et al., 1984). The elevated aluminium concentrations were attributed to the use of aluminium-contaminated dialysate. Hair aluminium was 10.5 mg/kg in 22 male and 10.1 in 29 female long-term haemodialysis patients who had elevated plasma and bone aluminium concentrations (Winterberg et al., 1987b). There was a significant correlation between hair and bone aluminium in the males, but not the females. Hair aluminium in 18 chronic haemodialysis patients who received an average of 6.3 kg of an aluminium-containing phosphate binder that released aluminium averaged 15.4 mg/kg compared to a hair concentration of 10.5 mg Al/kg in a group of 18 chronic haemodialysis patients who received an average of 6.6 kg of an aluminium-containing phosphate binder that released less aluminium (Winterberg et al., 1987b). The latter group of patients had lower aluminium in their plasma (34.6 vs.101.4 μg/L) and bone (11.6 vs. 26.2 mg/kg). Higher hair aluminium concentration was seen in 39 haemodialyzed patients (6.1 ± 2.8 mg/kg), than in 49 control subjects (3.4 ± 1.6) mg/kg (Chappuis et al., 1988). Hair aluminium did not correlate with serum or bone aluminium, leading the authors to conclude that hair aluminium levels do not predict aluminium-induced osteomalacia (Chappuis et al., 1989). The aluminium hair concentration of 12 home haemodialysis patients tested before introduction of water treatment by reverse osmosis was above that in controls, whereas it was not elevated in 16 patients undergoing continuous ambulatory peritoneal dialysis (Wilhelm et al., 1989b). Because hair aluminium concentration did not relate to daily or cumulative aluminium intake or to bone or plasma aluminium concentrations, the authors concluded that hair aluminium is of very limited value for diagnosis of aluminium exposure.

The concentration of aluminium in the hair of 10 AD patients averaged 7.5 mg/kg compared to 6.2 in 10 age-matched controls (Shore & Wyatt, 1983). In another study, hair aluminium concentration in 35 cases of AD was significantly lower than in 71 comparably-aged controls (Kobayashi et al., 1989). There was no significant effect of age on hair aluminium levels in either group (Kobayashi et al., 1989).

Milk

A literature review of studies published the previous 30 years suggested the reference value was 39 to 250 μg/L (Caroli et al., 1994). Another review concluded the value was 4 to 65 μg Al/L (American Academy of Pediatrics, 1996). Other studies reported means of 30 μg/L (Weintraub et al., 1986), < 5 to 45 (median 14) μg/L (Koo et al., 1988), 49 μg/L (Simmer et al., 1990), 3 to 79 μg/L (Baxter et al., 1991), a median of 161 μg/kg (Coni et al., 1990), a mean of 40 μg/L (Bougle et al., 1992), 9.2 μg/L (Hawkins et al., 1994), a mean of 86 μg/L (Vinas et al., 1997) and a median of 67 μg/L (Krachler et al., 2000). One report found a mean of 380 μg Al/L (Mandic et al., 1995) and another a mean of 350 μg Al/L (Bergerioux & Boivert, 1979); these two values are suspect compared to the others. Speciation calculations suggest ~ 88% of aluminium in human (and bovine) milk at pH 6.8 is Al(citrate)(OH)₂^-2 and 11% is Al(citrate)(OH)^-1, suggesting to the authors that little would be absorbed due to the net charge (Findlow et al., 1990). However, this assumes no change in speciation in the GI tract.

Elimination and Excretion

Biological Indices of Exposure, Body Burden and Organ Concentration

Human tissue and fluid aluminium concentrations are low compared to those in most exposure sources due to its very low bioavailability by most routes and effective urinary clearance.

Biological monitoring of human exposure to aluminium has been conducted with urine, which has been thought to indicate recent exposure, or serum, which has been thought to better reflect the aluminium body burden and long-term exposure (Alessio et al., 1989; Apostoli et al., 1992; Nieboer et al., 1995). However, neither is a very good predictor of the aluminium body burden.

As noted in Toxicokinetics, Distribution (Including Compartmentalization), Human Studies, Plasma Aluminium Concentrations, the serum aluminium level in normal humans has been reported to be ~ 1 to 2 μg/L. In dialysis patients, it was noted that > 30 μg Al/L serum has been associated with osteomalacia and related disorders (Nieboer et al., 1995). This conclusion was based on serum or plasma aluminium levels of 30 to 575 μg Al/L in 11 patients with osteomalacia and 27 to 160 μg Al/L in 11 patients with osteitis fibrosa (Cournot-Witmer et al., 1981), 65 to 360 μg Al/L in 18 patients with osteomalacia (Parkinson et al., 1981), 22 ± 12 μg Al/L in 18 patients with renal osteodystrophy (Gokal et al., 1983), 224 ± 22 μg Al/L in 40 patients with aluminium-related bone disease and 75 ± 13 μg Al/L in 21 patients with osteitis fibrosa (Norris et al., 1985) and 138 to 497 μg Al/L in 7 patients with moderately severe osteomalacia (Smith et al., 1987). It was noted in dialysis patients that concentrations > 80 μg Al/L serum have been associated with encephalopathy (Nieboer et al., 1995). This conclusion was based on serum aluminium concentrations of 89 to 486 μg Al/L in 17 patients (Parkinson et al., 1981) and 175 to 700 μg Al/L in 10 patients (McKinney et al., 1982). Ten dialysis patients who died with acute encephalopathy associated with aluminium contamination of the dialysate had a mean serum aluminium concentration of 808 μg/L. Seventeen patients who were similarly exposed but did not develop encephalopathy or die had a mean of 225 μg Al/L (Berend et al., 2001).

It has been suggested that the serum aluminium concentration should be kept below 30 μg/L (Razniewska, 2005). Psychomotor function in long-term dialysis patients whose mean serum aluminium was 59 μg Al/L was significantly impaired compared to that in controls (Altmann et al., 1989). Others suggested that a level of 40 to 50 μg Al/L warrants discontinuation of aluminium gels, 60 μg Al/L might indicate increased body burden, and > 100 μg Al/L indicates potential encephalopathy and the need for increased monitoring in dialysis patients. An AUC > 100 or 150 μg/L could present a risk of aluminium toxicity in these patients (Alfrey, 1986) and overt aluminium toxicity has been seen when the serum aluminium concentration was > 200 μg/L (Berend et al., 2002; Spencer, 2000). Mild neurophysiological and neuropsychological adverse effects were seen in MIG welders. The body burden threshold associated with these detrimental effects was estimated to be ~ 4 to 6 μM (108 to 162 μg/L) in urine and 0.25 to 0.35 μM (7 to 9 μg/L) in serum (Riihimaki et al., 2000).

Although the serum aluminium level increases with increased aluminium body burden, it does not directly reflect the aluminium body burden from long-term exposure. This is better measured by bone aluminium, the DFO challenge test, which has a high rate of false negative results, or the combined measurements of serum iPTH and the DFO test (Cannata-Andia & Fernadez-Martin, 2002; Mazzaferro et al., 1992; Pei et al., 1992). The peak and increment in serum aluminium after the DFO test in 28 chronic dialysis patients correlated significantly with bone aluminium (De Vernejoul et al., 1989) and in 11 other chronic dialysis patients correlated significantly with skin aluminium (Subra et al., 1991), suggesting utility of this test to indicate the aluminium body burden. Five mg/kg of DFO or less once or twice weekly has been shown to be safe and effective for long-term treatment of aluminium overload, and is as effective when given 1 hr before as after dialysis (Barata et al., 1996; Berend et al., 2002; Cannata-Andia & Fernadez-Martin, 2002). The serum aluminium increase in response to mobilization of aluminium from storage sites, such as produced by DFO, is used to test for aluminium body burden. A positive correlation was found between the oral dose of aluminium in a phosphate binder and bone aluminium content, but not with serum aluminium during the 6 months before the determination of aluminium in bone (Channon et al., 1988).

Conventional x-rays have not been shown useful to reveal aluminium-induced pneumoconiosis (Kraus et al., 2000; Letzel et al., 1996; Saia et al., 1981). High-resolution computed tomography was shown to reveal aluminium-induced lung fibrosis changes (Kraus et al., 2000).

A method using in vivo neutron activation analysis to measure aluminium in the bones of the hand has been developed and is ready for pilot studies in human subjects (Pejovi-Mili et al., 2005).

Intratracheal/intrapleural exposure

Intratracheal instillation of aluminium compounds in laboratory animals has been used as a simple and relatively inexpensive method for screening aluminium for fibrogenicity or other types of pulmonary toxicity, including carcinogenesis. Experimental studies to evaluate fibrogenic potential in rats i.e., the ability to induce pulmonary fibrosis of 4 different types of aluminium fibres including alpha (uncalcined form) and gamma alumina (calcined form) following single intratracheal injections, were performed by Dalbey & Pulkowski (2000). Six months after dosing, pulmonary function tests (functional residual capacity, deflation pressure-volume curves, maximal forced deflation, single breath carbon monoxide diffusion capacity, and pulmonary resistance) and histopathological evaluation were performed. Rales were noted during the first week of instillation in aluminium treated groups, but not in the groups given glass beads or quartz. Standard lung volume and maximal forced exhalation parameters were decreased at 6 months after instillation in aluminium treated groups as compared to animals injected with saline and glass beads (controls). Single-breath carbon monoxide diffusing capacity was significantly decreased in aluminium treated animals compared to both types of controls which indicated the presence of a physical barrier between the air in the alveoli and the blood. The weight of the postcaval lung lobe was significantly increased for all groups administered aluminas, and the most marked increase was seen in the quartz group. The histopathological changes were similar in all treated groups and consisted of areas of granulomatous inflammation with early collagenization (fibrosis). The presence of multinuclear giant cells and the infiltration of macrophages were suggestive of a foreign body type reaction. Interstitial fibrosis was also apparent and was characterized by a thickening of alveolar walls with collagen. Both groups treated with uncalcined aluminas tended to have a higher incidence and severity of granulomas with fibrosis. Although minor pulmonary changes were noted in the aluminium treated groups, these effects were significantly less pronounced than the changes induced by the instillation of the positive control (quartz). Results are consistent with previously published work (Ess et al., 1993; King et al., 1955; Stacy et al., 1959) and point to the variation in responses to material within the class of alumina compounds. In interpreting these results, it must be considered that large doses were instilled with the intent of overloading normal clearance mechanisms in the lung to exaggerate any reaction that might occur. The dose of 50 mg is equivalent to about 30 mg/g lung, well above the 1 mg/g generally associated with the onset of overloading during long-term studies (Oberdorster et al., 1992). Influx of alveolar macrophages (AM), accumulation of particles, inflammation, and fibrosis are changes which would be expected following the administration of a large dose of relatively insoluble particles producing low toxicity to rats. The main goal of these instillations was to rank several alumina samples for their general potential to induce pulmonary fibrosis.

Lindenschimdt et al. (1990) examined the effects of aluminium on the development of pulmonary fibrosis and histological changes/inflammatory responses in the lungs of rats instilled with 1 or 5 mg Al₂O₃/100 g body weight. A dose-dependent minimal and generally transient increase in inflammatory responses was measured in the bronchoalveolar lavage fluid (BALF) including; activity of lactate dehydrogenase (LDH) an index of cell membrane damage; beta-glucurnidase and N-acetylglucosaminidase, markers of macrophage/polymorphonuclear membrane damage; and levels of total protein, an index of potential fibrotic activity and/or vascular damage. Increase in total cells at this dose was primarily due to elevation in neutrophils and lymphocytes. At low dose, the only significant change was an increase of neutrophils on day 1 which returned to the control level by day 7. The changes observed at high doses returned slowly to normal values during the 2-month study period. Although intratracheal instillation is not the normal route of exposure, the minimal and generally transient changes induced by Al₂0₃ are consistent with the lack of significant lung toxicity found in both humans and animals. Significant pathologic response at high doses might be due to the overload phenomenon of aluminium oxide dust (~9.1 mg/g lung tissue). Morrow (1996) showed that deposition of large amounts of inert dust in the lungs (> 1-2 mg/g lung tissue) resulted in inhibition of phagocytic removal of dust, leading to a delayed clearance from the lung.

Tornling et al. (1993) administered intratracheal instillations of aluminium oxide (primary alumina), aluminium oxide with adsorbed fluorides (secondary alumina), and saline to three different groups of rats. The alumina dust (40 mg) was suspended in saline. BALF was obtained and histological examination of the lungs was performed 1, 4, and 12 months after exposure. No signs of fibrosis were found in any of the animals. No significant changes in alveolar cell concentrations were noted for the group treated with primary aluminium; however the secondary aluminium group exhibited increased concentrations of macrophages and neutrophils one month and one year after exposure. This suggests that fluoride plays an important role in early changes to alveolar cell populations. One year after exposure both the aluminium treated groups exhibited significantly raised concentrations of fibronectin, which indicates that alumina, not fluoride, is essential for this observed effect. The biochemical properties of fibronectin support the formation of an extracellular matrix network, and therefore fibronectin may be an early marker of fibrosis. Due to the administration of aluminium by intratracheal instillation, which is not a physiological route, the results observed in this study need to be confirmed by further investigations in which inhalation is used as a route of exposure. Instillation may have led to pulmonary overload which could have contributed to the development of the observed effects.

Pigott & Ishmael (1992) assessed the effects of a single intrapleural injection (0.2 mL suspension/20 mg suspended solids) of refractory alumina fibres (Saffil fibres) obtained immediately after manufacturing or, later, after extensive thermal ageing. The potential for these fibres, which had different diameters, to result in the development of mesotheliomas in groups of rats was examined. No mesothelioma was detected in any of the rats dosed with the Saffil fibres, or in the negative controls. Malignant mesothelioma was diagnosed in 7 rats in the asbestos group (positive control) and in 3 rats in one of the aluminosilicate groups. However, it must be considered that intra-cavity injections result in a high deposition of the test material directly on the target tissue. This does not reflect inhalation exposures in which the fibres must first be deposited in the alveolar region of the lung and penetrate lung tissue before it reaches the pleural space. The increased mesothelioma proliferation and malignant mesotheliomas detected in the aluminosilicate B group as compared to the aluminosilicate A group is likely a reflection of the size of the fibres. Coarse fibres were found to be more irritating than fine fibres. The results of this study suggest that Saffil alumina fibres are inert and are not associated with mesothelioma induction. This study supports the results from a previously conducted inhalation study (Pigott et al., 1981).

After a single intrapleural instillation of Al(OH)₃ at a dose of 0.3 g/mL saline to rats there was an increase in chest wall elastic properties and viscoelastic pressure accompanied by pleural inflammation after 7 days (Albuquerque et al., 2002). The pleural adherence was associated with a marked increase in the type I/type III collagen ratio after 30 days. Histological examination demonstrated no significant differences in lung parenchyma in the aluminium hydroxide treated and control groups.

Oral exposure

As discussed in Toxicokinetics, Absorption, Animal Studies, Oral Administration, the oral bioavailability of silicon and aluminium from Zeolite A (30 mg/kg), sodium aluminosilicate (16 mg/kg), magnesium trisilicate (20 mg/kg), and aluminium hydroxide (675 mg) in dogs was examined by Cefali et al. (1995). Twelve female dogs received a single oral dose of each compound at one-week intervals. One of the 12 dogs receiving aluminium hydroxide displayed frothy emesis, and two dogs excreted soft stool.

Intracellular binding of aluminium was examined in the mucosa of the stomach, duodenum, jejunum and ileum of adult rats following a single oral administration (300 mg/kg) of aluminium hydroxide (More et al., 1992). A second group of rats received a daily oral administration of Al(OH)₃ (300 mg/kg) for 5 days. No marked differences in body weight and no microscopic lesions in the GI tract (body and antrum of the stomach, duodenum, jejunum and ileum) were observed in either group of animals. Six hr after single Al(OH)₃ administration, aluminium deposits were observed in the gastric lumen, in the duodenum, and in the lumen of both the jejunum and ileum. After repeated administration, the presence of aluminium-reactive deposits was noted only in the lumen of the stomach (at the bottom of the antral glands) and in the lumen of the intestine from day 3 to day 7. Other data (discussed in Toxicokinetics, Absorption, Animal Studies, Oral Administration, The Site and Mechanisms of Oral Aluminium Absorption) have demonstrated that aluminium absorption occurs in the small intestine by a paracellular pathway process via the tight junctions (Garbossa et al., 1998b; Provan & Yokel 1988a). The results also suggest that after repeated administration of large oral doses, aluminium accumulates in the antral mucosa of the stomach and is released slowly in the digestive tract.

Injection

Aluminium hydroxide is one of the adjuvants most commonly used in routine human vaccines against hepatitis B virus (HBV), hepatitis A virus, and tetanus toxoid (TT) and in veterinary vaccines (see also Effects on Humans, Effects from Non-Occupational Exposure, Irritation, Irritation after Injection of Aluminium-Adsorbed Proteins (Vaccines and Hyposensitization Regimens)). Although it has been investigated since 1926, the mechanisms of action of aluminium adjuvants are not yet fully understood. It is likely that aluminium adjuvants induce immune activation which includes interleukin (IL) -1 production by monocytes, induction of eosinophilia, compliment activation and increased specific and non-specific immunoglobulin (Ig) G1 and IgE antibody responses (Gupta & Siber, 1995; HogenEsch, 2002; Jensen & Koch, 1988; Larsen et al., 2002; Norimatsu et al., 1995; Shi et al., 2001). Limitations of aluminium adjuvants for human vaccination include local reactions, augmentation of IgE antibody responses, ineffectiveness against some antigens and inability to augment cell-mediated immune responses, especially cytotoxic T-cell responses (Gupta, 1998).

Gherardi et al. (2001) administered a single i.m. injection of an aluminium hydroxide-containing HBV vaccine (GenHevac, 250 μL) to rats in an attempt to reproduce lesions characteristic of MMF (also see Effects on Humans, Case Reports). The aluminium hydroxide-containing vaccine induced a large necrotic area containing damaged muscle fibres and neutrophils, surrounded by abundant lymphocytes and macrophages (days 7-15), that progressed to a mature lesion (21 and 28 days). The focal infiltration of densely packed PAS-positive macrophages, without giant cell formation or muscle fibre damage, was similar to the macrophage infiltrate seen in MMF. Crystalline inclusions similar to those of MMF were detected by electron microscopy. It was proposed that aluminium hydroxide forms a deposit which damages the injected tissue, subsequently eliciting a signal from stressed cells. This signal attracts inflammatory and antigen presenting cells and the aluminium hydroxide deposit is then subject to phagocytosis (Balouet et al., 1997; IPCS, 1997; Schijns, 2000). Phagocytized aluminium hydroxide increases survival of macrophages and enhances the effects of granulocyte/monocyte stimulating factor (Hamilton et al., 2000). A number of aluminium loaded macrophages accumulate locally, resulting in the characteristic granuloma formation, while others migrate to the regional lymph nodes (IPCS, 1997). A recent study in monkeys showed that macrophage accumulation persisted more that 1 year after injection (Verdier et al., 2005). A residence time longer than 6 months was observed in rats (Gherardi et al., 2001).

Verdier et al. (2005) evaluated the local reaction and aluminium concentration following i.m. injection of aluminium adjuvant vaccines in Cynomolgus monkeys. Two groups of 12 male monkeys received a single i.m. injection of either aluminium phosphate adjuvant diphtheria-tetanus vaccine or aluminium hydroxide adjuvant diphtheria-tetanus vaccine. Four monkeys from each of the two groups were sacrificed 85, 169, or 366 days after the single i.m. injection, and macroscopic examination of the injected site was performed to detect any sign of local intolerance. Macrophage aggregation was graded as moderate to marked and was accompanied by a lymphoid infiltration in all cases following the initial sacrifice. Analysis of the injection site revealed high aluminium content for both aluminium treated vaccine groups; however, the aluminium concentration of the reactive zones of animals treated with aluminium hydroxide was 4 times higher than in those treated with aluminium phosphate. The size of the inflammatory lesion was greater in the monkeys given the aluminium hydroxide adjuvant. Six months after the vaccine injection 3 out of 4 monkeys exhibited appreciable lesions composed primarily of macrophages. One of the lesions had an extensive cyst-like structure which contained degenerate macrophages. Two of 4 monkeys in the aluminium hydroxide group had persistent macrophages aggregations with associated minor lymphocytic infiltrations one year following the injection. The histological appearance and persistence of the lesion observed at the injection site is similar to the lesions observed in human cases of MMF. Therefore these results suggest that this type of lesion is a usual reaction following the injection of an aluminium adjuvant vaccine by the i.m. route, and can occur in normal healthy animals following the administration of both aluminium phosphate and aluminium hydroxide containing vaccines.

A field trial involving 45 pigs was conducted to validate the hypothesis of aluminium-induced granulomas (Valtulini et al., 2005). The animals were randomly allocated to receive the same aluminium hydroxide adjuvant vaccine which induced the formation of nodules in the muscles of pigs from one particular farm; the adjuvant alone, distilled water, or the adjuvant and distilled water. The pigs were injected twice i.m. and slaughtered at about 165 kg weight. Granulomas located within muscular tissue were observed for all the aluminium containing vaccine groups; granulomas were not detected in any of the pigs who received only water. Granulomas were characterized by aggregates of macrophage-derived epithelioid cells, with some containing oval nucleus, a pale pink cytoplasm, and indistinct cell borders. These cells were surrounded by an infiltrate of mixed inflammatory cells, including large multinucleated giant cells, macrophages, lymphocytes, plasma cells and eosinophilic elements. In most samples, multiple granulomas were joined by a unique fibrous shell. X-ray microanalysis and atomic absorption revealed the presence of considerable amounts of Al, both within and outside the cells. These results indicate that high amounts of aluminium hydroxide have the potential to produce granuloma formulation.

Administration of endotoxin and aluminium hydroxide adjuvant (0.85 mg) s.c. to Sprague-Dawley rats demonstrated that aluminium hydroxide was able to protect animals from the adverse effects of a 15 μg/kg dose of endotoxin. Shi et al. (2001) suggested that the detoxification of endotoxin by Al-containing adjuvants occurs due to the irreversible binding of endotoxin to the surface of Al. These results suggest that all of the surface aluminium in the aluminium hydroxide adjuvant was able to covalently bind to the phosphate groups of endotoxin, and absorption of endotoxin was subsequently inhibited due to phosphate binding. The pro-inflammatory cytokines tumor necrosis factor (TNF)-α and IL-6 were not detected in the serum of animals receiving endotoxin with the aluminium hydroxide adjuvant (Shi et al., 2001).

Dermal exposure

The comparative irritancy of several aluminium salts was assessed by Lansdown (1973) in three different species. Groups of 5 mice, 3 rabbits and 2 pigs were treated daily for 5 consecutive days with applications of 10 % w/v aluminium chloride, aluminium nitrate, aluminium chlorhydrate, aluminium sulphate, aluminium hydroxide (the pH of the solution was highest at 7.2 among these chemical species of Al tested) or basic aluminium acetate. Twenty-four hr after the final treatment with aluminium hydroxide, signs of erythema, thickening, scaling hyperkeratosis, acanthosis, microabsecesses and the presence of aluminium in keratin were not observed. After single dermal application of aluminium hydroxide (10%) on mouse, rabbit and pig skin no signs of dermal irritation or inflammation were found (Lansdown, 1973).

Repeated Exposure

Oral exposure

Several repeated dose toxicity studies have been conducted in order to assess the effects of oral exposure to aluminium hydroxide on clinical signs, food and water consumption, growth, haematology and serum chemistries, tissue and plasma concentrations of aluminium, and histopathology.

In a study conducted by Hicks et al. (1987) there were no treatment-related effects in rats fed up to 288 mg Al/kg b.w./day as aluminium hydroxide in the diet for 28 days.

Berlyne et al. (1972a) investigated the effects of repeated oral, s.c., and i.p. aluminium hydroxide administration in normal and uraemic rats. Groups of nephrectomised rats were administered 1 or 2% AlCl₃ or Al₂(SO₄)₃ in the drinking-water or oral Al(OH)₃ (150 mg of elemental aluminium/kg/day) by gavage. Groups of non-nephroctomised rats received the same treatments. The duration of the treatment was not indicated. Groups of nephrectomized and normal rats also received i.p. and s.c. injections of Al(OH)₃. The clinical signs of intoxication in nephrectomized animals observed following i.p. administration (90 mg/kg b.w.) included periorbital bleeding, lethargy, anorexia and death. Plasma, liver, muscle, heart, brain, and bone levels of aluminium were markedly elevated in the i.p.-treated group. S.c. injection was apparently less toxic, resulting in no mortality, but periorbital bleeding occurred in nephrectomized animals. Aluminium levels were elevated in all tissues, the highest concentration being in the brain. Administration of high doses of aluminium chloride (180 mg/kg b.w.) and aluminium sulphate (300 mg/kg b.w.) in drinking water to nephroectomized rats produced periorbital bleeding and 100% death in treated animals. Periorbital bleeding was noted for the rats which received drinking water supplemented with Al(OH)₃. In normal rats only aluminium sulphate produced periorbital bleeding in 3 of 5 rats, but no mortality.

Thurston et al. (1972) examined aluminium deposition in the tissues of rats following dietary aluminium hydroxide exposure in order to assess whether the toxicity of this compound was modified when hypophosphatemia was prevented. Weanling rats (6 per group) were assigned to either a whole meal diet; a whole meal diet with aluminium hydroxide (3.2 g/kg) added; or a whole meal diet with added aluminium hydroxide (3.2 g/kg) plus 10 g/kg disodium hydrogen phosphate. An additional group underwent partial nephrectomy and was assigned to the whole meal diet with added aluminium hydroxide.

The duration of the experiment was 4 weeks; after the treatment the animals were sacrificed, blood samples were taken and a complete post-mortem examination was conducted. Animals in the aluminium hydroxide group exhibited a significant impairment of growth, while animals receiving both aluminium hydroxide and the phosphate supplement showed a normal rate of growth. The adverse effects on growth were more severe in uraemic rats but the pattern was the same for aluminium hydroxide treated and untreated animals. Skeletal aluminium content was raised in the normal animals given aluminium hydroxide or aluminium hydroxide and phosphate; however, the uraemic animals showed the most marked increase in skeletal aluminium levels. These results suggested that some aluminium accumulation is seen following oral exposure but that adverse effects are not exhibited if hypophosphatemia is avoided.

The accumulation of aluminium in bone and various regions of the CNS in rats treated with aluminium hydroxide (100 mg/kg b.w./day) or aluminium citrate (100 mg/kg b.w./day) i.g. for either 4 or 9 weeks (6 times a week) was studied by Slanina et al. (1984). However, a decrease in weight gain was observed after 4 weeks of aluminium hydroxide treatment indicating the presence of subacute adverse effect.

Subchronic oral administration (18 days) of aluminium hydroxide (271.3 μg Al/g diet) resulted in significantly increased tibia weight compared to rats fed aluminium phosphate (272 μg/g), aluminium lactate (262 μg/g), or aluminium palmitate (268 μg/g) (Greger et al., 1985).

Body weight of weanling and adult rats was not affected after repeated oral exposure to high doses of aluminium hydroxide mixed with sucrose in the diet (2000 ppm for 67 days) (Sugawara et al., 1988). Rats were fed test diets that had been supplemented with aluminium hydroxide at levels of 989 and 1070 μg Al/g diet. An additional group of rats was fed a control diet containing 26 μg Al/g. No aluminium-induced anaemia or hypophosphatemia was observed in young or adult rats and serum aluminium did not exceed the normal level. Aluminium concentration in the intestinal tract mucosal membrane increased significantly but no effect on inflammatory infiltration or necrosis was noted in the intestine. Serum and hepatic triglyceride levels and adipose weight were decreased significantly in young rats, but neither serum cholesterol nor phospholipid levels was affected by aluminium ingestion. In the adult group, aluminium hydroxide produced a decrease in only hepatic glycogen content (Sugawara et al., 1988).

The body burden of aluminium in weanling rats fed one of 4 diets for 29 days was assessed by Greger & Powers (1992). Rats were assigned to receive a diet containing 40 μmol Al/g diet with or without citrate, a diet containing 100 μmol Al/g diet with citrate, or a control diet containing 0.39 μmol Al/g diet. Rats were injected with DFO or buffer 24 hr prior to sacrifice. Rats fed Al-supplemented diets accumulated significantly more metal in their tissues than rats fed the basal diet, the accumulation was greatest in the rats fed aluminium with citrate. Haematocrit levels following oral aluminium exposure were inversely correlated to tissue aluminium concentrations. It was expected that DFO might mobilize aluminium from tissues subsequently increasing serum and urinary aluminium levels proportionately to bone aluminium concentrations. However, the changes induced by DFO were small and the elevated serum and urine aluminium concentrations were not more correlated to the body load of Al, as indicated by tibia aluminium concentrations. It was estimated that approximately 0.01 to 0.04% dietary aluminium was absorbed. Aluminium hydroxide added to the diet (0.05%) of rats during the 30 days did not affect vitamin A bioavailability (Favaro et al., 1994). Hicks et al. (1987) found no treatment related effects in rats fed up to 302 mg Al/kg bw as aluminium hydroxide for 28 days.

Oral administration of high doses of aluminium hydroxide (1513, 2697 or 3617 mg/kg) in rats for 30 days did not produce any clinical signs or gross symptoms of intoxication, or any significant differences in body weight and food intake. However, in treated animals, behavioural changes (memory and learning ability disturbances) associated with elevated brain aluminium content were observed (Thorne et al., 1986).

Dlugaszek et al. (2000) examined the effects of long term exposure to aluminium in drinking water, including the distribution of the ingested aluminium and changes in the tissue levels of essential elements. Aluminium was administered in drinking water as aluminium chloride, dihydroxy aluminium sodium carbonate, or aluminium hydroxide. Animals in the Al(OH)₃-treated group exhibited an increase in Mg concentration in bones, a decreased Fe concentration in the stomach, and a decline of copper in the kidneys and liver. The group which received AlCl₃ exhibited the highest elevation of aluminium in the tissues following oral exposure.

Bilkei-Gorzo (1993) investigated neurotoxic effects following daily oral administration (90 days) of insoluble aluminium hydroxide (300 mg/kg Al(OH)₃), water soluble AlCl₃ (30 or 100 mg/kg) and chelated aluminium hydroxide (100 mg Al(OH)₃/kg + 30 mg citric acid/kg) in rats. The ability to learn (determined by the number of runs necessary to learn the labyrinth) was affected in all aluminium treated groups; the learned performance was altered to a greater extent in the Al(OH)₃ and AlCl₃ treatment groups. The aluminium content of the brain was elevated in each treatment group; however, the elevation was highest in the groups treated with soluble aluminium compounds. Similarly, all treatments resulted in elevated acetylcholinesterase activity, with significant increases in the AlCl₃ group, and in Al(OH)₃ chelated to citric acid. No relevant differences in body weights, general conditions, or water and food intake were noted between control and treated groups. These results suggested that, although water-soluble aluminium compounds exhibit greater neurotoxicity, the highly insoluble aluminium hydroxide compound appeared to be absorbed subsequently producing some effect on nervous system functions.

Ecelbarger et al. (1994b) conducted a study to assess the impact of chronic exposure to dietary aluminium on aging rats. Male rats were fed diets containing 0.4 or 36.8 μmol Al/g diet in the form of aluminium hydroxide for 8 months until they reached 23 months of age. One day prior to sacrifice, one-half of the rats in both treatments were i.p. injected with DFO, and the remaining rats were injected with saline in order to investigate the usefulness of DFO for estimating body burden of Al. The rats exhibited little evidence of aluminium toxicity as body weight, feed intake, or changes in the relative size of tissues did not appear to be affected by the treatments.

The possible relation between aluminium intake, levels of aluminium in the brain, and dementia was investigated in rats and dogs following chronic oral aluminium hydroxide exposure (Arieff et al., 1979). Clinical signs of intoxication were not apparent in rats with normal renal function (n=10) or rats with chronic renal failure (n=14) exposed to an oral daily dose of 300 mg aluminium hydroxide, for 5 months. Brain Al³⁺ was significantly greater than normal for both groups of rats, the most marked increase being in the group with renal failure. The effects of aluminium were also investigated in two groups of mongrel dogs. One group of dogs received a diet which included 3 g of added aluminium hydroxide daily for 5 months, while the other group received the same diet without Al. In the aluminium loaded dogs the content of Al³⁺ in the cerebral cortex was significantly greater than in that of the controls. Electroencephalograms (EEG) were conducted in the exposed dogs and the results for the aluminium treated dogs were within the normal range. It must be considered that the number of animals in each treatment group was not clearly reported.

A significant increase in tubular phosphate reabsorption with an increase in the apparent velocity of maximal tubular transport was reported in rats following aluminium i.v. administration (Mahieu et al., 1998). Proximal tubule damage was reported in rats (Ebina et al., 1984) and rabbits (Bertholf et al., 1989) following i.v. administration of aluminium. Rats consuming a high aluminium diet (36.8 μmol Al/g diet) for 8 months excreted significantly more protein in urine which is indicative of renal damage (Ecelbarger et al., 1994b).

Studies on the effects of oral administration of aluminium on pregnant animals and their offspring are presented in Effects on Laboratory Mammals and In Vitro Test Systems, Reproductive and Developmental Toxicity.

Injection

Repeated s.c. injection of aluminium hydroxide for 5 or 10 months to rabbits and guinea-pigs did not induce any apparent symptoms or visceral alterations. Limited local lesions characterized by the presence of monocytes and macrophages were observed (Levatidi et al., 1968).

S.c. implants of aluminium foil induced subcutaneous tumours in 8 of 18 rats (O’Gara & Brown, 1967). Histologically, the tumours were fibrosarcomas, rhabdomyosarcomas, or combination of these types. One fibrosarcoma had metastasized to the lungs. The significance of aluminium in tumour induction is unknown as the smooth surface of the aluminium implant may be responsible for the induction of the observed lesions.

Mahieu et al. (2000) examined the effects of chronic i.p. administration of aluminium hydroxide on haematological parameters in rats. Male rats received 80 mg/kg body weight aluminium hydroxide three times a week for 6 months. Control rats were injected with saline solution for the same period. Animals in the aluminium treatment group developed a progressive microcyotosis which became more prominent with time. Significant decreases in mean corpuscular haemoglobin and mean corpuscular volume were noted throughout the duration of the experiment. The haematocrit levels were significantly reduced during months 1, 3, or 4 for the treated animals. The haemoglobin levels of aluminium-treated animals decreased in months 3 and 4 and then began to increase in months 5 and 6. These results suggest that chronic i.p. aluminium administration may interfere with different stages of red blood cell synthesis. Persistent exposure appears to trigger a compensating mechanism leading to restored haematocrit and haemoglobin concentrations, with persistence of microcytosis and depressed mean corpuscular haemoglobin levels.

Bazzoni et al. (2005) studied the effect of repeated i.p. injection of aluminium hydroxide on red blood cell parameters in rats. Male rats were injected with 80 mg/kg body weight of aluminium hydroxide 3 times a week, for a duration of 3 months. A control group of rats was injected with saline at the same frequency and volume. Significant decreases in haematocrit and haemoglobin concentrations were noted in the treated rats as compared to controls. The treated animals showed a high quantity of abnormally shaped (stomatocytes) erythrocytes which became more resistant to haemolysis, and the rigidity index was also substantially higher in the aluminium treated rats. It was reported that the observed effects may have been due to aluminium induced alterations in the mechanical properties of red blood cells resulting in disorganization of the erythrocyte membrane. These alterations likely result in a reduction of the viability of the circulating red blood cells which may lead to anaemia in aluminium intoxicated animals.

Effects of chronic parenteral aluminium administration on parameters of renal function, P and Ca movements at the tubular level, and the possible proximal cell mechanisms were studied by Mahieu et al. (1998). Rats were treated with an 80 mg/kg body weight i.p. dose of aluminium hydroxide (three times per week for 6 months). Control rats received saline for the same duration and frequency. Treated rats showed a significant decrease in asymptotic weight gain and in the initial efficiency of food conversion compared to the controls. Treated rats also exhibited a significant decrease in the haemoglobin level, haematocrit index, and serum Fe concentrations in the peripheral blood compared to control rats. The Ca balance in treated rats was significantly less than in the control group and was accompanied by a significant increase in Ca excreted in faeces, correlated with less intestinal absorption. Accumulation of aluminium on the surface of the trabecular bone and a reduction in the skeletal Ca mass (without any changes in the bone Ca resorption rate), were observed in all treated rats; however, there were no differences when expressed as per 100 g body weight. The reduction in bone turnover was accompanied by a lower recovery velocity from calcemia in the aluminium treated group. The fractional reabsorption of P and sodium was significantly lower in treated animals compared to controls. There did not appear to be any significant differences in the acid-base balance nor in the Ca and P concentrations in the plasma between treated and control animals. It was postulated that aluminium exposure interferes with P excretion by decreasing PTH or by diminishing the affinity for PTH receptors at the level of the renal tubule. These results are in agreement with other studies which demonstrated that aluminium is a potent inhibitor of PTH secretion (Morrisey et al., 1983; Morrissey & Slatopolsky, 1986). Smans et al. (2000) reported that aluminium inhibited PTH secretion in vivo, implying potential for lowered Ca release from the bone to blood.

Cointry et al. (2005) examined aluminium effects on the diaphyseal structure and biomechanics of rat bones. After subchronic i.p. injection of 27 mg Al(OH)₃ in 20% glycerol/water solution for a duration of 26 weeks, blood aluminium levels were significantly higher in treated rats. A significant difference was observed between tibial aluminium concentrations in control vs. Al-treated rats. A significant delay in femur length growth was noted in treated animals. The volumetric bone mineral density and cortical bone material properties (elastic modules and the stress at the yield point) were significantly reduced in treated animals compared to controls. The bone geometric and structural properties were unaffected by treatment. The serum Ca and P levels were not affected by aluminium treatment. No changes were observed in urine volume (diuresis) and urinary excretion of Ca and Pi. No significant differences were noted in the body weights of treated and control groups at the end of the study (Cointry et al., 2005).

An aluminium-induced inhibition of bone mineralization (Cointry et al., 2005) is consistent with other studies showing similar aluminium effects on bone mineralization in vitro and in vivo (Ballanti et al., 1989; Bellows et al., 1995; Blumenthal & Posner, 1984; Chan et al., 1987; Mjoberg et al., 1997; Posner & Blumenthal, 1985). Potential mechanisms proposed include; changes in the biosynthesis of collagen, increased collagen cross-links, altered osteocalcin levels (Blahos et al., 1991; Chan et al., 1987; Mjoberg et al., 1997), and physicochemical dissolution of mineral crystals (Bushinsky et al., 1995). Different combinations of these effects were described as aplastic bone disease (Ballanti et al., 1989; Malluche, 2002) and osteomalacia (Galceran et al., 1987; Lieuallen & Weisbrode, 1991; Malluche, 2002; Mjoberg et al., 1997; Quarles et al., 1985). Chronic oral or parenteral aluminium hydroxide administration was found to produce aluminium accumulation in bone and on the bone surface accompanied by a decrease of P and Zn concentrations in tibias (Berlyne et al., 1972a; Ecelbarger et al., 1994b; Greger et al., 1985; Greger & Donnabauer, 1986; Greger & Powers, 1992; Mahieu et al., 1998; Thurston et al., 1972).

Fiejka et al. (1996) conducted a study to determine the compartmentalization of aluminium in mice following i.p. injection of aluminium hydroxide. One group of mice (n=30) received aluminium hydroxide containing 1 mg of elemental aluminium every two weeks; another group (n=30) received an injection of aluminium hydroxide which contained 0.1 mg elemental aluminium 5 days a week. Controls were treated with saline. Ten animals from each group were sacrificed 48 hr after a cumulative dose of 2, 4 or 6 mg of elemental Al. In the liver the development of aluminium containing granulomas represented by swollen macrophages and multinucleated giant cells were observed; the number of granulomas increased with the aluminium dose. These results suggest that the physiochemical nature of the Al(OH)₃ is an important factor in the development of this type of foreign-body response (Fiejka et al., 1996).

Histochemical and immunochemical studies were carried out after repeated intracerebroventricular aluminium injections (5.4 or 0.68 μg/day for 5 days) in the adult rat brain in animals allowed to survive for a period of either 1 or 6 weeks. Platt et al. (2001) demonstrated that aluminium concentrated in the white matter of the medial striatum, corpus callosum, and singulate bundle. Inflammatory responses and damage in the singulate bundle was noted in Al-treated animals which led to a severe anterograde degeneration of cholinergic terminals in the cortex and hippocampus. These findings suggested that the enhancement of inflammation and interference with cholinergic signalling may be the modes of action through which aluminium results in learning and memory deficits in mammals (Platt et al., 2001).

The injection of 0.3 ml of a 1% suspension of aluminium (metallic) powder into the CSF (cisterna magna) of adult rabbits induced a slowly progressing encephalopathy characterized by alterations of posture, myoclonic jerks and muscle weakness (Bugiani & Ghetti, 1982). The presence of neurons with neurofibrillary degeneration (NFD) and proximal axon swelling was observed between 1 and 81 days after injection. In some animals, large axons with thin or no myelin sheath were also observed. In treated animals pathological changes in the peripheral nerves and muscles were also found. The neuropathological investigation confirmed that the vulnerability of the brain to aluminium is a time-related event. Within the observed time interval (1-81 days), NFD developed in every nuclear complex of the CNS other than the striatum and the amygdala. The factors regulating the time related degeneration of adult rabbit neurons resulting from exposure to aluminium were not clarified by this study. No relation was found between the chronology or the topography of the NFD and the size of the cell bodies, the site of the injection, and the distance from the subarachnoid space and the ventricles (Bugiani & Ghetti, 1982). These data are consistent with other studies which have demonstrated that aluminium salts induce an encephalopathy with NFD after subarachnoid injection in receptive animals (rabbits, cat, ferret) (Crapper et al., 1973; Klatzo et al., 1965; Wisniewski & Kozlowski, 1982). Although the mechanism by which aluminium induces the accumulation of neurofilaments is unknown, it has been demonstrated that these neurofilaments are morphologically identical to filaments that accumulate following the administration of tubulin-binding agents. These compounds bind to tubulin, induce a disruption of microtubules (Ghetti & Ohcs, 1978; Remillard et al., 1975) that is followed by the accumulation of filaments in the cell pericarion, in dendrites and proximal axonal segments.

Reproductive and Developmental Toxicity

Protective treatments on aluminium-induced developmental effects

Albina et al. (2000) conducted a study to determine if a chelating agent, deferiprone, could protect against aluminium-induced maternal and developmental toxicity in mice. Pregnant mice were randomly divided into 5 groups. One group was administered 1,327 mg/kg b.w. of aluminium nitrate nonahydrate by gavage on gestation day 12, a second group was given 24 mg/kg b.w./day of deferiprone on days 12-15 of gestation, and a third group was given 1,327 mg/kg b.w. of aluminium nitrate nonahydrate on gestation day 12 followed by deferiprone (24 mg/kg b.w. at 2, 24, 48 and 72 hr following aluminium exposure. The controls received sodium nitrate or deionised water. Administration of deferiprone alone did not produce any apparent signs of developmental toxicity. Aluminium-induced maternal toxicity included significant reductions in body weight gain, absolute liver weight and food consumption compared to controls. Administration of deferiprone did not offer protection against these aluminium induced maternal effects. In contrast, deferiprone administration following aluminium exposure resulted in a more pronounced decrease in maternal weight gain and corrected body weight change. Developmental toxicity was manifested by delayed ossification of a number of bones in the aluminium treated groups compared to controls. The group treated with aluminium nitrate and deferiprone exhibited a higher number of litters with foetuses showing skeletal deficiencies. These results suggested that deferiprone is not an effective agent to protect against aluminium-induced developmental toxicity and might increase the severity of aluminium-induced maternal and developmental adverse effects in mice.

Silicon-containing compounds were shown to limit absorption of ingested aluminium (Edwardson et al., 1993). Therefore, it was proposed as exerting a protective effect against aluminium-induced toxicity (see also Toxicokinetics, Absorption, Animal Studies, Oral Administration, Factors Influencing Oral Aluminium Absorption, Silicon-Containing Compounds). Bellés et al. (1999) conducted a study to test this hypothesis. Aluminium nitrate monohydrate was administered to three groups of pregnant mice by gavage (398 mg/kg b.w./day) on gestation days 6-15. These animals received silicon in drinking water at concentrations of 0, 118 or 236 g/L on days 7-18 of gestation. Three additional groups of pregnant mice received 270.6 mg/kg of sodium nitrate and the same concentrations of silicon in drinking water as the aluminium-treated groups. The percentage of aluminium-induced deaths, abortions and early deliveries was significantly reduced in the group administered 236 mg/L silicon. However, no significant differences were noted at 118 or 236 mg/L silicon on aluminium induced foetotoxicity.

Neurotoxicity

The scientific literature is replete with reports linking (or dissociating) various chemical forms of aluminium with neurotoxicity and neurodegeneration. Interested readers can gain a grasp of the literature and the differing views as to the potential mechanisms of aluminium toxicity, through a number of existing reviews and hypotheses papers (Atchison, 2003; Becaria et al., 2002; Campbell, 2002; 2004; Campbell & Bondy, 2000; Domingo, 1996; Elmore et al., 2003; Emmett, 2004; Exley, 1999; 2005; Flaten, 2001; Gupta et al., 2005; Kagan et al., 2002; Kumar, 1999; Oteiza et al., 2004; Priest, 2004; Rao et al., 1998; Reinke et al., 2003; Rob et al., 2001; Shin et al., 1995; Savory et al., 2003; Sim & Benke, 2003; Solfrizzi et al., 2003; Soni et al., 2001; Szutowicz, 2001; Tanaka, 2004; Van Landegham et al., 1998; Yokel, 2000). Here, we will focus on some of the literature from the last 10 years that addresses the potential for aluminium to cause neurotoxicity and/or provides insight into the mechanisms of toxicity.

In examining the literature on in vivo studies, it is useful to consider several factors in determining whether a given study may provide significant insight. For studies in animals, the over-riding considerations are dose, mode of administration, speciation of the metal, and measures of outcome. Human Exposure, Total Human Uptake from All Environmental Pathways (Combined Exposure) summarizes the most common sources of human exposure to aluminium. For the majority of individuals, these various sources of exposure result serum concentrations of 1-2 μg of aluminium per Litre of plasma. It is notable that daily doses can be much higher in individuals who use aluminium-based antacids (Lione, 1983). Typically, toxicological studies in rodents utilize doses that are 10-20 times the anticipated human dose. This calculation is based on increased metabolic rates for rodents and is meant to account for differences in the metabolism of the toxin. Therefore, one potential approach to relate experimental studies in rodents, which are most commonly used, to humans is to assume that studies in rodents that achieve serum levels of >10-20 μg/L would mimic normal environmental exposure levels of humans. Studies that achieve higher concentrations of aluminium in serum could be viewed as challenging the animal to determine the toxic potential of the metal.

Investigators have used a variety of routes of exposure, including oral, i.v. injection, and i.p. injection. A variety of aluminium salts have been used in these studies. As discussed in Toxicokinetics, soluble aluminium salts that are introduced into the body are subject to re-speciation with most aluminium in serum being bound to Tf, a minority found as aluminium citrate, and smaller amounts bound to other molecules. Aluminium citrate is usually excreted by the kidney rather quickly, and thus one could expect that Tf would be the major carrier of aluminium into the brain. In CSF, which has higher concentrations of citrate, aluminium dissociates from Tf with the majority bound to citrate. In studies where soluble aluminium salts have been given orally or by injection (i.p. or i.v.), the most likely species of aluminium in brain would be bound to Tf and citrate. In studies where aluminium has been injected directly into the brain there would be also be re-speciation but there would be opportunity for exposure to novel aluminium salts. For the purposes of review, studies involving oral and i.p, or i.v., injection methods of exposure are probably most informative regarding potential effects in humans. Studies in which aluminium salts have been injected directly into the brain have a greater probability of producing non-physiologic effects.

Alzheimer’s disease

Studies in both cell culture models and in vivo have clearly established the potential for aluminium to cause significant neurotoxicity. Likewise, well documented cases of encephalopathy associated with long-term dialysis establish a connection between aluminium and neurotoxicity in humans. One of the first proposed consequences of aluminium exposure in humans was an elevated risk of developing AD. AD is defined by both clinical and pathologic symptoms. Clinically, the symptoms include loss of cognitive ability, psychiatric symptoms, and emotional changes, progressing to profound motor dysfunction and death. Pathologically, AD is defined by the presence of senile plaques, composed of fibrillar extracellular deposits of amyloid β peptide, and NFTs, composed of fibrillar intracellular accumulations of hyperphosphorylated tau protein. Several disorders can produce clinical symptoms that overlap with, or duplicate, the clinical symptoms of AD, including frontotemporal dementia and vascular dementia.

The scientific literature contains numerous reviews that discuss the extant literature regarding AD and aluminium exposure. The initial lines of evidence linking AD to aluminium exposure were numerous reports of elevated levels of aluminium in the brains of AD patients and an association of aluminium with disease-specific lesions (amyloid senile plaques and NFTs). However, subsequent studies utilizing more sophisticated technology demonstrated that aluminium levels are not particularly elevated in the brains of AD patients, and aluminium is not disproportionately distributed to senile plaques or NFT (Lovell et al. 1993; Xu et al., 1992c) (see Tables 19-​-21).21). A well controlled study by Makjanic et al. (1998), demonstrated that the detection of aluminium in tissues from AD patients may partially result from methods of tissue preparation as measures of aluminium in frozen-untreated tissue preparations failed to detect aluminium (<20 ppm) (Makjanic et al., 1998). Studies of a very limited number of autopsied brains suggested that aluminium may preferentially accumulate in lipofuscin granules (Tokutake et al., 1995), which are a common pathology in the aged, and thought to represent remnants of lysosomes that contain undigestable-material.

In regard to animal studies designed to directly test the potential role of aluminium in the development of AD-related pathologies, Pratico et al. (2002) utilized a transgenic mouse model of Alzheimer amyloid pathology. Animals were fed diets enriched in aluminium (dose and chemical form not clarified in the publication) to determine whether aluminium induced changes in the rate or severity of amyloid deposition in this model. The authors reported significant increases (~ 2-fold) in brain loads of soluble and insoluble amyloid peptide and in the levels of isoprostane 8,12-iso-iPF2α-VI, which was examined as a marker of lipid peroxidation. Huang et al. (1997) also reported increases in the expression of the precursor to amyloid peptide, termed amyloid precursor protein (APP), in cervical spinal cords of rabbits given intracisternal injections of aluminium maltolate (2.5 mmole). Intracellular accumulations of APP immunostaining in neurons of the medulla have also been reported by Zhang et al. (2004), who described increased numbers of neurons immunoreactive for APP in the cortex and hippocampus of rats given aluminium chloride in drinking water (3 mg/mL) for 90 days (Zhang et al., 2003). Together, these data suggest that aluminium may modulate the expression and, or, processing of APP. If a life-time of constant exposure results in sustained elevations in APP expression or production of amyloid peptide, then increases in amyloid peptide could accelerate the rate of senile plaque formation and hasten the onset of AD (Jankowsky et al., 2004).

In a study in which rats were given large doses of aluminium sulphate by oral gavage, El-Rahman (2003) reported histological changes that resemble those found in AD patients. Animals were given daily doses of aluminium ranging from 4.59 to 17 mg/100 g b.w./day for 35 days. The brains of animals receiving the highest doses were reported to have developed both classic amyloid deposits and NFTs in the cortex. The report, however, did not quantify the prevalence of either pathology and relied on histological stains, rather than antibody immunostaining, to define pathology. The highest dose in this study would equate to a daily exposure for humans of 170 mg/kg b.w./day. Unfortunately, serum levels of aluminium were not reported, making if more difficult to assess exposure to tissue. Although this dose is far and above normal exposure levels, humans who take aluminium-containing antacids may be exposed to between 15 and 50 mg/kg b.w./day (Lione, 1983). Therefore the dose used in this study was within the 10 to 20-fold excess that is routinely used in rodent toxicological studies. If replicated, and examined with more precise methods, this study could be construed as evidence that very high exposure to aluminium could lead to AD-related pathologic changes in the brain.

Another potential mechanistic connection between aluminium and AD comes from work of Silva et al. (2002) who examined cholesterol levels in the brains of rats fed high doses of aluminium (1 g per day for 10 days or 0.03 g per day for 4 months). Under both conditions, there were significant and robust changes in the ratio of cholesterol to total phospholipids. There were also robust diminutions in membrane fluidity, attributed to the loss of cholesterol. Recent studies from a number of laboratories (Refolo et al., 2000; Smith et al., 2001) demonstrated that the proteolytic generation of amyloid peptide (the principal component of senile plaques in AD patients) is elevated by hypercholesterolemia. The findings by Silva noted above would suggest that aluminium exposure might indirectly reduce amyloid production. However, such beneficial effects could be offset by the reported ability of aluminium to increase the rate at which amyloid peptides (Kawahara et al., 1994; 2001; Mantyh et al., 1993), and other pathogenic proteins such as α-synuclein (Uversky et al., 2001), to assemble into more stable and pathologic fibrillar structures. Additionally, regardless of whether there is a direct action of aluminium on CNS membranes or an indirect metabolic consequence of aluminium exposure, changes in membrane fluidity and cholesterol content could account for the numerous reports of deleterious effects of aluminium on neurotransmitter systems and synaptic function (Csoti et al., 2001; Dave et al., 2002; El-Rahman, 2003; Nayak & Chatterjee, 2001; Wang et al., 2002; Zatta, 1997; Zhang et al., 2004).

The most convincing data to dissociate AD from aluminium exposure are detailed histological analyses of patients who have undergone long-term haemodialysis and succumbed to encephalopathy. To control serum phosphorus levels, patients take high doses of hydroxyl-aluminium gel over long periods, leading to elevated levels of aluminium in many tissues at autopsy. The brains of these patients have characteristic accumulations of aluminium in sub-cellular compartments of both neurons and glia (Reusche et al., 2001a). Morphologically, these structures are most commonly ovid in nature, intracellular, and appear to be lysosome-derived. In rare cases, where pathologic changes that meet criteria for diagnosis of AD in patients with dialysis-encephalopathy (one case out of 127 analyzed), the pathology associated with each disorder remains morphologically distinct (Reusche, 1997). Importantly, in the study by Reusche et al. (2001a), there was no association between AD-like pathology and long-term ingestion of aluminium. The most informative group of patients included 10 who received high doses of aluminium (> 500 g total intake over at least 5 years), for whom the frequency of AD-related pathology was no greater than controls. Moreover, in patients over 60 years of age the degree of AD-related pathology (subclinical levels) in dialysis patients was similar to that in controls. Collectively, these studies in humans support the view that exposure to aluminium poses no direct risk for development of AD. However, caution in over-stating the importance of the data from DAE patients is warranted given the rather small number of highly definitive cases. The animal studies cited above suggest that a life-time of exposure to low doses of aluminium, or exposure to a single-high-dose, might pose a risk for developing AD or contribute to the aetiology of a subset of cases.

Motor neuron disease

It has been known since 1965 that the rabbit is uniquely vulnerable to aluminium toxicity and produces both clinical and pathologic features of motor neuron disease (Klatzo et al., 1965). Intracisternal (directly into CSF) injection of aluminium salts (chloride, phosphate, maltolate) can be used to cause both acute and chronic neurotoxicity. Single high doses of aluminium (1 mg) cause rapidly progressing motor neuron dysfunction with hindlimb paralysis, whereas repeated low dose injections (100 μg) can induce a chronic motor dysfunction (for review see Savory et al., 2001). Detailed neuropathological studies of the chronic rabbit model have defined the similarities and differences between aluminium intoxication in the rabbit and human disease. First and foremost, aluminium intoxication in the rabbit is reversible. Symptoms and pathology associated with chronic low dose aluminium intoxication can be ameliorated if aluminium exposure is reduced at the first signs of symptoms. In humans, motor neuron disease is progressive and almost uniformly fatal. Pathologically, the major resemblance between aluminium intoxication in the rabbit and human disease is the appearance of intracellular inclusions in motor neurons of the spine and brain stem. These inclusions are formed by filamentous aggregations of neurofilament protein (He & Strong, 2000a; Wakayama et al., 1996). Absent from the rabbit model are significant numbers of inclusions that are immunoreactive with cystatin C, ubiquitin, and tau (Wakayama et al., 1996). Also conspicuously absent is evidence of astrocytic gliosis and activation of microglia, both of which are hallmarks of human ALS (He & Strong, 2000a; 2000b; Wakayama et al., 1996). In a chronic aluminium intoxication model, Ghribi et al. (2002) found evidence of caspase-3 activation, an important step in programmed cell death; however, others failed to detect the definitive marker of apoptosis, DNA fragmentation (He & Strong, 2000a), despite ~50% losses in the numbers of lumbar spinal motor neurons. Relevant to AD, Muma & Singer (1996) reported alterations in tau protein phosphorylation and accumulation, the principal component of NFT pathology of AD, in the spinal motor neurons of rabbits chronically given moderate intracisternal doses of aluminium (~ 200 μg of elemental aluminium). Notably, DAE patients do not show increased levels of neurofibrillary pathology (see below).

In vivo models

Behavioural studies of laboratory animals exposed to aluminium

Before reviewing the literature concerning the cognitive behaviour of laboratory animals exposed to aluminium, it is worth noting that there are detailed longitudinal studies of humans exposed to elevated levels of aluminium in the workplace. Buchta et al. (2003) examined a battery of neuropsychological and motor skills in a large cohort of auto manufacturing workers. They recorded average urine levels of aluminium of 70 μg/L, which compares with 1-2 μg/L in most individuals (Buchta et al., 2003). Individuals who had experienced at least 6 years of exposure were selected for analysis and their results were compared with those of co-workers of similar age, gender, and education who worked in other areas of manufacturing. The only measure by which the workers exposed to aluminium could be distinguished was a small reduction in reaction time (speed to respond to a question or perform a motor task). In all other cognitive measures including intelligence quotient (IQ), verbal intelligence, and the European Neurobehavioral Evaluation System, workers exposed to aluminium were no different from control populations (Buchta et al., 2003). Thus in humans, exposure to significant levels of aluminium does not lead to robust changes in cognitive function. However, the above study did not assess the levels of aluminium in blood, which would indicate absorption. Though there are clearly too few human studies, much of the data from studies in adult animals also suggests aluminium exposure does not lead to significant reductions in cognitive function.

As mentioned above, a lifelong exposure of Swiss Webster and C57BL/6J mice to high doses of aluminium (~100 mg/kg/day in feed) was utilized as a paradigm by Golub et al. (2000). With the caveat that brain levels of aluminium were not elevated in the treated mice, suggesting poor absorption, there were little or no deleterious effects of the aluminium-laced diet on several behavioural measures, including grip strength (slight reduction ~10%), temperature sensitivity (slight increase), and spatial reference memory. In the latter task, data from the C57Bl/6J mice are most informative where no adverse impact on acquisition or retention of memory was noted. A prior study by Golub et al. (1995) fed Swiss Webster mice food supplemented with either 500 μg or 1000 μg/g aluminium lactate (calculated dose 200 mg/kg/day) from conception to sacrifice at 150-170 days of age. In several measures of cognitive function, mice fed the aluminium laced diet performed as well as controls. However, 10 to 15% reductions in fore and hindlimb grip strength were noted.

Similarly, although Esparza et al. (2003) reported increased levels of several oxidative markers in the brains of rats given high doses of aluminium (5 and 10 mg/kg/day by i.p. injection), no changes in performance in a passive avoidance task were noted in animals treated with either dose. The classic passive avoidance task involves an electric shock deterrent in which rats are required to remember that the more preferred location (a dark enclosure next to the lighted open space) is associated with shock. Retention of the memory is usually tested 24 hr after conditioning. Hence the task is a measure of memory function.

In contrast to the results of the study by Esparza et al. (2003), Zhang et al. (2003) reported dramatic reductions in performance in passive avoidance in rats exposed to aluminium through drinking water. In the latter paradigm, aluminium chloride was provided through drinking water at a concentration of 3 mg/mL for 90 days. Serum levels of aluminium were not reported. Interestingly, in this study, an extract of Dispsacus asper (a herbal medicine) and vitamin E were shown to alleviate the memory deficits. The authors suggested that the anti-inflammatory and/or anti-oxidant properties of these drugs contributed to the improvements.

Domingo et al. (1996) examined passive avoidance in rats provided drinking water containing aluminium nitrate at concentrations that would equate to doses predicted to be 50 and 100 mg/kg/day for 6.5 months. No effects of the high aluminium exposure on measures of spontaneous motor activity or learning in passive avoidance tasks were noted.

Struys-Ponsor et al. (1997) studied rats given i.p. injections of 667 μg of aluminium gluconate 3 times per week for 60 days prior to assessment of spatial memory in a radial water maze task. Although a slowing of reaction time was noted, there were no statistically significant deficits in the ability of the aluminium-injected animals to perform the task.

Two studies by Miu et al. (2003; 2004), the pathological findings of which are described above, also assessed neuropsychiatric parameters. The 2003 work reported slight reductions in performance in a passive-avoidance memory task and in a spatial reference memory task. The latter work (2004) reported changes in behaviour in open fields which were interpreted as reductions in spontaneous activity and emotional responses.

Overall, the data on neuropsychological measures in rodents given high doses of aluminium by oral routes are not suggestive of profound toxicity. However, none of these animal studies is able to reproduce life-time exposures that could occur over the life-span of humans. It is clear that humans with compromised renal function develop neuropsychological symptoms upon exposure to elevated levels of aluminium. However, from the study of individuals exposed occupationally to aluminium fumes, humans with normal kidney function seem to tolerate high levels of exposure relatively well (see above). The degree to which the chemical form of aluminium, the route of exposure, and the age/health of the individual could modulate the neurotoxicity of aluminium is uncertain and has not been extensively modeled in animals.

Effects on Bone

The bone constitutes a primary site for the deposition of aluminium (Mahieu et al., 2004) (see also Toxicokinetics, Distribution (Including Compartmentalization), Animal Studies, Bone). Elevated aluminium levels in humans, primarily in individuals with impaired renal function, have been associated with several bone disorders including osteomalacia (excess unmineralized osteoid) and aplastic bone disease which is characterized by normal or decreased osteoid (Firling et al., 1999). The mechanism by which aluminium exerts its effects on bone tissue has not been fully elucidated (Cointry et al., 2005). Experimental evidence in a number of different animal models has led to a variety of proposed ways in which aluminium might influence new bone development. It has been suggested that aluminium may directly interfere with osteoblast activity thereby influencing the production or mineralization of osteoid (Firling et al., 1999). Bone formation may be impaired due to aluminium induced reductions in the total number of osteoblasts (Sedman et al., 1987). Direct physiochemical inhibition of mineralization sites has also been proposed as a potential mechanism (Firling et al., 1999). Aluminium-induced alterations in the PTH-calcium axis have also been extensively investigated with respect to aluminium-induced bone toxicity (Mahieu et al., 2004). One of the functions of the PTH is to stimulate bone resorption by increasing osteoblast activity (Quarles, 1990). It has been proposed that aluminium impairs the secretion of this hormone from parathyroid glands (Morrisey et al., 1983). Numerous studies, using a variety of animal models, have been conducted to investigate the effects of aluminium on bone. In interpreting the results of these studies, it is important to consider the difference in bone remodelling physiology between species. It is thought that larger animals such as the dog and pig approximate the bone physiology of humans more closely than rats and mice (Quarles, 1990). The route of aluminium administration and the duration of the study period may also have had significant impacts on the overall results of these studies. It should also be noted that the use of large doses of aluminium in some of these experimental studies may have resulted in a generalized toxicity to the animals, which could complicate the interpretation of aluminium-induced bone toxicity endpoints (Quarles et al., 1988). Some of the in vitro and animal studies investigating the effects of aluminium and bone have been reviewed and are summarized below.

PTH is considered to enhance osteoblast-directed osteoclast activity, and it has been proposed that aluminium may inhibit the production or secretion of this hormone (Jeffery et al., 1996). Morrissey et al. (1983) used dispersed bovine parathyroid cells to determine if aluminium directly affects PTH secretion. Digested bovine parathyroid glands were placed in media containing varying concentrations of aluminium, ranging from 0.5 to 2.0 mM. The incubations were terminated after 2 hr and the amount of hormone secreted into the medium was determined by radioimmunoassay. The secretion of PTH decreased with increasing amounts of aluminium. Hormone secretion decreased by an average of 68% in cells incubated with 2.0 mM aluminium compared to the cells incubated in the absence of aluminium. To examine if there was an irreversible toxic effect of aluminium with respect to PTH secretion, the cells were incubated for 1 or 6 hr with 2.0 mM aluminium, washed with low calcium buffer, and re-incubated in media without aluminium. Hormone secretion appeared to be restored and was comparable to that of cells which had not been incubated with aluminium. Cells were also incubated with radio-labelled leucine to examine the effect of aluminium on the biosynthesis of proparathyroid hormone, PTH, and parathyroid secretory protein. Examination of the incorporation of this radio-labelled amino acid into these proteins revealed that the biosynthesis of these compounds was not affected by aluminium incubation. Therefore, the results of this study suggest that aluminium directly affects the secretion of protein from parathyroid cells.

Ellis et al. (1979) investigated the effects of aluminium on bone toxicity in a group of 20 rats given daily i.p. injections of aluminium chloride for periods of up to three months. Sixteen rats received daily i.p. injections of 0.27 mg Al/day (as aluminium chloride), increasing gradually to a dose of up to 2.7 mg Al/day. The periods for the injections ranged from 48 to 85 days and, in 5 animals, no further injections were given after 63 or 84 days of treatment until sacrifice, 27 or 49 days later. The total dose of aluminium ranged from 38 mg to 109 mg. Four controls were i.p. injected with saline. The whole femur bone aluminium content was higher in the 16 rats given aluminium chloride (176 ± 8.2 ppm/ash compared with the controls (15.4 ± 4.7 appm). A mineralization defect of bone was detected in the rats after 53 days of aluminium treatment, and this increased in severity as the injections were continued. This was marked by an excess of osteoid on the surface of normally mineralized cartilage at the usual site of endochondral ossification. The excess osteoid was typical of osteomalacia with abnormally wide seams and no calcification front. Endochondral ossification was restored to normal, but osteomalacia persisted for up to 49 days after the cessation of treatment.

Chan et al. (1983) investigated the effects of i.p. aluminium chloride (1.5 mg/kg/day) for a duration of 9 weeks in normal (n=16) and uraemic (n=23) rats. Eight rats in the nonuraemic group and 10 rats in the uraemic group received the aluminium treatment, while no injections were given to the remaining control animals. At the end of the treatment period, the rats were euthanized and tissue aluminium, serum vitamin D metabolites, and quantitative bone histology were measured. Bone aluminium concentrations were higher in uraemic rats (121 ± 27 mg/kg) than in normal rats (47 ± 4 mg/kg), and liver aluminium values were higher in the normal group (175 ± 47 mg/kg) than in the uraemic rats (100 ± 36 mg/kg). Aluminium did not appear to have an effect on the levels of any vitamin D metabolites; however, serum concentrations of 25-hydroxyvitamin D and 24,25-dihydroxyvitamin D were reduced as a direct result of uraemia. The nonuraemic aluminium treated animals did not exhibit any significant skeletal changes as compared to the controls. Marrow fibrosis and osteomalacia developed in some of the uraemic, non-aluminium treated animals. However, osteomalacia as defined by (1) an increase in osteoid area (29 ± 13% uraemia + aluminium vs. 3 ± 6% uraemia, no aluminium), (2) an increase in osteoid surface (42 ± 16% vs. 7 ± 11%), and (3) an abnormal pattern of tetracycline uptake at the calcification front, was more severe in uraemic animals treated with aluminium than in untreated uraemic animals.

Robertson et al. (1983) conducted a study to investigate the effects of aluminium on bone histology and PTH levels, and to determine if chronic renal failure accentuates aluminium toxicity. Male Wistar rats were divided into 5 groups. The first group (n=5) received a low dose of aluminium (i.p. injection 0.1 mg aluminium as aluminium chloride 5days/week), the second group (n=5) received a high dose of aluminium (i.p. injection of 1.0 mg aluminium as aluminium chloride 5 days/week); the control group (n=4) was administered an i.p injection of an equal volume of diluent over the same injection schedule. One group (n=6) of rats underwent partial nephrectomy and received an i.p injection of diluent, and another group (n=5) underwent partial nephrectomy and received an i.p. injection of 1.0 mg aluminium as aluminium chloride 5 days/week. The treatment lasted for 120 days for the control and low dose aluminium group, but for a shorter period (between 90-100 days) for the other three groups due to the need for early sacrifice as a result of high attrition in these groups. The trabecular bone of the ischium and the iliac wing was obtained from each animal and examined histologically; the bone mineralization process was evaluated by double tetracycline labelling. There were no differences in bone parameters between the low dose aluminium group and the controls. In the other three groups, the relative osteoid volume (p < 0.02) and the osteoid seam width (p < 0.001) were significantly increased as compared to the controls. These parameters are indicators of osteomalacia. The number of osteoclasts/mm² increased in nephrectomized rats not exposed to aluminium (p < 0.02) and decreased (p < 0.05) in rats with normal renal function exposed to high doses of aluminium. The number of osteoclasts/mm² was not significantly different from that of the control level for the aluminium treated nephrectomized rats. Animals with normal renal function given the high dose were the only group to exhibit a decrease in PTH level, which may explain the reduced osteoclast numbers in this group. However, the increase in osteoclasts/mm², as compared to controls (8.13 ± 2.92 vs. 3.93 ± 100 osteoclasts/mm²), in the non-exposed renal failure group, compared to the reduction of osteoclasts in the exposed renal failure group (2.34 ± 1.38 osteoclasts/mm² vs. 3.93 ± 100 osteoclasts/mm²), indicates that aluminium may have a direct toxic effect on osteoclastic activity.

Cortical bone growth was measured in rats given aluminium to study the early effects of aluminium on bone (Goodman, 1984). Thirty weanling male rats were assigned to one of three groups: control (n=10), experimental (n=10), or basal (n=10). Rats in the experimental group were given i.p. injections of aluminium chloride (2 mg aluminium, 5 days per week); animals in the control group received an injection of saline vehicle at the same frequency, and the basal animals did not receive any treatment. The treatment period lasted for 44 days. Bone growth was assessed over two consecutive periods of 28 and 16 days in the control and experimental rats, using tetracycline labelling of bone. Rats in the basal group were sacrificed on the first day of the experimental period. Histological measurements in sections of bone obtained from the basal group were interpreted to represent the status of the bone at the beginning of the bone-labelling period in rats from the control and experimental groups. Bone (0.017 ± 0.004 mm³/day) and matrix formation (0.017 ± 0.004mm³/day) in the experimental group remained at control levels (bone formation: 0.020 ± 0.004, matrix formation 0.02 ± 0.004 mm³/day) during the first period of assessment (28 days after treatment initiation); however, both these measurements were significantly lower (p <0.01) than control values at the end of the entire 44 day study (bone formation: 0.014 ± 0.003 mm³/day vs. 0.022 ± 0.003mm³/day, matrix formation:0.014 ± 0.003mm³/day vs. 0.022 ± 0.003mm³/day). Bone and matrix apposition at the periosteum in aluminium treated animals was significantly reduced (p < 0.0001) from control levels at the end of the 44 day treatment period, but was not significantly different at the assessment following 28 days of treatment. Aluminium treatment did not induce a state of osteomalacia as no significant effect on the osteoid width or the mineralization front width was apparent. The results of this study indicate that aluminium reduces bone and matrix formation early in the course of aluminium exposure, prior to the development of histologically apparent osteomalacia. Aluminium may affect matrix synthesis by reducing the total number of active osteoblasts or diminishing the cellular activity of individual osteoblasts.

Goodman et al. (1984a) examined the effect of short-term (4 week exposure) aluminium administration on bone growth and histology, and evaluated the role of renal insufficiency in mediating the skeletal effects of aluminium. Thirty rats underwent partial nephrectomy and were assigned to one of three groups: control, aluminium-treated, or basal control. Thirty additional rats with intact renal function were divided into the same treatment assignments. Aluminium treated rats received i.p. injections of aluminium chloride (AlCl₃) in saline 5 days/wk for 4 weeks; the aluminium dose was 2 mg/day. Control rats received injections of saline only, according to the same treatment schedule. Bone growth, bone formation, mineralization, and resorption were measured using double tetracycline labelling of bone. Total bone and matrix formation and periosteal bone and matrix formation were reduced in both nephrectomized and normal renal function rats treated with aluminium as compared to the respective controls. Periosteal bone and matrix formation were similarly reduced in both groups. There was no difference in bone parameters between the control rats of the nephrectomized group as compared to the normal renal function control rats; however, total bone, total matrix, periosteal bone, and periosteal matrix formations were all less in the nephrectomized aluminium treated rats as compared to the non-nephrectomized aluminium treated rats. Resorption surface was greater in both aluminium- (1.70 ± 0.41mm vs. 1.33 ± 0.34 mm) and nephrectomized-aluminium-treated (1.87 ± 0.6 mm vs. 165 ± 0.43 mm) rats compared to the respective controls, and resorptive activity at the endosteum was greater (p < 0.05) in the nephrectomized aluminium treated group (12.2 ± 6.3 μm/d) than in the controls (7.9 ± 4.9 μm/d). Serum calcium and phosphorus concentrations were similar in aluminium-treated and control animals, suggesting that PTH secretion was not substantially affected by aluminium administration. Osteomalacia was not detected, but it should be considered that the duration of this study may not have been of sufficient length for this condition to develop. The results of this study suggest that aluminium exerts is toxic effects on bone by acting directly to reduce new bone and matrix synthesis. In addition, it appears that aluminium may act to increase bone resorption. The enhanced effect of aluminium in nephrectomized rats may be a result of increased aluminium accumulation due to an impaired ability to excrete the metal; however, aluminium in bone was not quantified in this study.

In order to investigate the effects of aluminium on the vitamin D-dependent mineralization process, aluminium chloride (1 mg/kg) was administered i.v. 3 times per week for 3 weeks to normal (n=5) and vitamin D-deficient (n=5) beagle puppies (Quarles et al., 1985). Vitamin deficiency was induced in the 5 dogs by providing a diet deficient in vitamin D and calcium for a period of 15 weeks before treatment initiation. Bone biopsies and plasma were obtained before and after the 3 week treatment period in each group. In the next phase of this study, aluminium chloride administration was continued at a lower dose (1 mg/kg twice a week) and both groups received the diet fortified with calcium and vitamin D. This phase of the study lasted for 11 weeks. The vitamin D deficient dogs displayed biochemical and bone biopsy evidence of osteomalacia before administration of aluminium. Plasma phosphorus, PTH, and 25-hydroxyvitamin D concentrations did not appear to be affected by the 3 weeks of aluminium administration in either group. However, bone aluminium content increased to a greater extent in the vitamin D-deficient dogs (390 ± 24.3 μg/g) than in the normal dogs (73.6 ± 10.6 μg/g). After 3 weeks of aluminium treatment, the bone histology in both groups revealed changes consistent only with aging. Provision of a calcium/vitamin D replete diet to the deficient dogs, and reduction of aluminium dosage for 11 weeks resulted normalization of their plasma biochemistry and healing of the osteomalacia. The bone aluminium content of the non-vitamin D deficient dogs increased during the additional 11 weeks of aluminium exposure to 151.0 ± 11.3 μg/g as compared to a bone aluminium content of 73.6 ± 10.6 μg/g after 3weeks of exposure. In contrast, the bone aluminium content in the vitamin D deficient dogs decreased from 390 ± 24.3 μg/g to 173.5 ± 5.6 μg/g at the end of the additional 11 weeks of exposure. Histochemical staining for bone aluminium revealed no apparent aluminium in the normal group and, in the vitamin-D deficient group, there was identifiable aluminium at the mineralization fronts after the 3 weeks of treatment. At the end of the 11 weeks of additional exposure staining for aluminium was detectable only in the cement lines, indicating that mineralization occurred over the previous sites of aluminium deposition. The results of this investigation indicate that aluminium accumulates preferentially in pre-existent osteomalacic bone and localizes at the calcification front (osteoid-bone interface). The presence of aluminium at the calcification front did not impair vitamin D-dependent mineralization as remineralization occurred in the bones of the osteomalacic pups following vitamin-D repletion. Osteomalacia did not occur in the normal pups; however, this does not eliminate the possibility that aluminium administration at higher doses or for prolonged periods might cause bone toxicity in these animals. Therefore, although aluminium may have the potential to cause osteomalacia, its presence at mineralization fronts does not appear to be the mechanism through which this occurs.

Alfrey et al. (1985) studied aluminium compartmentalization in rats with an induced state of uraemia or hypoparathyroidism. Seventy four male rats were divided into two main study groups. The first group consisted of 10 animals that underwent selective PTX, 10 animals were rendered uraemic by uninephrectomy, 11 animals underwent both nephroctomy and PTX, and 8 animals served as controls. 1,25(OH)₂D₃ was administered to normalize serum calcium levels. All animals received 1.5 mg/kg aluminium (as AlCl₃) by i.p. injection 5 days/wk for 79 days. There was a significantly greater (p < 0.001) trabecular bone osteoid area (percent total bone area) in the PTX uraemic group than in the uraemic group (45 ± 9.7% compared to 13.4 ± 10.6%). Plasma calcium levels were significantly higher in the uraemic PTX group (10.7 ± 0.9 mg/dL) and lower in the control PTX group (9.0 ± 0.5 mg/L) than in their respective controls (9.9 ± 0.6 g/L: uraemic group; 9.8 ± 0.3 mg/L: control group), and phosphorus levels were significantly higher in both the PTX uraemic group (9.3 ± 1.5 mg/L) and PTX control group (8.1 ± 0.6mg/L) as compared to the controls (5.9 ± 0.3mg/L). The second study group was comprised of 35 uraemic rats; 8 of these rats had previously undergone selective PTX and were given 1,25(OH)₂D₃. The uraemic PTX animals and 8 of the uaremic animals received 1.5 mg AlCl₃/kg by i.p. injection 5 days/week for 35 days, and were killed at the end of this treatment period. Aluminium injections were discontinued in the remaining animals, and nine underwent selective PTX at this time. These animals were followed for an additional 30 days. Bone aluminium levels in the animals killed after 35 days were lower (p < 0.05) in the PTX uraemic group than in the uraemic control group (37 ± 7.6 and 47 ± 7.6 mg/kg, respectively). The bone aluminium levels in the uraemic group that underwent PTX after aluminium loading (53 ± 8 mg/kg) and in the uraemic group (48 ± 5.7 mg/kg) 30 days after aluminium discontinuation were not significantly different. The results of this study suggest that PTX affects the compartmentalization of aluminium in bone, especially in animals in a uraemic state. It also appears that PTX may intensify aluminium-induced osteomalacia as the PTX group had significantly greater bone osteoid area than the uraemic group of rats.

Galceran et al. (1987) conducted a study to further characterize the mechanism of aluminium induced osteomalacia. One group of dogs (n=7) was treated i.v. with 0.75 mg aluminium, 5 days a week for 3 months; 7 additional dogs served as controls. At the end of the treatment period, the dogs were killed, and the tibiae were obtained and perfused in vitro. PTH and methylxanthine (an inhibitor of phosphodiesterase) were added to the perfusate. The serum aluminium level was 20.4 ± 2.3 μg/L before treatment, and rose significantly to 206.2 ± 28.3 (p < 0.01) after aluminium administration. No significant difference in PTH levels was detected before and after aluminium treatment. Because PTH acts to stimulate cAMP release from bone, cAMP levels were measured in both groups before and after PTH administration to examine the effect of aluminium on this process. Although the basal cAMP secretion was the same in both groups of dogs, cAMP increased to a peak of 188.2 ± 30.6 pmol/min in the normal dogs vs. 113 ± 8.15 pmol/min in the aluminium treated dogs after PTH was added to the perfusate (p < 0.05). Examination of bone biopsies taken before and after aluminium administration revealed that the number of osteoblasts had decreased 8-fold (p < 0.01) following aluminium treatment. Aluminium treatment led to an increase in the percent total osteoid surface (44.3 ± 4.7% vs. 22.1 ± 4.1%), decreased mineral apposition rate (0.5 ± 0.4 μm/day vs. 1.3 ± 0.2 μm/day), and aluminium deposition at the mineralization front. Despite the decrease in osteoblast number, the histological features of the post aluminium treatment biopsies indicated that aluminium stimulated osteoblastic activity at some point during its administration. This was apparent from the presentation of new woven bone formation in two dogs, and a layer of newly deposited lamellar bone covering all trabecular surfaces in another. Aluminium also appeared to stimulate the activity of fibroblasts as indicated by the presence of extensive marrow fibrosis in five of the treated dogs. These data support the possibility that aluminium is capable of both stimulating and suppressing matrix synthesis at different times throughout the exposure period. Although the levels of PTH were similar in the aluminium exposed and control animals, the decreased generation of cAMP following the addition of PTH to the perfused bones of the aluminium treated group suggests that aluminium may cause bones to be resistant to the effects of PTH.

Sedman et al. (1987) examined the effects of i.v. aluminium injections for a duration of 8 weeks on various bone parameters in growing piglets. Four piglets were administered 1.5 mg AlCl₃/kg/day parenterally, and 4 control piglets received daily injections of deionised water for the same period. Quantitative bone histology and measurements of bone formation were assessed at three skeletal sites (two in the proximal tibia and one in the distal femur) in both the experimental and control groups. Bone aluminium was significantly higher (p < 0.001) in experimental animals (241 ± 40 mg/kg) as compared to the controls (1.6 ± 0.9 mg/kg). Osteomalacia, as defined by histological criteria, was documented in all aluminium-treated animals. The rate of mineralized bone formation was also lower in the aluminium-treated group compared to that in the controls at all three sites. However, it was found that, at sites of continued osteoblastic activity, total osteoid production did not differ between the two groups. These results suggest that bone mineralization is inhibited by aluminium via a decrease in the number of active osteoblasts rather than by an inhibition of the calcification of osteoid.

Ott et al. (1987) conducted a study to investigate the development and reversibility of aluminium-induced bone toxicity in weanling and adult rats. Four groups of weanling rats and one group of adult rats were used in this study. Each group consisted of 18 rats. Half the rats in each group received daily i.p. injections 10mg Al/kg as aluminium chloride. The other half of the group was given normal saline solution. Weanling rats were sacrificed after 3, 6, or 9 weeks of treatment. The remaining group of weanling rats received injections for 9 weeks, and was allowed to recover for 3 weeks. The adult group received aluminium treatment for 9 weeks. The effects of aluminium on blood serum levels of various compounds, and on aluminium bone content, and rate of bone formation were assessed after intervals of 3, 6, and 9 weeks. The calcium, phosphate, creatinine, and PTH levels were similar in aluminium-treated rats and controls. Aluminium was detectable by histochemical stain after 6 weeks in the aluminium treated animals; however, other bone parameters did not differ significantly at this time between the treated animals and the controls. A decrease in bone formation (measured by tetracycline labelling) on trabecular and endosteal surfaces was apparent by 9 weeks in the aluminium exposed groups. The weanling rats had a bone formation rate of (0.15 ± 0.2 mm²/100 days vs. 0.46 ± 0.14 mm²/100 days) which was significantly lower (p < 0.02) than the rate in the controls, while the adult rats had a rate of (0.04 ± 0.06 mm²/100 days) vs. (0.23 ± 0.12 mm²/100 days) in the controls. One group of rats was allowed to recover for 3 weeks without any aluminium administration. The bone formation rate in the younger group was similar to that of the controls after 3 weeks of recovery. Adult rats showed signs of early osteomalacia as evidenced by an increase in the length (4.75 ± 2.3 vs. 2.15 ± 1.5% surface) and width (18.4 ± 9.7 vs. 4.8 ± 1.3 μm) of the trabecular osteoid as compared to the controls. In this study it appeared that aluminium administration led to decreased rates of bone formation in rats despite normal calcium and parathyroid levels, and normal renal function. It is possible that aluminium induced decreased bone formation by inhibiting osteoblast formation or activity.

Quarles et al. (1988) conducted a study to define the primary effects of aluminium on bone in the mammalian species, and to examine the dose/time-dependent actions of aluminium on bone. Two year-old beagles were assigned to one of three treatment regimens. The first group (n=6) received 0.75 mg Al/kg as aluminium chloride i.v. three times per week, the second group (n=6) received 1.20 mg Al/kg at the same dosing schedule, and the third group (n=6) received sodium chloride i.v. and served as controls. The treatment period lasted for 16 weeks. Transcortical bone biopsies were taken from each group after 8 weeks and 16 weeks. In both the low and high dose aluminium groups, serum aluminium levels were significantly elevated compared with those of controls, but calcium, or PTH levels were not altered by treatment. Bone biopsies taken at 8 weeks in the low dose group displayed characteristics of a low turnover state, as marked by a reduction of bone resorption (2.6 ± 0.63% vs. 4.5 ± 0.39%) and osteoblast–covered bone surfaces (2.02 ± 0.51% vs. 7.64 ± 1.86%) as compared to the controls. The mineralized bone formation rate was also found to be significantly decreased. Biopsies taken at week 16 of aluminium administration in the low dose group displayed evidence of de novo bone formation, as well as an increase of bone volume (38.9% ± 1.35 vs. 25.2% ± 2.56) and trabecular number (3.56/mm ± 0.23 vs. 2.88 ± 0.11) compared to the controls. The 16 week biopsies displayed a persistence of inactive osteoid marked by a diminished mineralization front in the low dose group compared to the controls (46.0 ± 4.2% vs. 71.9 ± 2.92%). The bone biopsies obtained from the high dose aluminium group at 8 weeks displayed changes similar to those exhibited after 16 weeks of the low dose treatment. De novo bone formation was evidenced by an increase in trabecular number as compared to the controls (3.41 ± 0.18/mm vs. 2.88 ± 0.11/mm) and increased bone volume (36.5 ± 2.38% vs. 25.2 ± 2.56%). Poorly mineralized woven bone accounted for a large proportion of the newly synthesized tissue, comprising 11.5 ± 4.6% of the bone volume. High dose treatment for 16 weeks further enhanced bone volume (50.4 ± 4.61%) and trabecular number (3.90 ± 0.5/mm). The woven osteoid volume at 16 weeks decreased to 2.43 ± 0.96% of the total bone volume, indicating that heterogeneous calcification of this tissue was more complete. The observation of histological changes similar to those observed in disorders characterized by low bone turnover in the low-dose aluminium group after 8 weeks, combined with the observation of new bone formation and stimulation of cellular activity after longer treatment, suggests that aluminium may exert both inhibitory and stimulatory effects on osteoblasts.

To further examine the influence of osteoblast function on aluminium-induced of de novo bone formation, Quarles et al. (1989) compared the effects of aluminium in TPTX beagles (n=4) with beagles which underwent thyroidectomy but had intact parathyroid glands (n=4). The animals underwent TPTX as a means to reduce osteoblast number and activity. The treatment procedure began 2 months after surgery, when sufficient time had elapsed to achieve a new steady state of bone remodelling activity. 1.25 mg AlCl₃/kg was administered to both groups of animals by i.v. three times per week for 8 weeks. The TPTX animals received supplements of calcium carbonate and calcitrol in order to maintain normal plasma 1,25-dihydroxyvitamin D levels as well as normocalcemia. Both groups were administered thyroxine to sustain normal free thyroxine concentrations. Although both groups of animals received the same aluminium treatment, TPTX beagles exhibited a significantly higher (p < 0.05) serum aluminium level (2386.1 μg/L) as compared to the controls (1087.0 μg/L). Aluminium administration did not alter the plasma calcium, creatinine, or PTH from baseline levels in either group of animals. Bone biopsies taken from the control animals after 8 weeks of treatment displayed evidence of de novo bone formation as compared to baseline bone parameters. This was evidenced by an increased bone volume (47.0 ± 1.0 vs. 30.4 ± 09%) and trabucular number (4.1 ± 0.2 vs. 3.2 ± 0.2/mm). Deposition of poorly mineralized woven bone accounted for much of the enhanced bone volume (9.9 ± 2.7%). TPTX animals demonstrated significantly less evidence of bone formation. Bone volume (35.5 ± 1.7% vs. 27.7 ± 1.9% at baseline) and woven tissue volume (1.4 ± 0.8% vs. 9.85 ± 2.66%), as well trabecular number (3.3 ± 0.1/mm vs. 4.2 ± 0.2/mm) were significantly less than those of the aluminium treated non-TPTX controls. It appears that the diminished functional osteoblast pool in the TPTX beagles limited the ability of aluminium to stimulate neo-osteogenesis.

Bellows et al. (1999) examined the effects of aluminium on osteoprogenitor proliferation and differentiation, cell survival, and bone formation in long-term rat calvaria cell cultures. The cells obtained from foetal rats were incubated in medium with or without aluminium added. The aluminium treated cells were incubated at various concentrations ranging from 1 uM to 1 mM of aluminium for up to 19 days. The numbers of mineralized and unmineralized bone or osteoid nodules in each culture dish were quantified by in situ staining. Alkaline phosphatase activity, cell viability, and cytotoxicity were also determined. Nodule formation was significantly increased (p < 0.001) by 30-1000 μM aluminium incubation, in a dose dependent manner, at 11 days but not at 17 days. Control and aluminium-treated cultures appeared similar with respect to nodules and cell layers at day 13 of culture. However, at day 17 of culture, aluminium concentrations of 30 μM and above resulted in reduced cellularity and an increased fibrilar appearance of the matrix that had formed outside of, or adjacent to, nodules. Aluminium also increased alkaline phosphatase activity at all time points in a dose-dependent manner. Significantly fewer viable cells were present in the 300 μM aluminium-containing cultures after 13 and 17 days. The results of this experiment indicate that aluminium has a stimulatory effect upon existing osteoprogenitor cells leading to an accelerated rate of osteoblastic differentiation and nodule formation, while inhibiting nodule mineralization. The concentration of aluminium at which this effect occurred resulted in decreased cell viability and enhanced cytotoxicity.

The effect of aluminium administration on bone, in a model of osteopoenia induced by chronic acid overload in rats with normal renal function, was examined by Gomez-Alonso et al. (1999). Thirteen male rats with induced osteopoenia were divided into two groups. The first group (n=8) received 10 mg/kg of AlCl₃ i.p. 5 times per week for 4 months, the second group (n=5) did not receive any aluminium treatment. At the end of the experiment, both tibias from each animal were extracted for the determination of aluminium content, in vitro bone densitometry, and histological analysis. Bone mineral density, measured at the proximal end of the tibia, was found to be significantly higher (<0.05) in the aluminium-treated group (0.292 ± 0.01 g/cm²) as compared to the controls (0.267 ± 0.02 g/cm²). Histomorphometric analysis showed a significant increase (p < 0.01) in bone volume (18.59 ± 5.66% vs. 7.69 ± 3.08%), cortical thickness (0.52 ± 0.06 mm vs. 0.36 ± 0.07 mm), osteoid thickness (14.05 ± 4.72 μm vs. 5.25 ± 090 μm), and osteoclast number (2.44 ± 0.52 N Oc/mm² vs. 1.30 ± 0.01 N Oc/mm²) in the aluminium-treated group as compared to the controls. No significant differences in serum calcium, phosphorus, creatinine, hydroxyproline, or PTH were apparent between the aluminium-treated and control animals. There was no evidence of osteomalacia in the aluminium-treated rats. These findings indicate that aluminium is able to induce bone formation in rats with normal renal function even when osteopoenia is present.

Firling et al. (1999) examined the influence of aluminium citrate administration on tibia formation and calcification in the developing chick embryo. Tibia formation and mineralization were assessed by radiology, total bone calcium content, calcium incorporation rate, collagen synthesis rate, bone alkaline phosphatase activity, and serum levels of osteocalcin, procollagen carboxy-terminal propeptide, and PTH. The chick embryos derived from White Leghorn strain eggs were divided into three treatment groups (aluminium citrate, sodium citrate, sodium chloride), and were treated acutely or chronically. Acutely treated embryos received 100 μL of 60 mM aluminium citrate, 60 mM sodium citrate or 0.7% sodium chloride via injections into the air sac of the egg on day 8 of incubation. Chronically treated embryos received a daily 25 μL dose of the solutions beginning on day 8. The embryos were incubated for an additional 2 to 8 days following treatment. Radiographic analysis of the tibias and femurs of the embryos revealed that the mineralization of the aluminium treated animals was less dense and restricted to a shorter length of the mid-diaphysis as compared to the other two groups. The bone calcium content of embryos acutely or chronically administered aluminium and incubated for 10 to 12 days was significantly lower (p < 0.05) compared to that of the other treatment groups. The calcium content of tibias from embryos chronically treated with aluminium remained lower than the controls for 12 day and 16 day embryos while, by day 14, there were no significant differences in the total calcium content from acutely treated embryos compared to the controls. Significantly higher (p < 0.05) levels of alkaline phosphatase activity were found in the tibias collected from embryos chronically treated with aluminium incubated from 12 (2.16 units/tibia/hr vs. 1.32 units/tibia/hr) to 16 days (10.38 units/tibia/hr vs. 6.78 units/tibia/hr) as compared to the sodium chloride control group. Aluminium did not have a significant effect on the rate of tibia collagen, non-collagenous protein synthesis or serum levels of procollagen carboxy terminal propeptide, osteoclacin or PTH. The lack of change in these parameters suggests that embryonic osteoblast number and activity is not markedly diminished by aluminium exposure at these doses. The authors suggested that the observed under-mineralization of the tibias in the aluminium treated embryos may be a manifestation of the production of defective osteoid, an inhibited terminal maturation of osteoblasts, or physiochemical inhibition of mineralization nucleation sites.

Zafar et al. (2004) investigated the effect of chronic exposure to dietary aluminium on calcium absorption and calbindin concentrations in male weanling rats fed various levels of calcium and aluminium for 3 and 6 week periods. One group of rats (n=40) was fed a calcium adequate diet, three other groups of 40 animals each were fed a calcium-deficient diet with 0, 0.05 or 0.1% aluminium, as aluminium chloride. After 3 weeks, 20 rats per group were fasted overnight and 10 rats per group were given an oral dose of 25 mg calcium labelled with 6 μCi ⁴⁵Ca by gavage. The other 10 rats in each group were given 6 μCi ⁴⁵Ca as an interperitoneal injection. Rats were anaesthetized the following day, blood was collected and the femurs were obtained. The remaining animals (20 per group) were switched to a calcium adequate diet containing the same level of aluminium they had been fed previously, and were maintained on these new diets for another 3 weeks. No difference in ⁴⁵Ca absorption was observed among the 4 groups at either 3 or 6 weeks. Aluminium supplementation at 0.05 and 0.1% of the diet reduced calbindin concentrations (compared to the group receiving a calcium deficient diet without aluminium). Total bone calcium decreased with aluminium supplementation. The bone calcium content was significantly different (p < 0.05) in the calcium deficient-no aluminium group, and in the calcium deficient groups supplemented with 0.05 and 0.1% aluminium, as compared to the calcium adequate group at both 3 and 6 weeks of treatment. In addition, the bone calcium content was significantly lower (p < 0.05) in the calcium deficient, 0.1% aluminium group as compared to all the other groups. Aluminium treatment reduced the breaking strength parameters of the bones from rats on the calcium-deficient diets. When the animals were switched to a calcium adequate diet for 3 weeks, there were no differences in the resistance to breaking due to aluminium intake. The results of this study suggest that dietary aluminium has detrimental effects on bone quality when calcium is deficient.

Cointry et al. (2005) analyzed the effects of aluminium accumulation on whole-bone behaviour in rats. Rats (n=14) received i.p. doses of 27mgAl/day, as aluminium hydroxide (Al(OH)₃), for a period of 26 weeks. Fourteen control rats received a 20% glycerol/water solution at the same dosing schedule. At the end of the experimental period, the left tibiae was obtained from each animal for bone ash determination. Both femurs from each animal were also dissected and examined for the volumetric mineral density of the cortical bone region and for cross-sectional properties of the cortical bone region. Mechanical testing of the femurs was also conducted. The Young’s modulus of elasticity (a measure of stiffness) was calculated, as well as the stress of the cortical tissue at the yield point, which is an indicator of the tissue’s ability to support loads before any crack initiation. Aluminium concentration was significantly higher (p < 0.001) in the tibias of treated animals (103 ±18 μg/g) as compared to those of control animals (6 ±1 μg/g). The volumetric bone mineral density was significantly reduced in the treated animals with respect to the controls. Up to the yield point, the structural stiffness and strength of the bones did not differ between groups; however, an aluminium-induced impairment of the ability to resist loads beyond the yield point was observed. Treatment had a negative impact on the bending stiffness (Young’s elastic modulus) and the yield stress of cortical bone. These parameters were decreased by 18 and 13% respectively in treated animals as compared to those of controls (p< 0.05). The cortical second moment of inertia, which is a measure of the architectural efficiency of the cortical bone, was significantly improved in aluminium-treated rats (+10% change, p <0.01) as compared to that of the control group. This suggests that an adaptive response to aluminium treatment may have caused an improvement of the spatial distribution of the available cortical tissue resulting in an enhanced ability of the bone to resist anterior-posterior bending. The results of this study suggest that the apparent adaptive response of the bone to aluminium may have maintained normal stiffness and strength, but aluminium may have reduced the ability of the bone to resist loads beyond the yield point, or the ultimate strength of the bone.

Mineral Metabolism

The effects of aluminium on metabolic parameters in humans are not well understood. There have been relatively few studies of the effects of occupational exposure to aluminium on mineral metabolism. Ulfvarson & Wold (1977) examined the levels of trace metals in the blood of welders of aluminium and stainless steel. Although the authors did not report the levels of aluminium in blood, which would have been useful in gauging exposure, other studies of aluminium welders have noted relatively high levels of exposure (Buchta et al., 2003). In the study by Ulfvarson & Wold (1977), the levels of lead, strontium, rubidium, bromine, gallium, zinc, copper, cobalt, iron, manganese, chromium, calcium, potassium, sulphur, phosphorus, silicon, and magnesium in the blood of aluminium welders were not statistically different from the levels of controls. However, significant variability in the data was noted. Little is known regarding the effects of aluminium on other metabolic parameters in humans.

Most of the data on the effects of aluminium on trace metal, and general, metabolism originates from studies on animals given aluminium by various routes. There is clear evidence that aluminium can influence iron metabolism in the context of haematopoiesis (see Effects on Laboratory Mammals and In Vitro Test Systems, Effects on Haematopoieses). However, as described below, alterations in iron levels in solid organs have also been noted, but with inconsistent outcomes. Studies of the metabolism of trace metals have yielded similarly conflicting outcomes. There have also been reports of alterations in a variety of metabolic pathways in response to aluminium exposure.

Non-haem iron and trace metals

Golub et al. (1995) reported that mice fed diets containing as much as 1 mg Al/g for 150 days had small, but statistically significant, reductions in the levels of iron in spinal cord and liver. The calculated dose of aluminium exposure in this experiment was 200 mg/kg b.w./day, which is about 100-fold greater than dietary exposure in humans. Animals exposed to these levels of aluminium had no significant differences in body weight.

Ward et al. (2001) injected rats (i.p.) with aluminium gluconate (2 mg/kg b.w.) 3 times a week for 8 weeks and then examined iron levels in liver, kidney, heart, spleen, and brain. Two to three-fold increases in tissue iron levels were noted in liver, spleen, and brain. Blood levels of aluminium were not reported, but the levels of aluminium in liver and spleen were recorded as 100 ng/mg protein and 150 ng/mg protein, respectively, indicating substantial exposure.

Han et al. (2000) examined non-haem iron levels in kidney, liver, and intestine of chicks fed diets containing 0.15 and 0.3% (by weight) aluminium for 3 weeks. Chicks receiving the higher dose of aluminium were reported to have gained 28% less weight. Levels of iron in liver were reduced by 40%, with 30% reductions in iron levels in intestine.

Esparza et al. (2003) examined the levels of manganese, iron, and copper in the brain and liver of rats given i.p. injections of aluminium lactate (5 mg/kg b.w.) 5 days per week for 8 weeks. Rats exposed to this level of aluminium showed 5-fold increases in aluminium levels in liver (8.45 μg/g compared to 48.35 μg/g) with a 2-fold increase in cerebellum (5.49 μg/g compared to 12.51 μg/g). The cortex and hippocampus of brain showed trends towards increased aluminium accumulation, but were not statistically different from those of controls. In liver, manganese levels were reduced by 40% with no alterations in iron or copper levels. In cerebellum, there were no statistically significant changes in the levels of any of these metals while, in cortex and hippocampus, the levels of copper were reduced 25 to 40%. Cortex also showed 25% reductions in manganese. Animals injected with aluminium lactate gained 25% less weight than control animals injected with vehicle.

The levels of copper, zinc, and manganese have been measured in the serum of rabbits exposed to aluminium by s.c. injection of aluminium sulphate (600 μmol aluminium/kg b.w./day) five times per week for 3 weeks (Liu et al., 2005). The total aluminium exposure per rabbit for the entire study was 243 mg/kg b.w. Aluminium levels in serum increased from 0.25 μg/mL to 1.3 μg/mL after 21 days of treatment before levelling to 1.2 μg/mL by day 42 of treatment. There were no statistically significant changes in the levels of zinc, copper, or manganese in serum after either 21 days or 42 days of treatment. The authors reported weight loss in the aluminium-treated group but did not specify the degree of loss.

Fattoretti et al. (2004) examined the levels of copper, zinc, and manganese in the brains of rats exposed to aluminium through drinking water (2 g AlCl₃ /L) for 6 months beginning at the age of 22 months. Serum levels of aluminium were not reported, making it difficult to assess the level of exposure. The authors sampled 3 domains of the brain; fore- and mid-brain together, pons and medulla together, and cerebellum. In cerebellum, there was no significant change in the levels of any metal. The authors reported accumulations of aluminium in fore- and mid-brain (94% increase) with increases in copper, zinc, and manganese (32, 41, and 50%, respectively). Pons and medulla showed 53% increases in aluminium accumulation with 46, 46, and 41% increases in copper, zinc, and manganese, respectively.

Sanchez et al. (1997) described an analysis of calcium, magnesium, manganese, zinc, copper, and iron levels in rats of 3 age groups exposed to aluminium in drinking water. The doses used were 50 and 100 mg/kg b.w./day of aluminium nitrate with added citrate at 355 and 710 mg/kg/day, respectively, to increase absorption of the metal. The ages at which studies were initiated were at 21 days, 8 months, and 16 months. Studies were terminated after 6.5 months of exposure and the levels of each trace element were measured in liver, bone, testes, spleen, kidney, and brain. The authors did not report the levels of aluminium in serum or bone, so it is difficult to assess the relative exposures of the animals. The amounts of aluminium in the water were roughly 10,000 times the average exposure level of humans and the addition of citrate would likely have increased absorption. Under these conditions, for each trace element, there were statistically significant effects of aluminium for multiple tissues; however, there were not always dose-dependent responses and the response of young vs. older animals sometimes differed. Among the most striking changes were the 25 to 50% reductions in calcium levels in kidney and brain of the mid and old age groups given the highest dose of aluminium. In young animals given the highest dose, however, the levels of calcium in these organs were ~2-fold higher than in those of controls. The testes of older animals also showed 2 to 3-fold increases in calcium levels. The data on copper levels showed significant fluctuations among age and treatment groups. The most robust finding was that, in young animals exposed to both doses of aluminium, the level of copper was reduced 30 to 50% in both kidney and brain. However, in older animals there were no changes in these organs. Magnesium, manganese, iron, and zinc levels fluctuated less among age and treatment groups in the various tissues, and there was less definitive evidence that aluminium caused consistent changes the amount of these elements in the tissues examined. There was evidence of a robust and consistent effect on levels of manganese in spleen, where, in all age groups, the highest dose of aluminium correlated with the highest tissue levels of manganese. There was also evidence of an effect on iron levels in spleen of young animals given the highest dose (30% increase in iron) and in kidney of older animals (25% reduction).

The authors also examined urinary excretion of trace metals at 2 time points within the treatment (at 3 and 6.5 months). The most robust and consistent changes in excretion were noted for zinc with a 3-fold reduction in excreted zinc at both doses in middle-aged group. In all other groups and for all other elements, the changes were either far less robust or inconsistent across age and treatment groups.

Yasui & Ota (1998) examined the levels of magnesium and calcium in serum, spinal cord, and bone in rats fed diets low in calcium (3 mg/100g diet) and high in aluminium (194 mg/100g diet) as aluminium lactate. Control diets contained 1250 mg/100g diet calcium and 10 mg/100g diet aluminium. After 60 days of exposure, the levels of aluminium in serum increased from 0.25 μg/dL to 1.25 μg/dL. Control diet low in calcium had no effect on aluminium levels and only slightly lowered serum levels of calcium. By contrast, there was a 2-fold reduction in serum calcium levels in animals exposed to both low calcium and high aluminium. None of the diets affected serum levels of magnesium. In spinal cord, the combined low calcium/high aluminium diet led to very modest reductions in the levels of magnesium (~10% reduction). However, in lumbar vertebra, magnesium levels were reduced by 25% on the combined diet. Diets low in calcium had no effect in either tissue. The authors did not comment on whether these diets affected weight gain.

In a study of much shorter duration (18 days) in which lower doses of aluminium were used (~270 μg/g of food), Greger et al. (1985) examined the levels of phosphorus, calcium, magnesium, iron, manganese, zinc, and copper in bone, kidney, and liver of rats exposed to aluminium in diet. Two food formulations provided trace metals at levels that were at the minimum requirement and at 2-3 times the minimum requirement. The chemical form of aluminium was varied, using aluminium palmitate, aluminium lactate, aluminium phosphate, and aluminium hydroxide. No significant differences in levels of aluminium accumulation in bone were noted among the different chemical forms of exposure (control 1.9 μg/g; treated 13-15 μg/g). The authors found no significant changes in the levels of any of the studied minerals in the tissues examined in animals exposed to aluminium by any of the methods.

Boudey et al. (1997) examined the effects of low doses of aluminium on growth rates and calcium metabolism of young weanling rats; an additional variable in the study included reductions in calcium levels. Animals were given control diets (8.4 mg Al/kg b.w. or supplemented diet (10.6 mg Al /kg b.w.), with or without altering calcium levels (7.6 g Ca /kg b.w. vs. 0.4 g Ca/kg b.w). Animals exposed to the higher dose of aluminium in calcium deficient diets weighed 40% less at the end of the 90 day study. Aluminium levels in brain, liver, and bone were approximately 3-fold higher than those of animals given diets containing normal levels of calcium and the lower dose of aluminium. The authors also noted that, in the presence of normal levels of calcium, the higher dose of aluminium caused 25 to 40% reductions in the levels of calcium in bone, liver, and brain. One conclusion of the study was that young animals may be more sensitive to the effects of aluminium if diets are deficient in calcium.

In comparison to these outcomes, Julka & Gill (1996) reported that young (100-150 g) rats exposed to very high doses of aluminium by i.p. injection (10 mg/kg b.w. per day) for 4 weeks had elevated levels of calcium (2 to 3-fold) in cortex and hippocampus of brain. Other physiological consequences related to calcium included decreased (~40%) calcium influx in isolated synaptic preparations from the aluminium-treated animals and reduced ability of calmodulin to stimulate cAMP phosphodiesterase (~25% reduction in activity).

Mahieu & Calvo (1998) examined renal function in rats exposed to aluminium by i.p. injection of aluminium hydroxide (80 mg/kg b.w., 3 times per week, for 6 months). After 6 months of exposure, aluminium levels in serum reached 800 μg/L as compared to controls (10 μg/L). These exposure levels are 400 to 800-fold higher than typically found in human serum (see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity). Deposition of aluminium in trabecular bone was also observed, indicating significant exposure levels. Similar to results of other studies, the authors reported that treated animals showed reductions in weight (25%) and reduced efficiency of nutrient utilization. No obvious loss of renal function was noted; there were significant reductions in the excretion of phosphorus (40% less) with significant increases in calcium excretion. The suggested mechanism for the effects on renal function involved loss of PTH response, either by lower release of the hormone or reduced receptor sensitivity.

In another study of renal function, Braunlich et al. (1986) reported that aluminium exposure (i.p. injection of 0.5 mg/kg b.w., 5 times weekly for 12 weeks) led to increased urine output (45%) and increased sodium excretion (57%). Notably, the dose used in these studies was more in line with what would be used to achieve 10-fold increases in aluminium loads in rodents.

Mahieu et al. (2004) examined the intestinal absorption of phosphorus and its subsequent deposition in bone in rats exposed to aluminium. Animals were injected i.p. with aluminium lactate (5.75 mg/kg b.w.) 3 times per week for 3 months. By the end of the study, serum levels of aluminium were reported to be 600 μg/dL as compared to 10 μg/dL in controls. Slight reductions in body weight (10%) at 3 weeks of treatment were noted. In serum, there were no differences in calcium or phosphorus levels between control and treated groups at any age tested. At 1, 2, and 3 months of exposure, 25 to 30% reductions in the level of phosphorus excreted in urine were noted. Small, but statistically significant, reductions in the levels of phosphorus absorbed by the intestine were also noted. Similar reductions, small but statistically significant, in calcium absorption were also reported. Slight increases in calcium urinary excretion were detected along with 10% reductions in calcium levels in bone. A significant increase in the bone accretion of phosphorous (³²P deposited/³²P absorbed) was also noted in treated animals as compared to controls (27% increase). Bone calcium was significantly (p < 0.03) reduced in treated rats (10% decrease). These findings indicate that phosphorous metabolism may be modified by aluminium through direct action on the intestine, kidney, and bone.

General metabolic effects

There are several reports that provide evidence that aluminium may have effects on multiple metabolic pathways. There have been two studies on the effects of aluminium on metabolic enzymes. Rats exposed to 100 μM aluminium chloride in drinking water (2.6 μg Al/mL) for 12 months showed 2-fold elevations in aluminium in brain with reductions in the activities of hexokinase and glucose-6-phosphate dehydrogenase (G6PDH) (73 and 80% of normal) (Cho & Joshi, 1989). Reductions in erythrocyte activities of G6PDH and glutathione reductase (82% of normal for each) were reported in rats injected i.v. with 5 mg AlCl₃/kg for 3 consecutive days and then harvested at 4 weeks post treatment (Zaman et al., 1993).

Another study examined nutritional effects of aluminium in drinking water (administered as aluminium nitrate at doses of 360, 720, 3600 mg/kg b.w./day) for 100 days (Domingo et al., 1987). Young female rats exposed to the highest dose gained 50% less weight than animals on lower doses or the controls. The largest difference in weight gain occurred in the first weeks of the study. Animals on the highest dose drank less water, consumed less food, and showed less urine and faecal output. No changes in blood uric acid, cholesterol, glucose, creatine, or urea levels were detected in any treatment group.

Gonzalez et al. (2004) reported that rats given aluminium hydroxide by i.p. injection (27 mg/kg b.w.) 3 times per week for 3 months show 25% reductions in bile flow, 39% reductions in bile salt output, 43% reductions in bile cholesterol output, and 38% reductions in total bile protein output. These effects were correlated with 40% reductions in the expression of multidrug-resistance-associated protein 2, which is the main multi-specific organic anion transporter of the bile duct. Plasma concentration of aluminium in these animals reached 750 μg /L (controls 9 μg/L), indicating very significant exposure. No change in weight was detected in the treated animals.

Effects on Haematopoiesis

One of the most common abnormalities associated with renal failure and haemodialysis is anaemia. Many patients with this disease receive high doses of hydroxyl aluminium gel over long periods to control serum phosphorus levels. As described above in Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, the majority of aluminium found in serum is bound to Tf, which is responsible for the transport of iron. As described below, Tf and iron are crucial regulators of erythropoiesis; thus an immediate suggestion of a mechanism behind anaemia in patients afflicted with renal failure is evident that implicates aluminium.

However, in the majority of these patients, treatment with erythropoietin, a hormonal stimulator of haematopoiesis, is effective in restoring haematocrits (percentage of cells in whole blood that are red blood cells), at least partially (Eschbach et al., 2002; Sakiewicz & Paganini, 1998). In most patients, one physiological basis of anaemia stems from compromised production of erythropoietin by diseased kidneys, leading to chronic anaemia. The condition is often exacerbated by iron deficiency, which can arise from decreased red blood cell t_½, chronic loss of blood, decreased uptake, and other nutritional deficiencies (Drüeke, 2001; Sakiewicz & Paganini, 1998; Winearls, 1998). In the majority of haemodialysis patients, treatments with erythropoietin and iron supplements are sufficient to raise haematocrits to levels >30% of normal, a level that alleviates most symptoms of anaemia. Although some patients are hyporesponsive to erythropoietin, the levels of iron-saturated Tf and aluminium do not provide a correlative explanation (Eschbach et al., 2002); unresponsive patients do not have higher serum levels of aluminium or lower levels of iron saturated Tf. Thus, in patients who are exposed to high levels of aluminium, the physiological basis for anaemia is complex and involves, at least in part, a loss of hormonal stimulation. However, as outlined below, there are both cell culture and animal model data to indicate that aluminium has the capacity of perturb haematopoiesis and thus effects on this system should be considered in assessing aluminium toxicity.

In vivo animal studies

The effects of aluminium on haematopoiesis have been investigated in various animal models. In these studies, aluminium exposure has been accomplished by both oral and injection routes. Data from selected studies involving direct injection of aluminium salts will be summarized first. As a point of reference, the average human exposure to aluminium in drinking water, is ~ 2.3 μg/kg b.w./day, with steady-state serum levels of aluminium averaging ~2 μg/L in most individuals (see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity).

Chmielnicka et al. (1996) exposed rats to high doses of aluminium chloride (4 mg/kg b.w./day) for 21 days by i.p. injection. Although the levels of aluminium in serum were not reported, this exposure level equates to 800 μg/kg b.w./day of elemental aluminium. At 3, 7, 14, and 21 days of exposure, the authors measured platelet counts, red blood cell counts, serum levels of iron, total haemoglobin levels, and haematocrits. No significant changes in platelet or red blood cell counts were noted at any age. By 21 days of exposure, small, but statistically significant, decreases in total haemoglobin and haematocrit were reported. The most robust effect identified was a 25 to 30% reduction in the level of iron in serum.

Farina et al. (2002) exposed rats to aluminium sulphate (50 μmol/kg b.w. = 2.6 mg of aluminium/kg b.w.) through i.p. injections 5 times a week for 3 months. The levels of aluminium in serum were not reported. At the end of the exposure period, the authors reported finding several indications of toxicity to the haematopoietic system, including 32% reductions in total haemoglobin levels, 24% reductions in haematocrit, and 30% reductions in serum levels of iron. However, total iron-binding capacity of serum from exposed animals was not statistically different from controls.

I.p. injections of aluminium hydroxide (80 mg/kg b.w. – 3 times per week) have also been used as an experimental model of chronic exposure to aluminium. Mahieu et al. (2000) studied animals exposed for up to 28 weeks, while Bazzoni et al. (2005) examined animals treated for 12 weeks. Serum levels of aluminium were not reported in these studies and thus the sustained body burden is not known. The elemental aluminium content of this exposure was calculated to be ~27 mg/kg b.w. at each injection. Several haematological factors were measured in each study. Mahieu et al. (2000) noted progressive decreases in mean corpuscular volume (microcytosis) to a maximum of 28% reductions. However, minimal reductions in red blood cell counts and haematocrit were reported. Modest reductions in total haemoglobin (20% reduction) and small increases in red blood cell fragility were reported. Bazzoni et al. (2005) reported 20% reductions in total haemoglobin with 7% reductions in haematocrit. This latter study reported significant increases in red blood cells with deformed morphology (not quantified) with slight increases in fragility. Bazzoni et al. (2005) also reported that red blood cells in the treated animals were 3 times more rigid and less prone to aggregate.

Farina et al. (2005) conducted a second study of rats exposed to aluminium citrate (30 mM aluminium sulphate with 35 mM sodium citrate) in water for a total of 18 months. The daily exposure to aluminium was estimated to be 54.7 mg/kg b.w. Citrate would be expected to increase absorbance; however serum levels of aluminium were not reported. The authors reported that chronic exposure at this level led to reductions of 20% in red blood cell counts, 13% in haematocrit, 15% in total serum haemoglobin, and 40% in total levels of iron in serum. However there was no change in total iron binding capacity and no increase in red blood cell fragility. Turgut et al. (2004) exposed mice to aluminium for 3 months through drinking water containing aluminium sulphate. The estimated dose was 877 μmol/kg b.w./day = 47 mg of aluminium/kg b.w./day. Serum levels of aluminium were not reported. Reductions in serum haemoglobin (14%) and haematocrit (13%) were described. The levels of iron in serum were elevated by 59% with a small increase in the levels of Tf.

Garbossa et al. (1998a) have reported evidence that aluminium exposure may have direct effects on erythroid differentiation, which may account for the reductions in red blood cell counts and haematocrit in animals exposed to aluminium. Garbossa et al. (1998b) conducted a study in which rats were exposed to aluminium citrate at two levels by two routes for 15 weeks; 1 μmol/g b.w./day by oral gavage 5 days/week and by drinking water containing a 100 mM solution (estimated exposure 14-17 μmol/g b.w. day). Serum levels of aluminium were measured in this study with the dose given by oral gavage leading to serum aluminium levels of 2.9 μmol/L (78 μg/L), as compared to controls with 0.8 μmol/L (21.5 μg/L). The serum levels of aluminium in animals exposed through drinking water were 3.6 μmol/L (97 μg/L). The authors reported that animals given aluminium in drinking water had reductions in haematocrit (11%), increased osmotic fragility of red blood cells (20% more fragile), and 24% reductions in red blood cell t_½. At both the lower and higher dose, there were reductions in the ability of isolated bone marrow stem cells to differentiate into erythrocytes after exposure to erythropoietin (colony-forming units – erythroid). In a later study by the same group, Vittori et al. (1999) exposed rats to aluminium via drinking water containing 80 mM aluminium citrate for 8 months. At the end of the study, the serum level of aluminium in the treated rats was 205 μg/L (range 120-790 μg/L) whereas the level in control animals was 15 μg/L (range 5 to 90 μg/L). These levels are between 10 (control) and 200 (treated) times the average level of aluminium in human serum (1-2 μg/L - see Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity). At this high exposure level, very significant reductions in erythrogenesis were found ~60% reduction in erythroid colony forming units (CFU-E) and decreased uptake of iron (~30% reduction). The authors also reported significant increases in abnormal red blood cell morphology.

In vitro studies

Studies of the effects of aluminium on erythroid cell differentiation have suggested potential mechanisms by which aluminium could impact haematopoiesis. In a study by Vittori et al. (1999), direct evidence that aluminium inhibits erythroid differentiation was reported. Erythroid progenitor cells were concentrated from human blood and exposed to 100 μmol aluminium citrate (8 mg/L elemental aluminium) for 10 days and induced to differentiate with erythropoietin. The authors reported 30% reductions in CFU-E activity. However, in a similar study, Mladenovic (1988) reported that aluminium at levels of 1,035 ng/mL (1.035 mg/L) in medium did not significantly diminish CFU-E. However, if Tf was added to the medium along with aluminium, then CFU-E was diminished by 90%. If Tf was first saturated with iron, aluminium had no effect. The negative effects of Tf-aluminium were not overcome by adding excess levels of erythropoietin. One conclusion that could be drawn from this study was that a primary effect of aluminium on erythroid differentiation was mediated by competition for the Tf receptor. If excess Tf saturated with aluminium is present, then erythroid differentiation is inhibited by the binding of aluminium-metallated-Tf (aluminoxamine) to its receptor. However, the affinity of Tf for iron is 5 orders of magnitude greater than its affinity for aluminium. Hence excess aluminium alone cannot displace the iron from Tf-iron complexes that pre-exist in culture medium. Adding demetallated Tf to the medium allows aluminium to bind and thus creates the opportunity for Tf-aluminium-complexes to compete for binding to the Tf receptor. The extent to which a similar scenario may occur in vivo depends upon the levels of free Tf in serum at the time of aluminium exposure and on the level of iron in the serum. If sufficient iron is present, then aluminium binding to Tf would be less favoured. Individuals with iron deficiency could therefore be at greater risk for developing haematopoietic abnormalities upon exposure to aluminium.

Direct effects of aluminium salts on the morphology of red blood cells have been reported (Suwalsky et al., 2004; Vittori et al., 2002; Zatta et al., 1997). However, as outlined above in Effects on Laboratory Mammals and In Vitro Test Systems, Neurotoxicity, most studies of aluminium metabolism indicate that most of the aluminium, absorbed from drinking water or food, present in plasma would be bound to Tf. Thus, the amount of soluble aluminium salts that red blood cells would be exposed to is limited. However, Tf-mediated delivery of aluminium to erythroid progenitors does provide a means to expose these cells directly to the metal. Vittori et al. (2002) noted that red blood cells exposed to aluminium citrate in vitro acquired abnormal morphology and showed increased degradation of a protein involved in maintenance of cell shape (band 3 protein). Suwalsky et al. (2004) similarly reported that red blood cells exposed to aluminium fluoride in vitro developed abnormal morphology.

Irritation

Dermal application

Aluminium was applied in 0.5 mL solution as 2.5, 5, 10 and 25% aluminium chloride, or 10 or 25% aluminium chlorhydrate to 2 cm² of shaved skin on the back of TF1 strain mice for 5 consecutive days. Aluminium was similarly applied for 5 days to mice, New Zealand rabbits and white strain pigs (1 mL of solution to 4 cm²) as 10% aluminium chloride, nitrate, chlorhydrate, sulphate, hydroxide (in suspension) and basic acetate (in suspension), and as 25% aluminium chlorhydrate. The 10% solutions of aluminium chloride and nitrate produced epidermal changes that included slight to severe hyperplasia with focal ulceration, epidermal damage, dermal inflammatory cell infiltration, hyperkeratosis, acanthosis, microabscesses, aluminium deposition and abnormal keratin (Lansdown, 1973). The other 4 aluminium forms did not produce these changes. The author suggested that the aluminium ion interacted with keratin to denature it, making the stratum corneum more permeable, thereby allowing aluminium penetration through the stratum corneum to cause toxicity to the epidermal cells. Magnesium aluminium silicate was a weak primary skin irritant in rabbits (CFTA, 1970). Therefore, aluminium can produce dermal irritation that is aluminium species-dependent.

Injection

The i.p. injection of ground aluminium phosphate in rats produced extensive fibrosis in the peritoneal cavity (Evans & Zeit, 1949). Zeolite injection produced a peritoneal fibrosis (Suzuki & Kohyama, 1984). Sub-plantar injection of bentonite produced granulomas in rats (Marek & Blaha, 1985).

Aluminium is used as an adjuvant in vaccines and hyposensitization treatments for allergies. Precipitation of toxins and toxoids by alum was found to enhance their antigenic properties and reduce the rate of absorption and elimination of the antigen (Glenny et al., 1926; 1931). However, an in vitro study demonstrated that proteins present in interstitial fluid that have a larger adsorption coefficient can displace aluminium-adsorbed proteins with a smaller coefficient, with > 50% protein displacement within 15 minutes. It has been suggested that this result does not support the proposed mechanism of aluminium-enhanced antigenicity by producing a persistent depot of antigen (Heimlich et al., 1999). In addition to delaying release, aluminium strengthens the immunological properties of weak antigens to improve antibody response, as shown by addition of aluminium hydroxide to triple vaccine, 2.5 mg in the 0.5 mL injection. This reduced toxicity and increased potency in laboratory tests, and reduced reactions in children. However it produced s.c. nodules (Butler et al., 1969) (see also Effects on Humans, Effects from Non-Occupational Exposure, Irritation, Irritation after Injection of Aluminium-Adsorbed Proteins (Vaccines and Hyposensitization Regimens)). Aluminium adjuvants may also increase antigenicity by increasing the production of a local granuloma which contains antibody-producing plasma cells (White et al., 1955); by increasing the immunogenicity of the aluminium-antigen complex; by increasing antigen specific as well as total IgE antibodies (Gupta, 1998); and by stimulating the body’s immune competent cells through activation of complement, induction of eosinophilia, and activation of macrophages, lymphocytes and lymph nodes (Gupta, 1998; Hunter, 2002). Aluminium hydroxide-stimulated macrophages contain aluminium and differentiate into mature, specialized antigen-producing cells that express surface molecules similar to those seen in cultured dendritic cells, including HLA-DR^high, CD-86^high, CD14^-, and CD83 (Rimaniol et al., 2004). These effects seem to be aluminium-dependent, because aluminium, phosphate-adsorbed TT produced a prolonged synthesis of specific IgE, whereas calcium phosphate adsorbed TT did not (Vassilev, 1978). This is discussed further in Effects on Humans, Effects from Non-Occupational Exposure, Irritation, Irritation after Injection of Aluminium-Adsorbed Proteins (Vaccines and Hyposensitization Regimens).

There is other evidence for aluminium induction of the immune system. I.p. administered ovalbumin-adsorbed-aluminium silicate enhanced IgE antibody production (Fujimaki et al., 1984). Aluminium silicate is a component of fly ash. IgE and IgG production were increased by intratracheal instillation of ovalbumin adsorbed onto 0.02 or 0.2 mg of aluminium silicate, or onto alum or kaolin (Fujimaki et al., 1986).

Aluminium phosphate-precipitated toxoid prolonged the response to the toxoid in rabbits and guinea pigs, producing granulomas and antibody-containing cells 4 to 7 weeks after injection (White et al., 1955). The intradermal injection of alum-precipitated antigens produced lymph node infiltration of histiocytes (Turk & Heather, 1965). Intradermal injection of aluminium chlorhydrate into guinea pigs produced granulomas consisting of aggregations of undifferentiated macrophages that followed lymph drainage to the regional lymph nodes where they collected (Gaafar & Turk, 1970). Intradermal injection of 0.1 mL of alum-precipitated protein or aluminium hydroxide emulsion into Hartley guinea pig ears produced histiocyte infiltration within 4 days, persisting up to 28 days, that was mainly localized in close proximity to the entrance to the afferent lymphatics (Gaafar & Turk, 1970). I.m. injection of aluminium hydroxide or aluminium-adsorbed TT (0.1 mL containing 0.05 mg aluminium) in mice produced interstitial oedema and degenerative/necrotic changes and polymorphonuclear leukocyte infiltration at 24 hr. This acute reaction decreased over 72 hr when macrophage infiltration began, leading to a chronic granulomatous reaction beginning 1 to 2 weeks later that peaked at 8 and persisted for 16 to 20 weeks (Goto & Akama, 1982). Intradermal injections of 0.05, 0.5 or 5 mg aluminium chlorhydrate or zirconium aluminium glycinate (ZAG) or 0.065, 0.65 or 6.5 mg of aluminium hydroxide were given to Hartley strain guinea pigs. The aluminium hydroxide injections produced granulomas after the higher 2 doses. Those produced by 6.5 mg aluminium hydroxide persisted for longer than 28 days. These granulomas demonstrated undifferentiated macrophages and occasional giant cells around the injection. There was little evidence of infiltration and none of fibrosis. Aluminium chlorhydrate and ZAG injections produced skin thickness increases that reached a maximum at 28 days, and granulomas consisting of shredded bundles of basophilic collagen, giant cells and histiocytes which were pleomorphic, strongly hyperchromatic and sometimes phagocytic, followed by intense fibrosis (Turk & Parker, 1977). The authors suggested aluminium induced persistent nodule formation by a nonallergic direct toxic effect (foreign body reaction). I.m. injection of aluminium hydroxide-adsorbed TT into mice produced muscle fibre necrosis and eosinophil infiltration 4 days later (Walls, 1977). When aluminium chlorhydrate or ZAG were injected into the toe pad of New Zealand rabbits twice weekly for 6 weeks for a total dose of 1.4 mg, foreign-body granulomas, but not positive skin reactivity, were induced (Kang et al., 1977). Intradermal injection of 0.25 mL of vaccine containing ~ 0.3 mg aluminium into the backs and sides of rabbits produced s.c. nodules at all sites 8 days later, which persisted in most rabbits for 56 days and in 1 for 72 days (Pineau et al., 1992). The nodules had greater aluminium concentration than did normal skin. The infiltrate intensity and aluminium concentration in the nodules positively correlated. However, this is not the typical route for such injections. Aluminium hydroxide gel and suspension (aluminium concentration, 1 or 3 mg/mL) were injected i.m. (0.5 mL) or s.c. (1 mL) as 3 mg aluminium with ovalbumin into the hind legs of Hartley strain guinea pigs. The aluminium hydroxide gel produced granulomatous inflammatory reactions characterized by macrophages with foamy cytoplasm, small lymphocytes and giant cells at the injection sites. These effects persisted at least 8 weeks (Goto et al., 1997).

Aluminium compounds are the only adjuvants widely used in routine human vaccines and are the most commonly used adjuvants in veterinary vaccines. Two production methods have been used. One is the addition of alum to the antigen to form a precipitate of protein aluminate, termed alum-precipitated vaccines, which are similar in composition and physicochemical characteristics to aluminium phosphate adjuvants. The second is the addition of the antigen solution to preformed aluminium hydroxide, aluminium phosphate, mixed aluminium hydroxide and phosphate, or gamma aluminium oxide to produce aluminium-adsorbed vaccines (Clements & Griffiths, 2002; HogenEsch, 2002). Aluminium hydroxide (chemically: crystalline aluminium oxyhydroxide) and aluminium phosphate (chemically: aluminium hydroxyphosphate) are most commonly used (Hem, 2002). The former has an isoelectric point of 11.4 and is positively charged in interstitial fluid, at pH 7.4, thereby adsorbing negatively charged antigens by electrostatic attractive forces. Aluminium phosphate has an isoelectric point between 4.5 and 6, is negatively charged at neutral pH and this adsorbs positively charged antigens. Rational selection of the aluminium form is based on the charge of the protein to be adsorbed.

When aluminium starch octenylsuccinate was prepared in suspension and injected intracutaneously into a depilated site on the back of guinea pigs and rabbits, thrice weekly the first week and weekly for 7 more weeks, no abnormal skin reactions were observed (Nair & Yamarik, 2002). Subarachnoid (cisternal magna) injection of kaolin into foetal lambs and monkeys produced a fibrotic reaction and inflammatory cell response of the meninges and infiltration of kaolin-containing macrophages into the subarachnoid space (Edwards et al., 1984).

In vitro test systems

Many reports are cited in Effects on Laboratory Mammals and In Vitro Test Systems, Irritation, Inhalation Exposure / Intratrachael Exposure, and Effects on Humans, Effects from Occupational Exposure, Irritation, Inhalation Exposure of the ability of aluminium to cause pneumoconiosis, an inflammation of the lung that can progress to fibrosis, which is typically caused by inhalation of dust. There are also many reports of negative findings. The discrepancy may be due to the chemical form (species) of the inhaled aluminium, granular vs. flake-like particles. In this section, studies addressing the mechanism(s) of action of these irritant effects are discussed.

Corrin (1963a;1963b) noted that aluminium reacts with water but is not able to do so when coated with inert aluminium oxide. Granular aluminium coated with aluminium oxide is produced without use of lubricating agents, such as spindle oil and stearine. The author attributed many of the reports of negative effects to exposure to aluminium oxide or aluminium oxide-coated aluminium. The author also found that spindle oil- or stearine-coated aluminium powder reacted with water, presumably forming reactive aluminium hydroxide, whereas stearine-coated aluminium powder did not, and noted that the substitution of mineral product for stearine coincided with the onset of aluminium-induced pneumoconiosis. The explanation offered was that respirable aluminium particles that can react with water in the lung can be toxic.

Aluminium oxide was shown to release histamine from rat peritoneal mast cells (Casarett et al., 1968). The greatest effect, seen with 4.4 μm maximal median diameter particles at pH 6.8, was comparable to that see with iron oxide but less than seen with chromium oxide. This may contribute to bronchoconstriction caused by inhaled aluminium particles. Four clays containing aluminium silicate (montmorillonite, bentonite, kaolinite and erionite), caused lysis of human umbilical vein endothelia, N1E-115 neuroblastoma and ROC-1 oligodendroglial cells (Murphy et al., 1993). The authors suggested these clays might disrupt the BBB, allowing their entry into the brain. Exposure of mouse peritoneal macrophages, human type II alveolar tumour (A549) cells and Chinese hamster V79-4 lung cells to 11 minerals, including short and long fibres of attapulgite (a hydrated magnesium aluminium silicate) revealed some dusts that were non-toxic to all three cell types, some that were toxic toward mouse peritoneal macrophages, as shown by LDH release, and some that were toxic to all 3 cell types, additionally causing increased diameter of the A549 cells and reduced survival of 79-4 cells (Chamberlain et al., 1982). The results suggest differential sensitivity of cells to toxicity produced by these dusts and that long-fibred dusts are more toxic than short-fibred dusts. Exposure of rabbit AM to hydrated aluminium silicate resulted in toxicity, as evidenced by reduced viability and ATP (Hatch et al., 1985). Hydrated aluminium silicate caused concentration-dependent haemolysis of erythrocytes (Woodworth et al., 1982). Murine neuroblastoma cells exposed to hydrated aluminium silicate showed an increase in membrane electrical conductance and loss of excitable activity, as evidence of toxicity (Banin & Meiri, 1990).

Immunotoxicity/Immunosuppression

The immune system and its reactions involve interactions between various cell types and soluble mediators. These responses can be clustered into innate (natural and non-specific) and acquired (adaptive) responses for which the reaction is directed to an antigenic determinant or epitope. Non-specific responses involve effector cells such as macrophages, natural killer cells, granulocytes, and mediator systems such as the complement system. Biologically, components of the immune system are present throughout the body and interactions between the immune system and other organ systems are a normal component of immunoregulation. While a number of metals have been demonstrated to have immunotoxic properties, little has been reported for aluminium (IPCS/WHO, 1996). In addition, aluminium hydroxide has been used as an adjuvant in many human vaccines (Roit et al., 1998) (see Table 7). Vaccine efficiency is enhanced by aluminium’s capacity to absorb antigen particles forming granulomas in the point of injection. Early studies suggested altered immune responses following excess aluminium exposure. Pregnant Swiss-Webster mice exposed to aluminium (500 or 1,000 μg Al/g diet; as aluminium lactate) showed a lower resistance to bacterial Listeria monocytogene infection while non-pregnant mice showed the reverse (Yoshida et al., 1989). Acute injection of aluminium (1-10 mg/kg body weight to non-pregnant mice) resulted in a lower mortality rate to L. monocytogenes as compared to controls (Yoshida et al., 1989). In these studies, the offspring showed no differences in mortality rates. However, Golub et al. (1993) suggested that excess aluminium exposure (1000 μg Al/g diet; as aluminium lactate) in Swiss Webster mice from conception to 6 months of age resulted in alterations in immune effector cell function. Splenic lymphocytes showed a depressed response to concanavalin A. When the spleen weight was measured in mice fed 1000 ppm in the food from weaning to adulthood (4 week and 8 week exposures) no changes were detected relative to controls (Golub & Keen, 1999). Based upon these studies, Tsunoda & Sharma (1999) examined pro-inflammatory cytokine mRNA levels in the brain and immune organs of mice following a 1-month exposure to 125 ppm aluminium ammonium sulphate in the drinking water. Isolated splenic macrophages and lymphocytes showed no aluminium-related changes in the basal mRNA levels of TNFα, IL-1β, or IFNγ. However, the authors suggested that, while low and somewhat variable, basal mRNA levels for TNFα were increased in the brain of aluminium exposed mice (Tsunoda & Sharma, 1999).

Effects on the Endocrine System

With the exception of experimental studies in which the effect of aluminium exposure on the reproductive system has been examined (Effects on Laboratory Mammals and In Vitro Test Systems, Reproductive and Developmental Toxicity, Reproductive Toxicity), those designed to examine adverse effects on the endocrine system have been limited to the parathyroid response given that aluminium overload leads to PTH suppression.

Many of the PTH receptors of interest for aluminium toxicity are present in both the bone and kidney; thus, much of the data with regard to the effect of aluminium exposure on the bone discussed in Effects on Laboratory Mammals and In Vitro Test Systems, Effects On Bone is related to the alterations in serum PTH levels and calcium homeostasis. For example, Pun et al. (1990) demonstrated that lower concentrations of aluminium (4μm and 40 μM) inhibited the cyclic AMP response to PTH challenge via a decrease in PTH receptor binding in both clonal osteoblastic UMR-106 cells and in dog renal cortical membrane. Bourdeau et al. (1987) examined the endocrine response of porcine parathyroid gland tissue slices to aluminium at concentrations of 20 to 500 ng/mL. High concentrations of aluminium inhibited induced PTH release in a calcium-dependent manner. Gonzalez-Suarez et al. (2003) reported that 8 weeks of exposure to aluminium chloride (AlCl₃) (i.p. daily) reduced serum PTH levels and cell proliferation in the parathyroid glands, yet did not alter serum phosphorus levels, cell apoptosis or the calcium sensing receptor expression in young adult male Wistar rats surgically nephrectomized (7/8^th tissue excised). In a similar study, Diaz-Corte et al. (2001), surgically nephrectomized adult male Wistar rats maintained on a high dietary phosphorus intake received 2 daily ip. injections of AlCl₃ five weeks after surgery and examined 2 weeks post-injection. While significant decreases in serum PTH levels and mRNA levels for PTH in the parathyroid gland were seen in the aluminium-injected group, no differences were seen in serum calcium and phosphorus levels, renal function or body weight. Similar decreases in plasma PTH concentrations have been reported in cats with stable chronic renal failure when maintained on a diet restricted in phosphorus and protein with aluminium hydroxide included as an intestinal phosphate binding agent (Barber et al., 1999).

Genotoxicity and Mutagenicity

Aluminium compounds have produced negative results in most short-term mutagenic assays. As early as 1976, aluminium (Al₂(SO₄)₃) had been shown to decrease DNA synthesis without affecting replication fidelity (Sirover & Loeb, 1976). At concentrations from 20 μM to 150 mM, the accuracy of DNA synthesis in vitro was maintained. In a rat osteoblast cell line, UMR 106, DNA synthesis as determined by 3H-thimidine incorporation was shown to be decreased in the absence of an increase of protein synthesis as determined by ³H-leucine incorporation by exposure to 30 μM aluminium (Blair et al., 1989). Aluminium concentrations of 0.01 mM to 0.1 M, as AlCl₃, showed no potential to induce depurination of DNA as measured by the release of adenine or guanine in calf-thymus DNA (Schaaper et al., 1987). Calf-thymus DNA was used by Ahmad et al. (1996) to examine alterations in DNA binding by AlCl₃ (0.6-25 mM). These authors reported that aluminium was bound to the backbone PO₂ group and the guanine N-7 site of the G-C base pairs by the process of chelation.

In bacteria, aluminium compounds have been considered, in general, to be non-mutagenic. Aluminium showed no mutagenic activity as measured by the Rec-assay using Bacillus subtilis (Nishioka, 1975). At concentrations of 1 to 10 mM, Al₂O₃, AlCl₃, and Al₂(SO₄)₃ were also negative in the Rec-assay with the Bacillus subtilis H17 rec+ and M45 rec- strains (Kada et al., 1980; Kanematsu et al., 1980). Both aluminium and hydrated aluminium chloride (AlCl₃-6H₂O) at concentrations ranging from 10 to 100 nM per plate failed to induce reverse mutations in the Salmonella typhimurium TA102 strain as identified by his gene mutations (Marzin & Phi, 1985). Positive results have been obtained in studies using dye-alumina complexes; however these (positive) results have been attributed to impurities in the complexes rather than an effect of aluminium (Brown et al., 1979). The absence of mutagenic effects of aluminium compounds on various bacterial strains including Salmonella typhimurium and Escherichia coli demonstrated in these early studies has been supported by the findings of more recent studies (Ahn & Jeffery, 1994; Gava et al., 1989; Marzin & Phi, 1985; Olivier & Marzin, 1987; Prival et al., 1991; Seo & Lee, 1993; Shimizu et al., 1985; Venier et al., 1985; Zeiger et al., 1987). In their assessment of mutagenicity, Ahn & Jeffrey (1994) failed to find positive findings with aluminium chloride (0.3 and 3.0 ppm) using the biological endpoint of his mutation induction in the absence of S9 metabolic activation in the TA98 strain. No induction of his mutations were seen by Marzin & Phi (1985) in the 102 strain exposed to aluminium chloride hexahydrate (10-100 nmol/plate), and by Gava et al. (1989) in the TA104, TA92, TA98, TA1000 strains exposed to aluminium acetylacetonate (1.9-48 μmol/plate), aluminium lactate (1.8 - 5.5 μmol/plate), or aluminium maltolate (0.5-3.7 μmol/plate). The Salmonella typhimurium strain, TA104, showed a his gene mutation to aluminium acetylacetonate (1.8 – 48 μmol/plate); however, negative responses were reported when comparisons were made between the absence and presence of S9 metabolic activation. Shimizu et al. (1985) showed no positive response to aluminium fluoride (0.02-119 μmol/plate) in TA98, TA100, TA1535, TA1537, and TA1538 strains. Zeiger et al. (1987) showed similar results in the TA98, TA100, TA1535, and TA1537 strains following exposure to sodium aluminium silicate (0.96 - 38.5 μmol/plate). Prival et al. (1991) showed no positive responses to sodium aluminium silicate (0.36 - 108.1 μmol/plate) or calcium aluminosilicate (0.033 - 10 mg/plate) in TA98, TA100, TA1535, TA1537, and TA1538 strains. Studies with Escherichia coli (WP2 strain) have shown negative responses in trp mutations with aluminium chloride (0.8 mol) (Seo & Lee, 1993); aluminium fluoride (12 nmol-1.6 mmol/plate) (Shimizu et al., 1985); calcium aluminosilicate (0.33-10 mg/plate), and sodium aluminium silicate (0.96.1 μmol/plate) (Prival et al., 1991). Using the SOS chromotest, Raabe et al. (1993) evaluated the genotoxicity potential of 36 characteristic waste products resulting from aluminium plasma etching. While a majority of the products show some genotoxic activity, a correlation between the organic constituents and biological effect could not be established. Using the common soil bacterium Rhizobium, Octive et al. (1991) reported preliminary findings showing no change in cell survival following an 18 hr exposure to 50 mM aluminium in both RDG 2002 and NZP2037 strains. In each strain, exposure to rafampicin for 3-14 days decreased cell growth; the data suggested a slight resistance in the RDG 2002 strain exposed to aluminium.

Studies have shown that aluminium compounds can inhibit cell division and produce chromosomal aberrations in plants. The relevance of data derived from plant studies to assess the carcinogenic potential in mammalian systems has come under question given that metal salts of various carcinogenic potential can give similar results in short-term plant assays (Léonard & Gerber, 1988). An early study by Gelfant (1963) showed that mammalian cell division can be inhibited by aluminium salts; however, later studies showed that this did not translate to morphological transformation of Syrian hamster embryo cells. In these cells, AlCl₃ and Al₂(SO₄) at concentrations up to 20 μg/mL medium neither caused morphological transformation nor enhanced their transformation induced by a simian adenovirus SA7 (Casto et al., 1979; Di Paolo & Casto, 1979). In the L5178Y mouse lymphoma assay, AlCl₃, 500-620 μg/mL, did not induce forward mutations at the thymidine kinase locus (Oberly et al., 1982). Chromosome aberrations have been reported in spermatocytes of grasshoppers (Phloeoba antennata) 48 – 60 hr post aluminium chloride (10 mg/0.21 g b.w.; Manna & Parida, (1965)) and in mammalian peritoneal cells (Nashed, 1975). Aberrations in the bone marrow cells of mice injected with aluminium chloride (0.1M aluminium chloride 1 mL/30 g b.w., acute ip. dose) have also been reported in a single study (Manna & Das, 1972).

Aluminium is known to act as a cross-linking agent for various cellular filaments. Using ascites hepatoma cells from Sprague-Dawley rats, Wedrychowski et al. (1986a; 1986b) reported that AlCl₃ could serve as a stimulator for the crosslinking of chromosomal proteins. However, in an Epstein-Barr virus-transformed Burkitts human lymphoma cell line, AlCl₃ showed no ability to crosslink DNA protein at either cytotoxic or non-cytotoxic concentration levels (Costa et al., 1996). Human blood lymphocytes showed positive responses for both micronuclei formation (Migliore et al., 1999; Roy et al., 1990) and sister chromatin exchange (Roy et al., 1990) at AlCl₃ levels from 11.6 μmol/mL (Roy et al., 1990) to 500 - 4000 μM (Migliore et al., 1999). Roy et al. (1990) reported that the increase in micronuclei formation was significant only in cells obtained from adult donors, and sister chromatin exchange was significant only in cells from females. The chromatid aberrations reported in the Roy et al. (1990) study were limited to an increase in gaps and breaks and the study was done in phosphate-free media. One additional study (Alfaro Moreno et al., 1997) conducted using BALB/c mouse 3T3 cells reported anaphasic alterations as a possible step in the process of chromosomal dysfunction following exposure to Mexicali dust (98% potassium aluminium silicates and 2% sodium dioxide- 0.67 mMl/L). The work of Karlik et al. (1980a; 1980b) suggested that interactions of aluminium with DNA would be dependent upon pH.

During the 1990s, a number of investigators conducted experiments to determine the genotoxic potential of aluminium compounds administered to the whole animal. Short-term studies examining changes occurring within the first 24 - 48hr of systemic dosing with aluminium compounds showed contradictory results. Isolated DNA from the liver of male Wistar rats showed no changes in the formation of 8-hydroxydeoxyguanosine (Umemura et al., 1990) following a high dose of aluminium nitrilotriacetate complex (7 mg Al/kg [259 μmol/kg] i.p.). Takagi et al. (1990) exposed F-344 rats to 1.2% aluminium clofibrate in the diet; within 1-12 months of exposure the rats showed an elevation in hepatic peroxisomal beta-oxidation enzyme activity. One early preliminary study using high levels of aluminium chloride (0.01 - 0.1 mol/mouse, i.p.) reported the induction of chromosomal aberrations (Manna & Das, 1972). Two later published studies used high dose levels of aluminium sulphate (100 - 500 mg/kg bw (0.3 - 1.5 mmol Al/kg b.w.)) as a known inducer of either micronucleated polychromatic peripheral erythrocyte (mnPCE) formation (Roy et al., 1992) or SCEs (Dhir et al., 1993) in murine bone marrow cells. As expected, a significant increase in mnPCEs was induced 24 hrs after a second aluminium dose of 500mg/kg b.w. in Swiss albino mice. No changes were seen at the 250 mg/kg b.w. dose level (Roy et al., 1992). The work of Dhir et al. (1993) showed an induction of SCEs in bone marrow from male Swiss albino mice in a dose-related fashion 24 hr after a single dose of aluminium sulphate. The induction of SCEs was detected by bromodeoxyuridine (BrdU) pre-labelling (50 mg BrdU paraffin-coated tablet implanted subcutaenously) followed by a single dose of aluminium sulphate (100, 200, or 400 mg/kg b.w.) and a single i.p. injection of colchicine (4 mg/kg b.w.) 22 hrs later.

Studies in which in either male rats (Rattus norvegicus) or sheep were exposed to complex mixtures containing aluminium were indicative of positive changes. Bauer et al. (1995) reported mnPCE formation in male and female Wistar rats that received vacuum pump oils contaminated by waste products from a BC13/C12 aluminium plasma etching process (dose of 1000 mg/kg/day in a gavage dosing volume of 0.54 mL/kg/day). Numerous compounds and elements were identified in the waste; however, the aluminium contribution to the waste was at approximately 2000 ppb. MnPCE formation in bone marrow was not detected until the animals had been exposed for 28 days. In another mixture study, the emissions of aluminium and other ionic forms of metals from an aluminium refining plant were administered orally in distilled water for one year to sheep (Sivikova & Dianovsky, 1995). The total calculated concentration of aluminium delivered was 1.1 or 2.4 mmol Al/animal/day. A significant increase in SCEs was found in the cultured lymphocytes of the high dose group only. A mitotic delay in both dose groups was reported by the authors based upon differences in cell metaphases relative to that seen in the controls. Spotheim-Maurizot et al. (1992) reported on the antigenotoxicity of aluminium. The in vitro frequency of single- and double-stranded breaks in plasmid DNA, as induced by radiation, was significantly inhibited by low concentrations of aluminium chloride. A 50% reduction in double-stranded breaks occurred with 0.2 mM aluminium chloride. A similar reduction was seen in single-stranded breaks with 0.04 mM concentration levels. The authors speculated that the reduction was due to structural changes occurring in the DNA that prevented the access of OH⁺ radicals.

Carcinogenicity

Overall, experimental animal studies have failed to demonstrate carcinogenicity attributable solely to aluminium compounds (for reviews, see ATSDR, 1999; Furst, 1971; Furst & Haro, 1969; Haddow & Hornig, 1960; IPCS, 1997; Shubik & Hartwell, 1969). The data that exist for aluminium compounds support the relevance of the physical characteristics of aluminium in relation to the adverse endpoint under study as was suggested by Krueger et al. (1984). As an example, in a very early study, O’Gara & Brown (1967) reported an increase in sarcomas (8 out of 18 rats) in NIH black rats implanted s.c. with 0.05 mm thick aluminium foil. However, intrapleural or i.p. administration of 3.5 μm diameter aluminium fibres to rats showed no indication of carcinogenicity (Pigott & Ishmael, 1981; 1992; Stanton, 1974). It has been proposed that the results of these studies support the hypothesis put forth by Bischoff & Gyson (1964) that the dimension of the implant rather than chemical composition is related to carcinogenicity (for review, see Krueger et al. (1984)). A similar pattern is evident in inhalation studies. Steinhagen et al. (1978) reported a dose-related increase in lung lesions in both rats (Fischer 344 males and females) and guinea pigs (Hartley) following inhalation of aluminium chlorhydrate for six months. Fifty-percent of the animals showed lesions following exposure to 2.5 mg/Al/m³ increasing to 100% of animals that had been exposed to 25 mg Al/m³. In guinea pigs, approximately 10% showed an increase in AM at a low dose of 0.25 mg Al/m³. The pathology was characterized by an increase in mononuclear inflammatory cells and large macrophages in alveoli around the termination of air passageways. These data suggest a progression in the inflammatory response of the lungs with aluminium chlorohydrate exposure that may be related to the tissue response to foreign bodies and the associated irritant response. A later study showed no increased tumour incidence in rats (male and female Wistar) following inhalation of alumina fibres (2.18 or 2.45 mg/m₃; Al₂O₃ and approx. 4% silica) of a similar diameter for 86 weeks (Pigott & Ishmael, 1981). When Al₂O₃ dust was intratracheally instilled in hamsters, Stenback et al. (1973) concluded that Al₂O₃ was not carcinogenic for the respiratory system. Of two additional early studies Schroeder & Mitchner (1975a) reported an increased incidence in gross tumours of approximately 52% as compared to 16% in control male Long-Evans rats exposed to aluminium (KAl(SO₄)₂) at a concentration of 5 ppm in drinking water for approximately 2-2.5 yr. This was observed in males only. In the second study, no increase in the incidence of neoplasms in either male or female Wistar rats exposed to aluminium phosphide/ammonium carbamate in the diet for 2 years was reported (Hackenberg,1972).

A limited number of studies have been conducted in mice. Two studies examined the effects of oral exposure to aluminium potassium sulphate in male and female mice. Schroeder & Mitchener (1975b) reported an increase in the incidence of gross tumours in female Swiss Webster mice (46 vs. 30%) following exposure to 5 ppm aluminium in drinking water (1.2 mg/kg b.w./day for 2-2.5 years). A more recent study by Oneda et al. (1994) showed no increase in the incidence of gross tumours, neoplastic lesions, or other proliferative lesions in B6C3F₁ mice following dietary exposure to up to 979 mg Al/kg/day for 20-24 months. Interestingly, in all groups, the incidence of spontaneous hepatocellular carcinoma was significantly decreased in the females. This was also significantly decreased in the high-dose group males (5.5 vs. 20.5% in controls) and the incidence of myocardial eosinophilic cytoplasm showed a dose-dependent decrease with aluminium exposure. When the route of delivery was changed and an i.p. injection of Al₂O₃ was given in male and female mice at 2 and 3 months of age, examination at the end of normal lifespan showed an increase in mesothelioma in the peritoneum (Frash et al., 1992).

Aluminium compounds have been proposed as possible chemotherapeutic agents. Experimental animal studies showed that aluminium nitrate (50 - 400 mg/kg; optimal dose: 150mg/kg b.w.) could significantly reduce the growth of i.p. transplanted Walker 256 carcinosarcomas in female Sprague-Dawley rats (Adamson et al., 1975; Hart & Adamson, 1971). A similar effect was not evident against P388 leukaemia cells, L1210 leukaemia, K1964 leukaemia, YPC-1 plasma cells, or Ehrlich ascites carcinoma cells. These effects may be related to differential uptake of aluminium by these cells (Adamson et al., 1975) or, alternatively, by direct alterations of the immune response (Pauwels et al., 1979). Pretreatment with a s.c. injection of 0.1 mL 0.2% aluminium chloride (AlCl₃ - 6H₂O) reduced the number of developing nodules induced by dimethyl nitrosamine (Yamane & Ohtawa, 1979).

Effects from Occupational Exposure

Irritation

One mineral form of aluminium oxide, a hexagonally, very closely packed form, corundum, is extensively used as an abrasive and polisher in sandpaper, emery, etc. This can cause abrasion of the skin, eyes or any other tissue against which it is rubbed.

Skin and eye

Foreign body entry into the eye of material containing bentonite resulted in anterior segment inflammation. Based on a similar response to bentonite, introduction into animal eyes, it was concluded that bentonite was the causative agent (Austin & Doughman, 1980).

Inhalation exposure

Several lung diseases have been associated with work in the aluminium industry: pulmonary fibrosis, alveolar proteinosis, granulomatosis, chronic obstructive lung disease, asthma, and lung cancer. There have been numerous reports of diffuse pulmonary fibrosis and interstitial emphysema, but without nodule formation. Some cases were fatal, attributed to inhalation of metallic aluminium dust in workers in aluminium processing or manufacturing. Most of the exposures that resulted in the more severe outcomes occurred in Germany during World War II, in industrial environments that were heavily contaminated with airborne aluminium flake powder (used in explosives), and where there was unsatisfactory ventilation due to blackout. It is estimated that the highest air levels of aluminium occurred during production of pyro powders in the 1940s when concentrations reached 50 to 100 mg/m³. This lead to a fibrotic lung disease, aluminosis (Sjögren, 2000). The case reports and studies are summarized in Table 22.

Table 22

Inhalation exposures to aluminium in the workplace – case studies.

These observations led to the conduct of studies in animals, reviewed above, to further understand the causes and conditions of the aluminium-associated pulmonary diseases. The apparently contradictory findings in some of the studies may be explained by the different types of aluminium, as noted by Corrin (1963b) (see also Effects on Laboratory Mammals and In Vitro Test Systems, Irritation, In Vitro Test Systems).

A review and historical perspective of the outcomes of these exposures to high concentrations of airborne aluminium follows. Four men who worked for 1.5 to 16 years in an aluminium powder factory demonstrated dyspnoea, lymphocytosis and lung x-rays similar to those seen in silicosis (Goralewski, 1941). Goralewski (1939; 1940; 1941; 1943) studied 700 workers who made a stamped aluminium powder that comprised 98% aluminium coated with a thin film of paraffin-like substance. Some workers had a dry cough and radiographs with focal shadows in the apical region, 4 had spontaneous pneumothorax and 1, on autopsy, had collagenous fibres enclosing phagocytes containing granular particles. Similar effects were produced in animals (Jötten & Eickhoff, 1942). It was suggested that aluminium dissolved in the lung and then co-precipitated with protein to cause this problem (Goralewski, 1947; 1948; Jäger & Jäger, 1941). It has also been suggested that the paraffinic substances were the causal agents (Perry, 1947), although Van Marwyck & Eickhoff (1950) found paraffin-coated aluminium dust produced less diffuse interstitial fibrosis than uncoated dust (see Effects on Laboratory Mammals and In Vitro Test Systems, Irritation, Inhalation Exposure).

Shaver & Riddell (1947) described a lung disease seen in 23 furnace feeders and ore bin operators, who had worked for 23 months to 15 years in a facility manufacturing the aluminium oxide abrasive, corundum. These workers were exposed to fumes containing considerable amounts of very fine aluminium oxide as well as silica and smaller quantities of many other substances. The disease was characterized by dyspnoea, sudden attacks of extreme breathlessness, pneumothorax, and diffuse, irregular, lace-like, granular shadows rapidly progressing to a non-nodular fibrosis. This became known as Shaver’s disease, a pulmonary fibrosis seen in workers in bauxite refining or exposed to finely divided aluminium powders, especially flake (pyro) powders. The fumes were shown to contain 30% silicon dioxide and 55% aluminium oxide; the silicon particle size (0.02 to 05 μm) was smaller than that associated with classical silicosis (Anon, 1956). More, similar, cases were described by Riddell (1948) who was not able to identify the cause and Wyatt & Riddell (1949) who found the majority of the fume particles to be amorphous and ≤ 0.5 μm in diameter. They believed the initial lesion was intracellular oedema with fibroblastic proliferation, followed by inflammatory cell infiltration within the thickened alveolar walls and then by collagen deposition. They hypothesized that intense exposure to amorphous aluminium dust may play a role in this disorder.

A fatal case described by Mitchell (1959) concerned exposure for 2.75 years to finely divided aluminium powder having a wide range of particle sizes, with ~ 10 mg/m³ being < 5 μm diameter (pyro powder). The powder was stearine coated and produced by a stamping machine. Mitchell et al. (1961) also described another fatal case of a worker who performed the same job for 3.5 years after the first case. These two workers were exposed to 615 and 685 mg/m³ of pyro powder and ~ 50 mg/m³ of respirable (< 10 μm) dust. Four additional cases identified among the 27 workmen of this small factory were investigated (Mitchell et al., 1961). The case described by Jordan (1961) worked for 5 years as a “flash filler”, hand filling fireworks with pyro powder, then frequently sweeping the floor, raising a considerable amount of dust. Pneumoconiosis was seen in 5 workers exposed to aluminium pyrotechnic flake powder in an aluminium powder mill (Swensson et al., 1962). One of these men died 6 years after exposure due to pulmonary oedema caused by cor pulmonale and aluminosis, another died 20 years after exposure due to respiratory insufficiency due to aluminosis, and another died 34 years later with pulmonary fibrosis. When the remaining 2 survivors were studied 40 years later they had no respiratory symptoms and had no further vital lung capacity deterioration (Sjögren et al., 1996b). McLaughlin et al. (1962) provided a frequently cited report of a 49 year old who worked for 13.5 years in the ball-mill room of an aluminium powder factory and who died with interstitial and nodular fibrosis and dementia. Two samples of dust in the workroom contained 0.94 and 1.46 mg/m³ total dust, of 71 and 60% aluminium, and 0.24 and 0.38/m³ respirable dust, of 48 and 42% aluminium. Chen et al. (1978) described an aluminium welder who they claimed to be the first reported case having pulmonary granulomatosis associated with aluminium inhalation, but not associated with aluminium processing. The case presented as a diffuse interstitial infiltrate and mild dyspnoea. Microscopic examination of material from an open lung biopsy showed extensive interstitial granulomas composed of macrophages, foreign body giant cells, and crystals containing aluminium.

Pneumoconiosis has been attributed to kaolin and montmorillonite exposure (Gough et al., 1956; Lynch & McIver, 1954; Sakula, 1961). A case of multiple kaolin-induced granulomas was reported (Herman et al., 1982) in which large amounts of kaolinite were found in a pneumoconiotic lesion of a worker exposed to kaolin. The presence of silica could not be detected (Lapenas & Gale, 1983). Pulmonary interstitial inflammation, fibrosis and granulomatosis have also been reported in those exposed to aluminium silicates (Musk et al., 1980; Sherwin et al., 1979); this was later concluded not to be caused by the aluminium silicates (Musk et al., 1988). The prevalence of simple and complicated pneumoconiosis in kaolin workers was 3.2 and 0.63%, respectively (Morgan et al., 1988). China clay workers who had a high quartz content of the dust recovered from the lung demonstrated nodular fibrosis whereas those who had a high content of kaolinite dust in the lungs had interstitial fibrosis (Wagner et al., 1986).

Desquamative interstitial pneumonia, a non-specific response to inhaled particles, was observed in an aluminium welder. Lung biopsy showed large amounts of particles containing aluminium, particularly in AM (Herbert et al., 1982).

Aluminium oxide refinery workers exposed to some low temperature transitional aluminas demonstrated little evidence of pulmonary fibrosis and pneumoconiosis. The noted adverse effects were consistent with non-specific chronic industrial bronchitis associated with excessive, prolonged exposures to nuisance dust (Townsend et al., 1985). The aluminium lung concentrations in 4 workers exposed to aluminium oxide for 10 to 37 years were 400 to 1080 mg/kg, respectively and correlated well with the extent of fibrosis (Gaffuri et al., 1985). About 1/3 of 197 workers engaged in electrolytic reduction of aluminium oxide, exposed > 3 years to dust ranging in size from 0.5 to 5 μm and in concentration from 0.4 to 3.1 mg/m³, and to gaseous fluoride at concentrations of 0.13 to 0.96 mg/m³ (below the TLV-TWA for fluorides of 2.5 mg/m³ (ACGIH, 2005), demonstrated bronchitis. X-rays indicated a slowly evolving pneumoconiosis (Clonfero et al., 1978). Similarly, the incidence of chronic bronchitis increased with duration of occupational exposure in aluminium foundry workers, to 38% in those with > 20 years exposure. Dust, fluoride and sulphate levels were half of their TLVs (Clonfero et al., 1980a). This group provided a case report of aluminium smelter workers who developed lung fibrosis or fibronodular changes, a pneumoconiosis (Clonfero et al., 1980b). An aluminium arc welder for 17 years presented with fibrotic nodules with seemingly necrotic centres in his lungs (Vallyathan et al., 1982). A 44 year old who had worked for 6 years as an aluminium rail grinder, often in an extremely dusty work environment, presented with shortness of breath, diffuse x-rays infiltrates and restrictive lung disease. Lung tissue was found to contain > 342,000,000 particles/g dry weight, of which 94.8% were pure aluminium. It was concluded that the inhaled aluminium particles caused the pulmonary alveolar proteinosis (Miller et al., 1984). Similarly, a 50 year old who had worked for 14 years in a potroom and died 3 years later of respiratory insufficiency had ~ 1,300,000,000 fibrous particles containing aluminium oxide/g dry lung and 15,000,000,000 non-fibrous Al-containing particles/g dry lung, respectively. The Al present was α Al oxide, not γ Al oxide or Al metal. This is ~ 1000-fold more aluminium in the lung than seen in the general population. The authors speculated that the fibres themselves may have played a role in aluminium-induced fibrosis/pneumoconiosis in cases of massive aluminium exposure (Gilks & Churg, 1987). Avolio et al. (1989) reported 13 cases of lung disease in workers in bauxite refining, preparation of aluminium oxide abrasives, cold grinding of aluminium and other jobs which exposed them to airborne dust. They felt their cases supported the possibility that exposure to aluminium can cause interstitial fibrosis but noted that this was associated with a relatively benign lung condition.

Population studies of aluminium smelter operators suggested minimal (Saia et al., 1981) or no fibronodular disease (Chan et al., 1983; Discher & Breitenstein, 1976) or increased mortality associated with pneumoconiosis (Gibbs, 1985). Doese et al. (1938) found no effects in 70 workers in an aluminium foundry or in 16 in a fabricating plant. However, it was noted that aluminium and aluminium bronze dusts produced upper respiratory tract lesions manifested as a head cold, irritation and atrophy of the nasal mucosa and septal perforation. Crombie et al. (1944) studied 125 workers in an aluminium stamp mill who had been exposed for 6 to 23 years and found no chest radiographic evidence of abnormalities attributable to dust inhalation. The authors concluded that grinders of aluminium airplane propellers, exposed to a mixture of 3 mg/m³ aluminium dust (1/3 between 2 and 7 μm and 2/3 > 7 μm diameter) and aluminium oxide (mostly < 7 μm diameter) and workers in a factory making aluminium powder showed no evidence of adverse effects (Hunter et al., 1944). In a study of 9 workers who had abnormal chest x-rays and a mean exposure of 25 years to aluminium oxide in the production of aluminium oxide abrasives, biopsies from 3 indicated interstitial fibrosis, an absence of silicotic nodules and aluminium levels several orders of magnitude above background. Aluminium constituted > 90% of the metals in the lung. This suggested that aluminium oxide was the cause of the pulmonary fibrosis (Jederlinic et al., 1990).

In reviewing the experimental studies and clinical reports on aluminium oxide, Dinman (1988) suggested that increased thermodynamic instability and enhanced bioreactivity occurs as particle surface area increases, contributing to pulmonary fibrogenicity, and that the thermodynamically active catalytic activity of the low temperature transitional forms also contributes to pulmonary bioreactivity. In their review of evidence for a relationship between work in the aluminium industry and lung disease, Abramson et al. (1989) concluded that pulmonary fibrosis had not been shown to be a significant problem. The difficulty of determining whether the pulmonary fibrosis and pleural abnormalities associated with working with aluminium are due to aluminium and/or asbestos was discussed (Kilburn & Warshaw, 1992).

Nemery presented a case of aluminium exposure causing a sarcoid-like lung granulomatosis and suggested that this might be evidence that aluminium causes a cell-mediated immune response (Nemery, 1990). A 32 year old aluminium polisher who was extensively exposed to metallic aluminium, aluminium oxide powders and other metals for 8 years developed severe lung fibrosis with T-lymphocyte alveolitis and sarcoid-like epitheloid granulomas which contained aluminium. Aluminium comprised ~ 1/3 and silicon ~ ¼ of the mineral species in his lung. The aluminium particles were 0.5 to 5 μm in diameter (De Vuyst et al., 1986). A case of pneumoconiosis in a stoneworker was presented with a high lymph node aluminium content that was attributed to aluminium oxide inhalation (Sunami et al., 1997). Assessment of bronchial and dermal responsiveness was conducted in 127 aluminium smelter workers. Hypersensitive reactions to pure aluminium and aluminium smelter dust were seen in some workers, but bronchial hyperresponsiveness to aluminium aerosol was not seen. The authors suggested that these changes might be have been due to an immunological disturbance in the aluminium smelter workers who had experienced long exposures to toxic compounds, mainly those of aluminium (Hosovski et al., 1998).

Observations in guinea pigs exposed to aluminium hydroxide (Gardner et al., 1944) led to issuing a warning that large amounts of aluminium inhalation may lower the resistance of patients to tuberculosis (Council on Industrial Health, 1946).

In reviewing metal-induced pulmonary toxicity, Nemery (1990) suggested that the rarely-observed pulmonary granulomatous fibrosis seen following aluminium exposure (aluminosis), is a form of induced pneumoconiosis, similar to that from exposure to beryllium and its compounds; it was also suggested that this may also be a cell-mediated immune response. This was also suggested by De Vuyst et al. (1986).

Observations that aluminium inhalation reduced the ability of quartz to produce lung fibrosis in some experimental animals lead to its application in humans. Initially a number of uncontrolled studies were conducted. Crombie et al. (1944) treated 34 silicotic coal miners with 200 to 300 inhalations of a fine aluminium powder, initially for 5 minutes, then increasing to 30 minutes. Clinical improvement was reported in 19 cases, evidenced by lessening or disappearance of shortness of breath, cough, chest pain and fatigue, whereas 6 of 9 control cases that did not receive aluminium reported progression of their disease. There were no apparent adverse effects, leading the investigators to conclude that inhalation of fine particulate aluminium powder is not harmful to human lung. Aluminium treatment was reported to produce symptomatic improvement in 135 of 143 industrial workers who had developed silicosis from exposure to siliceous dust (Hannon, 1944). Symptomatic improvement was reported in 49 non-ferrous metal workers treated with metallic aluminium. Forty-three percent of those treated with hydrated aluminium oxide reported improvement compared to 30% of those treated with metallic aluminium oxide. There was no objective measure of improvement in these cases (Bamberger, 1945). Improvement was reported in 56% of 75 silicosis patients treated with metallic aluminium powder (Johns & Petronella, 1945). In 1946, E. J. King & C. L. Sutherland (as noted in Berry, 1948) reviewed the evidence for the use of aluminium to prevent or treat silicosis and concluded that there was no definitive evidence that aluminium prevented further development or caused reversal of established disease. It was concluded, after reviewing the reports to that time, that the improvements reported by many silicotics were subjective and that there had been no adequately controlled studies (Berry, 1948). Twenty-six silicotic patients inhaled a dust containing 300,000,000 particles of hydrated aluminium oxide per cubic foot or air, equivalent to 35 mg/m³, for 20 minutes daily, 5 days weekly for 4 week periods alternating with 4 week periods of no inhalation, for up to 1 year; comparison was made with 16 controls who inhaled room air. All subjects were blinded. Subjective improvement was seen in both groups. No objective improvement was seen in those inhaling aluminium, suggesting the beneficial effects were psychological (Berry, 1948). It was noted that the observation of improved function coincided with improved dust control in these industries (Anon, 1956). Carefully planned and properly controlled studies were recommended (Kennedy, 1956). This led to an investigation that was conducted in pottery workers and coal miners of the efficacy of inhalation of 50 mg pure metallic aluminium dust, of which > 90% was respirable, for 15 minutes, 3 times weekly, for up to 3.5 years; comparison was made with inhalation of 5 mg of carbon black containing 1 mg/kg aluminium powder (Kennedy, 1956). No regression of the x-ray picture or improvement of functional capacity was seen. It was claimed that no nodulation was seen in 11 workers in a refractory plant who inhaled aluminium for 11 years, whereas this was seen in 11 of 12 controls (Hannon, 1953). Five percent of 164 workers who inhaled aluminium for 5 minutes a week in a foundry manufacturing large castings showed a reduction of respiratory function after 6 years; 9% of 125 who originally had functional incapacity deteriorated while 90% improved (Osmond, 1955).

Asthma was seen in workers in the aluminium industry, but this has been largely attributed to the presence of fluorine and substances other than aluminium. However, there are a number of reports suggesting that aluminium can cause asthma and related symptoms.

An airway obstruction called potroom asthma was seen in workers in the aluminium industry (Chan-Yeung et al., 1983; Frostad, 1936; Kongerud et al., 1994). Exposure to emissions from aluminium industry pots was thought to be the cause of bronchoconstriction among susceptible workers (Field, 1984). As the pot emissions contained aluminium oxide, carbon dusts, particulate polycyclic organics, gaseous and particulate fluorides, carbon monoxide, carbon dioxide, sulphur dioxide and nitrogen oxides, it has so far not been possible to identify the causative agent(s) of potroom asthma; however, fluoride compounds have been implicated (Söyseth et al., 1994). Determination of the contribution of fluoride, inspirable dust, sulphur dioxide, oil mist and the benzene-soluble fraction of coal tar pitch volatiles to respiratory symptoms in aluminium smelter workers suggested the first two were causative agents (Fritschi et al., 2003). Respirable sodium aluminium tetrafluoride fibres were found in the work environment of the primary aluminium industry (Gylseth et al., 1984). Fibres found in the lungs of previous workers in the aluminium industry who had pulmonary problems were similar and evidently persisted in the lung for years (Voisin et al., 1996).

Occupational airway hyperreactivity and nocturnal asthma were seen in 16 young male workers, representing 20% of workers, exposed to aluminium fluoride and in 3 exposed to aluminium sulphate in another plant. Although inhalation challenge tests were negative in 3 workers, intracutaneous injections produced positive responses in 3 of 8 workers. The authors concluded that the aluminium salts were the most likely cause of the occupational nocturnal asthma (Simonsson et al., 1985). Increased bronchial hyperreactivity in aluminium welders was attributed to the aluminium (Nielsen et al., 1993). Park et al. (1996) described a worker who had been exposed for 5 years to aluminium powder and who developed dyspnoea, cough and sputum production which began 1 hr after beginning work. These symptoms and a decreased FEV₁ were seen to a challenge with 10 mg aluminium powder and 10 mg/mL of aluminium chloride solution. The worker’s neutrophil chemotactic activity increased during the challenge. Skin tests to aluminium chloride produced no immediate reaction. The authors suggested that aluminium powder might have produced bronchoconstriction mediated by a non-immunological mechanism that could have involved neutrophils. A welder who had intermittently welded aluminium for 4 years developed symptoms of asthma 1 to 4 hr after ceasing welding (Vandenplas et al., 1998). Although skin tests with aluminium sulphate and nitrate did not show an immediate reactivity, the worker exhibited a large decrease in FEV₁ following aluminium welding. As positive skin tests to aluminium usually appear after 24 to 48 hr (see Effects on Humans, Effects from Occupational Exposure, Irritation, Dermal Exposure) the absence of immediate reactivity may not rule out hypersensitivity to aluminium. Asthmatic symptoms in this subject appeared after aluminium welding in the absence of detectable fluorides. A positive response was seen to a bronchial provocation test with aluminium chloride in a foundry worker who had 19 years exposure to molten aluminium and who experienced potroom asthma (Burge et al., 2000). An aluminium soldering flux containing potassium aluminium tetrafluoride has been shown to cause a potroom asthma reaction (Hjortsberg et al., 1994).

Dermal exposure

Four cases of contact dermatitis to dural, an alloy of 95 to 95.5% aluminium, 3.5 to 4% copper, 0.5% manganese and 0.5% magnesium, and 2 cases of contact dermatitis to aluminium were seen in 202 cases of contact dermatitis among aircraft workers (Hall, 1944). Itching from an irritant dermatitis that has been reported among potroom workers in factories using recycled aluminium was probably due to aluminium dust mixed with sulphur dioxide and traces of other metals (Johannessen & Bergan-Skar, 1980).

The use of a compressed air pistol to accidently impel newly milled narrow aluminium particles into the hand for 1.5 years of a worker resulted in erythema, hyperkeratosis, fissuring and partial desquamation of the hand. Patch testing was positive for aluminium filings in petroleum lubricant. Testing was not conducted with aluminium alone, so it is not known if this reaction was due to aluminium (Peters et al., 1998).

Cancer

The potential for carcinogenic effects from exposure specifically to aluminium or its compounds in the workplace has not been studied. However, according to IARC, processes used in the production of aluminium are carcinogenic in humans (IARC 1984; 1985; 1987; Straif et al., 2005). See below for further discussion on research papers relevant to the IARC ranking.

A synthesis of different epidemiological studies from 5 countries on the risk of cancer from occupational to aluminium is presented in Table 23. These studies show excess cancers at multiple sites, especially the bladder and the lung. In the occupational environment, certain PAHs, such as tar-pitch volatiles, which are emitted when organic matter is burned, are established carcinogens. However, uncertainty exists regarding the exposure-response relationship with respect to PAH. Two review articles have been published on epidemiological studies in which the relationship between exposure to PAH and lung and bladder cancers in occupationally exposed subjects was investigated (Armstrong et al., 2004; Mastrangelo et al., 1996).

Table 23

Epidemiological studies of occupational exposure to aluminium and cancer.

In Canada, a significant excess of mortality or morbidity from lung or bladder cancers has been observed in three studies (Armstrong et al., 1994; Gibbs, 1985; Spinelli et al., 1991). Gibbs (1985), in a study of 5,406 Canadian aluminium smelter workers, found that mortality from malignant neoplasms of the bladder and the lung was meaningfully related to the number of tar-years (an exposure index calculated for each year 1950-1977, which corresponds to a weighted sum of the total number of years in the tar-exposed occupation) and number of years of exposure (Gibbs, 1985).

Edling et al. (1987) in Sweden found no significant increase in cancer morbidity or mortality in 521 workers exposed to aluminium oxide in an abrasive-manufacturing plant and the cohort was followed up between 1958 and 1983. Selden et al. (1997) in Sweden investigated the risk of developing cancers at different sites in workers employed in secondary aluminium (scrap) smelters. They obtained significantly increased risk estimates for specific sites or organ-systems such as the rectum and anus (standardized incidence ratio (SIR) =2.13, 95% CI: 1.47-2.99), the nose and paranasal sinuses in a small number of cases (SIR= 4.70, 95% CI: 1.28-12.01), and the lung and pleura (SIR=1.49, 95% CI: 1.11-1.96). They also observed an increase in anorectal cancer, but this was not etiologically related to occupational determinants of risk. In contrast to previous findings in primary aluminium smelters, there was no overall increase in urinary bladder cancer in this study including workers from secondary aluminium smelters.

Two mortality studies on this topic have been conducted in France. In the first, Mur et al. (1987) investigated workers at 11 plants of the Aluminium Pechiney Company; the follow-up period was from 1950 to 1976. A slight but non-significant lung cancer excess of 1.14 was found based on 37 observed deaths. The second study conducted by the French National Institute for Research and Safety by request of the Aluminium Pechiney Company was an update of the mortality study of workers in the largest plant from the previous study (Moulin et al., 2000). An improvement in the study design was made by collecting the causes of death from death certificates over the period 1968-1994. The statistical analysis, based on 7 deaths, showed a non-significant excess of mortality from bladder cancer (standardized mortality ratio (SMR) =1.7, 95% CI: 0.71-3.64) and the observed mortality from lung cancer (19 deaths) was significantly lower than expected (SMR=0.63, 95% CI: 0.38-0.98).

The most recent Norwegian studies showed no elevated risk of lung cancer even among the most heavily PAH-exposed workers (Romundstad et al., 2000a; Ronneberg et al., 1999). However, these investigators obtained inconsistent results regarding bladder cancer. Limitations in study size have been problematic in interpreting the results of these studies and the latter one (Romundstad et al., 2000b) was initiated to clarify possible associations between PAH exposure and the incidence of lung, bladder, kidney and pancreatic cancers in a combined update of 6 Norwegian aluminium cohorts, i.e., 11,103 workers including 130 with bladder and 189 with lung cancers. This study showed an association between bladder cancer and exposure to PAH, but gave no support to an association with lung, kidney or pancreatic cancers.

Higher than expected mortality from other cancers, including lymphoreticular and genitourinary malignancies (Rockette & Arena, 1983) or brain cancer (Spinelli et al., 1991) have also been reported in aluminium reduction plants but again concomitant exposure to PAHs was likely to have been involved in these plants.

Occupational exposure to aluminium occurs not only in the primary refining of this metal but also in user industries such as automobile manufacturing. For example, inhalation exposure can occur during metalworking. In a nested case-control study of 667 cases (members of the cohort (n=45,000) who died before January 1, 1985) and 3,041 matched controls, Schroeder et al. (1997) analyzed the effect of aluminium machining on the risk of lung cancer. Semiquantitative estimates, based on years of exposure, were available for additives including sulphur and biocides, for the metals, steel, iron and aluminium, and for nitrosamines. A positive association between exposure to aluminium and lung cancer mortality was observed, particularly after an exposure period of 20 years. There was no evidence of a dose-response relationship between years of exposure and the outcome. However, ORs from these models indicate there are risks from exposure to aluminium together with soluble oils (a class of machining fluid), rather than from exposure specifically to aluminium. The likelihood that Al exposure is causally related to lung cancer in this cohort is small.

Armstrong et al. (2004) has reviewed the risk of lung cancer from inhalation exposure to PAHs in different aluminium-related industries. Meta-analysis of the data from eight of these industries showed the risk of lung cancer to be equivalent to 1.16 (95% CI: 1.05-1.28) for exposure to 100 μg/m³ benzo(a)pyrene.

Several epidemiological studies have shown that workers in the aluminium industry have an increased risk of developing lung cancer or bladder cancer; however, in all these studies this risk has been attributed to exposure to the PAHs generated during aluminium production, rather than to exposures to aluminium compounds. Benke et al. (1998) recommended that more attention be given to exposures to other contaminants and agents such as aromatic amines and nitro compounds, asbestos, heat stress and magnetic fields in future studies of cancer in aluminium smelter workers.

Effects from Non-Occupational Exposure

Neurotoxicity

It has been shown that aluminium is neurotoxic, based on the literature on dialysis encephalopathy (see Effects on Humans, Subpopulations at Special Risk, Individuals with Impaired Renal Function). This section reviews studies, in chronological order, of aluminium exposure and cognitive impairment, and dementia, including AD.

Aluminium in drinking water

A number of studies have examined the effects of aluminium in drinking water on cognitive impairment and dementia including AD, as well as other neurological conditions such as Parkinson=s disease and amyotrophic lateral sclerosis. These studies are listed in Table 24.

Table 24

Summary of epidemiological studies of aluminium in drinking water and cognitive impairment, dementia and AD.

Martyn et al. (1989) conducted an ecological study of the incidence of AD in men and women aged 40-69 in 88 county districts from the records of 7 CT scanning units to which patients had been referred for diagnosis, in relation to levels of aluminium in drinking water over the previous 10 years. CT scans covered a 3-year period, either 1983-85 or 1984-86. There was a wide range of water aluminium concentrations. Rates were adjusted to compensate for different rates of referral according to the distance from the nearest CT scanning unit. Dementia was classified as probable AD (242 men, 203 women), possible AD (112 men, 109 women), cerebrovascular dementia (162 men, 111 women) and other dementia (154 men, 92 women). To examine differences in referral practices in different areas, a group of 2,936 epilepsy patients was also included. The risk of probable AD was 1.5 times higher in districts where the mean aluminium concentration exceeded 0.11 mg/L than in districts with concentrations of <0.01 mg/L. There was no evidence of a dose-response relationship where there were intermediate levels of aluminium in water. There was no evidence of a relationship between other types of dementia or epilepsy and aluminium concentrations in water. It should be noted that this study included an unusually young population in relation to risk of AD (rates increase dramatically from ages 65-69 - the oldest people in this study – to those aged 85 and over). The authors justified exclusion of older age groups in which AD is more common on the basis that younger people presenting with clinical symptoms of dementia would more likely be more aggressively assessed and more likely to have a CT scan. The study was not population-based, i.e. not all people developing AD had an equal probability of being included in the analysis, only those under age 70 receiving a CT scan. It is not known how this sample differed from all AD cases in the area studied, but it was not representative.

Flaten (1990) examined geographical associations in Norway between aluminium in drinking water and death rates (underlying or contributing cause) from dementia including AD, as well as Parkinson=s disease and amyotrophic lateral sclerosis. Water samples were collected seasonally from October 1982 to August 1983 from 384 Norwegian waterworks supplying 70.9% of the Norwegian population, covering 193 municipalities. Death rates were standardized to the 1979 Norwegian population, for 3 5-year periods (1969-73, 1974-78, 1979-1983) and for the 10-year period, 1974-83. Municipalities were grouped into low, medium and high levels of aluminium (<0.05 mg/L, 0.05-0.2 mg/L and >0.2mg/L respectively), and correlation and regression analyses were conducted. Age-adjusted rates per 100,000 population (1974-1983) grouped by aluminium concentration in water showed relative risks (RR) for dementia in males of 1.00, 1.15 and 1.32 for low, medium and high levels of aluminium, respectively; for women, the corresponding values were 1.00, 1.19 and 1.42. This relationship did not hold true for mortality from Parkinson=s disease or amyotrophic lateral sclerosis. Flaten appropriately cautioned that these associations were ecological, serving to generate hypotheses for further study. The study has the strength of being population-based. The association could be stronger than reported if there was under-reporting of dementia on death certificates; however, since both underlying and contributing causes were used, this problem would be reduced. Typically, report of dementia on death certificates is specific but not as highly sensitive; that is, where dementia is recorded, it is highly likely that the decedent had dementia, but where it is not recorded, it will have been missed in some proportion of the death records.

Frecker (1991) examined the 1985-86 dementia mortality in Newfoundland to determine if there were unusual geographic distributions of disease. Data included age, sex, birth place, place of death, cause of death and surname of parents, as well as aluminium concentration, pH and colour of water. Two areas of high aluminium levels in drinking water had different rates of dementia mortality. No conclusive evidence of family clustering was found.

Neri & Hewitt (1991) conducted a case-control analysis, using hospital discharge records of 2,344 people aged 55 or over, hospitalized for AD or presenile dementia, and 2,232 people with non-psychiatric diagnoses, matched by age and sex to the cases. Drinking water data included the level of aluminium in treated municipal water. They found an OR of 1.46 for aluminium concentrations of ≥ 0.200 mg/L compared to <0.01 mg/L. While the analysis was conducted as a case-control study, the study was really ecological in that it did not have any data on potential confounders or effect modifiers.

In the Ontario Longitudinal Study of Aging (Forbes et al., 1991), 2,000 men were followed for about 30 years and, of the 782 who remained in the study in 1991, 400 demonstrated mental impairment at age 76. Aluminium and fluoride levels in drinking water were obtained from provincial sources and allocated by place of residence of the study participants. In their first publication, Forbes et al. (1991) noted only that ‘low’ and ‘high’ referred to below and above the 50^th percentile; there was no mention of what concentrations of aluminium or fluoride were considered to be ‘low’ and ‘high’. They found that men living in areas with high fluoride were more likely to be cognitively normal; men in areas with high aluminium were less likely to be cognitively normal. The combination of high aluminium and low fluoride gave the lowest likelihood of being cognitively normal (OR=0.37); conversely, the OR for low aluminium and high fluoride was 1.5. No CIs or levels of statistical significance were given. Two further papers in 1992 and 1994 expanded on the first report (Forbes & Agwani, 1994a; Forbes et al., 1992). The median concentration of aluminium in drinking water was 0.0847 mg/L, and, for fluoride, 0.88 mg/L in treated drinking water. Drinking water levels were classified as “high” if the concentration aluminium or fluoride was greater than, or equal to, the median value, and “low” if it was less than the median value. The OR was 2.72 (p < 0.01) for subjects in areas with high concentrations of aluminium and low levels of fluoride in the drinking water, compared to those exposed to low aluminium and high fluoride concentrations. An analysis of the risk of impaired mental functioning according to pH, in addition to aluminium and fluoride, levels resulted in small numbers, which makes interpretation difficult (Forbes et al., 1994). There was a lower risk of impaired mental functioning with a neutral pH, between 7.85 and 8.05. Additional caveats include that the sample was not representative of the general population (a sample of males aged 45 in 1959), the possibility that those with mental impairment were more likely to drop out of the study, and the difficulty of accurately identifying impaired mental functioning based on a brief screening test and subjective impressions of the responses to the overall interview.

Wettstein et al. (1991) conducted a well-designed population-based study of 805 residents, aged 81 to 85, living for more than 15 years in districts with high (98 μg/L) or low (4 μg/L) aluminium concentrations in their drinking water. They measured memory skills using the Mini Mental Status test. Mnestic and naming skills did not differ between the high and low exposure groups. Urinary aluminium and serum aluminium levels were measured in 10 clinically diagnosed Alzheimer=s patients and 10 controls in both areas; no significant difference was found, adding strength to the negative findings. However, the highest concentration of aluminium in drinking water was below 100 μg/L. Perhaps it would be more accurate to say that no association was seen between memory loss (or cognitive impairment) and low and medium levels of exposure.

There have been reports of adverse neurological effects resulting from the accidental contamination of drinking water in the Camelford area in Cornwall, England in 1988 (Altmann, 1999; McMillan et al., 1993). The water aluminium concentrations recorded at the time of the contamination event ranged from 30 to 620 mg/L (Eastwood et al., 1990). McMillan et al. (1993) reported evidence of memory impairment in 10 individuals; this was not an epidemiological study, but a description of 10 cases referred by their general practitioners. Altmann et al. (1999) reported on a self-selected group of 55 adults who were considering litigation due to the alleged effects of the water contamination three years after the incident. A control group consisted of 15 siblings (all who were available) who had not lived in the Camelford area since before the water contamination incident. An additional control group of 42 similarly aged adults underwent the same psychological tests. Subjects completed the national adult reading test as a method of assessing IQ, and the Bexley Maudsley automated psychological screening to measure cognitive impairment. All affected participants complained of short term memory loss and impaired concentration. The mean pre-morbid IQ as assessed by the national adult reading test was above average. The IQ measure of the affected subjects was similar to their unaffected siblings. Affected subjects performed poorly on the symbol digit coding test as compared to the 42 unrelated controls. The symbol digit coding test is thought to be a measure of organic brain disease. The 15 unexposed siblings had significantly better symbol digit coding test results and flash-pattern differences than did their affected siblings. Many methodological issues must be considered in interpreting the results of this study. The cases were self selected and were pursuing litigation for cerebral damage claimed to be a result of the Camelford incident. It is possible that these cases may have already had unexplained symptoms and cognitive problems, and the incident served to highlight a potential cause (David, 2000). In addition, individuals who believed that they had suffered damage as a result of the incident may have been more likely to have shown errors in psychological tests which they knew were designed to detect abnormalities (Esmonde, 2000). The level of aluminium in tap water in the period following the incident and the current levels of aluminium in the tap water were not reported in the study. There was also no indication of aluminium concentrations in the drinking water of the controls (Esmonde, 2000).

Jacqmin et al. (1994) investigated the relationship between aluminium, fluorine, calcium, and pH in drinking water and the risk of cognitive impairment. This investigation used data collected from the population-based Paquid study. The study sample included 3,777 individuals 65 years of age and older. To assess their cognitive status, subjects completed the MMSE. Seventy-eight areas of distribution of drinking water were defined and aluminium, calcium, fluorine, and pH were measured in two separate surveys. A weighted mean of exposure was calculated for each water supply. Of the 3,697 subjects who competed the MMSE, 24.5% were defined as cognitively impaired. A mixed effects logistical regression was performed to estimate the association between the exposure variables and cognitive impairment while adjusting for age, sex, educational level, and occupation of the participants. Higher calcium levels in drinking water appeared to have a protective effect for cognitive impairment; the OR was 0.8 (p = 0.015, CI not provided) for a calcium concentration of 75 mg/L or greater compared to less than 75mg/L. The association between aluminium levels and cognitive impairment was not significant when pH was ignored. A weak interaction was detected between aluminium and pH levels. The association between aluminium and cognitive impairment was reported to be positive up to a pH of 7.0, and negative above that pH level. For example an OR of 1.35 was calculated for a concentration of aluminium of 100 ug/L compared to 5 ug/L at a pH of 7.0, while an OR of 0.5 was obtained at the same concentration level with a pH of 8.5. The CIs for these measures of association were not reported; thus the stability and significance of these results is not known.

In a subsequent analysis, the effect of silica in water on cognitive impairment was assessed in 3,450 participants in 71 parishes (Jacqmin-Gadda et al., 1996). A weighted mean of the measures obtained from two different surveys of each component of drinking water (pH, and concentrations of aluminium, calcium, and fluorine) was computed for each of the 71 areas. Three categories of silica levels were used (low, medium and high), defined by the first and last quartiles of the distribution. The prevalence of cognitive impairment for each of the categories was 24.0 (n=775), 21.8 (n=1,828), and 28.3% (n=847), respectively. In the regression model, in which first quartiles of exposure for the different water components were used as cut points, a significant relationship between silica, aluminium, and cognitive impairment was detected. In this model, when the level of silica and the pH were both low, subjects exposed to an aluminium concentration of 3.5 ug/L or more were more likely to have cognitive impairment compared to subjects in the lowest quartile of aluminium exposure (OR=3.94, 95% CI: 1.39-11.2). This CI is wide, indicating that the results were based on small numbers. ORs were calculated for eight different categories defined by the distributions of aluminium, pH, and silica. Low aluminium, high silica, and high pH was chosen as the reference category. An OR for cognitive impairment of 1.30 (95% CI: 0.75-2.24) was obtained for water districts with high aluminium, but low silica levels, and low pH. A lower OR (0.74, 95% CI: 0.53-1.02) was obtained for subjects living in areas with high aluminium levels, low pH, and high silica. These results support the hypothesis of a protective effect of silica in the development of cognitive impairment.

Forster et al. (1995) conducted a case control study to investigate the relationship between presenile dementia of the Alzheimer type (PDAT), family history, medical history, cigarette smoking, and exposure to aluminium. Cases had been diagnosed between 1981 and 1989 as having dementia before the age of 65 years, and were referred for specialist hospital services. Subjects were required to meet the DSM-III-R criteria for dementia (American Psychiatric Association, 1987), and the NINCDS-ADRDA criteria for AD (McKhann et al., 1984). Subjects completed the MMSE (Folstein et al., 1975); those scoring from 17 to 24 underwent another test (Geriatric Mental State Examination) to exclude any possible non-organic psychiatric conditions. Each case was paired with a control randomly selected from the population matched for age and sex. Informants of both cases and controls answered questionnaires designed to obtain information on general health, demographic factors, possible risk factors for AD, and family history of dementia. Historical data on aluminium levels in the drinking water were related to the place of residence at which the individual had lived the longest in the 10 years before the onset of symptoms of dementia. A total of 109 case-control pairs, for which there were available data on aluminium concentrations, were included in the analysis. The levels of aluminium in drinking water examined were <50 ug/L, 50-99 ug/L, 100-149 ug/L and 150 ug/L and over. No significant relationship was detected between the mean aluminium concentration in drinking water at place of residence and the onset of dementia. This study did not examine the effects of other water constituents, such as silica, fluoride, calcium or pH on the risk estimate.

Silicon-containing compounds in drinking water are thought to reduce the bioavailability of aluminium, and is therefore considered to be protective against the development of AD. Taylor et al. (1995) conducted an analysis of the same 109 case-control pairs included in the study described above in order to determine if an inverse relationship exists between dissolved silicon and aluminium in the drinking water. Water samples were collected from the tap water in the residences of the 109 cases and 109 controls. Historical data of the silicon levels in water were not available; therefore the case-control pairs were compared based on the current levels of silicon at their residences. An inverse relationship between aluminium and dissolved silicon was found. Silicon values at the residences of the cases were not significantly different than in the controls (p = 0.65). The aluminium levels in the current water samples were also not significantly different (p = 0.60). Using a threshold value for soluble silicon of 3 mg/L or above, the OR in the case-control pairs was calculated as 0.8 (95% CI: 0.34-1.83). It must be considered that the potential protective effect of silicon was measured after the onset of presenile dementia of the Alzheimer=s type, however, silicon levels in the same water source are thought to be relatively stable over time.

McLachlan et al. (1996) found an elevated risk of AD where levels of aluminium in drinking water were ≥100 μg/L, based on 296 neuropathologically confirmed cases of AD and two series of controls (n = 125 with no pathology and 170 with other neurodegenerative diseases). Municipal drinking water data were obtained from the Ontario Ministry of Environment and Energy=s Drinking Water Surveillance Program. Residential histories were collected from next of kin; one analysis was based on drinking water at the last place of residence before death (OR = 1.7, p < 0.05), and a second analysis, on the weighted residential histories (OR = 2.5, p < 0.05). This is the only study based on neuropathologically confirmed cases of AD, which is a strength. However, there were two weaknesses: only aluminium levels (not fluoride, silica or pH) in drinking water were obtained, and the analysis did not control for possible confounders or effect modifiers.

Sohn et al. (1996) conducted a cross-sectional study to investigate the relationship between aluminium, calcium and magnesium and cognitive impairment in a sample of elderly individuals living in rural communities in Korea. Two hundred and forty-nine men and 309 women aged 60 years and older participated in this study. The MMSE was used to measure cognitive status. The geographic region from which the participants were selected was divided into 66 areas, and the levels of aluminium, calcium, and magnesium were measured in two separate surveys. The mean concentrations of the 3 drinking water components, as well as the pH, were determined for each area. No significant correlation was detected between cognitive impairment and the concentrations of the drinking water components or the pH level of the drinking water. The cross-sectional design of this study is a weakness; no information was available on residential history or change in the levels of the drinking water components over time.

Martyn et al. (1997) conducted a case-control study to investigate the effects of aluminium and silicon in drinking water on the risk of AD. Subjects were identified from the computerized tomographic records of eight neuroradiology centres in England and Wales. Male patients with a possible diagnosis of dementia or primary brain cancer were selected from the files. For each possible case of dementia the next male patient in the file who was born during the same 5-year period and had a diagnosis other than dementia (brain tumour, epilepsy, or chronic disabling disease) was also selected. One hundred and six men with AD were identified for inclusion in the study. Ninety-nine men with other dementing illnesses, 226 men with brain cancer, and 441 men with other diseases of the nervous system were included in the analysis as controls. Subjects or informants completed a questionnaire which was designed to obtain information regarding residential history. The response rate was 58% due to failure to obtain permission from the general practitioner to approach the patient or next of kin (303 subjects), difficulty in tracing the patient or next of kin (112 subjects), and refusal of the patient or next of kin (209 subjects). The average aluminium and silica concentration for each address and period of residence was calculated by obtaining data from the water plants supplying each area. Aluminium levels in the water supplies ranged from 0.004 to 0.481 mg/L, with a median level of 0.043 mg/L. Cases were compared to each of the control groups to examine the relationship between aluminium and silica levels and AD. The risk of AD did not appear to increase with increasing aluminium concentration. The average levels of silicon in the water ranged from 1.2 to 23.0 mg/L and were inversely related to the average aluminium concentrations. No significant association for AD risk was found for subjects living in areas where the concentration of silica was above 6 mg/L as compared to men exposed to silica concentrations below this threshold. In summary, this study did not find an association between the risk of AD and aluminium in drinking water (at an average concentration of up to about 0.2 mg/L), and the concentration of silica in drinking water did not appear to exert a protective effect. The low response rate was a weakness of the study; its effect on the results is unknown.

Gauthier et al. (2000) conducted a study (68 cases, 68 matched controls) in the Saguenay-Lac-Saint-Jean region of Quebec to examine aluminium speciation in drinking water and genetic characteristics in relation to AD. AD diagnosis was established through a three shage process. Subjects were first screened for cognitive impairment, those screening positive underwent a standardized clinical interview by a nurse, and finally the diagnosis of dementia was established according to DSM IV criteria by two external evaluators. Municipal water data collected in 1995-1996 was used to estimate long term exposure to aluminium species in the drinking water. On average, the subjects had lived 44 years in the residence they occupied at the time of disease onset. They found that only exposure to organic monomeric aluminium at the onset of disease was associated with AD (OR = 2.678, 95% CI: 1.04-6.90) after adjusting for education, family history of AD and the apoliprotein E gene (ApoE) ε4. They did not see any effect of silica or fluoride, as has been hypothesized (Jacqmin-Gadda et al., 1996; Kraus & Forbes, 1992; Taylor et al., 1995).

Rondeau et al. (2000) conducted an analysis of the Paquid data to examine the effect of aluminium and silica in drinking water on the risk of dementia and AD. The analysis included 2,698 participants aged 65 and over living in the community in 75 civil parishes in Gironde and Dordogne in south-western France, for whom drinking water data and covariates were available. The mean length of residence in the same parish was 41 years. Over an 8-year follow-up, 253 incident cases of dementia, including 182 with AD, were identified. The effect of consumption of bottled water was examined separately (data collected at the 3-year follow-up). The RR of dementia adjusted for age, gender, educational level, place of residence and wine consumption was 1.99 (95% CI: 1.20 - 3.26) for those exposed to an aluminium concentration >0.1 mg/L. For AD the adjusted RR was 2.14 (95% CI: 1.21 - 3.80). There was an inverse association with silica (RR = 0.74, p = 0.021) (see also Effects on Humans, Effects from Non-Occupational Exposure, Modifying Factors). There was no statistically significant interaction between aluminium and silica concentrations or between aluminium and mineral water consumption.

Overall, many of the studies of aluminium in drinking water and risk of cognitive impairment, dementia and/or AD have had methodological weaknesses. Early studies were ecological, taking into account only the level of cognitive impairment, dementia, or AD in relation to aluminium in drinking water. Most of the case-control and cohort studies took few or no possible confounders or effect modifiers into account in their analyses. The exception to this is the analysis by Rondeau et al. (2000), which is by far the strongest methodologically of the epidemiological studies conducted to date. However, one limitation of this study is a lack of an intermediate exposure category, and therefore the inability to examine the possibility of a dose-response relationship.

Aluminium in antacids and antiperspirants

Graves et al. (1990) conducted a case-control study to investigate the association between exposure to aluminium, from the lifetime use of antiperspirants and antacids, and the risk of AD. Subjects diagnosed with AD were selected from one of two clinics. Selection of cases was limited to those who were married at the time of diagnosis since the spouse served as an informant. Friends of the case and/or informant were selected as controls; when this was not possible, a non-blood relative of the case was identified. The controls were within 10 years of age of the cases and did not suffer from memory loss. Case and control informants were interviewed about the subjects’ antiperspirant/deodorant use and antacid use. Subjects were defined as an antiperspirant/deodorant user if they were reported to have used these products at least once a month for one year prior to the reference year. Frequency, duration, and the name of the most common brands used were also obtained. A subject was considered to have been exposed to antacids if he or she had used antacids daily or almost daily for at least one year prior to the reference year. One hundred and thirty case-control pairs were included in this analysis. The age-adjusted OR for use of aluminium-containing antiperspirants vs. no use of antiperspirant was 1.7 (95% CI: 1.1-2.4). This component of the analysis included only 63 matched pairs as a result of missing information. There was a statistically significant dose-response relationship between the lifetime use of aluminium-containing antiperspirants and the risk of AD, adjusted for age (p = 0.01). The age-adjusted OR for antacid use vs. no use was 3.1 (95% CI: 1.3-7.5). A dose-response gradient was found for any antacid use (p = 0.009), and the adjusted OR for the highest level of exposure was 11.66 (95% CI: 1.28-106.4). However, no association was found when only aluminium-containing antacids were considered (OR=0.8, 95% CI: 0.3-2.1). There are some methodological issues which should be considered in interpreting the results of this study. The necessary use of surrogate informants may have resulted in misclassification of exposure data and a high frequency of missing information. Also the small numbers of subjects in the exposure categories, especially in the subanalyses, may have led to unstable estimates of the ORs.

Flaten et al. (1991) compared the rate of mortality from dementia (as an underlying or contributory cause) of a cohort of 4,179 patients who had surgery for gastro-duodenal ulcer disease between 1911 and 1987 to national rates. It was assumed that the majority of the patients operated on after 1963 were heavy users of antacids containing aluminium, although no precise exposure information was available. No statistically significant results were seen. The SMR for dementia was 1.10 (95% CI: 0.85 – 1.40, n=64) for all patients, and for those operated on in the period 1967-78 it was 1.25 (95% CI: 0.66 – 2.13, n=13). Limitations of the study include misclassification both in diagnosis and exposure; in addition, the observation period may have been too short to detect an effect. Thus the possibility of an effect cannot be ruled out based on this study.

The Canadian Study of Health and Aging included a prospective analysis of risk factors for AD (Lindsay et al., 2002). The baseline population of this study was comprised of 6,434 subjects aged 65 years or older in 1991. Of these, 4,615 subjects participated in the follow-up study in 1996. The analysis included 194 AD cases, and 3,894 cognitively normal controls. Detailed information regarding a variety of risk factors was obtained through a self-administered questionnaire completed by the subjects at baseline. No association was found between the regular use of antiperspirants OR=0.77 (95% CI: 0.55-1.08), or antacid use OR=0.91 (95% CI: 0.65-1.26) and the risk of AD.

In addition to investigating the relationship between PDAT and exposure to aluminium in drinking water, Forster et al. (1995) also investigated the effects of aluminium exposure from medicinal sources. In an analysis of 109 case-control pairs matched for age and sex, there was no significant association between use of antacids for 6 months or more and PDAT (OR=1.6, 95% CI: 0.77-3.51).

Exley (1998) suggested that little is known about absorption of aluminium from antiperspirants, and that more study is required. There have been no epidemiological studies showing an association between antacid consumption and AD (Doll, 1993).

Aluminium in food

Although the main source of exposure to aluminium for most people probably is food, it is believed that very little is absorbed (Marcus et al., 1992) (see also Toxicokinetics, Absorption, Studies in Humans, Oral Administration). There have only been a few epidemiological studies of the effects of aluminium in food in the general population. A small study (23 cases, 23 matched controls) reported positive results (Rogers & Simon, 1999). While they reported on some aspects of medical history, there is no indication that they considered possible differences between cases and controls in renal function or vitamin deficiencies relevant to cognitive decline. A comprehensive review of the role of diet in cognitive decline (Solfrizzi et al., 2003) referred to the possibility of a link between aluminium in the diet and AD as “controversial” (see Effects on Humans, Subpopulations at Special Risk).

Aluminium in vaccines

Verreault et al. (2001) examined the association between past exposure to conventional vaccines and the risk of AD in order to determine if changes to the immune system could be a factor in the development of this condition. The study population in this investigation consisted of subjects 65 years of age or older who participated in the Canadian Study of Health and Aging, which is a prospective cohort study of dementia. Subjects were screened for dementia at baseline, and cognitively normal subjects completed a risk factor questionnaire which included information about exposure to vaccines. Respondents were asked whether they had ever received vaccinations for tetanus, diphtheria, poliomyelitis, or influeneza. Subjects were screened again for dementia 5 years later. Multivariate logistic regression models were used to determine the association between vaccine exposure and the risk of AD, and to adjust for a variety of other suspected risk factors. One hundred and eighty three subjects, with information about past exposure to vaccines, were diagnosed with probable or possible AD during the 5 year follow up period; 3,682 subjects remained cognitively normal at follow-up. A statistically significant lower risk of AD was demonstrated for subjects who reported at least one vaccination against diptheria/tetanus (OR=0.40, 95% CI: 0.25-0.65), and at least one vaccination against poliomyelitis (OR=0.54, 95% CI: 0.30-0.97). Exposure to influenza vaccine was also associated with a lower risk of AD but did not reach statistical significance (OR= 0.81, 95% CI: 0.55-1.19). These results suggest that past exposure to vaccines may protect against the development of AD.

Health Effects

Dose response

This section identifies quantitative benchmarks of toxicity for the health effects for which there is modest or strong supporting evidence (see Evaluation of Human Health Risks, Health Effects, Hazard Identification). The discussion is organized by exposure pathway.

Exposure characterization

Human Exposure comprehensively reviews the available data quantifying human exposure to aluminium in the environment.

Ambient respirable exposures for the general population average as much as nearly 2 μg/m³ (PM_2.5 fraction) to 7 μg/m³ (PM₁₀ fraction) of aluminium in ambient dust fractions (see Table 13). Occupational exposures for this route average as much as 1 to 6 mg/m³ (see Table 15).

Table 16 summarizes estimated intake rates for food, drinking water, ambient air (followed by swallowing), antacid use, buffered aspirin use, and from antiperspirants. The exposures associated with each of the exposure routes listed span approximately five orders of magnitude. The largest exposures are associated with the use of antacids (7,200 mg/day) and aspirin (1,000 mg/day). The smallest exposures are associated with inhalation of ambient air and consumption of drinking water. Using that information, Table 26 estimates body weight normalized aluminium uptake. Differences in bioavailability for the various routes of exposure change the ranking of the exposure routes, but only to a limited degree.

The values in this table indicate that intake of aluminium from food dominates that from drinking water. Even inhalation contributes more aluminium than does consumption of drinking water. As expected, occupational exposures exceed general population exposures by more than two orders of magnitude. However, consumption of medications containing aluminium makes the greatest potential contribution to internal aluminium exposure. Contributions from these two sources exceed the contribution from food by a factor of about 100 (aspirin) or 1,000 (antacid). The contribution of anti-perspirant use to aluminium uptake could potentially be of the same order of magnitude as the contribution of diet, although the degree to which aluminium in anti-perspirants is absorbed has not been well demonstrated (see Toxicokinetics, Absorption, Studies in Humans, Dermal Exposure).

In 2004, the McLaughlin Centre for Population Health Risk Assessment at the University of Ottawa was asked by the International Aluminium Institute to conduct a comprehensive review of the potential human health risks associated with aluminium, aluminium oxide, and aluminium hydroxide. The U.S. Environmental Protection Agency agreed to co-sponsor the risk assessment. An international scientific advisory committee was formed to provide independent oversight the risk assessment. The scientific advisory committee was comprised of: Jose L Domingo, Rovira i Virgili University, Spain, Anders Glynn, Swedish National Food Administration, Vesa Riihimaki, Finnish Institute of Occupational Health, and Thomas Wisniewski, New York University School of Medicine.

The authors wish to acknowledge Ian Arnold, Eirik Nordheim, Chris Bayliss and Ed O’Hanian for providing helpful scientific background information on aluminium and Mari Golub and Wesley Harris for providing comments on the manuscript. The assistance of Nicole Boom and Nataliya Karyakina who served as research assistants, contributing to the development of background material on toxicological and epidemiological aspects is also gratefully acknowledged. Finally, Fan Mo helped to create a database of reference material, Nagarajkumar Yenugadhati provided editorial assistance, and Robert Clarke helped with organizational aspects of the work.

D. Krewski is the NSERC/SSHRC/McLaughlin Chair in Population Health Risk Assessment at the University of Ottawa.