HEALTH EFFECTS OF ALUMINUM: A CRITICAL REVIEW WITH EMPHASIS ON ALUMINUM IN DRINKING WATER

E. Nieboer. Department of Biochemistry, McMaster University, 1200 Main Street West, Hamilton, ON L8N 3Z5, Canada.
B.L. Gibson. Department of Clinical Epidemiology and Biostatistics, McMaster University, 1200 Main Street West, Hamilton, ON L8N 3Z5, Canada.
A.D. Oxman. Department of Family Medicine, McMaster University, 1200 Main Street West, Hamilton, ON L8N 3Z5, Canada.
J.R. Kramer. Department of Geology, McMaster University, 1280 Main Street West, Hamilton, ON L8S 4M1, Canada.

This is not a complete copy of the paper which appears in Environmental Reviews, Vol. 3: 29-81 (1995). This document has been edited to appear as a Topical Review with the permission of Evert Nieboer.

Introduction

Background

The possibility that aluminum exposure might be an etiological factor in Alzheimer's disease (AD) was raised by Klatzo et al. (1965) and Terry and Pena (1965) on observing neurofibrillar degeneration in rabbits following exposure of the central nervous system to aluminum salts. Aluminum exposure has been linked for some time to other health effects, especially osteomalacia (OM) (Ward et al. 1978) and encephalopathy (Alfrey et al. 1976) in patients on dialysis or total parenteral nutrition (TPN) (Klein et al. 1982a,1982b). Several epidemiological studies have provided evidence with respect to a possible link between aluminum in drinking water and dementia (Anonymous 1992). Recently, the case has been made that the evidence is strong enough to indicate that a major reduction in aluminum exposure would significantly reduce the prevalence of AD and that public policy measures to achieve this end, including guidelines and standards for the reduction of aluminum in drinking water, should be undertaken (McLachlan et al. 199 lb). The justification was based on animal and human evidence of aluminum neurotoxicity, aluminum accumulation in AD-affected brains, epidemiological studies, and some evidence of slowed progression of AD when treating patients with chelating agents. Although the general topic of aluminum and health has been extensively reviewed (e.g., Gitelman 1989), a comprehensive review that does so within the context of aluminum in drinking water is not available.

Scope of the review

The routes of human exposure to aluminum are considered first. Drinking water and other dietary sources are examined in some detailed, and occupational exposures are also reviewed. Studies of aluminum absorption and mass balance are then summarized, with emphasis on the bioavailability of aluminum in the diet, drinking water, and auxiliary sources. Issues with respect to measuring aluminum concentrations in biologic fluids are critically appraised and the speciation of aluminum in drinking water is supplemented by an indepth review (see Appendix).

Reported aluminum concentrations in body fluids and tissues, including blood, serum, plasma, urine, bone, and brain are critically evaluated to indicate which values in the literature can be considered with confidence for making comparisons between different groups of individuals. The toxicokinetics of aluminum excretion are also considered.

In reviewing the evidence for the effects of aluminum on human health, bone disease and dialysis encephalopathy (DE) are examined in some depth, as well as the recent literature on occupational exposures. The epidemiological evidence with respect to aluminum in drinking water and dementia is formally treated as a subproject of the study and thus constitutes a research overview (Oxman and Guyatt 1988). A strict statistical meta-analysis was not justified by the quality of the epidemiologic studies. Further, the aluminum-related epidemiological evidence is discussed in the context of the current status of the etiology of AD. The evidence from chelation treatment inferring possible causal roles for aluminum in disease is also examined.

Approach to the literature and information sources

Databases of the National Library of Medicine (NLM) Medical Literature Analysis and Retrieval System (MEDLARs) were searched online for references to aluminum. General searches on aluminum and analysis, tissue levels, diseases, epidemiology, or drinking water were conducted on MEDLINE for 1986 through to February 1992 and on T0XLINE and T0XLIT for 1981 to February 1992. CATLINE lists books and was searched in May 1992. MEDLINE and T0XLINE were subsequently searched at various times until August 1994 for specific aspects of aluminum toxicity such as AD, OM, encephalopathy, amyloid proteins, and food aluminum levels. Articles by specific authors known to conduct significant research on AD, aluminum measurement, or toxicity were also sought. Relevant journals were kept under surveillance for new articles throughout 1992-1994, during which time significant new work came to light. Agency documents (Ontario Ministries of Health, Labour and Environment, the U.S. Environmental Protection Agency) were also retrieved. Personal contact was made with a number of researchers to obtain preprints or unpublished documents describing their most recent work.

The analytical and toxicologic literature was critically appraised according to the guidelines outlined in Rang et al. (1992). The epidemiological literature was reviewed according to the quality of evidence guidelines and research-overview methodology outlined by Oxman and Guyatt (1988).

Summary and conclusions

Dietary and related sources of aluminum

Nonoccupational sources of aluminum include air, drinking water, diet and food additives, health-care products, and pharmaceuticals. Disregarding differences in bioavailability, food (including additives) is the major source of aluminum (estimated as 9-l4 mg/day) for individuals not using aluminum-containing medications such as antacids.

Total aluminum in drinking water varies among jurisdictions and regions within them. The annual mean levels reported for the period 1984-1992 in six regions of the Province of Ontario are typical and illustrate this: 3.9 ± 1.3 umol/L (105 ± 34 ug/L) with a range of 0.02-230 umol/L (0.5-6200 ug/L). It is clear that avoiding alum as a flocculant in water treatment does not guarantee low residual aluminum concentrations. Raw water can contain significant levels.

Aluminum is present in drinking water in a low molecular mass form that is labile (i.e., chemically reactive). A mean intake of aluminum from this source is estimated at 160 ug/day for Ontario residents based on a consumption of 1.5 L/day.

The transfer from cooking utensils or foil of aluminum to foods on contact, handling, or cooking has been estimated to be <0.1 mg/100 g for about 47% of food items and <1 mg/l00 g for 85% of foods. Acidic foods leach out the largest amounts.

Aluminum in urine and serum

It is clear that both urinary and plasma aluminum levels reflect body burden and current exposure (nonoccupational or occupational). Of the two, urinary concentration is the more responsive. Current reference intervals for healthy individuals are estimated to be 0.04-0.12 umol/L (1-3 ug/L) or lower for serum or plasma, and 0.1-0.3 umol/L (3-8 ug/L) for urine. This range reflects the absence of obvious exposures such as occupational or ingestion of aluminum-containing antacids. Much of the published baseline data exceeds the suggested reference intervals and appears to be of limited use because of analytical uncertainties, especially resulting from inadvertent contamination. Nevertheless, current analytical protocols allow reliable measurements to be made on a routine basis.

Excretion

Although there is evidence that aluminum is secreted into bile, enterohepatic circulation appears to circumvent excretion in the faeces. Consequently, urinary excretion is the major pathway and for a healthy individual without obvious exposures to aluminum it amounts to approximately 0.20 ~mol/day (5.4 ug/day). Aluminum elimination from the plasma compartment appears to be at least biphasic, with half times (t,1/2) of 8-l4 h (rapid release) and 4-100 days (slow component). The dynamics of the slow compartment appears to depend on the nature of the exposure, e.g., t 1/2= 35-50 days for welders with aluminum flake exposure and 85 days for dialysis patients on Al(OH)3 medication.

Absorption and bioavailability

Accepting that the basal urinary output is 0.20 umol/day, the net biliary excretion of aluminum into faeces is small, and deposition into tissues is insignificant (estimated to be about 5%), aluminum absorption can be accounted for if it corresponds to 3.4% of the mean drinking water intake of aluminum (taken as 0.16 mg) or 0.054% of the daily dietary intake. Gastrointestinal absorption from a normal diet supplemented with sodium lactate was estimated as 0.8% (drinking water not included, based on urinary excretion). A similar value was found for orally administered dilute solution of aluminum citrate. Citrate is known to enhance uptake in humans and animal studies indicate that lactate might as well. In one study, aluminum uptake from antacid compounds containing aluminum was in the ratio of 1:7.5:50 when swallowed with water, orange juice, or a citric acid solution, respectively. (On average, about 4 g of citrate are present in an adult's daily diet, with citrus fruits a major source.) Although tea has high levels of aluminum, typically 2-6 mg/L, it appears to be present mostly in a nonbioavailable form. Renal disease and perhaps age and AD, as well as natural chelating agents like citrate, appear to enhance aluminum absorption in the gastrointestinal tract. Animal studies suggest that uptake of aluminum is passive and occurs via paracellular pathways. Citrate appears to open this route, unlike lactate.

Not much is known about the bioavailability of dietary aluminum or aluminum-containing food additives (e.g., buffers, emulsifiers, stabilizers, leavening agents), which are ubiquitous in processed foods. On average, it is estimated that about 0.1% or less of the published daily dietary aluminum intake (9-14 mg depending on age) is absorbed.

Aluminum toxicity in chronic renal failure

Patients on dialysis for chronic renal failure tend to develop OM or aplastic bone disease, for which aluminum exposure is a probable etiologic factor. The role of aluminum is unclear, because the disease depends on a complex set of metabolic interactions, including low bone turnover or formation and altered plasma levels of calcium and parathyroid hormone. DE has a distinctive clinical course and symptomatology and is often fatal. Aluminum intoxication is most likely the cause. Systemic aluminum accumulation in tissues, not just brain and nerve tissues, occurs.

Occupational exposure to aluminum

There is no evidence to support carcinogenicity or excess mortality related to aluminum exposure in the workplace. Collectively, the studies reviewed give some support to the hypothesis that aluminum exposure leads to mild neuropsychological deficits, but confounding exposures have not been ruled out.

Serum aluminum levels and disease

OM patients have serum aluminum of <1 umol/L (>30 ug/L), while DE patients have shown levels of >3 umol/L (>80 ug/L). There appear to be no proven health effects associated with serum aluminum in the critical range at or just above the basal level in healthy individuals of 0.10-1.0 umol/L (3-30 ug/L), although neuropsychiatric symptoms have been reported for aluminum welders with serum aluminum levels in this range.

Drinking water exposure to aluminum and dementia (including AD)

In four of five critically reviewed epidemiological studies, in which AD and (or) dementia was an outcome in relation to aluminum concentration in the municipal water supply, the overall OR and (or) dose-response gradient were statistically significant, but generally low (under 2.0). The studies inevitably underestimated any true aluminum exposure effect because of misclassification bias. All of the study designs had inherent weaknesses and the results could easily be produced by confounding factors. No good evidence for alternative hypotheses to explain the results exists. The epidemiological evidence indicates that a true association between drinking water aluminum concentrations and dementia (including AD) cannot be ruled out.

Etiology of AD

Alzheimer`s disease has a distinct set of clinical symptoms, clinical course, and brain pathology at autopsy. Advanced age, a family history of dementia, and having Down's syndrome are recognized risk factors. Deposition of B-amyloid protein in neuritic plaques and phosphorylation of the tau protein in neurofibrillar tangles are features of the disease process, but the precise pathological pathways have not been determined. Based on the current evidence, a genetic-environmental model is the best etiological paradigm. In this model, multiple genes interact with multiple exogenous factors. Consequently, AD has all the earmarks of a multifactorial genetic disease. Three genetic loci for AD have been identified, one each on chromosome 14, 19, and 21. Mutations in the gene on chromosome 21 appear to affect B-amyloid production and (or) processing and are important in a small number of early-onset families. The chromosome 14 locus is expected to account for a significant fraction of early-onset cases, although its gene product has not been characterized. By contrast, the chromosome 21 gene exhibits polymorphism, with one of three alleles constituting a risk factor for AD, including sporadic cases. The gene variant in question (E4) corresponds to an apoE isoform that binds B-amyloid protein avidly and is expected to enhance its uptake by neurons and astrocytes. The dominant or codominant genes involved in the three genetic loci appear to have an age-dependent penetrance, presumably related to environmental risk factors. Head trauma is the only environmental factor for which there is good evidence; the critical review of the epidemiologic data described in the present document suggests that aluminum in drinking water constitutes another possibility. It should also be considered that very late onset AD may be a sequela of the normal aging process.

Aluminum in bone and brain tissue

Aluminum in bone is elevated in patients with renal failure who are treated with Al(OH)3 or individuals on TPN receiving aluminum-contaminated nutrient solutions. Based on the lowest levels reported, reference values in brain tissue (mostly gray matter) are around 1-3 ug/g dry weight or 0.5 ug/g wet weight. Brain aluminum levels were clearly elevated by factors of two to three in patients who died of chronic renal failure (with or without dialysis, but with Al(OH)3 treatment), especially in patients with DE (enhancement of 5- to 10-fold). There is no evidence that aluminum is consistently elevated in brain tissues or neurons of AD patients; when results are positive, enhancement relative to age-matched controls is marginal. The preferential association (two- to three-fold excess) of aluminum with DNA in neurons from AD patients compared with non-AD demented individuals is somewhat more convincing. To strengthen consistency, more studies are needed.

Evidence from chelation therapy

Treatment of OM and DE patients with the chelating agent DFO has been effective. Reduction in the body burden of aluminum is consistent with the observed improvements, although direct effects of DFO or concomitant alterations in iron metabolism cannot be discounted. Evidence is not available to link the recently reported improvement in daily living skills of AD patients treated with DFO for 2 years with the removal of aluminum.

Conclusions

A critical review of the epidemiologic evidence indicates that a real association between aluminum in drinking water and dementia (including AD) cannot be dismissed. However, all of the study designs had inherent weaknesses and the results could easily be produced by confounding variables. Since AD appears to be a multifactorial genetic disease, there is room for environmental factors. However, it is very unlikely that a single environmental factor such as aluminum in drinking water constitutes a sufficient explanation. An interpretation favoured by the authors is that aluminum is neurotoxic and may have a dementing effect independent of the pathological processes associated with AD. The positive findings in the drinking water studies would, therefore, constitute an additive effect. Iatrogenic exposure during dialysis of chronic renal patients clearly indicates that aluminum is a potent neurotoxin. Although the evidence is weaker, there a are also suggestions of mild neuropsychological disturbances in occupationally exposed individuals.

A consideration of aluminum bioavailability and mass balance (i.e., apparent dose absorbed and amount excreted) indicates that the contribution to gastrointestinal uptake by aluminum in drinking water may not be insignificant. For example, an absorption of 3.4% of the mean drinking water aluminum consumed by a resident of, for example, Ontario would account for the estimated average daily uptake of about 5 ug. This absorption rate appears attainable, as it approaches levels observed for citrate-facilitated uptake from dilute solutions. Since renal disease and perhaps age and AD, as well as natural chelating agents, appear to enhance aluminum absorption in the gastrointestinal tract, aluminum in drinking water is a public health issue. Even though it is concluded that it is unlikely to be responsible for a major burden of illness from dementia, the level of concern warrants that the guideline of 100 ug/L used in some jurisdictions, such as the Province of Ontario, become a regulatory standard so that it is not exceeded. Although drinking water with higher levels are routinely distributed, achieving the indicated objective appears technically feasible.

Sources of aluminum other than drinking water do represent significant dietary or supplementary intakes, which may constitute a special health concern for the most elderly. Prudent avoidance is recommended of products containing and practices yielding potentially bioavailable quantities of aluminum (e.g., aluminum-based antacid compounds, processed foods containing aluminum compounds, or acid foods that are cooked or stored in aluminum utensils or foils). To this end, reliable information is needed about the bioavailability and gastrointestinal absorption of, for example, aluminum from food additives and teas.

Measurement of serum and especially urinary aluminum levels is technically feasible and, if done carefully, can serve as a reliable index of exposure including dietary sources. Biological monitoring studies are recommended to help in assessments of the bioavailability of aluminum in dietary components, as well as evaluating personal exposure or differences in exposure (i) between communities with different aluminum levels in their drinking water, (ii) between groups of AD patients (e.g., late onset, early onset, sporadic) and controls, and (iii) as a component of prospective cohort studies.