Aluminum salts final content: chapter 3.2

3. Assessment of toxic under CEPA 1999

3.2 CEPA 1999 64(c): human health

3.2.1 Estimated population exposure

The average daily intake of aluminum in six age groups in Canada is estimated on the basis of concentrations measured in: (a) indoor and outdoor air (section 2.3.2.1); (b) drinking water (section 2.3.2.2.2); (c) soil (section 2.3.2.4); and (d) food (section 2.3.2.6). Table 3.1 shows the overall estimate of average daily intakes by age group and different environmental media (water, indoor air, outdoor (ambient) air, soil, and food and beverages) for total aluminum.

The average daily intake values were derived using a deterministic exposure assessment, which provides a single point estimate of intake (in this case and estimate of the mean). Probabilistic exposure assessments, on the other hand, provide information on the full range of possible intakes in the study population, and may, as well, give a more accurate estimate of mean exposure. The potential influence of a probabilistic analysis on the current assessment, with regard to the daily total aluminum intake in food, is discussed in more detail in section 3.2.1.4.

Consideration of the environmental media--drinking water, air, soil and food--in the derivation of the average daily intake is consistent with other assessments of priority substances. Daily intake of other sources of aluminum (for example, antacids, vaccines and cosmetics) is difficult to quantify for the general Canadian population, both because of the limited data on exposure and absorption, and the variability in usage within the population. Therefore, these sources were not included in the estimation of the average daily intake. Similarly, occupational exposures to aluminum, such as inhalation exposure by aluminum welders, were not considered in the estimate of average daily intake. All of these additional sources may however, constitute non-negligible exposures to aluminum, and should be considered in the qualitative evaluation of uncertainty associated with the estimate of the average daily intake.

3.2.1.1 Air
3.2.1.1.1 Estimated Average Daily Intake of Total Aluminum in Outdoor Air

The estimated average daily intake of total aluminum in airborne particles in outdoor air was determined using more than 10,000 measurements taken over the past ten years at some 50 sites in Canada. The average provincial/territorital total aluminum concentration of 0.17 µg/m3 in PM10 in Canada was used in the daily intake estimate (section 2.3.2.1.1). By age group, average daily intakes for PM10 were very low, ranging from 0.03 µg/kg bw/d for seniors to 0.1 µg/kg bw/d for young children aged six months to four years old.

3.2.1.1.2 Estimated Average Daily Intake of Total Aluminum in Indoor Air

In the case of indoor air, only measurements conducted on PM10 samples were evaluated to estimate intake since the concentration of aluminum in PM2.5 was often below the detection limit. The concentration based on the average daytime and nighttime concentrations of total aluminum is estimated to be 1.49 µg/m3 (section 2.3.2.1.2). The estimated average daily intake from indoor air is therefore higher than that from outdoor air, ranging from 0.3 µg/kg bw/d in adults and seniors to 0.8 µg/kg bw/d in young children aged six months to four years old.

3.2.1.2 Water

On the basis of data provided by municipal drinking water treatment plants from across Canada (section 2.3.2.2.2), the mean total aluminum concentration was estimated to be 101 µg/L. This estimate applies to plants that use coagulant/flocculents containing aluminum salts and secure their water supply from surface water sources. The average daily intake for each age group ranged from 2.0 µg/kg bw/d for adolescents and adults to 10.8 µg/kg bw/d for non-breastfed infants.

3.2.1.3 Soil

The mean total aluminum concentration in soil of approximately 41,000 mg/kg (section 2.3.2.4) was used to estimate the exposure of the Canadian population via soil. The average daily intake of aluminum from soil among infants was 166 µg/kg bw/d, and significantly higher in young children aged six months to four years old, who were found to have an estimated average daily intake of 268 µg/kg bw/d. For the other groups, the average daily intakes of total aluminum are progressively lower from 87 µg/kg bw/d for children aged 5 to 11 years old to 17 µg/kg bw/d for seniors.

3.2.1.4 Foods

For each age group defined in the Canadian population, the estimated mean dietary intake of total aluminum was derived using the fifth Total Diet Study completed in 2000-2002 (Dabeka 2007). Daily intakes of aluminum from food and beverages are presented in Table 3.1. For breastfed infants aged zero to six months old, the exposure to aluminum from human milk was approximately 12 µg/kg bw/d, whereas an intake of 85 µg/kg bw/d was calculated in non-breastfed infants. Among young children aged six months to four years old, the estimated mean daily intake from food was approximately 268 µg/kg bw/d. In the other groups, the mean daily intake of total aluminum ranged from 341 µg/kg bw/d in children aged 5 to 11 years old to 113 µg/kg bw/d in adults over 60 years old.

The above mean intake values of total aluminum in food were derived using a deterministic exposure assessment, which provides a single point estimate of intake but does not provide information about the full range of possible exposures within a population. The deterministic approach in this case is expected to overestimate mean estimates of exposure, in part because the aggregation of food categories inflates the contribution of less frequently consumed foods having higher levels of contamination. Further, the deterministic assessment does not take into account the day-to-day variability in the types of foods consumed by individuals.

Probabilistic exposure assessments estimate the probability of a given exposure in a population. The distribution of intakes that is generated provides more information about the full range of possible intakes in that population. Such statistical modelling can also account for intra- and interindividual variability in eating behaviours. As such, probabilistic exposure assessments, when the datasets are available to allow such assessments, are considered to provide a more accurate picture of exposure than deterministic exposure assessments.

Given that food is recognized to be the predominant source of aluminum exposure in humans, a more thorough evaluation of dietary exposure to aluminum using probabilistic techniques would be warranted (see section 3.2.7 on recommendations for research).

3.2.1.5 Overall Estimate of Exposure in the Canadian Population

The estimated mean daily intake of total aluminum was lower in breastfed than in non -breastfed infants, with levels of 179 and 262 µg/kg bw/d, respectively. The highest EDI of total aluminum was found in young children aged six months to four years old with 541 µg/kg bw/d, whereas for other age groups this intake decreased significantly to 432 µg/kg bw/d in children aged 5 to 11 years old, 293 µg/kg bw/d in adolescents, 163 µg/kg bw/d in adults aged 20 to 59 years old and finally 133 µg/kg bw/d in adults over 60 years old.

The contribution from various environmental media was evaluated for each of the age groups (Table 3.2). In young children aged six months to four years, approximately 50% of the aluminum intake was from food, 50% from ingestion of soil, and less than 1% from the ingestion of drinking water and inhaled particles. The contribution from the ingestion of food increased in the other age groups to 80% or more, whereas the contribution from soil decreased with age to 20% in children aged 5 to 11 years old and approximately 10% in the older age groups. The contribution from the ingestion of drinking water and inhaled particles is very low, at less than 2% or 0.2%, respectively for all age groups other than infants.

In infants, for the exclusively breastfed group, more than 90% of the total aluminum intake was found to be from the ingestion of soil and approximately 7% from the ingestion of human milk. For those infants who consumed infant formula and different food groups and beverages, approximately 30% of total aluminum intake was from the ingestion of food and about 63% from the ingestion of soilFootnote 25.

With respect to the three salts--aluminum chloride, aluminum nitrate, and aluminum sulphate--the only media in which the mean concentration is significantly affected by these the use of these salts is drinking water, in which aluminum sulphate or aluminum chloride may be added during the treatment process. While aluminum sulphate is permitted as an additive in some food products, this use is infrequent and would be expected to have a very minor influence on the total aluminum intake from food. The question of the relative contribution of the three salts to overall exposure to aluminum is discussed in more detail in section 3.2.5.

For those who regularly use aluminum-containing over-the-counter oral therapeutic products (e.g., pharmaceuticals such as antacids), these products represent the major source of daily aluminum intake. Based on the manufacturers' maximum recommended daily doses, EDIs of aluminum from these products may reach  approximately 31,000 µg/kg bw/d.

Table 3.1: Estimated mean daily intake of total aluminum based on Canadian data
(µg/kg bw/d)
Source of exposure InfantsFootnote a(0-6 months)
Breastfed (exclusively)
InfantsFootnote a(0-6 months)
Non-breastfed
ToddlersFootnote b
(0.5-4 years)
ChildrenFootnote c
(5-11 years)
TeensFootnote d
(12-19 years)
AdultsFootnote e
(20-59 years)
SeniorsFootnote f
(> 60 years)
Drinking waterFootnote g 0 10.8 4.57 3.59 2.04 2.14 2.25
Food and beveragesFootnote h 12.2 85.0 268 341 270 143 113
Ambient airFootnote i 0.05 0.05 0.1 0.08 0.05 0.04 0.03
Indoor airFootnote j 0.37 0.37 0.78 0.61 0.35 0.30 0.26
SoilsFootnote k 166 166 268 87 21 18 17
Total 179 262 541 432 293 163 133
Table 3.2: Contribution (%) of each source of exposure based on Canadian mean daily intake of total aluminum
Source of exposure Infants
(0-6 months)
Breastfed (exclusively)
Infants
(0-6 months)
Non-breastfed
Toddlers
(0.5-4 years)
Children
(5-11 years)
Teens
(12-19 years)
Adults
(20-59 years)
Seniors
(> 60 years)
Drinking water 0.00 4.1 0.84 0.83 0.70 1.31 1.69
Food and beverages 6.80 32.4 49.5 78.9 92.2 87.7 85.0
Ambient air 0.030 0.02 0.02 0.02 0.02 0.02 0.02
Indoor air 0.21 0.14 0.14 0.14 0.12 0.18 0.20
Soils 92.7 63.4 49.5 20.1 7.17 11.0 12.8
Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00

3.2.2 Hazard characterization

The discussion in this section focuses on the broad characterization of the types of effects of concern for the human health risk assessment of aluminum, on the basis of both human and experimental animal data. The suitability of the different sources of data for the exposure-response analysis, presented in section 3.2.3, is evaluated as well.

3.2.2.1 Effects in humans

Aluminum has been shown to produce neurotoxic effects in humans as well as bone and blood toxicity, during medical treatment in which the gastrointestinal barrier is bypassed (for example, aluminum-induced encephalopathy through dialysis treatment in patients with renal failure). There is also some epidemiological evidence for long-term cognitive impairment, in pre-term infants receiving aluminum-containing nutritional solution intravenously, and associated with occupational exposures, as discussed in section 2.4.3.1. These exposure conditions are not applicable to the general population, particularly as the exposure to aluminum generally does not occur via ingestion, and therefore human studies have not been used as a basis for characterizing the dose-response relationship for environmental exposures (see section 3.2.3). However, this evidence does support the identification of neurotoxicity and developmental neurotoxicity as endpoints of concern in the human health risk assessment for aluminum.

With respect to the conditions of exposure in the general population, the most relevant available information is provided by the epidemiological investigations into the association between exposure to aluminum through drinking water and AD and other forms of dementia (see section 2.4.3.2). The use of these findings for first identifying an endpoint of concern (that is, hazard identification), and then for evaluating the exposure-response relationship is discussed below.

The hypothesis of aluminum in drinking water as a risk factor for AD or impaired cognitive function in the elderly is controversial in the scientific community, and has important implications for public health. Hence, it is important to evaluate in detail the weight of evidence for the observed associations, in the context of traditional criteria for causality. This evaluation, for studies published prior to 1998 is presented in the Guidelines for Canadian Drinking Water Quality - Technical Documents: Aluminum (Health Canada 1998b) and in the SOS report (Environment Canada and Health Canada 2000). In the SOS report the criteria of consistency and specificity, strength, dose-response, temporality, biological plausibility, and coherence of the observed association were evaluated, and the conclusion was as follows:

"Overall ... the weight of evidence for causality for the observed associations between aluminum and Alzheimer's disease is weak, at best. There is only limited consistency in the results of the analytical epidemiological studies. While the criteria for diagnosis were generally more stringent in the studies in which there was a positive outcome, there was more consistent control of potential confounding factors in the studies in which no associations were reported. Moreover, while there is some evidence of exposure-response in the individual available studies for the reported association between aluminum and Alzheimer's disease, there is little consistency in results among the different investigations in this respect, at least based on the limited extent of comparison permitted by the available data. There are also limited data to serve as a basis of the extent to which the observed association between aluminum and Alzheimer's disease meets the criterion of temporality. Most limiting, however, in the assessment of the weight of evidence for causality of the observed association is the lack of relevant data on biological plausibility; indeed, there is no hypothesized plausible pathway from exposure to effect with measurable key events, for which sufficient investigation has been conducted to assess weight of evidence against traditional criteria of causality, such as consistency, strength, specificity, dose-response, temporal patterns, biological plausibility and coherence."

Since the publication of the State of the Science (SOS) report, a significant positive association between AD and aluminum in drinking water has been observed in the additional analysis of the data from the principle lifetime occupation and cognitive impairment in a French elderly (PAQUID) cohort in southwestern France (Rondeau et al. 2000; Rondeau et al. 2001, as described in section 2.4.3.2). While the exposure assessment in this cohort study is improved in relation to previous case-control studies, it is still limited by two factors: the quantification of the aluminum exposure of individuals from other dietary sources and the relatively narrow range of aluminum exposure in the population studied.

Recent reviews of the epidemiological literature have reiterated the limitations of the epidemiological data base, in its entirety, in regard to the causality of the occurrence of aluminum in the environment and AD, while also maintaining that the hypothesis cannot be rejected at this time (InVS-Afssa-Afssaps 2003; ATSDR 2006; JECFA 2006; Krewski et al. 2007). As a result of these limitations, JECFA (2006) and ATSDR (2006) chose not to base their regulatory values for aluminum intake on epidemiological studies.

The epidemiological data on aluminum exposure in drinking water were not used in this assessment for developing the dose-response relationship (see section 3.2.3), because of the lack of evidence for a causal relationship between aluminum in drinking water and AD, and the lack of data on total exposure to aluminum, for which food is the predominant contributor. Nonetheless, the observed associations in some studies between aluminum in drinking water and the development of AD do support further consideration of neurotoxicity as an endpoint of concern in the human health risk assessment for aluminum.

3.2.2.2 Effects in experimental animals

The scientific community has primarily focused its investigations of aluminum toxicity on the endpoints of neurotoxicity and reproductive/developmental toxicity, principally because of the evidence from human case studies and epidemiological studies indicating that these effects may be of concern. A total of 138 toxicological studies, published from 1979 to 2007, reporting on neurotoxicity and reproductive/developmental effects of oral aluminum exposure in rodents, monkeys and dogs, have been evaluated for the present assessment.

The observations of the toxic effects of aluminum may be influenced by dose, aluminum salt, dosing regimen and exposure media as well as animal species and strain, age, sex, and health status. Considering the database evaluated for this assessment, the different studies vary with respect to all of these factors, and with respect to the specific endpoints investigated. Moreover, the majority of studies compare animals exposed at a single dose to a control group. In these single-dose studies, the dose corresponding to a lowest observed effect level (LOEL) or to a no observed effect level (NOEL) is strongly influenced by the researcher's choice of administered dose.

In 2000, in its SOS report, Health Canada summarized the experimental database on aluminum toxicity as follows (Environment Canada and Health Canada 2000):

"Altered performance in a variety of neurobehavioural tests and pathological and biochemical changes to the brain have been observed in studies of the oral administration (that is, drinking water, diet, gavage) of aluminum salts to mice, rats and monkeys for varying periods of time as adults or during gestation, weaning and/or post-weaning. Interpretation of the results of a number of these studies is limited by designs that focus on testing specific hypotheses rather than examination of a range of neurotoxicity endpoints, the administration of single doses or a lack of an observed dose-response, lack of information on concentrations of aluminum or bioavailability from basal diets, the use of specific ligands to enhance accumulation of aluminum and small group sizes. Indeed, there have been no studies in which a broad range of neurological endpoints (biochemical, behavioural and histopathological) have been investigated in a protocol including multiple dose groups."

Since 2000 the database for neurological and reproductive/developmental endpoints has been considerably expanded. Yet the same limitations apply, most notably in regard to an emphasis on testing specific hypotheses rather than examining a range of neurotoxicity endpoints, testing of single doses or lack of an observed dose-response relationship, and small group sizes. There is no single study that has investigated multiple dose groups for a broad range of neurological endpointsFootnote 26.

The database does, however, provide a broad range of studies carried out by researchers from many different laboratories. Considered in its entirety, it gives evidence for neurological, neurodevelopmental and reproductive toxicity in experimental animals, including motor (for example, rotarod test and grip strength), sensory (for example, auditory startle) and cognitive effects (for example, maze learning and passive avoidance tests) as well as neuropathological (e.g., neuronal degeneration), and biochemical changes (for example, alterations in energy metabolism, trace element tissue concentrations and neurotransmission systems).

While no single or limited number of studies provides an adequate basis for characterizing the dose-response relationship, consideration of the database, as a whole, does provide a basis for approximately determining the lower range of doses at which researchers have repeatedly observed statistically significant changes in neurological, neurodevelopmental and/or reproductive endpoints in experimental animals orally exposed to aluminum salts.

3.2.3 Exposure-response analysis

The objective of the exposure-response analysis was to identify the lower range of doses for which oral exposure to aluminum has been shown to produce toxicologically significant effects in multiple studies. This lower range of doses was then considered as the level of concern for the human health risk characterization presented in section 3.2.4.

In order to characterize the lower range of doses at which oral exposure to aluminum produces effects in experimental animals, two subsets of the studies, based primarily on exposure period, were evaluated: (a) neurotoxic effects in adults following subchronic or chronic exposure (greater than 90 days); and (b) neurodevelopmental and reproductive effects in prenatal/lactation exposure studies. The studies included in these subsets are briefly described in Tables C1 and C2 (Appendix C). These two exposure periods were considered to be of greatest relevance to the evaluation of risks from long-term exposure to aluminum. Studies pertaining to other age categories (juvenile or older animals) are discussed separately in section 3.2.3.1.

These subsets include studies with highly diverse experimental conditions, notably with respect to the animal species and strain, type of aluminum salt administered, exposure vehicle as well as other aspects of the experimental methodologyFootnote 27. There is also variability in the reporting of doses. Some researchers adjust the concentration in drinking water for a constant dose in mg Al/kg bw/d and report this value (for example, Colomina et al. 2005; Colomina et al. 2002; Roig et al. 2006), while others estimate doses in terms of mg Al/kg bw/d based on measures of animal body weight and food and water intake, but keep the same concentration in the diet throughout the experiment (for example, Golub and Germann 2001b; Golub et al. 2000). In other cases, the dose is reported only as a concentration administered via diet, drinking water or gavage, and the intake in mg Al/kg bw/d has been estimated using Health Canada (1994) reference values for animal body weight and intake.

In the case of the developmental studies, the LOELs are reported as the maternal dose at the beginning of gestation. In the studies where the concentration in drinking water or the diet remained constant, this dose would generally be lower than the received dose, due to increased food and water intake during gestation and lactation. For the purpose of human health risk assessment, however, the maternal dose at the beginning of pregnancy was considered, as this provided a common point of comparison between studies.

One condition that was applied to both subsets of studies was that the experimental administered dose constitutes the principal contribution to total aluminum. As previously discussed, the concentration of aluminum in standard laboratory rodent chow may be significant, contributing approximately 10 mg Al/kg bw/d in rats and 30 mg Al/kg bw/d in mice for a typical concentration of 250 ppmFootnote 28. In the majority of studies, this base diet concentration is not measured. Base diet concentration would considerably impact the exposure-response analysis if: (a) the bioavailability of the aluminum contained in the chow was of a similar magnitude to the bioavailability of the administered aluminum; and (b) the lab chow were to contribute a large percentage of the total aluminum exposure. While it could be hypothesised that the aluminum in the lab chow, associated with ligands in the food matrix, would be less soluble and therefore less bioavailable than added aluminum, no experimental data were identified to assess the relative bioavailabilities of aluminum in lab chow and added aluminum salts. Therefore, with regard to those studies where base diet was not quantified, studies were included in the two subsets only if the administered dose (Da) likely exceeded the base diet dose (i.e., Da > 10 mg Al/kg bw/d for rats and Da > 30 mg Al/kg bw/d for mice). This approach limits the influence of the unknown base diet aluminum concentration on the exposure-response analysis, but does introduce a bias against inclusion of low dose studies in the exposure-response analysisFootnote 29. This issue is considered further in the discussion of uncertainties (section 3.2.3.2).

Other conditions applied in the compilation of these subsets were that the doses and other experimental conditions be reported unambiguously. In addition, in the subset of adult studies, studies of juvenile and older animals were not included. Studies based on these other exposure periods are discussed in section 3.2.3.1.

The LOELs of the studies meeting the conditions described above are presented graphically in Figure 3.1. In the four studies in which a LOEL for a specific endpoint is also associated with a NOEL, this is so indicated. Six other studies listed in Tables C1 and C2 found no effects for any endpoints measured (von Linstow Roloff et al. 2002; Domingo et al. 1996; Roig et al. 2006; McCormack et al. 1979; Colomina et al. 1994 and Katz et al. 1984). Consideration of these studies is important in assessing the consistency of the database and are included in the evaluation presented below. However, the studies are not included in Figure 3.1 as no corresponding LOELs for the endpoints were observed.

Considering the studies of Tables C1 and C2 collectively, the following observations concerning the exposure-response relationship for aluminum may be made:

  • There is a wide variation in reported LOELs (from 1 to 663 mg Al/kg bw/d). As previously discussed, this variation would be expected, considering the diverse experimental conditions (species, strains, aluminum salt, dosing regimes, dosing vehicle, statistical power and endpoints measured).
  • There is a predominance of single dose studies or studies where the LOEL was observed at the lowest dose. Thus, the LOELs in Figure 3.1 may be elevated with respect to the effect levels that might be observed in multiple dose studies.
  • For the 16 subchronic and chronic exposure studies for neurotoxicity in adults, the LOELs range between 1 and 500 mg Al/kg bw/d (administered and combined doses--Da and Dc--considered together). Among these studies the neurobehavioural endpoints examined included Morris water maze performance and impaired learning in the shuttle box as well as effects on reflex and motor activity. Biochemical endpoints included alterations in neurotransmission systems, increased apoptosis in the brain, alterations in synaptosomal membrane fluidity and increased lipid peroxidation in the brain.
  • For the 22 studies of exposure during gestation and lactation, the LOELs (Da and Dc) vary between 29 and 663 mg Al/kg bw/d. Neurobehavioural endpoints included grip strength, auditory startle, negative geotaxis and other reflexes, maze learning, thermal sensitivity, and motor development. The observed reproductive/developmental effects included a decrease in the number of corpora lutea and the number of implantation sites, a decrease in placental and fetal weight or reduced pup body weight, an increase in skeletal malformations, and an increase in the number of days to sexual maturity. In addition, alterations in essential element metabolism, deficits in synaptic plasticity in the hippocampus, a decrease in myelin sheath width as well as increased lipid peroxidation and a decrease in superoxide dismutase and catalase activity in the cerebrum and cerebellum were reported in developmental studies.

In order to estimate the lower range of doses at which oral exposure to aluminum produces toxicologically significant neurological or reproductive/developmental effects, the individual studies presented in Tables C1 and C2 were critically reviewed. The limitations of the collective database previously described--including the use of a single exposure dose, examination of a limited number of endpoints, lack of information on base diet aluminum concentration and small group sizes--often apply to these studies as well. Nonetheless, some of the studies provided stronger evidence than others for establishing the dose range at which neurological and reproductive/developmental effects may occur. The following discussion focuses particularly on studies documenting LOELs at the lowest doses, and evaluates the findings in relation to three issues: (a) use of a low administered dose; (b) toxicological significance of different endpoints; and (c) methodological strengths and limitations and consistency of study findings.

(a) Use of a low administered dose

Of the studies included in Figure 3.1 the lowest LOEL was observed by Huh et al. (2005). This study reported apoptosis as well as the activation of the catalytic activity of monoamine oxidases A and B in the brains of Sprague-Dawley rats at a reported combined dose of 1 mg Al/kg bw/d. The aluminum-exposed group received aluminum maltolate in drinking water over a period of 12 months.

This study reported an aluminum concentration of 11.5 ppm in the base diet. Although this is a relatively low value for laboratory chow, it does constitute an aluminum dose (0.6 mg Al/kg bw/d) of nearly twice that of the administered dose (0.38 mg Al/kg bw/d). The use of an administered dose less than the base diet dose raises the question of exposure misclassification of individual animals, as the normal variability in intake between animals may create overlap between the two groups with respect to the dose received. This is considered to be a major limitation of this study.

In spite of the extremely low administered dose, the animals receiving aluminum maltolate were found, after one year, to have approximately four times the amount of aluminum in the brain (462 ng/g) as compared to the controls (110 ng/g)Footnote 30. This finding suggested a comparable increase in both the fraction of aluminum absorbed into the bloodstream and/or the amount of aluminum distributed to the brain when the aluminum is administered as the maltolate salt. Recently, Zhou et al. (2008) found differences in aluminum oral bioavailability, which were not statistically significant, between the citrate, maltolate and fluoride salts in drinking water. The measured bioavailabilities of all the salts were low (estimated means of 0.5%, 0.61% and 0.35% for maltolate, citrate and fluoride, respectively) and approximately twice the estimated bioavailability of aluminum in food (0.1% to 0.3%, as presented in Table 2.7) as measured with the same experimental protocol. These findings suggest that while aluminum maltolate may be more bioavailable, the increase would not be sufficient to explain the results of Huh et al (2006).

In light of the uncertainty associated with the reported increased brain concentrations in the Huh et al. (2005) study, in addition to the methodological limitation of testing an administered dose that is less than the base diet dose, the study by Huh et al. (2005) was not retained for the purpose of estimating the lower range of aluminum doses at which neurological effects may be expected to occur.

Other investigations with relatively low doses over periods of 12 weeks or longer have also reported neurotoxic effects. These studies were not considered in the exposure-response analysis as the aluminum content in the laboratory chow was not reported, and thus, unlike the study by Huh et al. (2005), the relative contribution of the aluminum in the base diet could not be evaluated. However, it should be noted that LOELs ranging from 0.07 to 22 mg Al/kg bw/d (administered dose) have been associated with a significant increase in brain aluminum levels as well as significant increases in neurobehavioural or histopathological effects (refer to Kaur and Gill 2006; Kaur et al. 2006; Varner et al. 1993; Varner et al. 1994; Varner et al. 1998; Somonova et al. 1997; Fleming and Joshi 1987; Kaneko et al. 2004; and Abd-Elgahaffar et al. 2005). These results were found for different species and for different aluminum salts, administered either in drinking water or by gavage. Thus, the possibility of toxicologically significant neurological effects in this low dose range cannot be discounted. However, the difficulty of interpreting the results of these studies underlines the importance of: (a) quantifying the aluminum content in the base diet and drinking water; and (b) using a purified low-aluminum diet in studies in which the administered dose is also very low.

Among the investigations mentioned above, the study findings with respect to aluminum fluoride are of particular concern, because of the presence of both of these ions in drinking water, either naturally or through addition during the treatment process. Varner et al. (1993), Varner et al. (1994) and Varner et al. (1998), in observing increased aluminum levels in the brain associated with a low administered aluminum fluoride dose, suggested that fluoride may enhance the uptake of aluminum by the brain. At present, the scientific database is very limited with respect to the toxicokinetics and health effects specific to aluminum fluoride.

(b) Toxicological significance of different endpoints

Considering the 16 subchronic and chronic adult exposure studies, the LOELs range between 19 and 500 mg Al/kg bw/d (administered and combined doses--Da and Dc--considered together, and excluding the Huh et al. (2005) study). For neurobehavioural endpoints (Morris water maze performance, impaired learning in the shuttle box and motor activity), the LOELs of the seven relevant studies vary between 40 to 500 mg Al/kg bw/d (Da and Dc), with four studies having LOELs at Das of 40 to 70 mg Al/kg bw/d (Commissaris et al. 1982; Lal et al. 1993; Gong et al. 2005; Mameli et al. 2006). The neurobehavioural endpoints examined constitute standard elements of neurobehavioural testing and impaired performance is considered to be toxicologically significant in the experimental animal.

The biochemical effects observed in the remaining studies included alterations in neurotransmission systems, alterations in synaptosomal membrane fluidity and increased lipid peroxidation in the brain, and were associated with LOELs varying from 19 to 420 mg Al/kg bw/d. These observations provide supportive evidence for neurotoxicity observed via other endpoints as well as information on mechanisms of action, but are more difficult to evaluate with respect to toxicological significance. For this reason, studies with these endpoints were given less weight in the exposure-response evaluation, in comparison to studies that include neurobehavioural endpoints.

Considering the 22 studies of exposure during gestation and lactation, the LOELs (Da and Dc) varied between 29 and 663 mg Al/kg bw/d. For neurobehavioural endpoints (grip strength, auditory startle, negative geotaxis and other reflexes, maze learning and thermal sensitivity, and motor development), the LOELs (administered doses) ranged from 50 to 155 mg Al/kg bw/d, with the LOELs of two studies falling in the range of 50 to 60 mg Al/kg bw/d (Colomina et al. 2005; Golub and Germann 2001b).

With respect to reproductive parameters, the lowest LOEL was reported by Belles et al. (1999), where aluminum nitrate was administered to pregnant mice via gavage at a dose of 29 mg Al/kg bw/d and observed an increase in the number of early deliveries and reduced fetal body weight. Reduced birth or fetal weight was also observed by Colomina et al. (1992) and Sharma and Mishra (2006) at LOELs ranging between 50 and 70 mg Al/kg bw/d. Morphological effects in offspring were also observed in the latter two studies.

The motor, reflex and learning endpoints examined in the developmental studies as well as the reproductive parameters of fetal growth and morphological variations are all standard endpoints included in neurodevelopmental testing procedures, and considered to be toxicologically significant.

(c) Evaluation of methodology and consistency of results in studies with LOELs of less than 70 mg Al/kg bw/d

The methodologies and findings of the abovementioned studies with LOELs of less than 70 mg Al/kg bw/d for neurobehavioural or reproductive/developmental endpoints were compared in order to characterize the strength of evidence for the effects observed at these dose levels. With respect to the neurobehavioural effects in adults at exposures greater than 90 days, four studies were evaluated: Mameli et al. (2006), Gong et al. (2005), Lal et al. (1993) and Commissaris et al. (1982). The reproductive/developmental studies included Sharma and Mishra (2006), Belles et al. (1999), Colomina et al. (1992), Colomina et al. (2005) and Golub and Germann (2001b). In addition, investigations in which NOELs were observed for these same endpoints are discussed.

Neurobehavioural effects in adults

Of the four neurobehavioural studies in adults, all were carried out in rats using aluminum chloride, in drinking water (Gong et al. 2005; Mameli et al. 2006; Lal et al. 1993), or in the diet (Commissaris et al. 1982), for periods varying between 90 days and 11 months.

Several weaknesses were identified in the investigations of Commissaris et al. (1982) and Gong et al. (2005). First, exposure information in these two reports was expressed as concentrations in the food or drinking water, and no information was included on intake rates or body weight of the animals. Thus the administered doses (50 and 60 mg Al/kg bw/d, respectively) were calculated on the basis of default intake and body weight values (refer to Health Canada 1994), and are therefore associated with greater uncertainty than had the doses been reported by the researchers on the basis of experimental observations. Moreover, the concentration of aluminum in the base diet was not reported in the two studies, and so the combined dose could not be calculated.

The investigations of Commissaris et al. (1982) and Gong et al. (2005) were also limited by the use of a single aluminum dose and the absence of a group receiving sodium chloride. Thus, a dose-response relationship could not be examined, and the observed effects could not be definitively attributed to the aluminum ion. It should be added that these two investigations were carried out with the primary objective of examining the influence of other test substances on aluminum toxicity--parathyroid homone and Ginkgo biloba leaf extract, respectively--and not for the purpose of evaluating aluminum toxicity at different dose levels for different endpoints.

In the study of Lal et al. (1993), adult male Druckrey albino rats were exposed to an administered dose of 52 mg Al/kg bw/d for 180 days in drinking water. Although this dose was not reported directly in this form, information on daily water consumption and average body weight was provided, allowing for calculation of the dose based on experimental data. The investigation included a range of behavioural, biochemical and histopathological endpoints. The researchers observed reduced spontaneous motor activity and impaired learning in the shuttle box and maze tests, in addition to increased lipid peroxidation and decreased Mg2+ and Na+K+-ATPase activities in the brain. The aluminum concentration in different brain regions was significantly increased in the aluminum-exposed animals, but no pathological alterations were observed.

In the context of evaluating the exposure-response relationship, the study by Lal et al. (1993) is more informative than the Commissaris et al. (1982) and Gong et al. (2005) studies, in that the dose is more accurately reported, brain aluminum content was measured and a range of endpoints were examined, with generally consistent findings reported for the different endpoints. Its limitations include the use of a single dose, the absence of a group exposed to sodium chloride and the lack of information on the aluminum concentration in the base diet. Assuming a concentration of 250 ppm of aluminum in the laboratory chow (ATSDR 2006), the corresponding approximate aluminum dose would be 13 mg Al/kg bw/d, leading to an estimated combined dose for the Lal et al (1993) study of 65 mg Al/kg bw/d.

It should be noted that NOELs for impaired learning in the maze and shuttle box tests in aluminum-exposed adults have been observed at doses of 100 and 140 mg Al/kg bw/d, respectively by Domingo et al. (1996) and VonLinstow Roloff et al. (2002). In the study by Domingo et al. (1996) the aluminum was administered to rats as Al nitrate, with added citrate, in drinking water for a period of 6.5 months. Von Linstow Roloff (2002) administered Al sulphate in drinking water to rats for a period of seven months.

Of these four studies, only Mameli et al. (2006) included more than one dose group, and were thereby able to establish a LOEL of 43 mg Al/kg bw/d and a NOEL of 22 mg Al/kg bw/d. At this administered dose the researchers found impairment of the vestibulo-ocular reflex in male rats of different ages (3, 10 and 24 months old) exposed to aluminum chloride in drinking water. Significant increases of aluminum were observed in brain regions (brainstem-cerebellum and cerebrum). This study, which used 20 animals per dose per age group, also included an exposure group for the salt, in this case sodium chloride, such that the observed effects could be more clearly attributed to the aluminum and not the chloride ion. It should be noted, however, that evidence from other studies supporting the effects of aluminum on the vestibulo-ocular reflex is not available, as this endpoint has not been evaluated by other researchers.

In the study by Mameli et al. (2006), the base diet aluminum concentration was measured but not clearly reported, nor was food intake measured. The LOEL of 43 mg Al/kg bw/d is thus the administered dose. The combined dose may be estimated at approximately 50 mg Al/kg bw/d, based on default values for rat dietary intake.

Considering the observations of LOELs and NOELs associated with neurobehavioural effects in adults as well as the probable combined doses, alterations in learning and reflexes may be observed at approximately 50 to 65 mg Al/kg bw/d, based on the LOELs of Mameli et al. (2006) and Lal et al. (1993) expressed as estimated combined dose.

Reproductive effects

With respect to reproductive effects, the lowest LOEL presented in Figure 3.1 is associated with the study of Belles et al. (1999). In this investigation, mice were exposed to aluminum nitrate via gavage from gestational day 6 to 15 at a dose of 29 mg Al/kg bw/d. In addition to the control group, one group received sodium nitrate at a similar nitrate dose. A high mortality (52%) in the aluminum-exposed pregnant mice was observed in this study, which was not observed in other developmental studies in which aluminum nitrate or other aluminum salts were administered at similar or greater doses. Other observations included reduced body weight gain in the dams during gestation and reduced fetal body weight. The number of early deliveries was also increased in the aluminum-exposed animals as compared to the control group, but there was no significant difference in this regard when compared to the sodium nitrate-exposed group.

This study is limited to a single dose, and the aluminum content in the base diet was not measured. The lack of information on base diet is particularly important in studies with mice because of their small body weight. A laboratory chow containing 250 ppm of aluminum would be equivalent to a dose of approximately 33 mg Al/kg bw/d, which is higher than the administered dose in this investigation.

Reduced maternal body weight gain and reduced fetal weight in aluminum-exposed animals were also observed at the LOELs associated with the Sharma and Mishra (2006) and Colomina et al. (1992) studies. A significant reduction in pup weight was also observed at the higher doses tested in the studies of Golub and Germann (2001b) and Colomina et al. (2005), at approximately 100 mg Al/kg bw/d.

In the study by Sharma and Mishra (2006), rats received 70 mg Al/kg bw/d as aluminum chloride via gavage during gestation and lactation. In addition to the effects on fetal weight, the authors observed an increase in skeletal malformations and in oxidative stress in the brains of mothers, fetuses and sucklings. The dose level in this study is based on the measured maternal weights. However, no information on base diet was included. The combined dose, based on a concentration of 250 ppm of aluminum in a typical lab chow and default values of Health Canada (1994), is estimated at approximately 83 mg Al/kg bw/d.

Colomina et al. (1992) administered aluminum lactate to mice through gavage. A LOEL of 57.5 mg Al/kg bw/d (administered dose) was observed for an increased incidence of morphological effects (cleft palate, delayed ossification of parietals), in addition to reduced fetal weight. This study did not report the aluminum content in the base diet. Considering the reported concentration in the laboratory chow used by this research group in other experiments of 42 ppm of aluminum, the estimated base diet dose would be approximately 5.5 mg Al/kg bw/d, based on Health Canada (1994) default values for body weight and food intake in mice. The combined dose would then be estimated at 63 mg Al/kg bw/d.

In contrast to the findings mentioned above, in the study of McCormack et al. (1979), rats were fed aluminum chloride in the diet at maternal dose levels of 25 and 50 mg Al/kg bw/d during gestation, and no differences in fetal growth or skeletal anomalies were observed. Colomina et al. (1994) found no differences in dam body weight, fetal growth or morphological variations in mice exposed via gavage to 104 mg Al/kg bw/d of aluminum hydroxide, during gestation. The latter finding may have resulted from the lower solubility and therefore the lower bioavailability of the hydroxide salt.

Considering the observations of LOELs and NOELs associated with reproductive effects, and the probable combined doses, reductions in fetal and pup body weight may be observed beginning at approximately 60 mg Al/kg bw/d (for example, Colomina et al. (1992)). The study of Belles (1999), in which a LOEL of 29 mg Al/kg bw/d was observed for reduced fetal weight, is given less weight in this evaluation, in light of the uncertainty associated with the high maternal mortality rate observed in the exposed animals, and the elevated contribution of the base diet to aluminum exposure as compared to the administered dose.

Neurodevelopmental effects

With respect to neurodevelopmental effects, the lowest LOELs presented in Figure 3.1 are associated with the investigations of Colomina et al. (2005) and Golub and Germann (2001b). Both of these studies included exposure through gestation and lactation. The experimental conditions of the two studies, however, differed in many other respects, and these are described briefly below.

Colomina, Roig et al. (2005) exposed female Sprague-Dawley rats to 0, 50, or 100 mg Al/kg bw/d as aluminum nitrate in drinking water with citric acids, in combination with a base diet dose of approximately 3 mg Al/kg bw/d. Aluminum exposure was maintained through gestation, lactation and the life of the dams.

The maternal effects of aluminum administration included decreased food intake (with reduced body weight) during gestation and lactation and decreased water intake during lactation in the 100 mg Al/kg bw/d dose group. No effects were observed with respect to the length of gestation, the number of litters or the number of fetuses per litter. With respect to the pups, there was a significant increase in the number of days until sexual maturation in males in the 100 mg Al/kg bw/d dose group and in females at both 50 and 100 mg Al/kg bw/d. A significant reduction in forelimb grip strength in males was observed in the 100 mg Al/kg bw/d dose group on PND 11 compared controls.

In the water maze task, assessing spatial learning, the performance of aluminum treated rats (50 mg Al/kg bw/d) was significantly improved in comparison to the control group. The pups in the 100 mg Al/kg bw/d dose group were not tested in the water maze test, because of altered maternal food and water intakes in this group. No differences in aluminum-exposed animals were observed with respect to surface righting, negative geotaxis or activity in an open field. The authors also measured aluminum concentration in brain regions but did not find increased levels in any regions in the aluminum-exposed animals.

The study of Golub and Germann (2001b) investigated the long-term consequences of prenatal exposures to aluminum in Swiss Webster mice, in conjunction with a suboptimal base diet. The base diet was designed to simulate the usual diet of young women in the U.S., with respect to estimated phosphate, calcium, iron, magnesium, and zinc intakes. Following breeding, dams were exposed to aluminum in the diet as aluminum lactate. The doses were equivalent to approximately < 1, 10, 50 and 100 mg Al/kg bw/d, as estimated at the beginning of gestation.

The dams were exposed throughout gestation and lactation. Following weaning at 21 days, the pups were fed the same diet as the dams for two weeks (although the per kg dose levels were higher). No effects were observed in the number of dams completing pregnancy, gestation length, weight gain of the dams (GD0 to GD15), litter size or birth weight. By weaning, both males and females in the two highest dose groups weighed significantly less than the controls, although by PND35 only the highest dose group showed this effect.

The female offspring of the highest dose group (maternal exposure of 100 mg Al/kg bw/d) were found to be slower in maze learning at three months old, as indicated by longer latencies during the first three sessions of the four-session learning series. All aluminum treated groups were similar to controls by the fourth session. Differences in aluminum exposed groups were also observed in the cue relocation trials, in which average trial latency was significantly increased at the two highest dose levels (50 and 100 mg Al/kg bw/d) as compared to the control group.

In the motor testing of male offspring at five months old, males in the highest dose group (maternal exposure of 100 mg Al/kg bw/d) had significantly lower hindlimb grip strength and greater number of rotations in the rotarod test (animal losing footing). When body weight was taken into account, only the findings for the rotarod test remained significant.

The investigations by Colomina et al. (2005) and Golub and Germann (2001b) are methodologically superior in many respects to the majority of the studies described in Tables C1 and C2. Both include two dose levels in addition to the control group, quantify the aluminum dose associated with the base diet, and examine a range of reproductive and neurodevelopmental endpoints. The Colomina et al. (2005) study includes measurement of aluminum concentration in different brain regions. The Golub and Germann (2001b) study, however, used an experimental protocol designed to test the influence of a suboptimal diet, which limits comparisons of the findings with other investigations of aluminum toxicity, particularly as no groups were included with equivalent aluminum dose levels and a standard diet.

Interpretation of cognitive and motor test findings in the studies investigating the effects of aluminum exposure is also complicated by a possible biphasic dose-response relationship. For example, in the study by Roig et al. (2006), rats received aluminum nitrate in drinking water during gestation and lactation at administered doses of 50 and 100 mg Al/kg bw/d. No difference in the motor activity of aluminum-exposed pups and controls was found. However, the animals exposed to 50 mg Al/kg bw/d showed an improved performance in maze learning. The performance of animals exposed to 100 mg Al/kg bw/d was significantly reduced as compared to the animals exposed to 50 mg Al/kg bw/d, but not significantly different from controls. Colomina et al. (2005) also observed improved maze performance in aluminum-exposed animals, although the highest exposure group in that study was not tested for this endpoint.

Considering the neurodevelopmental studies described above, diminished performance in learning or motor tests may be observed in animals exposed prenatally or through lactation at maternal combined doses beginning at approximately 50 mg Al/kg bw/d. There is, however, considerable variability in various study results with respect to these endpoints, which also suggest a possible biphasic dose-response relationship in relation to maze learning.

3.2.3.1 Studies pertaining to other life stages

Some experimental animal studies have focused on life stages not included in the subsets discussed above. These are described below.

Golub and Keen (1999) investigated the effects of aluminum lactate administered in the diet to pubertal mice for four- or eight-week periods at doses of 17, 78, 122 and 152 mg Al/kg bw/d. A significant association between aluminum intake and reduced brain weight was observed in the four-week cohort at 152 mg Al/kg bw/d, but not in the eight-week cohort, suggesting that effects in young animals are reversible, even as exposure continues. There were no consistent effects, however, on startle response or grip strength.

Rajasekaran (2000) administered 53 mg Al/kg bw/d of aluminum chloride via gavage to male pubertal Wistar rats for 30 days. Testing at the end of the exposure period showed a decrease in spontaneous motor activity in the exposed rats, but no effect on motor coordination. Acetyl cholinesterase activity was decreased in the cerebrum but not the cerebellum or brain stem.

Fattoretti et al. (2004) administered aluminum chloride in drinking water to 22-month-old rats, at a dose of 31 mg Al/kg bw/d for six months. They observed an increase in trace elements and aluminum in brain regions, and an increase in the area occupied by the mossy fibres in the hippocampal CA3 zone. No neurobehavioural endpoints were examined in this study.

Colomina et al. (2002) administered aluminum nitrate in drinking water (with citric acid) for 114 days to rats who were 18 months old at the start of the experiment. The weighted dose over the four months was 94 mg Al/kg bw/d. They found a decrease in mean body weight in aluminum-exposed older rats but no differences in brain aluminum concentration. No effects were observed in the passive avoidance test or in open-field activity. However, the percentage of perforated synapses in the brain increased with age and aluminum exposure.

A recent study by Walton (2007a, 2007b) of rats exposed from 12 months to the end of life investigated neurotoxicity endpoints at combined doses of 0.4 and 1.6 mg Al/kg bw/d, doses simulating current estimated low-end and high-end human exposures. Two of the six rats in the high exposure group developed significant impairment in memory tests in old age, and the brains of these rats were examined with respect to aluminum loading and inhibition of PPP2 activity (a major phosphate-removing enzyme active against tau hyperphosphorylation)Footnote 31. The study, limited by the small group size, did not report on differences between the two aluminum-exposure groups, and thus does not provide a basis for conclusions in regard to the relationship between observed biochemical and behavioural effects and aluminum exposure.

3.2.3.2 Identification of the level of concern and associated uncertainties

On the basis of the 43 studies presented in Tables C1 and C2, and considering additional studies on other age groups, it is recommended that a dose of 50 mg Al/kg bw/d, expressed as a combined dose of total aluminum, be considered as the level at which neurological and reproductive/developmental effects begin to be repeatedly observed in animal studies.

While the dose of 50 mg Al/kg bw/d is an estimation of the lower end of a broad range of LOELs observed under different experimental conditions, it is not considered to be an overly conservative estimate of the effect level of concern. As previously discussed, there are two sources of bias against consideration of lower values of LOEL in the above characterization: (a) low-dose studies were not considered if the administered dose was less than the probable base diet dose; and (b) LOELs from single-dose studies may be overestimates of the actual effect levels. The dose of 50 mg Al/kg bw/d has, however, produced neurotoxic, reproductive and developmental effects in laboratory animals more consistently under a wide range of experimental conditions, as compared to lower doses. This exposure level is therefore retained for the purpose of the characterization of human health risks as the level of concern for neurotoxic, neurodevelopmental and reproductive effects.

Figure 3.1: Compilation of the LOEL values from the two major subsets of studies (adult exposure > 90 days and reproductive/developmental) considered in the exposure-response analysis.

Compilation of LOEL values from two major subsets of studies

The numbers represent the 38 studies in which LOELs were observed, as summarized in Tables C1 and C2, and listed below. Where the base diet aluminum level is quantified, the LOEL is expressed as combined dose. NOELs associated with LOELs are indicated when observed.

Study references and endpoints

Reproductive and developmental studies

  1. Bernuzzi et al. 1986: Reduced body weight of pups, impaired negative geotaxis.
  2. Golub et al. 1987: Reduced birthweight, decreased body weight gain in pups.
  3. Bernuzzi et al. 1989:
    1. Impaired locomotor coordination;
    2. Impaired righting reflex;
    3. Impaired grasping reflex.
  4. Muller et al. 1990: Impaired negative geotaxis, impaired performance in suspension and locomotor coordination tests.
  5. Gomez et al. 1991: Reduced fetal body weight, increase in skeletal variations.
  6. Colomina et al. 1992: Maternal toxicity, reduced fetal body weight (aluminum lactate), increased incidence of morphological effects (aluminum lactate).
  7. Misawa and Shigeta 1993: Maternal toxicity, decreased pup weight, delay in pinna detachment and eye opening in females, delayed development of auditory startle in males.
  8. Golub et al. 1993: Effects on Mn metabolism.
  9. Golub et al. 1994: Reduced auditory startle response.
  10. Poulos et al. 1996: Delayed expression of phosphorylated high molecular weight neurofilament protein in tracts in diencephalon, maternal toxicity.
  11. Golub et al. 1996: Lower retention of both Mn and Fe.
  12. Verstraeten et al. 1998: Increased phospholipid and galactolipid contents in brain myelin, increased lipid peroxidation.
  13. Llansola et al. 1999: Decrease in pup body weight, decreased number of cells in cerebellum, disaggregation of microtubules and neuronal death in cerebellar neuron cultures.
  14. Belles et al. 1999: Increased mortality of dams and increased early deliveries, reduced fetal body weight.
  15. Golub and Tarara 1999: Decreased myelin sheath width.
  16. Golub et al. 2000: Reduced forelimb and hindlimb grip strength, decreased thermal sensitivity.
  17. Golub and Germann (2001b):
    1. Impaired performance in rotarod test (males);
    2. Decreased weight gain in pups, impaired learning of maze with respect to cue utilization (females).
  18. Wang et al. 2002a: Reduced body weight, deficits in synaptic plasticity in dentate gyrus of hippocampus.
  19. Chen et al. 2002: Deficits in synaptic plasticity in dentate gyrus of hippocampus.
  20. Nehru and Anand 2005: Increased lipid peroxidation, decreased superoxide dismutase and catalase activity in cerebrum and cerebellum.
  21. Colomina et al. 2005:
    1. Reduced forelimb strength in males;
    2. Increased number of days to sexual maturation.
  22. Sharma and Mishra 2006: Decreased number of corpora lutea, number of implantation sites, placental and fetal weight, increased skeletal malformations, increased oxidative stress in brains of mothers/fetuses and sucklings.

    > 90 days exposure studies in adults
  23. Commissaris et al. 1982: Reduced motor activity, impaired learning (shuttle box).
  24. Johnson et al. 1992: Decreased levels of microtubule associated protein-2 and spectrin in hippocampus.
  25. Golub et al. 1992: Decreased motor activity, hindlimb grip strength and auditory and air puff startle responsiveness.
  26. Lal et al. 1993: Reduced spontaneous motor activity; impaired learning (shuttle box, maze), increased brain lipid peroxidation, reduced Mg2+- and Na+K+-ATPase activities.
  27. Florence et al. 1994: Cytoplasmic vacuolization in astrocytes and neurons.
  28. Gupta and Shukla 1995: Increased lipid peroxidation in brain.
  29. Zatta et al. 2002: Increased acetylcholinesterase activity.
  30. Silva et al. 2002: Increased synaptosomal membrane fluidity, decreased cholesterol/phospholipid ratio in synaptosomes.
  31. Flora et al. 2003: Evidence of increased lipid peroxidation in brain.
  32. Jing et al. 2004: Impaired performance in Morris water maze, altered synapses in hippocampus and frontal cortex.
  33. Gong et al. 2005: Impaired performance in Morris water maze.
  34. Shi-Lei et al. 2005: Impaired performance in Morris water maze, decrease in long-term potentiation in hippocampal slices.
  35. Silva et al. 2005: Decreased Na+/K+-ATPase activity in brain cortex synaptosomes.
  36. Huh et al. 2005: Induced apoptosis in brain, increased efficiency of monoamine oxidases and increased level of caspase 3 and 12 in brain.
  37. Rodella et al. 2006: Decreased nitrergic neurons in the somatosensory cortex.
  38. Mameli et al. 2006: Impaired vestibulo-ocular reflex.

3.2.4 Human health risk characterization for total aluminum

As noted in the Introduction (section 1) three aluminum salts are specifically named for assessment on the Priority substances list (PSL) : chloride, nitrate and sulphate. Although the data available for the assessment do not allow for accurate quantification of exposure associated with specific salts, it is possible to qualitatively estimate their relative contribution to different environmental media (see Table 3.2).

Based on the use pattern of these three salts, described in section 2.2.1, the major use of sulphate and chloride salts is in water treatment, therefore exposure to these particular salts would be expected via drinking water. Aluminum sulphate has a minor use as a food additive; other aluminum-containing additives are much more widely used.  Aluminum nitrate use is limited in comparison to the sulphate and chloride salts.  It is used in fertilizers and as a chemical reagent in various industries and is not expected to contribute significantly to aluminum in food and soil, the principal media of total aluminum exposure.

Based on these use patterns, the only media in which the mean concentration is significantly affected by the use of these salts is drinking water.  Although the contribution of aluminum via these salts cannot be accurately quantified, in order to quantitatively compare the exposure level of concern with potential exposure to aluminum from the three salts, as a surrogate for exposure it is assumed that all aluminum in drinking water is derived from aluminum chloride and aluminum sulphate.

Therefore, the human health risk characterization for the three salts is based on the comparison of the exposure level of concern of 50 mg/kg bw/d, identified in the exposure-response analysis of section 3.2.3, and the age-group with the highest average daily intake of total aluminum from drinking water (10.8 mg/kg bw/d in non-breastfed infants, see Table 3.1). The ratio of these two levels, generally referred to as the margin of exposure (MOE), is greater than 4000.  This margin of exposure is considered adequate, taking into account the fact that aluminum exposure from the three salts is overestimated in this calculation, and the following considerations.

To account for toxicokinetic and toxicodynamic variability and uncertainty, a factor of at least 100 within the MOE is considered appropriate. As there is little consensus as to the mode of action, and multiple mechanisms are likely involved, the delineation of chemical-specific adjustment factors is not possible here.  Effects at the lower-bound were generally small changes in performance in motor activity and learning tests identified across a range of studies, and the MOE is considered adequate to account for uncertainties in the identification of this lower-bound.

The adequacy of the collective database for the neurotoxicity and reproductive/developmental toxicity of orally-administered aluminum was reviewed in section 3.2.2.2. As discussed, there is a clear need for further investigation in experimental animals, in which studies are designed to provide a basis for determining a critical dose for risk assessment. The existing database is nonetheless extensive, providing a basis for the determination of the lower range of LOELs observed in the different studies, carried out under different experimental conditions and for an array of aluminum salts. The neurobehavioural and neurodevelopmental effects most frequently associated with the range of LOELs may be characterized as small but statistically significant changes in performance in motor activity and learning tests.

Collectively the limited aluminum bioavailability data do not indicate that the relative bioavailabilities of aluminum in drinking water, soil and different types of food are significantly different (see section 2.3.3.1). Therefore, it is not anticipated that aluminum from drinking water would contribute relatively more bioavailable aluminum, in proportion to its external dose, as compared with other sources.  In addition there is no evidence to suggest that there are differences in relative bioavailability between humans and experimental animals.

3.2.5 Uncertainties and degree of confidence in human health risk characterization

There is a moderately high degree of confidence in the deterministic exposure assessment for aluminum, as it relates to the average external dose associated with food, drinking water, soil and air, due to a large database of experimental information for most media. There is more uncertainty with respect to the maximum or high-end exposures in the population for the different media due to the variability in measured levels.

For total aluminum, food is the principal source of exposure, followed by soil, while exposure via drinking water and air combined is less than 2% of total aluminum intake. Based on their use pattern, the three aluminum salts on the PSL are not significant contributors to the principal media of total aluminum exposure.  Given the importance of food in the total exposure to aluminum, a probabilistic analysis of the exposure to aluminum from foods accounting for intakes by different subsets of the Canadian population is warranted.  In addition, such an analysis should distinguish aluminum originating from food additives from natural aluminum sources in foods. 

The greatest uncertainty with respect to the exposure assessment is the uncertainty and variability relating to the extent to which different aluminum salts are absorbed from the different media. Although some experimental bioavailability data are available for food and water, collectively the limited aluminum bioavailability data do not indicate that the relative bioavailabilities of aluminum in drinking water, soil and different types of food are significantly different. However, further research in this area, particularly in regard to soil, could provide evidence for significant differences that would in turn influence the human health risk characterization.

3.2.6 Recommendations for research

Areas for further research are described briefly below, in order to identify the main avenues for reducing the uncertainties associated with the human health database for aluminum.

3.2.6.1 Exposure assessment

Consideration of bioavailability is important to the characterization of human health risks of aluminum if relative bioavailabilities for different exposure media and different species (that is, humans and experimental animals) differ from unity. This hypothesis could be explored through the determination of bioaccessibilities of aluminum in aluminum-treated drinking water, different soil and dust samples, in selected food items (for example, processed cheese and packaged bakery items), and in laboratory animal chow, followed by the comparison of these in vitro bioaccessibilities with the in vivobioavailability of aluminum determined in experimental studies for a given media.

In light of the wide use of aluminum-containing products applied to the skin, the dermal absorption of aluminum in humans should be more adequately characterized.

3.2.6.2 Exposure-response assessment

Further epidemiological study of aluminum exposure in the Canadian population is called for, to the extent that such research addresses the limitations of previous studies, including the characterization of aluminum exposure by dietary and other sources.

Additional experimental animal studies on toxicokinetics of different salts, including aluminum fluoride as well as the neurological and neurodevelopmental effects of aluminum, is necessary to provide information for better characterizing the exposure-response relationship. Following Organization for Economic Co-operation and Development (OECD) guidelines for neurotoxicity and neurodevelopmental toxicity, these studies would include adequate numbers of animals, multiple doses, and examination of a standard array of neurological and neurodevelopmental endpoints. Note that one such study is currently underway in Canada.

Page details

Date modified: