Introduction

Human exposure to aluminium is burgeoning (Exley 2013; Klotz et al. 2017; Stahl et al. 2017). Humans are living in the aluminium age (https://www.hippocraticpost.com/mens-health/the-aluminium-age/) and everyday exposure is impacting upon daily life (Cabral Pinto et al. 2018, 2019; Stahl et al. 2018; Handra et al. 2019). Systemic aluminium is excreted in urine (Michalke et al. 2018) and sweat (Genuis et al. 2011) and retained in tissues and especially brain tissue (Exley and House 2011). The longevity of neurones predisposes them to the accumulation of aluminium though very little is understood as to whether higher concentrations of aluminium in brain tissue are due to its greater access or its increased retention (Exley 2014; Bondy 2016). Ageing is considered a major risk factor for the accumulation of aluminium in brain tissue (Roider and Drasch 1999) and in itself may underlie a higher content of aluminium in brain tissue in sporadic Alzheimer’s disease (House et al. 2012; Yumoto et al. 2018). There are no other confirmed risk factors for the accumulation of aluminium in brain tissue though aluminium is shown to be elevated in a number of neurodegenerative and neurodevelopmental disorders including familial Alzheimer’s disease (fAD) (Mirza et al. 2017) and autism spectrum disorder (ASD) (Mold et al. 2018a). Recently, we made the first analyses of aluminium in brain tissue in individuals who died with a diagnosis of the neurological condition, multiple sclerosis (MS) (Mold et al. 2018b). Our previous research had suggested that individuals with MS had a higher-than-expected body burden of aluminium (Exley et al. 2006; Jones et al. 2017) and this prompted us to look for aluminium in MS brain tissue. The aluminium content of tissues in MS was found to be universally high with every donor brain having at least one tissue where the measured content of aluminium was considered as pathologically significant (≥ 3.00 μg/g dry weight). Detailed statistical analyses revealed no relationships with gender or, perhaps surprisingly, age of donor. Age was not a risk factor for aluminium in brain tissue in MS. Previously, we have asked the question as to how much aluminium in brain tissue is too much (Exley and Mold 2019) and this question is equally applicable in considering MS. Do we have evidence that the aluminium content of brain tissue in MS is high due to increased uptake or higher retention? To help in answering this question, The Multiple Sclerosis Society Tissue Bank provided us with brain tissues from donors who did not die with a diagnosis of MS and additionally showed no neurodegeneration beyond that which could be attributed to normal ageing. However, all donors died following a period of (usually traumatic) disease and primarily a cancer-related condition. There are no data in the current scientific literature describing aluminium in brain tissue in individuals who died of a non-neurodegenerative/non-neurodevelopmental disease. Herein, we have made these measurements and compared the data with our previous data on MS. We aimed to provide further information on aluminium in brain tissue in disease and to test if data for MS are, as was previously suspected, unusually high.

Materials and Methods

Tissues

Brain tissues were obtained from the Multiple Sclerosis Society Tissue Bank, Imperial College, London, following ethical approval (NRES Approval No. 08/MRE09/31). Donors included three females and nine males between the ages of 35 and 88 years. Cancer had played a role in the death of at least eight donors. Neuropathological investigation of all brains revealed no specific neuropathology beyond that described as age related. None of the donors had multiple sclerosis.

Quantitative Measurements

The aluminium content of tissues was measured by an established and fully validated method (House et al. 2012) which herein is described only briefly. Samples of cortex, between 0.6 and 5.0 g in weight, were thawed at room temperature and cut using a stainless steel blade into sections approximately 0.3–0.5 g in weight. Tissues were dried for 48 h, to a constant weight, in an incubator at 37 °C. Dry and thereafter weighed tissues were digested in a microwave (MARS Xpress CEM Microwave Technology Ltd.) in a mixture of 1 mL 15.8 M HNO3 (Fisher Analytical Grade) and 1 mL 30% w/v H2O2 (BDH Aristar). The resulting digests were clear with no fatty residues and, upon cooling, were made up to 5 mL volume using ultrapure water (cond. < 0.067 μS/cm). Total aluminium was measured in each sample by transversely heated graphite furnace atomic absorption spectrometry (TH GFAAS) using matrix-matched standards and an established analytical programme alongside previously validated quality assurance data (House et al. 2012). The latter included method blanks, detailed descriptions of which have been published recently (Exley and Mold 2019).

Statistical Analyses

Data for aluminium content of tissues were skewed and were not normally distributed. For descriptive summary statistics, the median and interquartile range were calculated for each donor and additionally per donor and lobe. For all test statistics and models, aluminium content data were log transformed. Due to unbalanced groups (non-MS donors = 14, MS donors = 12) and unbalanced numbers of samples per donor and lobe, mixed effect models including random effects for donors and lobe were used.

First, a model was calculated for the non-MS group to analyse differences between the factors lobes and gender and associations with the covariate age. The model contains all these factors as main terms without interactions. Additional random effects were nested and included the factor, number of samples, nested within lobe and donor.

A second model included the factor group to analyse differences between the MS and the non-MS groups (Zuur et al. 2009). Data on the hippocampus lobe exist only for the non-MS group and were therefore excluded from the mixed effect model comparing MS and non-MS donors. We considered a p-value smaller than 0.05 to be statistically significant. To obtain pairwise differences between lobes and disease groups, post hoc tests with Tukey’s correction were performed with the function glht from the R package multcomp. For univariate and descriptive analysis, SPSS Statistics v.22 (IBM Analytics, Armonk, NY, USA) was used; for the mixed effect model, RStudio Version 1.1.463 © 2009–2018.

Results

Aluminium Content of Brain Tissues

The aluminium content of all tissues ranged from 0.01 (limit of quantitation) to 17.26 μg/g dry weight (Supplementary Table 1). The latter being a significant outlier and originating from the temporal lobe of a 35-year-old male who died of cancer of the tongue. The majority of measurements across all four lobes (and occasionally the hippocampus) for all individuals were below 1.00 μg/g dry weight (189/291) though 59, 24 and 19 measurements were in the range, 1.00–1.99, 2.00–2.99 and ≥ 3.00 μg/g dry weight, respectively. There were statistically significant associations relating to brain aluminium content and age in non-MS donors (r(289) = 0.2, p = 0.008, n(samples) = 12, n(samples) = 291, Table 2). However the effect of age was weaker in the multivariate mixed effect model (mixed effect model, p = 0.014, n(donors) = 12, n(samples) = 291, Table 2). There were no significant differences in brain aluminium content of males and females (mixed effect model, p = 0.864, n(donors) = 12, n(samples) = 291, Table 2). There were no statistically significant differences in aluminium content between lobes (mixed effect model, df = 4, p = 0.203, n(donors) = 12, n(samples) = 291, Table 2). Regarding the latter, the data for all cases (median and IQR) were 0.74 (0.48–1.28), 1.23 (0.62–1.63), 0.84 (0.45–1.14) and 1.01 (0.62–1.65) μg/g dry weight for occipital, parietal, temporal and frontal lobes, respectively (Table 1). There was a significant trend for higher brain aluminium content in individuals who died with a diagnosis of cancer (8 individuals) compared to other conditions (4 individuals) (mixed effect model, df = 1, p = 0.004, n(donors) = 12, n(samples) = 291, Tables 1 and 2).

Table 1 Descriptive summary statistics for the non-MS group per lobe, showing median, interquartile range (IQR) and sample size per donor (n)
Full size table
Table 2 Mixed effect model for aluminium content in non-MS donors with the fixed factors disease group (non-cancer as reference), lobe (occipital as reference), gender (female as reference) and the covariate age
Full size table

Comparison with Brain Aluminium Content in Multiple Sclerosis

The non-MS group and MS group were gender matched (Fisher's exact test, n = 26, p = 0.13) though the former were older (t test, p = 0.003, n = 26). There was no statistically significant association relating to brain aluminium content and age for the MS group (r(330) = 0.03, p = 0.611, n = 14, observations = 332). The aluminium content across all lobes were significantly higher in MS donors (mixed effect model, n(samples) = 615, N(donors) = 26, p = 0.004, Tables 3 and 4) than non-MS donors (Fig. 1). The single mixed effect models per lobe showed a significant difference between MS donors and non-MS donors for the frontal lobe (p = 0.012, Table 4, Supplementary Table 2, Fig. 2).

Table 3 Descriptive statistics of aluminium contents according to the disease groups. Number of donors is shown in brackets, IQR (interquartile range). Data for the lobe hippocampus (n(samples) = 8) are excluded in this table
Full size table
Table 4 Full mixed effect model for aluminium content in all donors with the fixed factors disease group (non-MS as reference), lobe (occipital as reference), gender (female as reference) and the covariate age
Full size table
Fig. 1
figure 1

Boxplots for aluminium measurements stratified by lobes for non-MS and MS groups

Full size image
Fig. 2
figure 2

Boxplots for aluminium measurements stratified by lobes for non-MS and MS groups

Full size image

Discussion

We report the first measurements of aluminium in human brain tissue in donors who died of a non-neurodegenerative or non-neurodevelopmental disorder and whose brain showed no recognisable signs of neurodegeneration. These are not ‘healthy’ donors, they all died from a serious medical condition, primarily (8 of the 12 donors) relating to cancer. Data covered a wide range of tissue aluminium content and while a high proportion of tissues, ca 65%, could be considered as within a normal or non-pathogenic range (< 1.00 μg/g dry weight) ca 7% of tissues had an aluminium content considered as pathologically significant (≥ 3.00 μg/g dry weight) (Exley and Mold 2019). The donor’s age varied between 35 and 88 years and a statistically significant association showed higher brain aluminium content with increased age. This finding was opposite to that previously observed in MS (Mold et al. 2018b). While it confirmed a previous observation of age as a risk factor for brain aluminium content in a non-neurologically affected population (Roider and Drasch 1999), it also suggests that a correlation with age may not be significant in neurologically impaired populations. There was a clear significant trend for higher brain aluminium content in the eight cancer-related cases compared with the four non-cancer-related cases (Tables 1 and 2). The Multiple Sclerosis Society Tissue Bank provided the brain tissues investigated herein as possible control tissues for comparing with aluminium in brain tissue in MS. It is certainly true that none of the donors had MS or signs of neurodegeneration and while they were slightly older than the MS donor group, they were gender matched. Detailed statistical analyses showed unequivocally that the content of aluminium in brain tissue in MS was significantly higher than that in non-MS control group. This is the first unequivocal confirmation that aluminium is increased in brain tissue in MS (Mold et al. 2018b) and it further supports the contention that the body burden of aluminium is elevated in MS compared to individuals without MS (Exley et al. 2006; Jones et al. 2017). The data do not prove a role for aluminium in MS, but a higher content of a neurotoxin and powerful pro-oxidant (Verstraeten et al. 1997; Exley 2004) in brain tissue in MS cannot be discarded as a putative aetiological factor in a disease with both genetic and environmental components (Thompson 2017). The results also have translational aspects in that they support previous research in which silicon-rich mineral waters were suggested as potential long-term therapies in treating MS (Jones et al. 2017).