A Pilot Study Measuring Aluminum in Bone in Alzheimer’s Disease and control Subjects Using in vivo Neutron Activation Analysis

Abstract

Aluminum, being the most abundant metal in the earth’s crust, is widely distributed in the environment, and is routinely taken up by the human body through ingestion and inhalation. Aluminum is not considered an essential element and it can be toxic in high concentrations. Most of the body burden of aluminum is stored in the bones. Aluminum has been postulated to be involved in the causality of Alzheimer’s disease. A system for non-invasive measurement of bone aluminum using the in vivo neutron activation analysis technique has been developed and previously reported in the literature by our group. The results are reported as ratio of Al to Ca in order to eliminate the variations in beam parameters and geometry as well as the physical variations among the subjects such as size of the hand and bone structure. This pilot study included 30 subjects, 15 diagnosed with Alzheimer’s disease in mild and moderate stages and 15 control subjects, all of whom were 60 years of age or older. The mean value of aluminum for the control group was 2.7±8.2μg Al/g Ca (inverse-variance weighted mean 3.5±0.9μg Al/g Ca) and for the Alzheimer’s disease subjects was 12.5±13.1μg Al/g Ca (inverse-variance weighted mean 7.6±0.6μg Al/g Ca). The difference between the mean of the Alzheimer’s disease group and the mean of the control group was 9.8±15.9μg Al/g Ca, with a p-value of 0.02. An age-dependent linear increase in bone aluminum concentration was observed for all subjects. The difference in serum aluminum levels between the two groups did not reach significance.

Keywords

Aluminum Alzheimer’s disease bone gamma spectroscopy neutron activation analysis

INTRODUCTION

Alzheimer’s disease (AD) is the most common form of dementia accounting for up to 80% of dementia cases [1]. AD starts with a clinically-dormant phase followed by progressive loss of memory and cognitive functions [2]. According to World Alzheimer’s Report, as of 2015, an estimated 46.8 million people worldwide are suffering from dementia with an alarming projection of about 131.5 million by 2050 [3]. This translates into about 40 million cases of AD in 2015. Given the sharp increase in the projected number of AD cases in the coming years [3] and its significant impact on society, developing methods for understanding the cause and monitoring of symptoms of AD are of great importance.

Many studies have been conducted in an effort to identify the underlying causes of AD. Although genetic, lifestyle, and environmental components are known to be associated with the development of AD, the true cause remains unknown [4 –8].

Aluminum (Al) seems to be one of the most likely candidates among the environmental toxicants, partly due to its ubiquitous nature [9]. The contribution of Al to AD has been strengthened through multiple independent observations involving, among other factors, animal models, brain signaling pathways, environmental exposures, and Al-chelating pharmaceutical treatments of AD symptoms [10]. Al is the most abundant metal and the third most abundant element in the earth’s crust with widespread applications in industrialized societies, examples of which include food, medical, drinking water, cosmetic, and construction industries [11]. Moreover, Al has been classified as a neurotoxin [12, 13]. Diminished cognitive functions in rats have been observed following chronic exposure to Al [14]. Elevated levels of Al have been detected in neurofibrillary tangle-bearing neurons in experimental animals as well as postmortem analysis of human AD brains [15 –18]. Measuring Al has been of interest during the past few decades with the focus of the early work in the 1970s and 1980s being on dialysis encephalopathy due to the Al contamination in the water used for preparation of dialysate [19 –21]. In subsequent years, the growing need and interest to investigate Al with respect to occupational exposures [16, 22] and medical conditions such as parenteral nutrition [23 –25] led to the development of various in vitro techniques such as atomic absorption and emission spectrometry [17, 26] and neutron activation analysis involving sample processing (NAA) [27, 28], all of which were destructive methods or required sample processing.

Measuring and monitoring Al in the context of AD is a complex task as the organ of interest for measuring Al in AD is the brain. Penetrating the brain for the purpose of in vivo measurement of Al requires high doses of radiation as the brain is protected by the skull, rendering the method impractical, given the possibility of an Al signal being generated in the skull, masking that of the brain. At the current state of technological advancement, measurement of Al in the brain can only be performed postmortem. Blood, urine, cerebrospinal fluid, and bone have been investigated in vitro as possible proxies for the brain, with elevated levels of Al in these organs observed and reported in the literature [29]. Serum measurement is the most common method of monitoring Al in clinical practice and can be used in conjunction with other measurements. In vitro methods of measuring Al suffer from the high risk of contamination, due to abundance of Al. Furthermore, urine is considered as a better measure of acute exposure to Al [11]. About 60% of the body burden of Al is stored in the bone, rendering the skeleton the Al depot of the body [11]. Bone Al measures are considered to be less susceptible to short-term exposures and better reflect longer term aggregate exposure. Conventional bone Al measurement, however, requires the invasive sample collection procedure of bone biopsy, which inhibits subject participation.

In vivo neutron activation analysis (IVNAA) is a method of measuring various elements in the human body using low doses of ionizing radiation. As a non-invasive method, it can be used for monitoring Al levels in humans and screening patients at the early stages of the disease. Furthermore, it does not require sample preparation, which significantly reduces the risk of contamination. IVNAA has been in use for medical and occupational purposes for several decades. The first in vivo measurement of Al was performed in 1980 by Williams et al. [30]. IVNAA of Al has since been used by various research groups and has undergone major improvements to increase its sensitivity to match the current low levels of Al detected in humans [31 –36]. A comprehensive review of the history of in vivo Al measurement has been recently published in the Journal of Alzheimer’s Disease, which can provide further details about IVNAA methodology [37]. The present work is the first clinical study involving the in vivo measurement of Al in bone in AD patients and control subjects. The goal is to i) investigate whether or not differences in levels of bone Al are found in subjects with AD versus normal controls, ii) determine the viability of measuring bone Al using IVNAA as a possible proxy for monitoring the Al levels in AD subjects, especially in the early stages of the disease, and iii) determine the capacity of the available IVNAA facility for measuring current Al levels in the bone.

MATERIALS AND METHODS

IVNAA setup

The measurements were performed at the IVNAA facility at the McMaster Accelerator Laboratory (MAL). This facility, with a Tandetron accelerator (HVEC) and a 4π detection system, has been developed for the purpose of measuring trace elements in the human hand. It has been previously utilized to perform human measurements of Al in healthy and occupationally exposed individuals [35, 36], as well as other elements such as manganese and fluorine [38, 39]. The details of the irradiation and detection system have been discussed in previous publications [36 , 41]. The Tandetron is a 1.25 MV high-current tandem accelerator capable of accelerating protons to the energies of up to 2.5 MeV. Neutrons are produced through the ⁷Li(p,n)⁷Be reaction when high-energy protons impinge on the thick ⁷Li target enclosed in the irradiation cavity. The irradiation cavity is designed to maximize the thermal neutron field and, therefore, activation of the target, while reducing the radiation dose outside the cavity to below the regulatory dose limits. The former is achieved through using polyethylene as the moderating material to thermalize the neutrons further and graphite reflectors to reflect the stray neutrons back into the irradiation cavity. The latter is accomplished by adding shielding and lead filters to absorb the neutrons and the gamma-rays produced in the target and other unwanted radiation. Following thermalization, neutrons are incident on the sample for a specific period of time allowing for the neutron activation to occur. This is achieved inside a specially designed irradiation cavity that is large enough to accommodate the human hand [41]. The detector assembly consists of 9 NaI(Tl) scintillation detectors arranged in a 4π geometry with an opening for inserting the sample [40]. The digital processing system (DSP) then acquires and compiles the data from the detectors. The DSP system is designed such that it can collect data in coincidence and anticoincidence modes, in addition to single detector responses. This is an important feature in measuring Al, as anticoincidence significantly reduces the spectral interference of the adjacent chlorine peak located only 0.14 MeV away from the Al peak(Fig. 1).

Upon bombardment with thermal neutrons, ²⁷Al nuclei in the sample absorb neutrons through the reaction ²⁷Al(n, γ)²⁸Al and the radioisotope ²⁸Al is produced with a thermal neutron cross-section of 0.23 barn. The unstable ²⁸Al decays to ²⁸Si with a half-life of 2.25 minutes and a 1.78 MeV gamma-ray is emitted with an emission probability of 100%.

The choice of proton energy determines the maximum energy of the neutrons which not only has important implications in the context of radiation dose to the subject, but also greatly affects the activation of other elements in the biological sample. While multi-element activation can provide a unique opportunity in terms of investigating multiple elements, it can also be a cause for undesirable interferences. In the case of Al, the interference from phosphorus and silicon, naturally present in the bone, are of great importance as they share the same activation product as ²⁷Al. Phosphorus and silicon have cross-sections for fast neutron activation through the reactions ³¹P(n,α)²⁸Al and ²⁸Si(n,p)²⁸Al. According to ICRP 23 report on the reference man [42], the human body contains 780 g and 18 g of phosphorus and silicon, respectively. At the proton energy of 2.3 MeV, the maximum neutron energy is 0.55 MeV which is far below the activation thresholds of 2.01 MeV for the ³¹P(n,α)²⁸Al and 4.00 MeV for the ²⁸Si(n,p)²⁸Al reactions, respectively. Therefore, these elements are not activated and their interference with ²⁸Al is eliminated by the production of neutrons with energies well below the threshold energies.

Hand phantom measurements

Hand phantoms were made to test the system, generate the calibration line and find the minimum detectable limit (MDL) of the system. Most of the bone in the hand is cortical. The phantoms were made in triplicates and according to the composition of human hand comprising both the cortical bone and soft tissue as published in ICRP 23 [42], including only those elements with significant neutron activation cross-sections and abundances, namely Ca, Na, Cl, and Al with reactions ⁴⁸Ca(n, γ)⁴⁹Ca, ²³Na(n, γ)²⁴Na, ³⁷Cl(n, γ)³⁸Cl, and ²⁷Al(n, γ)²⁸Al, respectively. Each phantom contained 14.9 g Ca, 1.19 g Cl, and 1.25 g Na. Al was added in triplicates, varying concentrations corresponding to zero, 16.8, 33.6, 50.3, 67.1, 134.2, and 335.6μg Al/g Ca. It is noted, as mentioned above, that while phosphorus and silicon are both present in the skeleton and have cross-sections for fast neutron activation, they are not included in the phantom composition since there is no probability for their activation given the neutron beam energy.

As mentioned previously, one important consideration when measuring Al is contamination. Al has a wide presence in the environment and is found in trace amounts in many materials. Therefore, it is critical to remove all sources of Al contamination in the phantoms if the results are to be accurate and reproducible. The original high density polyethylene (HDPE) Nalgene^TM bottles used for preparing the phantoms were found to have significant amounts of Al. Further measurements on the supposedly aluminum-free low density polyethylene (LDPE) bottles revealed that there were still trace amounts of Al present. In order to remove the possibility of activating the aluminum in the bottles, the contents of the phantoms were transferred to un-irradiated bottles immediately after irradiation and prior to detection. Moreover, all the chemicals used in the production of the phantoms were chosen from high purity chemical compounds (Sigma Aldrich) to avoid Al contamination and reduce uncertainty in the phantom measurements and calibration line.

Phantoms or subjects were exposed to thermal neutrons for 45 s with the proton energy of 2.3 MeV and the proton current of 400μA following which they were transferred to the detection system located at some distance from the irradiation facility in order to avoid activating the detector crystals. The mean transfer time for the phantoms was 30 s. Once in the detector, the gamma spectra were collected for 10 cycles of 60 s. The total time of 10 min was chosen to allow for multiple half-lives of Al to be detected while minimizing inconvenience to the subjects. The effective dose to the subjects was 0.21 mSv, based on the recent dose simulations and measurements [43], which is in the same order of magnitude as a chest x-ray (0.1 mSv) [44].

In addition to the phantoms, the available National Institute of Standards and Technology’s (NIST) standard reference material 1400 which consists of bone ash with certified Ca content was measured in the same way as the calibration standards and subjects for the purpose of validating the measurements. The measured NIST standard weighed 23.1±0.1 g and contained 38.18% ± 0.13% of Ca and 530μg/g of Al. Although the concentration of Al in the sample is not certified, it still provided reassurance of theperformance of the system used in this work.

Clinical study

This in vivo study was approved by the research ethics boards of McMaster and Ryerson Universities in southern Ontario. After providing written informed consent, 15 AD (7 females and 8 males) and 15 control (9 females and 6 males) subjects were recruited for the study from the Hamilton, Ontario area by the collaborating geriatrician. Where it was felt that AD subjects were incapable of providing consent, consent was obtained from their legal proxies. AD subjects were diagnosed according to DSM IV criteria, and were in the mild to moderate stages based upon Clinical Dementia Rating Scale (CDR) and the Standardized Mini-Mental State Examination (sMMSE) [45] shown in Tables 1 and 2, respectively. The control group had no history of AD or other types of dementia. The mean age of the participants was 77.6 with a range of 63–89 years. The mean ages for AD and control groups were 80.2 and 75, respectively. The irradiation and counting protocols were the same as phantoms. All subjects received a brief orientation before and upon arrival at the facility. Each subject was seated in a wheelchair for convenience. Using a wheelchair also facilitates safer and faster transfer between the irradiation and counting rooms. Subjects were fitted with a water sleeve which was filled with water once the subject’s arm was correctly positioned in the irradiation cavity. Hydrogen, and therefore, water is an effective moderator for neutrons, thus providing additional reduction in the dose. It is important to fill the sleeve with water in such way that there is no or negligible gap between the irradiation cavity wall and subject’s arm so to avoid unnecessary radiation exposure to the rest of the body and the accompanying caretaker. Both the subject and caretaker were fitted with electronic dosimeters to measure their neutron and gamma doses actively. Following the 45-s irradiation and the removal of the water sleeve, the subject was wheeled to the detection room by the caretaker, where the irradiated hand was positioned in the detection system for counting. The median transfer time for the subjects was 33 s with a range of 21–69 s. The subjects were instructed to remove all items of jewelry, as well as nail polish from the hand being irradiated to avoid potential contamination, although the latter has been shown not to contain Al [46]. The hand irradiation did not produce any skin irritation, temperature change, and/or discomfort to the subjects.

In addition to bone Al measurements, serum Al was measured in the subjects prior to IVNAA of the bone. However, four of the study subjects refused to participate in the blood sample collection (2 AD subjects and 2 controls). For these individuals, only bone Al measurements were performed.

Data analysis

Data analysis was limited to the relevant regions in the spectra which included Al and Ca peaks at energy ranges of 1.5–1.9 MeV, and 2.75–3.2 MeV, respectively. Mathematical models using the Marquardt algorithm involving non-linear least squares fitting were used to find the areas under Al (1.78 MeV) and Ca (3.08 MeV) peaks. Analyzing the Al peak area is particularly challenging due to the presence of the more biologically abundant Cl at 1.64 MeV adjacent to the Al peak. For Al, the mathematical model involved two Gaussian functions plus a linear background to account for both Al and Cl peaks. The Ca peak is well separated, and therefore, was fitted with one Gaussian and a linear background. The spectra were analyzed for each of the 10 cycles and the inverse-variance-weighted mean of all cycles was taken as the Al area. Corrections were made to account for radioactive decay during transfer and data acquisition as well as the inherent delay in the electronics.

The activation, and therefore, content of radioisotopes found using the current method of IVNAA depends on the neutron fluence which may vary among different measurements due to variations in neutron production within the target. Additionally, variations in the size and shape of each individual’s hand, possible movement of subject, and positioning within the irradiation cavity introduce further complexities. These variations can be accounted for by normalizing the Al content to Ca which provides an index of Al levels per unit bone mass.

RESULTS

Spectra of a phantom and an AD subject are shown in Fig. 1, which demonstrates that the composition of the phantom mimics well the major and relevant neutron-activated radionuclides found in the hand.

Figure 2 illustrates the calibration line generated based on phantom measurements. The MDL is defined as twice the uncertainty of the zero concentration phantom divided by the slope of the calibration line. The achieved MDL in phantoms was 5.2μg Al/g Ca or 76.9μg of Al. The MDL can also be defined based on the in vivo measurements as twice the median of the uncertainty in all measurements which was 7.2μg Al/g Ca.

The NIST 1400 sample was measured on three separate occasions, the results of which are provided in Table 3. The mean of the three measurements was 1372.5±65.3μg Al/g Ca, which agrees with the NIST 1400 Al/Ca ratio of 1388.0μg Al/g Ca (uncertified) within the error margin of the measurements.

Figure 3a and 3b show the Al/Ca ratio in hand for control and AD subjects, respectively. The error bars represent the uncertainty of one sigma (68% confidence level). The range of concentrations for control subjects varied from –21.9±7.6 to 15.0±4.7μg Al/g Ca with a mean of 2.7±8.2μg Al/g Ca (inverse-variance weighted mean 3.5±0.9μg Al/g Ca). For the AD subjects, the mean concentration was 12.5±13.1μg Al/g Ca (inverse-variance weighted mean 7.6±0.6μg Al/g Ca) with a range of –3.9±2.4 to 37.4±5.3μg Al/g Ca. The difference in the Al/Ca ratio between AD and control subjects was 9.8±15.9μg Al/g Ca (inverse-variance weighted mean difference 4.0±0.1μg Al/g Ca), found to be significant at the 95% confidence level p-value of 0.02 based on the student’s t-test, suggesting higher bone Al content in the AD subjects (The six highest Al/Ca values were observed in the hand bone of AD subjects).

The relationship between age and Al content of the bone for all subjects measured in this study is shown in Fig. 4. A linear increase in the Al/Ca ratio was observed with increasing age which was significant with p < 0.005. The linear function provided the best fit with r = 0.515. A linear relationship also best described both AD and control groups.

No correlation between CDR (p = 0.24) or MMSE rating (p = 0.15) and Al/Ca ratio in the AD subjects was observed.

The mean serum Al concentrations in control and AD subjects were 190.2±123.5 nmol/L and 198.9±111.8 nmol/L, respectively, which did not reach significance with a t-test p-value of 0.86.

DISCUSSION

The spectra shown in Fig. 1 demonstrate the close agreement between the radioisotopes found in human subjects and that of ICRP’s reference man mimicked in the phantoms. There is a difference, however, in the total number of counts in the two spectra. This is due to the differences in hand size and positioning of the subjects compared to the phantoms, as well as the variations in the total number of neutrons emitted.

MDL values of 5.2μg Al/g Ca for phantoms and 7.2μg Al/g Ca for in vivo hand measurements achieved in this study are major improvements to the previously reported figures in the literature by our group [36]. Phantom and in vivo MDLs were reduced by factors of 1.61 and 1.66, respectively, from the previously reported values of 8.32μg Al/g Ca and 12μg Al/g Ca [36]. This improvement can be attributed to higher proton energy used in the current study, higher quality phantoms and improved analysis.

Al content of human bone has been previously measured [24 , 47]. However, the results of measurements have been reported in different ways, depending on the technique applied and method of normalization. Normalization is usually done in relation to dry weight, wet weight, or calcium content of the bone without a direct means of converting between methods, which in turn leads to discrepancies between the reported values. Comparison of the results is thus limited either to studies that use the same method of normalization or to using crude conversion factors between different methods. One such conversion factor is that of dry weight (dw) to calcium, calculated on the basis of the ratio of the inorganic component of the bone crystal, representing the dry weight, and the calcium content, yielding a conversion factor of approximately 4. Davis et al. [35] measured Al/Ca ratio in the hand bone of healthy adults living in southern Ontario using IVNAA. However, due to the relatively high MDL of the study (phantom MDL was 19.5±1.5μg Al/g Ca andin vivo MDL was 28.0μg Al/g Ca), no direct comparison of bone Al concentrations can be drawn between their results and the control subjects in the present work. Bone Al content of 1.06μg/g dw in healthy controls reported by Hellström et al. [47] agrees with the 2.7±8.2μg Al/g Ca found in this work, once the conversion factor of 4 is applied. Kruger et al. measured the Al content of bone samples from 7 parenteral nutrition patients and 18 control subjects [24]. They found an average Al content of 32.0±18.7μg/g dw and 2.6±1.8μg/g dw in the parenteral nutrition and control subjects, respectively. The bone Al content of the control group in Kruger’s study agrees with the results of the present work within the margin of uncertainty, although the mean value in Kruger’s study is higher.

The Al/Ca ratio can be treated as the index of elevated Al per unit mass of bone. For the number of participants and the MDL of the current experimental setup, the results indicate a significant difference in Al/Ca in the hand bone between the control and AD group at 95% confidence level, suggesting a possible association between accumulation of Al in bone and the presence of AD. Additionally, bone Al results signify that the skeleton can potentially be used as a proxy system for the brain; however, more research is required to investigate this outcome further. Furthermore, IVNAA has proven a promising tool for such measurements and also monitoring the Al levels in the bone. It is noted, however, that the present results should be treated with caution for several reasons. Firstly, these findings are based on the relatively small sample size of 30 subjects. Secondly, despite the major improvements in the MDL, most of the in vivo measurements are close to or below thein vivo detection limit. As such, for these subjects no individual interpretation of Al/Ca ratio can be drawn and the results would be reported as “not detectable” in a clinical setup. Lastly, higher Al/Ca ratios than those reported here have been observed in occupationally exposed, but otherwise healthy individuals [36]. Thus, the likely conclusion based on the present findings would be that Al could be considered as a contributing factor in the causality of AD; however, more work is required to investigate its role further.

The current findings challenge those reported by Hellström et al. [47] where no significant difference in the Al content of bone was observed following in vitro measurement of iliac bone in AD and control subjects.

The increase in the Al/Ca ratio with age in the study subjects is in agreement with the work of Hellström et al. [47]. They also reported an age-dependent increase in bone Al concentration which followed an exponential function. The linear relationship found in the current study, and therefore the discrepancy in the functions governing the increase, may be due to the smaller size of this study and the restricted age range of these subjects, compared to the work of Hellström et al. [47]. It is a known fact that human bone loses calcium and other minerals with age as the bone mass decreases. However, the extent to which bone Al is lost relative to Ca is not known and may be a factor in the increase in the Al/Ca ratio with age. The extensive whole body Ca IVNAA measurements conducted by Ellis [48] showed a decline in total body calcium in both male and female subjects with age. This decline, however, was not significant within the margins of error for each of the 5-year age groups relevant to the ages studied in the present work.

Serum Al is usually considered as an indication of acute exposures, whereas bone Al better reflects the Al body burden, which may explain the lack of significant difference in serum Al between control and AD groups. Furthermore, blood measurement of Al is subject to a high risk of contamination as Al is ubiquitous and special measures need to be taken to remove Al contamination from the laboratory environment. It is noted that serum Al measurements were performed at a hospital laboratory in Hamilton, ON, and not a research lab with Al-contamination measures in place. Therefore, the possibility of contamination cannot be ruled out. Additionally, 4 out of 30 subjects refused to participate in blood sample collection, which somewhat reduces the statistical significance of the serum Al results.

Another limitation of the current clinical study is that it does not include subjects with severe cases of AD. All AD subjects recruited for this study were categorized as mild and moderate, based on CDR and MMSE categories. This was mainly due to the challenges involved in recruiting subjects in advanced stages of the disease in the community, and the distress that measurement in the Tandetron Accelerator may have caused them. If possible, measuring bone Al in severe cases might shed more light on the pattern of Al accumulation through different stages of AD.

CONCLUSIONS

This work presents, for the first time, the results of in vivo measurements of Al in the hand bone of patients suffering from AD and control subjects. A significant difference with a p-value of 0.02 was observed in the Al/Ca ratio between the two groups. This finding was not inconsistent with the hypothesis of an association between raised levels of Al in the body and the risk of AD. However, these findings must be interpreted with caution. There was a small number of subjects involved in the study, n = 15 in each group. The levels of Al found in the AD remained within the range of values for normal subjects in a much larger study of biopsy samples [47]. There is evidence of higher levels of Al in bone in people with occupational exposure to Al and in people on long-term parenteral nutrition; and there is not strong evidence for an association between occupational exposure to Al or parenteral nutrition and the risk of AD. Larger studies are required to complement the current results further, hopefully with the inclusion of subjects with advanced AD. The majority of measurements reported here are close to or lower than the MDL of the system, warranting further investigation into possible ways of improving the sensitivity of in vivo bone Al measurements. The increasing trend in Al/Ca ratio with age observed in this study is consistent with other findings reported in the literature.

Footnotes

ACKNOWLEDGMENTS

The authors would like to extend their gratitude to Justin Bennett, the accelerator operator at the MAL for assisting in conducting experiments and patient irradiation. They also thank Dr. Farshad Mostafaei for his assistance in patient data collection and Michael Inskip for valuable advice and access to his laboratory. This work was made available through the funding provided by the Canadian Nuclear Laboratories (CNL) and Natural Sciences and Engineering Research Council of Canada (NSERC).

Authors’ disclosures available online (.

References

Alzheimer’s Association (2015) 2015 Alzheimer’s disease facts and figures. Alzheimers Dement 11, 332–384.

Amieva

, Le Goff

, Millet

, Orgogozo

, Peres

, Barberger-Gateau

, Jacqmin-Gadda

, Dartiques

(2008) Prodromal Alzheimer’s disease: Successive emergence of clinical symptoms. Annals Neurol 64, 492–498.

Prince

, Wimo

, Guerchet

, Ali

, Wu

, Prina

(2015) World Alzheimer Report, http://www.alz.co.uk/research/world-report-2015, Accessed 8 December, 2015.

Nee

, Eldridge

, Sunderland

, Thomas

, Katz

, Thompson

, Weingartner

, Weiss

, Julian

, Cohen

(1987) Dementia of the Alzheimer’s type- Clinical and family study of 22 twin pairs. Neurology 37, 359–363.

Raiha

, Kaprio

, Koskenvuo

, Rajala

, Sourander

(1996) Alzheimer’s disease in Finnish twins. Lancet 347, 573–578.

Räihä

, Kaprio

, Koskenvuo

, Rajala

, Sourander

(1998) Environmental differences in twin pairs discordant for Alzheimer’s disease. J Neurol Neurosurg Psychiatry 65, 785–787.

Bekris

, Yu

, Bird

, Tsuang

(2010) Genetics of Alzheimer disease. J Geriatr Psychiatry Neurol 23, 213–227.

, Tan

, Wang

, Jiang

, Tan

, Zhao

, Li

, Wang

, Yu

(2015) Meta-analysis of modifiable risk factors for Alzheimer’s disease. J Neurol Neurosurg Psychiatry 86, 1299–1306.

Walton

(2014) Chronic aluminum intake causes Alzheimer’s disease: Applying Sir Austin Bradford Hill’s causality criteria. J Alzheimers Dis 40, 765–838.

10.

Bhattacharjee

, Zhao

, Hill

, Percy

, Lukiw

(2014) Aluminum and its potential contribution to Alzheimer’s disease. Front Aging Neurosci 6, 62.

11.

Yokel

(2013) Aluminum. In Encyclopedia of Human Nutrition, 3rd ed. Academic Press, Waltham, MA, pp. 57–63.

12.

Simonsen

, Johnsen

, Lund

, Matikainen

, Midtgard

, Wennberg

(1994) Methodological approach to the evaluation of neurotoxicology data and the classification of neurotoxic chemicals. Scand J Work Environ Health 20, 1–12.

13.

Kopeloff

, Kopeloff

(1942) Recurrent convulsive seizures in animals produced by immunologic and chemical means. Am J Psychiatry 98, 881–902.

14.

Walton

(2012) Cognitive deterioration and associated pathology induced by chronic low-level aluminum ingestion in a translational rat model provides an explanation of Alzheimer’s disease, tests for susceptibility and avenues for treatment. Int J Alzheimers Dis 2012, 1–17.

15.

Crapper

, Krishnan

, Dalton

(1973) Brain aluminum distribution in Alzheimer’s disease and experimental neurofibrillary degeneration. Sci New Ser 180, 511–513.

16.

Exley

, Vickers

(2014) Elevated brain aluminium and early onset Alzheimer’s disease in an individual occupationally exposed to aluminium: A case report. J Med Case Rep 8, 41.

17.

Andrasi

, Pali

, Molnar

, Kosel

(2005) Brain aluminum, magnesium and phosphorus contents of control and Alzheimer-diseased patients. J Alzheimers Dis 7, 273–284.

18.

Walton

(2006) Aluminum in hippocampal neurons from humans with Alzheimer’s disease. Neurotoxicology 27, + 285–394.

19.

Ellis

, McCarthy

, Herrington

(1979) Bone aluminium in haemodialysed patients and in rats injected with aluminium chloride: Relationship to impaired bone mineralisation. J Clin Pathol 32, 832–844.

20.

Alfrey

, Hegg

, Craswell

(1980) Metabolism and toxicity of aluminum in renal failure. Am J Clin Nutr 33, 1509–1516.

21.

Parsons

, Davies

, Goode

, Ogg

, Siddiqui

(1971) Aluminium in bone from patients with renal failure. Br Med J 4, 273–275.

22.

Elinder

, Ahrengart

, Lidums

, Pettersson

, Sjogren

(1991) Evidence of aluminium accumulation in aluminium welders. Br J Ind Med 48, 735–738.

23.

Klein

, Targoff

, Ament

, Sherrard

, Bluestone

, Young

, Norman

, Coburn

(1980) Bone-disease associated with total parenteral nutrition. Lancet 2, 1041–1044.

24.

Kruger

, Parsons

, Galusha

, Morrissette

, Recker

, Howard

(2013) Excessive aluminum accumulation in the bones of patients on long-term parenteral nutrition: Postmortem analysis by electrothermal atomic absorption spectrometry. JPEN J Parenter Enteral Nutr 38, 728–735.

25.

Fewtrell

, Bishop

, Edmonds

, Isaacs

EB LA

(2009) Aluminum exposure from parenteral nutrition in preterm infants: Bone health at 15-year follow-up. Pediatrics 124, 1372–1379.

26.

House

, Esiri

, Forster

, Ince

, Exley

(2012) Aluminum, iron, and copper in human tissues donated to the medical research council’s cognitive function and ageing study. Metallomics 4, 56–65.

27.

Bogdanovich

, Krishnan

, Pei

, Hercz

, Lui

, Hancock

, Harrison

, Bone

(1992) Non-destructive bone aluminum assay by neutron activation analysis. J Radioanal Nucl Chem 164, 293–302.

28.

Sharif

AAM

, Ghafourian

, Ahmadiniar

, Waqif

, Saber-Tehrani

, Ghods

(2004) Determination of aluminum levels in serum and red blood cells from long-term haemodialysis patients using instrumental neutron activation analysis. J Radioanal Nucl Chem 262, 473–477.

29.

Virk

, Eslick

(2015) Aluminum levels in brain, serum, and cerebrospinal fluid are higher in Alzheimer’s disease cases than in controls: A series of meta-analyses. J Alzheimers Dis 47, 629–638.

30.

Williams

, Elliott

, Boddy

, Haywood

DJK

, Henderson

, Harvey

, Kennedy

(1980) Whole-body aluminum in chronic renal failure and dialysis encephalopathy. Clin Nephrol 14, 198–200.

31.

Ellis

, Kelleher

, Savory

, Wills

(1988) In vivo monitoring of skeletal aluminum burden in patients with renal failure. J Radioanal Nucl Chem 124, 85–95.

32.

Green

, Chettle

(1992) A feasibility study of the in vivo measurement of aluminum in peripheral bone. Phys Med Biol 31, 2287–2296.

33.

Wyatt

, Ryde

, Morgan

, McNeil

, Hainsworth

, Williams

(1993) The development of a technique to measure bone aluminium content using neutron activation analysis. Physiol Meas 14, 327–335.

34.

Pejović-Milić

, Byun

, Comsa

, McNeill

, Prestwich

, Chettle

(2005) In vivo measurement of bone aluminium: Recent developments. J Inorg Biochem 99, 1899–1903.

35.

Davis

, Pejović-Milić

, Chettle

(2008) In vivo measurement of bone aluminum in population living in southern Ontario, Canada. Med Phy 35, 5115.

36.

Aslam , Davis

, Pejović-Milić

, Chettle

(2009) Noninvasive measurement of aluminium in human bone: Preliminary human study and improved system performance. J Inorg Biochem 103, 1585–1590.

37.

Mohseni

, Chettle

(2016) A history of in vivo neutron activation analysis in measurement of aluminum in human subjects. J Alzheimers Dis 50, 913–926.

38.

Aslam , Pejović-Milić

, Chettle

, McNeill

(2008) Quantification of manganese in human hand bones: A feasibility study. Phys Med Biol 53, 4081–4092.

39.

Mostafaei

, McNeill

, Chettle

, Wainman

, Pidruczny

, Prestwich

(2015) Measurements of fluorine in contemporary urban Canadians: A comparison of the levels found in human bone using in vivo and ex vivo neutron activation analysis. Physiol Meas 36, 465–487.

40.

Byun

, Prestwich

, Chin

, McNeill

, Chettle

(2006) 4-pi NaI(Tl) detector array for in vivo neutron activation analysis. IEEE Trans Nucl Sci 53, 2944–2947.

41.

Pejović-Milić

, Arnold

, McNeill

, Chettle

(2000) Monte Carlo design study for in vivo bone aluminum measurement using a low energy accelerator beam. Appl Radiat Isot 53, 657–664.

42.

ICRP (1974) Report of the task group on reference man, Pergamon, Publication 23, Oxford.

43.

Darvish-Molla

, Prestwich

, Byun

(2015) Comprehensive radiation dose measurements and Monte Carlo simulation for the 7Li(p,n) accelerator neutron field. Radiat Prot Dosimetry. doi: 10.1093/rpd/ncv428

44.

NCRP (2009) Ionizing Radiation Exposure of the Population of the United States, Report No 160.

45.

Molloy

, Alemayehu

, Roberts

(1991) Reliability of a Standardized Mini-Mental State Examination compared with the traditional Mini-Mental State Examination. Am J Psychiatry 148, 102–105.

46.

Favaro

, Bode

, De Nadai Fernandes

(2005) Trace elements in nail polish as a source of contamination when used in epidemiological studies. J Radioanal Nucl Chem 264, 61–65.

47.

Hellström

, Mjoberg

, Mallmin

, Michaelsson

(2005) The aluminum content of bone increases with age, but is not higher in hip fracture cases with or without dementia compared to controls. Osteoporos Int 16, 1982–1988.

48.

Ellis

(1990) Reference man and woman more fully characterized. Variations on the basis of body size, age, sex, and race. Biol Trace Elem Res 26-27, 385–400.