J. Trace Elements Med. BioI. Vol. 10, pp. 6-11 (1996)
Determination of Aluminium in Samples from Bone and Liver of Elderly Norwegians D. HONGVE*, S. JOHANSEN*, E. ANDRUCHOW*, E. BJERTNESS**, G. BECHER* and J. ALEXANDER* *Department of Environmental Medicine, **Department of Population Health Sciences, National Institute of Public Health, Geitmyrsveien 75, N-0462 Oslo, Norway (Received September/Novemher 1995)
Recent reports associating aluminium with several skeletal and neurological disorders in humans suggest that exposure to aluminium may pose a health hazard. In connection with an epidemiological study on aluminium and Alzheimer's disease, information on the body burden of aluminium was needed. Aluminium is stored in the body mainly in bone and soft tissues such as the liver. Therefore, an analytical procedure was developed for the determination of aluminium in the head of femur and the liver using graphite furnace atomic absorption spectrometry. Problems in such analyses are associated with low levels of aluminium, risk of contamination during sample preparation, inhomogeneity of the sample tissues, and complexity of the matrices. Bone samples gave poorer repeatability compared to liver samples and up to four replicate analyses were performed. In the analyses of the digested bone samples, severe chemical interference occurred when these either contained much dissolved bone mineral or had a high AI concentration. It is possible that this interference is brought about by formation of aluminium phosphate in the graphite tube. The analytical results for bones are given relative to ash weight because the samples sometimes contained a lot of fat. The ash weight of the bone tissue varied between 8 and 69 per cent. The material consisted of 84 samples of head of femur and 95 liver samples from deceased elderly Norwegians. Two cases had high Al levels in bone (16.8 and 18.0 mg!kg ash weight) and liver (10 and 22.7 mg!kg dry weight) tissues. The remaining cases showed about ten to fifteen-fold variation of the Al level in both liver (0.4 - 5.7 mg!kg) and bone (0.5 - 5.8 mg!kg). There was no correlation between the level in liver and bone when the two cases with the highest levels were excluded. Keywords: Aluminium, bone, liver, human, electrothermal atomic absorption spectrometry.
Introduction It is well known that aluminium may cause neurotoxicity and osteomalacia when the gastro-intestinal barrier to absorption is by-passed or renal excretion impaired. Usually, the development of such toxic effects is accompanied by strongly raised levels of aluminium in serum, the brain and bone tissue (1-8). -----------------
Reprint reljuests to Dr. lan Alexander, Department of Environmental Medicine, National Institute of Public Health, Geitmyrsveicn 75, N-0462 Oslo, Norway. © 1996 by Gustav Fischer Verlag Stuttgart . lena . New York
In a number of epidemiological and chemico-neuropathological studies, aluminium has also been linked to Alzheimer's disease (AD) (4,6,9-13). On the other hand, recent investigations may have weakened this link and have questioned the role of aluminium in the etiopathogenesis of this disease (14-17). When carrying out epidemiological studies on the relationship between AD and aluminium, it is mandatory to have valid and reliable parameters of aluminium exposure. In the present study, analytical techniques were developed to determine levels of aluminium in bone and liver
Determination of aluminium in samples from bone and liver of elderly Norwegians
in connection with an epidemiological study on deceased AD patients and controls (Bjertnes et aI., sUbmitted). It has been suggested that aluminium in bone tissue might reflect long-tenn exposure, whereas the level in soft tissues such as liver may reflect more recent exposure (1,7). However, the half-lives of aluminium in these compartments are not known. Detennination of aluminium in the blood and tissues has primarily been done in relation to renal failure associated encephalopathy and bone disease. In these cases, where concentrations of aluminium are very high, staining of histological specimens with a complexing agent may be used for quantitation of Al (8,18). In contrast, tissue levels of aluminium found in the general population, including those with AD, are generally one to two orders of magnitude lower (1,2,4,18,19-24). At these low levels, detennination of aluminium is particularly demanding and contamination during sample handling and preparation may create serious problems (20). The most common methods for detennination of Al in samples from the general population have been atomic absorption spectrometry (20-23) and inductively coupled plasma emission spectrometry (24). Different procedures in the tissue preparation, such as ashing, wet digestion, extraction of aluminium with a chelating agent or solubilization with tetramethyl ammonium hydroxide (20), have been employed, depending to some extent on whether the sample is soft tissue or bone. In the present study we have used electrothennal atomic absorption spectrometry as the instrumental technique. We present here the validation of the methods used for aluminium detennination and the levels found in bone and liver. The results of the epidemiological study, which also includes selection of material, clinical evaluation, post mortem histopathological classification and data analysis of disease related to aluminium levels, will be presented elsewhere (Bjertnes et aI., submitted).
was not perfonned because of the contamination risk. Instead, four subsamples of each specimen were analyzed.
Chemicals Ultrapure re-distilled nitric acid (Scanpure, 65%) obtained from ChemScan AS, Elverum, Norway was used in the digestion of samples. All other chemicals were commercially obtained and of p.a. quality. Reference serum, Serononn Trace Element, was obtained from Nycomed Phanna, Oslo, Norway.
Digestion Liver: Freeze-dried subsamples of about 100 mg were acid-digested in 7 mL Teflon vials with screw caps. 1.0 mL of nitric acid (65%, ultrapure) was added and the tissue dissolved overnight in closed vials at 105°C. The solutions were transferred to polyethylene vials and diluted with 10 mL of water. The final volumes were determined by weight. Bone: The femur head consists of the cortical bone and a trabecular bone structure with fat in lacunae inside it. Due to a high frequency of osteoporosis in the study group, there was a great variation in the relative amount of bone tissue. The proportion of bone substance was detennined by ashing half of each femur head at 550°C for 20 h in platinum crucibles. The weight after ashing ranged from 8 to 69 per cent of the freeze-dried sample. Samples (70-500 mg) containing both the outer cortical bone and the inner trabecular structure were cut from each ashed specimen with a stainless steel knife. These samples were dissolved in the same way as the liver, except that 2.0 mL of the nitric acid was used. Ca in the di-
Materials and Methods
84 samples from the head of femur (thigh bone) and the liver plus 9 samples of the liver only were obtained from deceased non-demented and from Alzheimer's patients during post mortem examination. The age of the subjects ranged from 66 to 98 years with a mean of 85 years. The samples were sealed in polyethylene bags, frozen in liquid nitrogen, freeze-dried and stored at -70°C. Subsampling: Preliminary analyses showed that aluminium could be rather heterogeneously distributed within different parts of the analyzed tissues. Homogenization
Time (sec) Figure 1. Al signal (solid line) from graphite tube with pyrolytic coating (A) and without coating (B). Broken line: Background signal.
D. Hongve, S. Johansen. E. Andruchow. E. Bjertness, G. Becher and 1. Alexander
gest was determined by AAS and Sr by AES, using standard analytical procedures (25).
Instrumental equipment The analysis was perfonned using a Perkin-Elmer 5100 Zeeman atomic absorption spectrometer equipped with an AS-60 autosampler. The sharpest signal peaks were obtained with pyrolytically-coated graphite tubes (Figure I). The use of a L'Vov platfonn did not result in further improvements. The pyrolytically coated graphite tubes supplied by Perkin-Elmer contained traces of alu-
Table I. Optimized time/temperature program for electrothermal AAS analysis of bone and liver samples Temperature DC
Gas !low) mL/min
140 100 1700',1500' 20 2700 2600
5 30 20 5 0
30 10 10 5 5 3
300 300 300 300 50 300
'bone, 'liver, )Purge gas: argon Conditions: Wavelength: 309.3 nm, slit: 0.7 nm, injected volume 20 flL
minium. However, after about ten repeated firings, the background was reduced to an insignificant level. The
technique for calibration could not be used (Figure 2). In
optimized time/temperature programs for bone and liver
the analysis of digested bone samples, severe chemical
are given in Table I.
interference occurred when a sample had a high mineral content or had a high aluminium concentration. Phos-
Analytical prohlems in analysis of hone
phate is the major chemical constituent of bone tissue and the digested bone samples contained up to 22 g PO/'lL. It
The analyses of digested liver samples were straight-
is therefore probable that the observed chemical interfer-
forward without any interference or background prob-
ence occurs through formation of aluminium phosphate
lems. In contrast, during the analyses of bone sample in-
in the graphite tubes. Thus, samples with a lot of bone
creasing background values were observed with increas-
material and/or high aluminium concentration had to be
ing aluminium concentration in the digest, which sup-
diluted in order to increase the signal-to-background ratio
pressed signal peaks. Therefore, the standard addition
(Figure 3). Fumes evolving during the atomization caused severe degradation of both the graphite tube and the contact cylinders. After about 100 firings, a high electrical resistance was observed and new tubes had to be
~ ~l:-_J:. . J_'_>_\. :.;.= =~" "'i' 0.Wl-t_-,-,."_!_/"'_\_1,~....:,_",,._=_
~7061 !\_~, 08741-l---"--t_·v,---,,~,====.
Figure 2. Two examples of unsuccessful attempts to determine AI in bone by means of the standard addition technique. Two additions raising the concentration by 50 flg/L each time were made. The blank-corrected peak areas were: Sample I: I A: 0.2XO; I B, 1st addition: 00434: IC. 2nd addition: OAX7; Sample 2: 2A: 0.170; 2B, Ist addition: 00430: 2C, 2nd addition: 00403.
Figure 3. Examples of replicate analysis of a bone sample with high background and the same sample after 1:4 dilution. The concentrations calculated were: Replicate 2 Replicate I Undiluted: A: 44.7 flg/L B: 31.9 flg/L. Diluted (1:4): C: 2104 flg/L D: 22.0 flg/L.
Determination of aluminium in samples from bone and liver of elderly Norwegians
Table 2. Determination of aluminium in quality control samples Sample type
Serum (Ilg/L) 106 ±6 Water (Ilg/L) 36.0 Liver' (mg/kg)
Measured results mean range
106 103.8 -109.3 35.5 32 - 39 1.13- 1.85 1.44
2 1.8 28 0.24 15
'digest concentration 10 - 15 Ilg AI/L
~ ~oo oS %0<90
Validation To ascertain that there was no source of aluminium contamination, the sample preparation equipment was tested for leakage of aluminium to acid solutions and blank samples were run through the entire procedure. The limit of detection (LOD) was based from the standard deviation of signal intensities from repeated analysis of eight samples at an aluminium level of 0.6-0.7 mg/kg. The LOD defined at three times the standard deviation was 0.15 mg/kg. For control of the analytical quality, a liver sample of 10 g was homogenized and split into subsamples of 100
20 a -5
10 Al (mg/kg dry liver)
Figure 5. Relationship between aluminium in liver and in bone from 84 deceased elderly persons.
mg which were then digested separately. These samples, together with one laboratory-made aluminium solution and a certified serum standard, were analyzed at regular intervals between the test samples. The results are given in Table 2. An analogous quality control sample for bone could not be obtained without severe contamination of the sample during processing. It is reasonable to assume that the variation between subsamples in bone analyses exceeds the variation between repeated analyses of identical samples. Independent analyses of 16 samples of
10 Table 3. Concentration of aluminium in human tissues
Al concentration mg/kg Mean ± SO (Range)
2.2 ± 2.5
1.8 ± 1**
0.3 - 6.3**
Bone, Iliac crest
""-co 5 10 :;:;:
3.88 ± 1.73***
2.39 ± 1.18***
o j.amm1ilJlllll'i'il_ _111111 Samples Figure 4. Aluminium in bone ash (a) and liver (b) from elderly Norwegians. Each bar represents the mean and standard deviation of four subsamples separately digested and analyzed from each specimen.
Head of femur
3.3 ± 2.9***
5.7 ± 1****
2.6 ± 2.6****
*dry weight, **wet weight ***defatted dry weight, ****ash weight
D. Hongve, S. Johansen, E. Andruchow, E. Bjertness, G. Becher and J. Alexander
bone tissue at MRC Neurochemical Pathology Unit, Newcastle General Hospital, Newcastle upon Tyne, UK, showed very good agreement with the results of our laboratory (mean±SD: Oslo, 41.3±20.4llg/L and Newcastle, 42.0±23.4 Ilg/L; Pearson's coefficient of correlation: r2=0.992). A set of 24 liver samples was analysed twice in our laboratory at an interval of 2 months. The average difference between replicate analyses was 0.3 mg/kg dry weight, which is about 2% of the average concentration.
Results and Discussion
The relative amount of bone tissue in the femur head samples varied greatly due to osteoporosis. The ash weight relative to dry weight varied from 8 to 69 percent. Calcium and strontium, which accumulate exclusively in bone tissue, give fairly good measures for the amount of calcified bone in the samples. Thus, these elements were also determined in the bone samples. Linear regression analysis showed that for aluminium in bone tissue an almost identical distribution was obtained when the results were given relative to calcium or to ash content (r2=0.998) because the ratio of calcium to ash was very constant (409±1O mg/g). The correlation was less good with strontium (r2=0.888), possibly reflecting different exposure to this element. The strontium concentration in ash was 0.25±0.06 mg/g. Two samples with extreme Al concentrations were not included in the calculation of correlation coefficients. Since ash weight is a fairly good measure for the amount of bone tissue, Al in the head of femur was expressed relative to the ash weight. The results of our analyses, given as average values in mg/kg ash for bone and mg/kg dry matter for liver, respectively are shown in Figure 4. Due to occasional large variations between replicate analyses, particularly for bone, replicate analyses were performed for each subsample to reduce the level of analytical uncertainty to less than 5 per cent. In addition to this, there was a variation between subsamples from the same specimen. The variation coefficients between subsamples were 22% for bone and 18% for liver. For liver samples two cases showing high levels of Al (10.0 and 22.7 mg/kg dry weight) had a medical record of gastric ulcer and had probably used large amounts of antacids containing aluminium. The remaining cases show about a ten-fold variation in aluminium concentration (0.4 to 5.7 mg/kg dry weight). The two cases showing the high levels of Al in liver had also high concentrations in bone (16.8 and 18.0 mg/kg ash weight). The remaining cases show about the same variation as in liver (0.5 to 5.8 mg/kg ash weight). There was no linear correlation be-
tween the level in liver and bone when the two cases with the highest levels were excluded (Figure 5). Our material, which consisted only of elderly deceased persons in a relatively narrow age range, did not allow a correlation analysis between aluminium content and age to be made. In a previous study, however, we found a tendency towards increasing bone aluminium values with age (23). The values obtained for liver are in good agreement with those from previous studies (see Table 3). Regarding reported levels for aluminium in bone among controls, samples have been taken from the iliac crest and values have been reported relative to wet weight, dry weight, and ash weight (Table 3). Usually, the iliac crest samples contain very little connective tissue, which may explain the minor differences found between studies calculating the values on a different basis.
While the analyses of liver samples was not accompanied by major problems, the analyses of bone samples required special precautions to circumvent problems such as inhomogeneity of the samples and chemical interference during AAS analysis. Our results for concentration of aluminium in bone seem to agree very well with previously reported data. Due to the large variation in the relative amount of bone tissue in our samples, it seems appropriate to express our values in relation to ash weight or to calcium. There is a good agreement between the present data and our previous data on aluminium (24) when aluminium concentration is expressed relative to the calcium content of the sample. In the present study, we measured the aluminium content in the whole sample, consisting both of cortical and trabecular bone. It has been reported that in uremic patients trabecular bone consistently contained more aluminium than cortical bone, whereas for controls the values did not seem to differ significantly between cortical and trabecular bone (1). Hence, the procedure we chose, to measure aluminium in the whole sample, should be valid.
This study was supported by the Norwegian Research Council, Programme on Environmental Epidemiology. We thank G.A. Taylor, MRC, Neurochemical Pathology Unit, Newcastle General Hospital, for valuable advice in making the aluminium analyses and for replicate analyses of selected samples.
Detennination of aluminium in samples from bone and liver of elderly Norwegians
References 1. ALFREY, A.C., LEGENDRE, G.R. AND KAEHNY, W.D. (1976) The dialysis encephalopathy syndrome: possible aluminium intoxication. New Engl. 1. Med., 294, 184-188
14. CRAPPER-MCLACHLAND, D.R. AND DE BONI, U. (1980) Aluminum in human brain disease - an overview. Neurotoxicologyl,3-16
2. ALFREY, A.C. (1980) Aluminium neurotoxicology in uremia. NeurotoxicoloRY 1,43-53
15. CANDY, J.M., KLINKOWSKI, J., PERRY, R.H., FAIRBAIM, A., OAKLEY, A.E., CARPENTER, T.A., ATACK, J.R., BLESSED, G. AND EDWARDSON, J.A. (1986) Aluminosilicates and senile plaque formation in Alzheimer's disease. Lancet i, 354-357
3. WILLS, M.R. AND SAVORY, J. (1983) Aluminium poisening: dialysis encephalopathy, osteomalacia and anaemia. Lancet, ii, 29-34
16. WISNIEWSKI, H.M., MORETZ, R.C., STURMAN, J.A., WEN, G.Y. AND SHEK, J.w. (1990) Aluminum neurotoxicity in mammals. Environ. Geoehem. Health 12,115-120
4. GANROT. P.O. (1986) Metabolism and possible health effects of aluminum. Environ. Health Persp. 65, 363-441
17. LANDSBERG, J.P., MCDONALD, B. AND WATT, F. (1992) Absence of aluminium in neuritic plaque cores in Alzheimer's disease. Nature 360, 65-68
5. ACKRILL, P. (1990) Clinical aspects of dialysis encephalopathy. Environ. Geochem. Health 12, 87-88 6. ARIEFF, A.I. (1990) Aluminum and the pathogenesis of dialysis dementia. Em'iron. Gern·hem. Health 12, 89-93 7 . DAY, J.P. (1990) Biochemistry of aluminium in relation to the assessment of aluminium exposure in humans. Environ. Geochem. Health 12.75-76 8. HALSE, J., NORDAL, K.P., DAHL, E. AND THOMASSEN Y. (1990) Aluminium metabolism in chronic renal failure: Environmental influences and regional differences in Norway. Environ. Geochem. Health 12,77-82 9. OTT, S.M., MALONEY, N.A., KLEIN, G.L., ALFREY, A.C., AMENT, M.E., COBURN, J.w. AND SHERRARD, DJ. (1983) Aluminum is associated with low bone formation in patients receiving chronic parenteral nutrition. Ann. Internal Med. 98,910-914 10. OTT, S.M., MALONEY, N.A., COBURN, J.w., ALFREY,A.C. AND SHERRARD, DJ. (1982) The prevalence of bone aluminum deposition in renal osteodystrophy and its relation to the response to calcitriol therapy. New En!;. 1. Med. 307, 709-713 11. VOGT, T. (1986) Water quality and health - study of a possible relation between aluminium and dementia. Social and Economic Studies No. 61 (in Norwegian). Central Bureau of Statistics of Norway, Oslo. 12. MARTYN, C.N., OSMOND, C., EDWARDSON, J.A., BARKER, DJP, HARRIS, E.C. AND LACEY, R.F. (\989) Geographical relation between Alzheimer's disease and aluminium in drinking water. Lancet, i, 59-62 13. FLATEN, T.P. (1990) Geographical associations between aluminium in drinking water and death rates with dementia (including Alzheimer's disease, Parkinson's disease and amyotrophic lateral sclerosis) in Norway. Environ. Geoc·hem. Health, 12,152-167
18. COPESTAKE, P. (1993) Aluminium and Alzheimer's diseasean update. Food Chem. Toxieol. 31, 679-683 19. ELLIS, H.A., MCCARTHY, J.H. AND HERRINGTON, J. (1979) Bone aluminum in hemodialyzed patients and in rats injected with aluminum chloride: relationship to impaired bone mineralization. J. Clin. Path. 32, 832-844 20. STEVENS, BJ. (1984) Electrothermal atomic absorption determination of aluminum in tissues dissolved in tetramethyl ammonium hydroxide. Ciin. Chem. 30, 745-747 21. BOYCE, B.F., ELDER, H.Y., ELLIOT, H.L., FOGELMAN, I., FELL, G.S., JUNOR, BJ., BEASTALL, G. AND BOYLE, I.T. (1982) Hypercalcaemic osteomalacia due to aluminium toxicity. Lancet ii, 1009-1013 22. HEAF, J.G., P0DENPHANT, J. AND ANDERSEN, J.R. (1986) Bone aluminium deposition in maintenance dialysis patients treated with aluminium-free dialysates: role of aluminium hydroxide consumption. Nephron 42,210-216 23. ALEXANDER, J., GJESSING, E., NORDAL, K.P., DAHL, E., HALSE, J. AND THOMASSEN Y. (1990) A preliminary study of aluminium in serum and other human materials in subjects from different areas of Norway. Environ. Geoehem. Health 12, 83-86 24. SEBERT, J.L., MARIE, A., GUERIS, J., HERVE, M.A., LEFLON, P., GARABEDIAN, M., SMADJA, A. AND FOURNIER, A. (1985) Assessment of the aluminum overload and of its possible toxicity in asymptomatic uremic patients: evidence for a depressive effect on bone formation. Bone 6, 373-375 25. Analytical methods for atomic absorption spectrophotometry (l982)The Perkin-Elmer Corporation, Norwalk Connecticut, U.S.A.