9 1987 by The Humana Press Inc. All rights of any nature, whatsoever, reserved. 0163-4984/87/1300-0035502.00
Origin and Resolution of the Aluminum Controversy Concerning Alzheimer's Neurofibrillary Degeneration S. S. KRISHNAN, *'1'4 J. E. HARRISON,3'4 AND D. R. C R A P P E R M C L A C H L A N z3
1Departments of Applied Chemistry and Chemical Engineering, 2physiology, and Wledicine, University of Toronto and 4Toronto General Hospital, Toronto, Ontario, Canada ABSTRACT Elevated concentrations of aluminum are found in regions of neurofibrillary change in brains with senile or presenile dementia of Alzheimer's type. The concentrations of aluminum found in the human disease are comparable to those found in experimental animals with aluminum-induced neurofibrillary degeneration (NFD). Although there are a number of reports confirming these observations, two laboratories have been unable to detect elevated levels in Alzheimer's disease. We conducted an interlaboratory study to resolve this discrepancy and traced the discrepancy to difficulties in analytical procedures. We concluded that failure to detect elevated aluminum levels associated with NFD is the result of (a) lack of strict adherence to the criteria for sample selection; (b) selection of too large a sample for analysis; and (c) use of analytical methodology that has potential matrix interference for the measured signal.
Index Entries: Aluminum, and Alzheimer's dementia; Alzheimer's dementia, and aluminum; neutron activation analysis; atomic absorption spectrophotometry; brain; aluminum-induced neurofibrillary degeneration. *Author to whom all correspondence and reprint requests should be addressed.
Biological Trace Element Research
35
Vol. 13, 1987
36
Krishnan, Harrison, and ~cLachlan
INTRODUCTION Although aluminum is abundant in nature, and biological systems have probably evolved in the presence of relatively high concentrations in the environment, trace amounts of this element are toxic to the nervous system of mammals (1-3). However, in brain tissue of higher animals, aluminum concentrations that exceed the normal by three- to fivefold are associated with major deficits in the performance of learning and memory tasks, changes in electrical property of brain cells, and accumulation of excessive number of neurofilaments, a histopathological change called neurofibrillary degeneration (NFD). The most common human disorder involving aluminum accumulation in neurons is a memory disorder known as senile dementia of the Alzheimer type. We reported in 1973 (1) that trace quantities of aluminum, when injected into the brains of cats, induced a progressive encepalopathy, in which deficits in learning-memory tasks were followed by progressive deficits in motor functions and, ultimately, death, within about 28 d. The brains of these animals contained elevated levels of aluminum that were distributed in a patchy manner, and a small proportion of large neurons exhibited increased NFD. The similarity in the clinical course and the histopathology of the animal model to Alzheimer's disease led to an examination of human brains with this disease for aluminum. Our work demonstrated that the toxic range for aluminum in cat and rabbit brain appeared to range between 4 and 8 ~xg/g dry weight. About 30% of neocortical regions sampled in brains of patients affected by Alzheimer's disease had concentrations in this range. The aluminum measurements were performed by flameless atomic absorption spectrophotometry. Similar findings were reported by Trapp et al. (4), who also used atomic absorption spectrometry (AA). We later showed that the aluminum found in elevated levels were largely intranuclear (5). This observation was subsequently confirmed by Pearl and Brody, using scanning electron microscopy (6). In brain tissues (taken from patients from the island of Guam) with parkinsonism and dementia or amyotrophic lateral sclerosis, aluminum was found in both the nucleus and cytoplasm of neurons with neurofibrillary degeneration (7) by similar techniques. These techniques are capable of precise localization of aluminum within cells when present in high concentration, but are not quantitative. In contrast to these reports, assay for aluminum by neutron activation failed to demonstrate elevated levels of aluminum in Alzheimer's brain (8), and McDermott (9), employing atomic absorption, found an increase in aluminum with aging, but failed to find a specific increase in Alzheimer's disease. These latter reports have led to a controversy concerning aluminum in Alzheimer's disease and the possible role of this element in the disease. We investigated the reason for the different conclusions derived from the various analytical techniques employed for assay of trace amounts of aluminum in brain tissue. The results are now reported. Biological Trace Element Research
VoL 13, 1987
Aluminum and Alzheimer's Degeneration
37
EXPERIMENTAL Brain Samples A documented clinical history was available for all patient material employed in this study. Based on clinical criteria, control brains had no neurological disease, and Alzheimer-affected brains had the typical clinical course of the disease (2). Brains were bisected in the sagittal plane, and one-half were frozen for aluminum assay and the other half fixed in formalin and examined microscopically. All brain material was examined with the aid of the Bielschowshy stain for NFD. All control brains were free of NFD and all Alzheimer brains had NFD and the usual morphological changes of Alzheimer's disease.
Analytical Methods Two analytical methods were examined; neutron activation analysis (NAA); and flameless atomic absorption spectrophotometry.
Neutron Activation Analysis Two procedures were examined: (A) an instrumental method; and (B) the application of preirradiation separation of aluminum. (A) For the instrumental neutron activation analysis (INAA), the Canadian SLOWPOKE reactor at the University of Toronto was used. The samples were irradiated at a thermal neutron flux of approximately 10 ~2 n/s/cm~ for 60 s. After a decay for 60 s, the 28A1activity was counted for 200 s, using Ge(Li) detector system with associated gamma-ray spectrometer (Canberra Model 8100). (B) For the preirradiation separation of aluminum, the tissue was ashed, using a quartz tube furnace at 450~ and the residue taken in 0.5N nitric acid and passed through a Dowex 50 ion-exchange column. The column was washed with the same acid and deionized water. The eluate contained the phosphorus and the column retained the aluminum. The aluminum was eluted with 2.4 mL of 0.1M HF, which quantitatively removed aluminum (10). This was then irradiated with neutrons to induce 28A1.
Atomic Absorption Spectrophotometry A Perkin-Elmer atomic absorption spectrometer (Model 305B), equipped with a graphite furnace (HGA 2000), was used. The spectrometer was equipped with a deuterium lamp for background correction. The procedure is described elsewhere (11). Biological Trace Element Research
Vol. 13, 1987
38
Krishnan, Harrison, and /VlcLachlan
RESULTS AND DISCUSSION Sample Criteria and Size We investigated first the discrepancies of our results (2) (laboratory A) vs those of McDermott et al. (9) (laboratory B). Both the laboratories used AA. However, laboratory A used 10-20 mg dry weight of tissue, whereas laboratory B used 200-1000 mg dry weight of tissue. An interlaboratory comparison of aluminum assay was made. First, a homogenized, freeze-dried bovine brain powder was analyzed by both the laboratories, with laboratory A reporting an aluminum value of 2.88 I*g A1/g dry wt, _+ 0.48 (SD), n =21, whereas laboratory B reported 2.7 >g A1/g dry wt. This showed that the assay of aluminum gave essentially comparable values in both the laboratories. Further analysis of 12 control samples from nonAlzheimer's brain and free of NFD gave values of 2.42 I*g A1/g dry wt _+ 1.25 (SD) by laboratory A and 2.43 I*g A1/g dry wt + 2.27 (SD) by laboratory B. Thus, with normal control tissues, in which aluminum can be expected to be uniformly distributed, the concentrations were comparable when analyzed by both laboratories. The discrepancies arose with the analysis of brain samples with NFD, with laboratory A reporting an average value of 6.02 Ixg A1/g dry wt m 4.93 (SD) and laboratory B reporting 2.92 Ixg A1/g dry wt m 1.47 (SD). In order to resolve the above discrepancy, some of the brain samples used by laboratory B were examined by laboratory A to see whether they meet the sampling criteria. It was found that at least two brains classified as controls by laboratory B (#1402 and #1408) in fact contained NFD, which was detected under the microscope with the aid of Congo red stain and, therefore, did not constitute proper control samples. On the other hand, one brain classified as from Alzheimer's disease (#313) did not contain NFD, as seen by histological staining, and, therefore, did not constitute a proper Alzheimer's sample. This fact was also confirmed by other workers (12). Thus, some of the discrepancies are caused by the samples and controls not strictly meeting the criteria used. Further examination revealed that the sample sizes used had a significant bearing on the aluminum levels found, thus indicating another probable cause of the discrepancies between the two laboratories.
Sample Size In a brain with NFD, laboratory A reported an average value of 6.02 I~g A1/g dry wt • 4.93 (SD) for 11 areas of the brain analyzed, whereas laboratory B reported an average value of 2.92 I~g A1/g dry wt • 1.47 (SD). Thus, although laboratory A reported significant elevation of aluminum levels in the samples containing NFD, laboratory B essentially found no difference in the aluminum levels between controls and samples with NFD. These results are listed in Table 1. The only difference between the methods of assay employed by the two laboratories Biological Trace Element Research
VoL 13, 1g87
Aluminum and Alzheimer's Degeneration
39
TABLE 1 Comparison of A1 Assays in Brain with NFD Between Laboratories A and B" Aluminum, lag/g dry wt Sample No. 1 2 3 4 5 6 7 8 9 10 11
Area of brain samples Area 11 Area 11 (medial) Area 46 Area 6 Area 40 Area 76 Area 34 and 28 Area 36 Cingulate Gyrus Corpus Callosum Area 17
Laboratory A, 10-20 mg sample size 9.20 10.39 3.53 2.18 5.10 2.93 18.27 3.32 6.05 1.18 3.18 6.02 _+ 4.93
Laboratory B 200-1000 mg sample size 2.77 7.22 1.90 2.25 2.58, 2.49 1.81, 2.14 3.68, 5.71 2.51 2.64, 2.31 2.23, 2.26 2.18 ~?2.92 _+ 1.47
"Sample: Brain from senile dementia patient with NFD.
was the sample size, i.e., laboratory A used 10-20 mg sample sizes and laboratory B used 200-1000 mg sample sizes. Thus, it is apparent that although smaller samples show higher aluminum levels, the larger ones show much lower levels, which is the diluted average level of the tissue. This is consistent with our earlier findings, and it was confirmed by Pearl and Brody (6) that NFD and aluminum in Alzheimer's disease shows a patchy distribution in localized spots. Therefore, the larger the sample size taken for the aluminum assay, the smaller the probability of isolating the "hot spot." However, in normal brain tissue, aluminum, not being a biologically essential element, appears to be uniformly distributed, and, therefore, the concentration of the element in a given sample is independent of the sample size. Our continuing work to resolve the discrepancy with respect to our laboratory and that of Markesbery et al. (8) (laboratory C) shows that the diagnostic and histologic criteria between laboratories A and C are in agreement. However, laboratory C had used NAA for the aluminum assay, whereas laboratory A had used AA. Laboratory C had also used a sample size of 100-250 mg dry wt. We investigated the experimental parameters used in AA and NAA from the point of view of sensitivity, minimum sample size, and signal interference. The data is given in Table 2. It has already been pointed out that in order to be able to isolate the aluminum/NFD hot spots, it is preferable to take as small a sample as possible. The minimum sample size required would depend on the detection limit of the analytical procedure used. On the basis that about a
Biological Trace Element Research
Vol. 13, 1987
Krishnan, Harrison, and ?4cLachlan
40
Origin and Resolution of the Aluminum Controversy Concerning Alzheimer's Neurofibrillary Degeneration S. S. KRISHNAN, *'1'4 J. E. HARRISON,3'4 AND D. R. C R A P P E R M C L A C H L A N z3
1Departments of Applied Chemistry and Chemical Engineering, 2physiology, and Wledicine, University of Toronto and 4Toronto General Hospital, Toronto, Ontario, Canada ABSTRACT Elevated concentrations of aluminum are found in regions of neurofibrillary change in brains with senile or presenile dementia of Alzheimer's type. The concentrations of aluminum found in the human disease are comparable to those found in experimental animals with aluminum-induced neurofibrillary degeneration (NFD). Although there are a number of reports confirming these observations, two laboratories have been unable to detect elevated levels in Alzheimer's disease. We conducted an interlaboratory study to resolve this discrepancy and traced the discrepancy to difficulties in analytical procedures. We concluded that failure to detect elevated aluminum levels associated with NFD is the result of (a) lack of strict adherence to the criteria for sample selection; (b) selection of too large a sample for analysis; and (c) use of analytical methodology that has potential matrix interference for the measured signal.
Index Entries: Aluminum, and Alzheimer's dementia; Alzheimer's dementia, and aluminum; neutron activation analysis; atomic absorption spectrophotometry; brain; aluminum-induced neurofibrillary degeneration. *Author to whom all correspondence and reprint requests should be addressed.
Biological Trace Element Research
35
Vol. 13, 1987
36
Krishnan, Harrison, and ~cLachlan
INTRODUCTION Although aluminum is abundant in nature, and biological systems have probably evolved in the presence of relatively high concentrations in the environment, trace amounts of this element are toxic to the nervous system of mammals (1-3). However, in brain tissue of higher animals, aluminum concentrations that exceed the normal by three- to fivefold are associated with major deficits in the performance of learning and memory tasks, changes in electrical property of brain cells, and accumulation of excessive number of neurofilaments, a histopathological change called neurofibrillary degeneration (NFD). The most common human disorder involving aluminum accumulation in neurons is a memory disorder known as senile dementia of the Alzheimer type. We reported in 1973 (1) that trace quantities of aluminum, when injected into the brains of cats, induced a progressive encepalopathy, in which deficits in learning-memory tasks were followed by progressive deficits in motor functions and, ultimately, death, within about 28 d. The brains of these animals contained elevated levels of aluminum that were distributed in a patchy manner, and a small proportion of large neurons exhibited increased NFD. The similarity in the clinical course and the histopathology of the animal model to Alzheimer's disease led to an examination of human brains with this disease for aluminum. Our work demonstrated that the toxic range for aluminum in cat and rabbit brain appeared to range between 4 and 8 ~xg/g dry weight. About 30% of neocortical regions sampled in brains of patients affected by Alzheimer's disease had concentrations in this range. The aluminum measurements were performed by flameless atomic absorption spectrophotometry. Similar findings were reported by Trapp et al. (4), who also used atomic absorption spectrometry (AA). We later showed that the aluminum found in elevated levels were largely intranuclear (5). This observation was subsequently confirmed by Pearl and Brody, using scanning electron microscopy (6). In brain tissues (taken from patients from the island of Guam) with parkinsonism and dementia or amyotrophic lateral sclerosis, aluminum was found in both the nucleus and cytoplasm of neurons with neurofibrillary degeneration (7) by similar techniques. These techniques are capable of precise localization of aluminum within cells when present in high concentration, but are not quantitative. In contrast to these reports, assay for aluminum by neutron activation failed to demonstrate elevated levels of aluminum in Alzheimer's brain (8), and McDermott (9), employing atomic absorption, found an increase in aluminum with aging, but failed to find a specific increase in Alzheimer's disease. These latter reports have led to a controversy concerning aluminum in Alzheimer's disease and the possible role of this element in the disease. We investigated the reason for the different conclusions derived from the various analytical techniques employed for assay of trace amounts of aluminum in brain tissue. The results are now reported. Biological Trace Element Research
VoL 13, 1987
Aluminum and Alzheimer's Degeneration
37
EXPERIMENTAL Brain Samples A documented clinical history was available for all patient material employed in this study. Based on clinical criteria, control brains had no neurological disease, and Alzheimer-affected brains had the typical clinical course of the disease (2). Brains were bisected in the sagittal plane, and one-half were frozen for aluminum assay and the other half fixed in formalin and examined microscopically. All brain material was examined with the aid of the Bielschowshy stain for NFD. All control brains were free of NFD and all Alzheimer brains had NFD and the usual morphological changes of Alzheimer's disease.
Analytical Methods Two analytical methods were examined; neutron activation analysis (NAA); and flameless atomic absorption spectrophotometry.
Neutron Activation Analysis Two procedures were examined: (A) an instrumental method; and (B) the application of preirradiation separation of aluminum. (A) For the instrumental neutron activation analysis (INAA), the Canadian SLOWPOKE reactor at the University of Toronto was used. The samples were irradiated at a thermal neutron flux of approximately 10 ~2 n/s/cm~ for 60 s. After a decay for 60 s, the 28A1activity was counted for 200 s, using Ge(Li) detector system with associated gamma-ray spectrometer (Canberra Model 8100). (B) For the preirradiation separation of aluminum, the tissue was ashed, using a quartz tube furnace at 450~ and the residue taken in 0.5N nitric acid and passed through a Dowex 50 ion-exchange column. The column was washed with the same acid and deionized water. The eluate contained the phosphorus and the column retained the aluminum. The aluminum was eluted with 2.4 mL of 0.1M HF, which quantitatively removed aluminum (10). This was then irradiated with neutrons to induce 28A1.
Atomic Absorption Spectrophotometry A Perkin-Elmer atomic absorption spectrometer (Model 305B), equipped with a graphite furnace (HGA 2000), was used. The spectrometer was equipped with a deuterium lamp for background correction. The procedure is described elsewhere (11). Biological Trace Element Research
Vol. 13, 1987
38
Krishnan, Harrison, and /VlcLachlan
RESULTS AND DISCUSSION Sample Criteria and Size We investigated first the discrepancies of our results (2) (laboratory A) vs those of McDermott et al. (9) (laboratory B). Both the laboratories used AA. However, laboratory A used 10-20 mg dry weight of tissue, whereas laboratory B used 200-1000 mg dry weight of tissue. An interlaboratory comparison of aluminum assay was made. First, a homogenized, freeze-dried bovine brain powder was analyzed by both the laboratories, with laboratory A reporting an aluminum value of 2.88 I*g A1/g dry wt, _+ 0.48 (SD), n =21, whereas laboratory B reported 2.7 >g A1/g dry wt. This showed that the assay of aluminum gave essentially comparable values in both the laboratories. Further analysis of 12 control samples from nonAlzheimer's brain and free of NFD gave values of 2.42 I*g A1/g dry wt _+ 1.25 (SD) by laboratory A and 2.43 I*g A1/g dry wt + 2.27 (SD) by laboratory B. Thus, with normal control tissues, in which aluminum can be expected to be uniformly distributed, the concentrations were comparable when analyzed by both laboratories. The discrepancies arose with the analysis of brain samples with NFD, with laboratory A reporting an average value of 6.02 Ixg A1/g dry wt m 4.93 (SD) and laboratory B reporting 2.92 Ixg A1/g dry wt m 1.47 (SD). In order to resolve the above discrepancy, some of the brain samples used by laboratory B were examined by laboratory A to see whether they meet the sampling criteria. It was found that at least two brains classified as controls by laboratory B (#1402 and #1408) in fact contained NFD, which was detected under the microscope with the aid of Congo red stain and, therefore, did not constitute proper control samples. On the other hand, one brain classified as from Alzheimer's disease (#313) did not contain NFD, as seen by histological staining, and, therefore, did not constitute a proper Alzheimer's sample. This fact was also confirmed by other workers (12). Thus, some of the discrepancies are caused by the samples and controls not strictly meeting the criteria used. Further examination revealed that the sample sizes used had a significant bearing on the aluminum levels found, thus indicating another probable cause of the discrepancies between the two laboratories.
Sample Size In a brain with NFD, laboratory A reported an average value of 6.02 I~g A1/g dry wt • 4.93 (SD) for 11 areas of the brain analyzed, whereas laboratory B reported an average value of 2.92 I~g A1/g dry wt • 1.47 (SD). Thus, although laboratory A reported significant elevation of aluminum levels in the samples containing NFD, laboratory B essentially found no difference in the aluminum levels between controls and samples with NFD. These results are listed in Table 1. The only difference between the methods of assay employed by the two laboratories Biological Trace Element Research
VoL 13, 1g87
Aluminum and Alzheimer's Degeneration
39
TABLE 1 Comparison of A1 Assays in Brain with NFD Between Laboratories A and B" Aluminum, lag/g dry wt Sample No. 1 2 3 4 5 6 7 8 9 10 11
Area of brain samples Area 11 Area 11 (medial) Area 46 Area 6 Area 40 Area 76 Area 34 and 28 Area 36 Cingulate Gyrus Corpus Callosum Area 17
Laboratory A, 10-20 mg sample size 9.20 10.39 3.53 2.18 5.10 2.93 18.27 3.32 6.05 1.18 3.18 6.02 _+ 4.93
Laboratory B 200-1000 mg sample size 2.77 7.22 1.90 2.25 2.58, 2.49 1.81, 2.14 3.68, 5.71 2.51 2.64, 2.31 2.23, 2.26 2.18 ~?2.92 _+ 1.47
"Sample: Brain from senile dementia patient with NFD.
was the sample size, i.e., laboratory A used 10-20 mg sample sizes and laboratory B used 200-1000 mg sample sizes. Thus, it is apparent that although smaller samples show higher aluminum levels, the larger ones show much lower levels, which is the diluted average level of the tissue. This is consistent with our earlier findings, and it was confirmed by Pearl and Brody (6) that NFD and aluminum in Alzheimer's disease shows a patchy distribution in localized spots. Therefore, the larger the sample size taken for the aluminum assay, the smaller the probability of isolating the "hot spot." However, in normal brain tissue, aluminum, not being a biologically essential element, appears to be uniformly distributed, and, therefore, the concentration of the element in a given sample is independent of the sample size. Our continuing work to resolve the discrepancy with respect to our laboratory and that of Markesbery et al. (8) (laboratory C) shows that the diagnostic and histologic criteria between laboratories A and C are in agreement. However, laboratory C had used NAA for the aluminum assay, whereas laboratory A had used AA. Laboratory C had also used a sample size of 100-250 mg dry wt. We investigated the experimental parameters used in AA and NAA from the point of view of sensitivity, minimum sample size, and signal interference. The data is given in Table 2. It has already been pointed out that in order to be able to isolate the aluminum/NFD hot spots, it is preferable to take as small a sample as possible. The minimum sample size required would depend on the detection limit of the analytical procedure used. On the basis that about a
Biological Trace Element Research
Vol. 13, 1987
Krishnan, Harrison, and ?4cLachlan
40
