Neurocognitive Effects of Aluminum Karen I.
Bolla, PhD; Gary Briefel, MD; David Spector, MD; Brian S. Schwartz, MD, MS; Lisa Wieler; Janice Herron; Louis Gimenez, MD
The neurocognitive effects of aluminum (Al) were studied in 35 hemodialysis patients. Higher Al levels were associated with a decline in visual memory. As Al levels increased, \s=b\
patients
vocabulary intelligence) showed
with lower
scores
(a
measure
of pre-
decline in attention/ on several neuwhile those with measures, vocabulary higher rocognitive scores revealed no Al-related decline. These results suggest that individuals with lower verbal intelligence may possess less well-developed compensatory strategies to overcome the neurocognitive effects associated with Al. These data also indicate that Al is neurotoxic and, therefore, potential sources of environmental Al should be identified and eliminated. (Arch Neurol. 1992;49:1021-1026) morbid
a
concentration, frontal lobe functions, and
(Al) hypothesized be potential of Alzheimer's disease, Guarnían Aluminum agent the lateral sclerosis, Parkinson's dementia, and is
in
to
a
toxic
cause
amyotrophic dialysis encephalopathy.1"4 The possible neurotoxicity of Al warrants further investigation, not only as it relates to specific neurologic diseases, but also because of the poten¬ tial effects that Al may have on public health. Acid rain re¬ sults in increases of Al content in surface water,5 which could affect large populations detrimentally. Although the evidence suggests that Al may cause neuropathology, the data are not conclusive and the neurocognitive sequelae of Al have not been rigorously studied. The neurotoxicity of Al has been studied most exten¬ sively in dialysis patients who have developed dialysis encephalopathy (DE). Dialysis encephalopathy, first de¬ scribed in 1972, consists of dementia, speech dis¬ order, apraxia, myoclonus, asterixis, and abnormal electroencephalography.3'4-6"13 In autopsy studies of DE patients, high levels of aluminum were found in bone,
Accepted From the
for
publication June 9,
1992.
Departments of Neurology, Francis Scott Key Medical Cen-
(Dr Bolla and Mss Wieler and Herron), Psychiatry and Behavioral Science (Dr Bolla), Division of Renal Medicine (Drs Briefel, Spector, and Gimenez), and School of Hygiene and Public Health, Department of Environmental Health Sciences (Drs Bolla and Schwartz), The Johns Hopkins University School of Medicine, Baltimore, Md. Presented in part at the 41st annual meeting of the American Academy of Neurology, Chicago, Ill, April 1989. Reprint requests to the Department of Neurology, The Johns Hopkins University School of Medicine, Francis Scott Key Medical Center, 4940 Eastern Ave, Baltimore, MD 21224 (Dr Bolla).
ter
liver, and brain (gray matter).3-7 Sources of Al include
high Al dialysate
content in the water used to make up the
and ingestion of Al-containing compounds for purposes of phosphate binding.914 Although the associ¬ ation between increased levels of Al and DE have been reported, the association between the Al level and neu¬ rocognitive functioning has not been studied in dialysis patients without clinical evidence of DE. Former studies of the Al level and neurocognitive per¬ formance have failed to use objective, standardized neu¬ rocognitive tests.812 These studies have used baseline serum Al level (a reflection of recent Al dose) instead of using serum Al levels after the administration of deferoxamine (an Al chelating agent), which more accurately re¬ flects the body burden of Al.151617 The primary aim of this study was to investigate the association between alumi¬ num levels and neurocognitive functioning in patients on en¬ hemodialysis without neurologic evidence of cephalopathy. To our knowledge, this is the first study to examine this association using both an extensive battery of standardized neurocognitive tests and a measurement of body burden of aluminum (post-deferoxamine
dialysis
administration).
SUBJECTS
AND METHODS
Subjects Patients were recruited from dialysis units at three area hospi¬
tals and were not paid for their participation. Patients were excluded if they had a history of a neurologic or psychiatric dis¬ order or were seropositive for human immunodeficiency virus. We examined 35 patients on hemodialysis, with a mean age (± SD) of 51 ±14 years (range, 20 to 75 years) and a mean level of edu¬ cation of 11 ±4 years (range, 0 to 16 years). Seventeen men and 18 women were tested. The duration of time on dialysis ranged from 1 month to 10.5 years (mean [±SD] duration, 3.8±3 years). Mean serum creatinine concentration (±SD) for the group was 1299+309 µ /L (range, 574 to 1874 µ /L), and mean (±SD) serum urea nitrogen was 26.3±6.7 mmol/L (range, 13.5 to 41.0 mmol/L). Mean (±SD) baseline serum aluminum level before administration of deferoxamine was 42±40 µg/L (range, 5 to 162 µg/L), and post-deferoxamine aluminum levels ranged from 25 to 630 µg/L, with a mean of 170±167 µg/L.
Procedures Patients were dialyzed three times a week using water purified by a charcoal filter and reverse osmosis. The chemical analysis of treated water used for hemodialysis was performed twice a year, showing concentrations of possible contaminants well within the accepted ranges, of the Association for the Advancement of
Downloaded From: http://archneur.jamanetwork.com/ by a DALHOUSIE UNIVERSITY-DAL-11762 User on 06/20/2015
Medical Instrumentation published standards. Aluminum levels were 0.08 µg/L (routine accepted level, 0.1 µg/L). Baseline serum Al levels were taken just prior to dialysis. Patients were given a single dose of deferoxamine (40 mg/kg of body weight intrave¬ nously via an intravenous infusion pump at 70 mL per hour) ad¬ ministered over 2 to 3 hours. While most investigators report ad¬ ministering deferoxamine during dialysis, we administered deferoxamine after dialysis so that the maximal dose of deferox¬ amine was retained and not partially dialyzed out of the patient thus providing a more accurate measurement of Al body burden. Blood specimens were drawn 48 hours after administration of deferoxamine prior to starting the next dialysis session. For each blood sample, approximately 10 mL of whole blood was collected. Plasma Al levels were quantified using nameless atomic absorp¬ tion spectrometry.'8 Serum urea nitrogen and creatinine were measured within 1 week of the neurocognitive assessment. Each patient completed a demographic questionnaire and was administered a battery of neurocognitive tests prior to deferox¬ amine administration. Neurocognitive testing was performed 24 to 36 hours after a dialysis session because this has been shown to be the optimal time for cognitive testing since metabolic and electrolyte changes are at their minimum.19 A self-rated, 20-item symptom depression scale known as the (CES-D)20 was given at this time. The cognitive domains that were assessed and the tests used to assess them are summarized in Table 1. A brief neurologic examination was also administered at this time. This included the assessment of tremor, myoclonus, asterixis, smell, pin prick, and vibration. These were rated on a 0 (greatest symptom severity) to 4 (symptoms not present) scale. Informed consent was obtained from all patients and this research protocol was approved by the Institutional Review Board.
Data
Table
Tests
Intelligence Vocabulary Similarities
Orientation Mini-Mental State Examination Months backward
Attention/concentration Months backward Trails A Trails
Digit symbol Language Verbal
fluency (FAS)
Naming Test Repetition (words, phrases) Boston
Reading comprehension writing: copy writing: dictation
Sentence Sentence
Verbal memory Logical memory
Rey Auditory Verbal Learning Test Visual memory Visual reproduction Rey Osterich Complex Figure (memory) Symbol Digit Paired Associate Learning Test
Visuoperception/visuoconstruction Block design Rey Osterich Complex Figure (copy) Clock drawing Hooper
Analyses
To determine the specific cognitive functions that were associ¬ ated with Al body burden, we clustered the 30 individual neuro¬ cognitive tests into nine neurocognitive domain scores (Table 1). This clustering was done based on what functions each of the subtests measure.21 Since many of the neurocognitive tests assess multiple cognitive processes, some tests were assigned to more than one cognitive domain score. In addition, the correlations be¬ tween tests within a cognitive domain (Trails and digit symbol) were examined and found to be higher than the correlations be¬ tween tests in different cognitive domains (Trails and the Bos¬ ton Naming Test). Due to differences in maximal scores on each subtest, individual subtest scores were converted into scores using the means and SDs of the group and then summed to pro¬ duce the overall functional domain scores. This reduced the number of comparisons from 30 individual subtest scores to 9 cognitive domain scores and therefore reduced the probability of a type 1 error. Natural log transformations were performed on the variables that were not normally distributed (CES-D, Trails A, Trails B, and reaction time) and parametric statistics were applied. Centering was used to reduce multicolinearity between independent vari¬ ables before including such terms in the regression models. Mul¬ tiple linear regression was used to assess the relationship between Al level after deferoxamine administration and neurocognitive test performance. We identified a priori and then examined, age, vocabulary score (a measure of premorbid verbal intelligence22-23), neurologic symptoms, visual acuity, and depression as potential confounding variables. Measures of renal function (creatinine, serum urea nitrogen, and duration of time on dialysis) were not found to be important confounders of the association between Al level and neurocognitive performance. Creatinine did not corre¬ late significantly with any of the composite cognitive domain scores or any of the individual neurocognitive tests (r=.009 to .35). The correlations between serum urea nitrogen concentration and the neurocognitive measures ranged from r=.001 to .23 and du¬ ration of time on dialysis correlated with the dependent measures r=.002 to .219. We also evaluated the interactions between Al level and the confounding variables. Only age and vocabulary score
Domains Assessed and 1.—Cognitive Administered
Executive/motor Reaction time
Digit symbol
Trails A Trails
Apraxia Finger tapping Crip strength Frontal
Alternating fingers
Luria motor sequences Trails A Trails
found to be significant confounders; therefore, the final models included AI level (after deferoxamine administration), age, and vocabulary score. Interactions were retained in the final models if significant (P
Bolla, PhD; Gary Briefel, MD; David Spector, MD; Brian S. Schwartz, MD, MS; Lisa Wieler; Janice Herron; Louis Gimenez, MD
The neurocognitive effects of aluminum (Al) were studied in 35 hemodialysis patients. Higher Al levels were associated with a decline in visual memory. As Al levels increased, \s=b\
patients
vocabulary intelligence) showed
with lower
scores
(a
measure
of pre-
decline in attention/ on several neuwhile those with measures, vocabulary higher rocognitive scores revealed no Al-related decline. These results suggest that individuals with lower verbal intelligence may possess less well-developed compensatory strategies to overcome the neurocognitive effects associated with Al. These data also indicate that Al is neurotoxic and, therefore, potential sources of environmental Al should be identified and eliminated. (Arch Neurol. 1992;49:1021-1026) morbid
a
concentration, frontal lobe functions, and
(Al) hypothesized be potential of Alzheimer's disease, Guarnían Aluminum agent the lateral sclerosis, Parkinson's dementia, and is
in
to
a
toxic
cause
amyotrophic dialysis encephalopathy.1"4 The possible neurotoxicity of Al warrants further investigation, not only as it relates to specific neurologic diseases, but also because of the poten¬ tial effects that Al may have on public health. Acid rain re¬ sults in increases of Al content in surface water,5 which could affect large populations detrimentally. Although the evidence suggests that Al may cause neuropathology, the data are not conclusive and the neurocognitive sequelae of Al have not been rigorously studied. The neurotoxicity of Al has been studied most exten¬ sively in dialysis patients who have developed dialysis encephalopathy (DE). Dialysis encephalopathy, first de¬ scribed in 1972, consists of dementia, speech dis¬ order, apraxia, myoclonus, asterixis, and abnormal electroencephalography.3'4-6"13 In autopsy studies of DE patients, high levels of aluminum were found in bone,
Accepted From the
for
publication June 9,
1992.
Departments of Neurology, Francis Scott Key Medical Cen-
(Dr Bolla and Mss Wieler and Herron), Psychiatry and Behavioral Science (Dr Bolla), Division of Renal Medicine (Drs Briefel, Spector, and Gimenez), and School of Hygiene and Public Health, Department of Environmental Health Sciences (Drs Bolla and Schwartz), The Johns Hopkins University School of Medicine, Baltimore, Md. Presented in part at the 41st annual meeting of the American Academy of Neurology, Chicago, Ill, April 1989. Reprint requests to the Department of Neurology, The Johns Hopkins University School of Medicine, Francis Scott Key Medical Center, 4940 Eastern Ave, Baltimore, MD 21224 (Dr Bolla).
ter
liver, and brain (gray matter).3-7 Sources of Al include
high Al dialysate
content in the water used to make up the
and ingestion of Al-containing compounds for purposes of phosphate binding.914 Although the associ¬ ation between increased levels of Al and DE have been reported, the association between the Al level and neu¬ rocognitive functioning has not been studied in dialysis patients without clinical evidence of DE. Former studies of the Al level and neurocognitive per¬ formance have failed to use objective, standardized neu¬ rocognitive tests.812 These studies have used baseline serum Al level (a reflection of recent Al dose) instead of using serum Al levels after the administration of deferoxamine (an Al chelating agent), which more accurately re¬ flects the body burden of Al.151617 The primary aim of this study was to investigate the association between alumi¬ num levels and neurocognitive functioning in patients on en¬ hemodialysis without neurologic evidence of cephalopathy. To our knowledge, this is the first study to examine this association using both an extensive battery of standardized neurocognitive tests and a measurement of body burden of aluminum (post-deferoxamine
dialysis
administration).
SUBJECTS
AND METHODS
Subjects Patients were recruited from dialysis units at three area hospi¬
tals and were not paid for their participation. Patients were excluded if they had a history of a neurologic or psychiatric dis¬ order or were seropositive for human immunodeficiency virus. We examined 35 patients on hemodialysis, with a mean age (± SD) of 51 ±14 years (range, 20 to 75 years) and a mean level of edu¬ cation of 11 ±4 years (range, 0 to 16 years). Seventeen men and 18 women were tested. The duration of time on dialysis ranged from 1 month to 10.5 years (mean [±SD] duration, 3.8±3 years). Mean serum creatinine concentration (±SD) for the group was 1299+309 µ /L (range, 574 to 1874 µ /L), and mean (±SD) serum urea nitrogen was 26.3±6.7 mmol/L (range, 13.5 to 41.0 mmol/L). Mean (±SD) baseline serum aluminum level before administration of deferoxamine was 42±40 µg/L (range, 5 to 162 µg/L), and post-deferoxamine aluminum levels ranged from 25 to 630 µg/L, with a mean of 170±167 µg/L.
Procedures Patients were dialyzed three times a week using water purified by a charcoal filter and reverse osmosis. The chemical analysis of treated water used for hemodialysis was performed twice a year, showing concentrations of possible contaminants well within the accepted ranges, of the Association for the Advancement of
Downloaded From: http://archneur.jamanetwork.com/ by a DALHOUSIE UNIVERSITY-DAL-11762 User on 06/20/2015
Medical Instrumentation published standards. Aluminum levels were 0.08 µg/L (routine accepted level, 0.1 µg/L). Baseline serum Al levels were taken just prior to dialysis. Patients were given a single dose of deferoxamine (40 mg/kg of body weight intrave¬ nously via an intravenous infusion pump at 70 mL per hour) ad¬ ministered over 2 to 3 hours. While most investigators report ad¬ ministering deferoxamine during dialysis, we administered deferoxamine after dialysis so that the maximal dose of deferox¬ amine was retained and not partially dialyzed out of the patient thus providing a more accurate measurement of Al body burden. Blood specimens were drawn 48 hours after administration of deferoxamine prior to starting the next dialysis session. For each blood sample, approximately 10 mL of whole blood was collected. Plasma Al levels were quantified using nameless atomic absorp¬ tion spectrometry.'8 Serum urea nitrogen and creatinine were measured within 1 week of the neurocognitive assessment. Each patient completed a demographic questionnaire and was administered a battery of neurocognitive tests prior to deferox¬ amine administration. Neurocognitive testing was performed 24 to 36 hours after a dialysis session because this has been shown to be the optimal time for cognitive testing since metabolic and electrolyte changes are at their minimum.19 A self-rated, 20-item symptom depression scale known as the (CES-D)20 was given at this time. The cognitive domains that were assessed and the tests used to assess them are summarized in Table 1. A brief neurologic examination was also administered at this time. This included the assessment of tremor, myoclonus, asterixis, smell, pin prick, and vibration. These were rated on a 0 (greatest symptom severity) to 4 (symptoms not present) scale. Informed consent was obtained from all patients and this research protocol was approved by the Institutional Review Board.
Data
Table
Tests
Intelligence Vocabulary Similarities
Orientation Mini-Mental State Examination Months backward
Attention/concentration Months backward Trails A Trails
Digit symbol Language Verbal
fluency (FAS)
Naming Test Repetition (words, phrases) Boston
Reading comprehension writing: copy writing: dictation
Sentence Sentence
Verbal memory Logical memory
Rey Auditory Verbal Learning Test Visual memory Visual reproduction Rey Osterich Complex Figure (memory) Symbol Digit Paired Associate Learning Test
Visuoperception/visuoconstruction Block design Rey Osterich Complex Figure (copy) Clock drawing Hooper
Analyses
To determine the specific cognitive functions that were associ¬ ated with Al body burden, we clustered the 30 individual neuro¬ cognitive tests into nine neurocognitive domain scores (Table 1). This clustering was done based on what functions each of the subtests measure.21 Since many of the neurocognitive tests assess multiple cognitive processes, some tests were assigned to more than one cognitive domain score. In addition, the correlations be¬ tween tests within a cognitive domain (Trails and digit symbol) were examined and found to be higher than the correlations be¬ tween tests in different cognitive domains (Trails and the Bos¬ ton Naming Test). Due to differences in maximal scores on each subtest, individual subtest scores were converted into scores using the means and SDs of the group and then summed to pro¬ duce the overall functional domain scores. This reduced the number of comparisons from 30 individual subtest scores to 9 cognitive domain scores and therefore reduced the probability of a type 1 error. Natural log transformations were performed on the variables that were not normally distributed (CES-D, Trails A, Trails B, and reaction time) and parametric statistics were applied. Centering was used to reduce multicolinearity between independent vari¬ ables before including such terms in the regression models. Mul¬ tiple linear regression was used to assess the relationship between Al level after deferoxamine administration and neurocognitive test performance. We identified a priori and then examined, age, vocabulary score (a measure of premorbid verbal intelligence22-23), neurologic symptoms, visual acuity, and depression as potential confounding variables. Measures of renal function (creatinine, serum urea nitrogen, and duration of time on dialysis) were not found to be important confounders of the association between Al level and neurocognitive performance. Creatinine did not corre¬ late significantly with any of the composite cognitive domain scores or any of the individual neurocognitive tests (r=.009 to .35). The correlations between serum urea nitrogen concentration and the neurocognitive measures ranged from r=.001 to .23 and du¬ ration of time on dialysis correlated with the dependent measures r=.002 to .219. We also evaluated the interactions between Al level and the confounding variables. Only age and vocabulary score
Domains Assessed and 1.—Cognitive Administered
Executive/motor Reaction time
Digit symbol
Trails A Trails
Apraxia Finger tapping Crip strength Frontal
Alternating fingers
Luria motor sequences Trails A Trails
found to be significant confounders; therefore, the final models included AI level (after deferoxamine administration), age, and vocabulary score. Interactions were retained in the final models if significant (P
