1. Serum zinc is decreased in Alzheimer's disease and serum arsenic correlates positively with cognitive ability
L.W. Baum1, I. Chan2, S.K. Cheung1, W. Goggins3, V. Mok1, L. Lam4, V. Leung4, E. Hui1, C. Ng1, J. Woo1, H. Chiu4, B.Y. Zee5, W. Cheng6, M.H. Chan7, S. Szeto7, V. Lui4, J. Tsoh4, A. Bush8, C. Lam2, T. Kwok1
1Dept. of Medicine & Therapeutics, 2Dept. of Chemical Pathology, 3School of Public Health, 4Dept. of Psychiatry, 5Center for Clinical Trials, 6Clinical Trials Section, Institute of Chinese Medicine, Chinese University of Hong Kong, Shatin, Hong Kong; 7Dept. of Medicine and Geriatrics, Kwong Wah Hospital, Kowloon, Hong Kong; 8Genetics and Aging Research Unit, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA
Introduction
Metals may affect Alzheimer's disease (AD). Ions of zinc, copper, and iron can induce Ab aggregation and accumulate in plaques. AD serum was reported to contain altered levels of some metals. To examine the relation of metals to AD, we measured 12 elements in serum of AD and controls.
Methods
Subjects were Hong Kong Chinese. Patients were recruited from outpatient clinics (n=35) and old age homes (n=9) and had NINCDS-ADRDA diagnosis of probable or possible AD. Controls (n=41) lacked neurological disease. 43 AD and 1 control subject took the Cantonese MMSE. Serum was collected from all subjects. Aluminum, arsenic, beryllium, chromium, cobalt, copper, iodine, iron, manganese, nickel, selenium, and zinc were assayed in serum by inductively coupled plasma-mass spectrometry (ICP-MS 7500c, Agilent). Beryllium and nickel were often below detection limits and thus were not analyzed. Continuous variables were compared between AD and controls by testing for normality (Kolmogorov-Smirnov test) and using t tests if normal and Mann-Whitney otherwise. Of metals, only zinc was normally distributed. All other metals were log transformed for logistic regressions.
Positive correlation of serum arsenic with MMSE might be due, as with zinc, to absorption of arsenic by amyloid plaques, depleting arsenic from serum. Thus, more severe AD with more plaques may show less serum arsenic. However, no studies have yet been published on arsenic binding to plaques.
Alternatively, polymorphisms in glutathione S-transferase omega genes GSTO1 and GSTO2 were associated with AD onset age, and GSTO1 metabolizes arsenic, thus polymorphisms may affect both AD onset age and arsenic metabolism. However, arsenic levels did not significantly differ between AD and controls.
Finally, most arsenic in humans comes from seafood. Inorganic arsenic, such as from drinking water, can impair cognition, but seafood arsenic is generally in organic, non-toxic forms. Eating fish with DHA is associated with increased cognition and reduced AD risk, supporting the arsenic-MMSE association. But arsenic did not decrease in AD vs. controls, perhaps due to DHA improving cognition in humans developing AD but whose cognition has not yet deteriorated enough to be diagnosed, thus delaying AD onset and reducing the incidence in prospective studies but not affecting risk in case-control studies, like ours.
Acknowledgements
Supported by a grant from The Chinese University of Hong Kong Institute of Chinese Medicine, by a Chinese University of Hong Kong Direct Grant for Research, and by a BUPA Foundation medical research grant.
Conclusions
Serum zinc lower in AD.
Serum aluminum higher in AD.
Serum arsenic positively correlates with cognition, perhaps as marker of fish.
Results
The table shows descriptive statistics and metal concentrations for AD vs. controls. Results are median±interquartile range and unadjusted (Unadj.) p for Mann-Whitney test (except for a, which indicates mean±SD and unadj. p for t-test, or b, which indicates Fisher's exact test). Three logistic regression models were fit, with case-control status as the outcome: (1) one metal with age and sex as additional covariates; (2) all metals with age and sex; and (3) variables selected from model (2) by forward stepwise selection (p=.05, likelihood ratio test). Zinc significantly differed in all models. MMSE correlated with aluminum (Pearson’s correlation coefficient: -0.33, p=0.03) and arsenic (0.55, p<0.0001).
Controls (n=41)
AD (n=44)
Unadj. p
Model 1
Model 2
Model 3
Age (y)a
79.1±6.0
74.3±8.7
.0041
.027
.0056
Women/Men (n)b
20/21
29/15
.13
.0030
.0006
Metals (nmol/L)
Iron
23800±11000
17700±8200
.017
.037
.18
Copper
15300±2700
16200±3500
.23
.37
.72
Zinca
12300±1600
10900±1600
.0001
.0007
.0015
<.0001
Selenium
1390±240
1420±230
.56
.29
.0020
.0078
Iodine
610±170
560±140
.038
.069
.058
.032
Aluminum
580±620
905±630
.0055
.0023
.12
.0077
Arsenic
38.7±37
35.7±40
.57
.61
.096
Cobalt
23.5±28
9.1±12
.27
.55
.64
Chromium
17.2±12
22.7±21
.093
.025
.020
Manganese
13.3±9.2
21.4±21
.065
.26
.48
The scatterplot above shows MMSE score vs. logarithm of arsenic concentration (nmol/L).
Discussion
In most carefully designed studies, zinc increased in AD brain (and positively correlated with plaque numbers) but either did not change or decreased in AD serum. Zinc induces Ab aggregation, thus the decrease we observed in serum zinc in AD might be due to deposition and sequestration of zinc in brain amyloid, perhaps depleting zinc in other body compartments. Another possible cause of decreased serum zinc in AD may be the poor diet of severe AD patients. But a study of mild/moderate AD did not show decreased zinc intake, and excluding subjects with MMSE<10 did not eliminate the decrease in serum zinc in AD (p=0.002), which thus may not be due to deficient zinc intake in severe AD.
In contrast to zinc, aluminum decreased or did not change in AD brain but increased in serum, as we found. The serum increase may be due to more dietary aluminum uptake in AD, but the cause of uptake change is unknown.