Association of Seafood Consumption, Brain Mercury Level, and APOE ε4 Status With Brain Neuropathology in Older Adults

Abstract

IMPORTANCE

Seafood consumption is promoted for its many health benefits even though its contamination by mercury, a known neurotoxin, is a growing concern.

OBJECTIVE

To determine whether seafood consumption is correlated with increased brain mercury levels and also whether seafood consumption or brain mercury levels are correlated with brain neuropathologies.

DESIGN, SETTING, AND PARTICIPANTS

Cross-sectional analyses of deceased participants in the Memory and Aging Project clinical neuropathological cohort study, 2004–2013. Participants resided in Chicago retirement communities and subsidized housing. The study included 286 autopsied brains of 554 deceased participants (51.6%). The mean (SD) age at death was 89.9 (6.1) years, 67% (193) were women, and the mean (SD) educational attainment was 14.6 (2.7) years.

EXPOSURES

Seafood intake was first measured by a food frequency questionnaire at a mean of 4.5 years before death.

MAIN OUTCOMES AND MEASURES

Dementia-related pathologies assessed were Alzheimer disease, Lewy bodies, and the number of macroinfarcts and microinfarcts. Dietary consumption of seafood and n-3 fatty acids was annually assessed by a food frequency questionnaire in the years before death. Tissue concentrations of mercury and selenium were measured using instrumental neutron activation analyses.

RESULTS

Among the 286 autopsied brains of 544 participants, brain mercury levels were positively correlated with the number of seafood meals consumed per week (ρ = 0.16; P = .02). In models adjusted for age, sex, education, and total energy intake, seafood consumption (≥1 meal[s]/week) was significantly correlated with less Alzheimer disease pathology including lower density of neuritic plaques (β = −0.69 score units [95% CI, −1.34 to −0.04]), less severe and widespread neurofibrillary tangles (β = −0.77 score units [95% CI, −1.52 to −0.02]), and lower neuropathologically defined Alzheimer disease (β = −0.53 score units [95% CI, −0.96 to −0.10]) but only among apolipoprotein E (APOE ε4) carriers. Higher intake levels of α-linolenic acid (18:3 n-3) were correlated with lower odds of cerebral macroinfarctions (odds ratio for tertiles 3 vs 1, 0.51 [95% CI, 0.27 to 0.94]). Fish oil supplementation had no statistically significant correlation with any neuropathologic marker. Higher brain concentrations of mercury were not significantly correlated with increased levels of brain neuropathology.

CONCLUSIONS AND RELEVANCE

In cross-sectional analyses, moderate seafood consumption was correlated with lesser Alzheimer disease neuropathology. Although seafood consumption was also correlated with higher brain levels of mercury, these levels were not correlated with brain neuropathology.

Numerous studies have found protective associations between seafood consumption and dementia.^1–7 Little is known about the relationship between seafood consumption and brain neuropathology. Seafood is high in the long-chain n-3 fatty acid docosahexaenoic acid (DHA), which has established properties for normal neuronal function.^8,9 However, seafood is also a source of mercury, a neurotoxin that impairs neurocognitive development.¹⁰ Mercury toxicity is reduced by selenium, an essential nutrient present in seafood that has high binding affinity to mercury.¹¹ In this study, we examined the associations of dementia neuropathologies with dietary n-3 fatty acids and with brain levels of mercury and selenium.

Methods

Study Population

The analytic sample comprised autopsied cases of deceased participants in the Rush Memory and Aging Project (MAP) who died between November 2004 and November 2013 and who completed a dietary assessment before death. MAP is an ongoing clinical-neuropathological cohort study of older adults that began in 1997 and includes Chicago residents of retirement communities and subsidized housing.¹² At enrollment, participants were dementia free and agreed to undergo annual clinical neurological evaluations and brain autopsy at death. Annual dietary assessments of MAP participants began in February 2004. Written informed consent was obtained from all study participants, and the study was approved by the institutional review board of Rush University.

Brain Neuropathology

The methods for brain autopsies and pathologic evaluations are described in detail elsewhere.¹³ Briefly, slabs from 1 cerebral hemisphere were placed in a −80°C freezer and used for metal analyses. Slabs from the contralateral hemisphere were fixed in 4% paraformaldehyde, and then dissected tissue samples from brain regions were embedded in paraffin blocks, cut into 6-micron sections, and mounted onto slides. Alzheimer disease neuropathologies, including diffuse and neuritic amyloid plaques and neurofibrillary tangles, were identified using modified Bielschowsky silver-stained 6-micron sections in multiple cortical regions. Raw counts (greatest density in 1-mm² area) of the neuritic and diffuse plaques and tangles were standardized in each region and averaged across regions to create 3 summary scores that were then averaged for a global measure of Alzheimer disease pathology. We also analyzed 2 semiquantitative measures of Alzheimer disease pathology: CERAD (Consortium to Establish a Registry for Alzheimer Disease; [for neuritic plaque density])¹⁴ and Braak neurofibrillary tangle stage.¹⁵ CERAD scores ranged from 1 (no neuritic plaques) to 4 (frequent neuritic plaques). Braak staging scores ranged from 0 (no tangles in any region) to 6 (frequent tangles across multiple regions). Another variable described the level of Alzheimer disease pathology based on National Institute on Aging (NIA) and the Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association criteria¹⁶ with scores ranging from 1 (low Alzheimer disease pathology) to 4 (high Alzheimer disease pathology). A board-certified neuropathologist, blinded to participant ages and clinical data, determined the neuropathology diagnoses.

Counts of chronic macroscopic cerebral infarctions and microinfarcts were made as previously described.¹⁷ Macroinfarcts were identified through visual inspection and were histologically confirmed. Multiple regions were examined for microinfarcts using hematoxylin and eosin stain.

Lewy body staging was assessed in 6 regions through staining with antibodies to α-synuclein.^13,18

Mercury and Selenium Analyses

Brain metal concentrations were measured in 2 cortical regions (inferior temporal and midfrontal) involved in Alzheimer disease neuropathology and the cerebellum (a region unaffected by Alzheimer disease neuropathology) to facilitate the interpretation of study findings. Brain tissue was cut into 100-g samples using a ceramic blade (to avoid metal contamination) and sent to the University of Missouri Research Reactor to be assessed for selenium and mercury by instrumental neutron activation analysis.¹⁹ Samples were loaded into cleaned, high-purity quartz vials, lyophilized, and then irradiated for 50 hours in a neutron flux of 6.5 × 10¹³ n/cm²∙s and allowed to decay for 20 to 40 days before being counted using a high-purity germanium detector system. Measured concentrations of the metals in reference materials were analyzed for consistency with accepted standards and agreed well with certified values. For analyses, we first transformed (log₁₀) the highly skewed metal concentrations and then computed z scores of these values for each brain region. Mean z scores for each region were then averaged across the 2 cortical regions and across all 3 regions.

Dietary Consumption

Dietary intake was assessed by a semiquantitative food frequency questionnaire that was validated for use in older Chicago residents.²⁰ The questionnaire included 4 seafood items (tuna sandwich; fish sticks, cakes, or sandwich; fresh fish as a main dish; and shrimp, lobster, or crab) that were used to compute weekly seafood consumption. Daily intakes of long-chain n-3 fatty acids (eicosapentaenoic acid [EPA, 20:5 n-3] + DHA [22:6 n-3]) and α-linolenic acid (18:3 n-3) were obtained by multiplying the nutrient content of all food items by frequency of consumption and summing over all items. Nutrient intakes were energy adjusted by the regression residual method.²¹ The questionnaire also included a query about intake of fish oil supplements. Dietary intake levels were averaged over all valid food frequency questionnaires obtained during the years before death. Evaluators of neuropathological measures and brain concentrations of metals were blinded to dietary intake data.

Statistical Methods

We used linear regression models to investigate the associations of the dietary variables and brain metal concentrations with Alzheimer disease neuropathology (general Alzheimer disease pathology [log transformed] and CERAD, Braak, and NIA and Reagan Institute scores). We examined the diet/metal associations with Lewy bodies, macroinfarcts, and microinfarcts in logistic regression models with outcomes modeled present or absent because of the highly skewed distributions. Model assumptions were investigated graphically and statistically, and findings were confirmed through reanalyses using proportional hazards models. Analyses were conducted using SAS version 9.3. Models were adjusted for established predictors of neuropathology including age at death, sex, and education, and for the seafood model only, adjustment was made for total energy intake (kcal/d). Statistical interactions were tested in adjusted models with a multiplicative term between the diet/metal variable and the potential effect modifier. The level of statistical significance for all tests, including those for interactions, was 2-sided (P ≤ .05). We investigated both linear and nonlinear correlations of the diet and metal exposures to neuropathologic outcomes. Consistent with previous studies,^1,3,5,7,22 seafood consumption was not linearly correlated with the Alzheimer disease outcomes, and only the findings with seafood modeled as an indicator variable (≥1 meal[s] per week vs less) are presented. To make the data presentation succinct, we present only the linear model coefficients for analyses stratified by apolipoprotein E (APOE ε4). We examined the associations of brain metal concentrations to neuropathologic outcomes with the z score metal concentrations averaged over the 2 cortical regions and also the cortical and cerebellum regions. Because the findings were similar, we present averaged concentrations from the 3 brain regions.

Results

The study was conducted among 286 MAP participants who completed a dietary assessment, died between November 2004 and 2013 (n = 554), and had a brain autopsy performed (n = 447); thus representing 51.6% of all deceased MAP participants. The mean (SD) postmortem time interval was 8.3 (7.4) hours. Of the analyzed sample, 43.7% (125) completed 1 food frequency questionnaire, 20.3% (58) completed 2, and 36% (103) completed 3 or more. The correlation between the reported number of seafood meals consumed at the first and last obtained food frequency questionnaires was the Spearman rank-order correrelation coeffecient (ρ = 0.57; P < .001). The mean time interval between administration of the first questionnaire and death was 4.5 years, and there were 2.4 years between the last questionnaire and death. Participants were coded as fish oil supplement users if they reported taking the supplements in at least 1 dietary assessment (n = 49); most (n = 31) reported use at only 1 visit.

The analytic sample had a mean (SD) age of 89.9 (6.1) years at death, was largely composed of women (67%; n=193), had mean (SD) educational attainment of 14.6 (2.7) years, and 22.7% (65) were positive for the APOE ε4 allele. These characteristics were comparable with those of the total number of deceased participants (89.2 [6.2] years at death, 68% were women, 14.2 [3.0] years of education, and 23.6% were APOE ε4 positive). These and other characteristics were similar over the tertiles of dietary n-3 fatty acid intake levels with the exception that participants in the highest tertile of intake were more likely to be women and those in the lowest tertile more likely to be APOE ε4 positive (Table 1). Participants in the lowest n-3 fatty acid tertile also had somewhat higher levels of neuropathology.

Table 1.

Characteristics of Deceased MAP Participants by Tertile of Dietary Long-Chain n-3 Fatty Acids (DHA + EPA) (n = 286) and by Tertile of Brain Mercury Concentrations (n = 203), 2005–2013

	Dietary n-3 Fatty Acid Level by Tertile, g/d			Brain Mercury Level by Tertile, μg/d
	1	2	3	1	2	3
No. of participants	116	82	88	67	69	67
Tertile
Mean (SD)	0.04 (0.02)a	0.09 (0.02)a	0.24 (0.13)a	0.02 (0.01)b	0.09 (0.09)b	1.28 (1.62)b
Median (interquartile range)	0.04 (0–0.06)a	0.09 (0.06–0.12)a	0.19 (0.12–0.92)a	0.02 (0.01–0.03)b	0.06 (0.03–0.20)b	0.56 (0.20–6.88)b
Characteristics
Age at death, mean (SD), y	89.2 (6.3)	90.9 (6.6)	90.0 (5.2)	88.2 (6.7)	89.1 (5.2)	90.5 (5.7)
Women, No. (%)	77 (66.4)	51 (62.2)	65 (73.9)	44 (65.7)	42 (60.9)	47 (70.1)
APOE ε4, No. (%)	34 (29.3)	14 (17.1)	17 (19.3)	16 (24.6)	12 (17.4)	18 (28.1)
Education, mean (SD), y	14.3 (2.8)	14.8 (2.5)	14.7 (2.7)	14.6 (2.4)	15.0 (2.9)	14.3 (2.5)
Dietary Consumption
Seafood consumption, mean (SD), meals/wk	1.3 (1.1)	2.0 (0.9)	2.7 (1.3)	1.8 (1.1)	2.0 (1.4)	2.3 (1.3)
Fish oil supplement use,
No. (%)	14 (12)	20 (24)	15 (17)	11 (16)	13 (19)	8 (12)
α-Linolenic acid (18:3 n-3), mean (SD), g/d	1.14 (0.43)	1.16 (0.40)	1.10 (0.39)	1.12 (0.41)	1.13 (0.43)	1.14 (0.42)
Total energy intake, mean (SD), calories/d	1837 (611)	1891 (572)	1739 (505)	1837 (536)	1757 (547)	1807 (603)
Brain Measures
Global Alzheimer disease neuropathology, mean (SD), score	0.77 (0.58)	0.57 (0.50)	0.69 (0.55)	0.73 (0.54)	0.64 (0.60)	0.59 (0.47)
Neuritic plaque density, mean (SD), zc	2.92 (1.05)	2.56 (1.18)	2.81 (1.18)	3.01 (1.06)	2.56 (1.23)	2.71 (1.09)
Neurofibrillary tangle severity, mean (SD), scored	3.68 (1.25)	3.41 (1.06)	3.61 (1.24)	3.54 (1.25)	3.48 (1.29)	3.54 (1.28)
Alzheimer disease diagnostic, mean (SD), scoree	2.84 (0.72)	2.61 (0.60)	2.81 (0.74)	2.85 (0.70)	2.70 (0.79)	2.70 (0.70)
Brain macroinfarcts, No. (%)	52 (46)	24 (30)	29 (33)	29 (43)	28 (41)	20 (30)
Brain microinfarcts, No. (%)	36 (31)	21 (26)	21 (24)	28 (42)	14 (20)	15 (23)
Lewy bodies, No. (%)	27 (23)	22 (27)	13 (15)	14 (21)	11 (16)	13 (20)
Selenium, mean (SD), μg/g	1.43 (0.66)	1.38 (0.73)	1.36 (0.58)	1.14 (0.16)	1.20 (0.23)	1.85 (0.96)

Abbreviations: APOE 4, apolipoprotein E; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; MAP, Memory and Aging Project.

^aUnits of measure are reported as grams/day for dietary n-3 fatty acid levels.

^bUnits of measure are reported as micrograms/day for brain mercury levels.

^cData are reported as a z-score of square root counts. Consortium to Establish a Registry of Alzheimer Disease score range: 1 (no neuritic plaques) to 4 (frequent neuritic plaques).

^dBraak staging score range: 0 (no tangles in any region) to 6 (frequent tangles across multiple regions).

^eNational Institute on Aging and the Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association score range: 1 (low Alzheimer disease pathology) to 4 (high Alzheimer disease pathology).

The subsample of 203 autopsied cases for which brain concentrations of metals were analyzed had similar characteristics to the larger sample (Table 1). Brains of participants in the highest tertile of mercury concentrations were more likely to be older at death, to be women, and to have lower levels of global Alzheimer disease neuropathology and higher brain selenium concentrations.

Mean (SD) levels for brain mercury concentrations were 0.26 (0.87) μg/g in the midfrontal region, 0.25 (0.78) μg/g in the inferior temporal region, and 0.87 (1.80) μg/g in the cerebellum The levels were correlated across regions (Spearman ρ, 0.65 to 0.76 [P < .001]). Brain selenium concentrations were comparable across the brain regions (Table 2). Brain cortical mercury levels were positively correlated with cortical selenium levels (ρ, 0.35 [P < .001]) and with the number of seafood meals consumed per week (ρ, 0.16 [P = .02]).

Table 2.

Brain Concentrations of Total Mercury and Selenium by Brain Region for 203 Autopsied Brains of Deceased MAP Participants, 2005–2013

	Brain Region				P Value
	Inferior Temporal	Midfrontal	Cerebellum	All Regions
Mercury, μg/g
Mean (SD)	0.25 (0.78)	0.26 (0.87)	0.87 (1.80)	0.46 (1.09)
Median (interquartile [1–3] range)	0.03 (0.01–0.06)	0.03 (0.02–0.07)	0.07 (0.03–0.67)	0.65 (0.02–0.28)
Correlation, Spearman ρ				0.65–0.76	<.001
Selenium, μg/g
Mean (SD)	1.18 (0.48)	1.24 (0.47)	1.77 (1.22)	1.40 (0.65)
Median (interquartile [1–3] range)	1.16 (0.95–1.29)	1.18 (1.06–1.32)	1.36 (1.25–1.62)	1.23 (1.09–1.36)
Correlation, Spearman ρ				0.42–0.49	<.001

Abbreviation: MAP, Memory and Aging Project.

Brain Neuropathology and Dietary n-3 Fatty Acids

We first analyzed the data in the entire sample of 286 deceased participants. Dietary intake of α-linolenic acid was correlated with decreased odds of cerebral macroinfarcts and microinfarcts in models adjusted for age at death, sex, and education but was not correlated with other neuropathological markers (Table 3). Dietary intake of long-chain n-3 fatty acids had a nonlinear correlation with cerebral macroinfarcts in the adjusted model but was not correlated with microinfarcts or other neuropathologic markers. Seafood consumption and fish oil supplement use were not correlated with any brain neuropathology in the total sample.

Table 3.

Relations of Cerebral Infarcts and Lewy Body Neuropathology to Dietary Intakes of Seafood, Long-Chain n-3 Fatty Acids (DHA + EPA), and α-Linolenic Acid Based on Adjusted Logistic Regression Models Among 286 Deceased MAP Participants, 2004–2013^a

Model	Odds Ratio (95% CI)
	Macroinfarcts (n = 282)	Microinfarcts (n = 282)	Lewy Bodies (n = 283)b
Seafood (≥ 1 meal/wk vs <1 meal/wk)	0.99 (0.52–1.88)	0.98 (0.49–1.96)	0.70 (0.34–1.45)
DHA + EPA Food Sources
Tertile
1	1 [Reference]	1 [Reference]	1 [Reference]
2	0.50 (0.27–0.92)	0.74 (0.39–1.42)	1.15 (0.59–2.23)
3	0.60 (0.33–1.08)	0.67 (0.35–1.26)	0.54 (0.26–1.14)
P value for linear trendc	.12	.23	.09
α-Linolenic 18:3 n-3
Tertile
1	1 [Reference]	1 [Reference]	1 [Reference]
2	0.65 (0.36–1.18)	0.82 (0.44–1.54)	1.40 (0.69–2.86)
3	0.51 (0.27–0.94)	0.49 (0.25–0.96)	1.42 (0.69–2.90)
P value for linear trendc	.03	.04	.35

Abbreviations: DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; MAP, Memory and Aging Project.

^aModels were adjusted for age at death, sex, and education and total energy intake (seafood model only).

^bLewy body staging range: 0 (no disease), 1 (nigra-predominant disease), 2 (limbic-type disease), 3 (neocortical disease).

^cLinear trend variable in which values within each tertile were scored at the median intake level for that tertile.

Based on observed statistical interactions between APOE ε4 genotype and seafood or n-3 fatty acid intakes on cognitive decline in previous studies,^23–30 we investigated this potential effect modification on brain neuropathology. We observed significant interactions with APOE ε4 status and dietary intakes of seafood and long-chain n-3 fatty acids on every measure of Alzheimer disease neuropathology (all P ≤ .04) except for n-3 fatty acids and neurofibrillary tangle severity (eTable 1 in the Supplement). Tests for interaction effects between APOE ε4 and either α-linolenic acid or fish oil supplement use on the neuropathologic outcomes did not reach the significance threshold (P > .25), with the exception of macroinfarcts (P for interaction = .05) for which fish oil supplement use was correlated with lower risk in APOE ε4 carriers.

Further analyses examined correlations between the dietary variables and brain neuropathology stratified by APOE ε4 status (Table 4). Individuals who were APOE ε4 positive and consumed at least 1 seafood meal per week or had higher intakes of long-chain n-3 fatty acids had less Alzheimer disease neuropathology compared with those who consumed lower amounts. For example, among the APOE ε4 carriers who consumed seafood weekly, the likelihood score for Alzheimer disease was lower by 0.53 score units (Table 4). By contrast, there was no correlation between seafood consumption and neuropathology among those who were APOE ε4 negative. There were no material differences in the estimated effects with further adjustment for vegetable consumption except that the correlation between seafood consumption and neurofibrillary tangle severity was no longer statistically significant (β, −0.76 score units [95% CI, −1.54 to 0.03]).

Table 4.

Relations of Alzheimer Disease Neuropathology to Dietary Intakes of Seafood, Long-Chain n-3 Fatty Acids (DHA + EPA), and α-Linolenic Acid Based on Adjusted Linear Regression Models in Analyses Stratified By APOE ε4 Status Among 286 Deceased MAP Participants, 2004–2013^a

Model	Global Alzheimer Disease Pathology	Neuritic Plaque Densityb	Neurofibrillary Tangle Severityc	Alzheimer Disease Diagnostic Scored
Seafood (≥1 meal/wk vs < 1 meal/wk)
APOE ε4-positive, No.	65	65	65	65
β (95% CI)	−0.77 (−1.53 to −0.01)	−0.69 (−1.34 to −0.04)	−0.77 (−1.52 to −0.02)	−0.53 (−0.96 to −0.10)
APOE ε4-negative, No.	221	218	218	218
β (95% CI)	0.13 (−0.30 to 0.55)	0.26 (−0.14 to 0.65)	0.20 (−0.19 to 0.59)	0.20 (−0.04 to 0.43)
DHA + EPA Food Sources
APOE ε4-positive, No.	65	65	65	65
β (95% CI)	−4.80 (−8.60 to −0.99)	−5.04 (−8.20 to −1.88)	−2.36 (−6.25 to 1.53)	−2.46 (−4.65 to −0.28)
APOE-ε4-negative, No.	221	218	218	218
β (95% CI)	0.33 (−2.24 to 2.91)	1.24 (−1.17 to 3.65)	0.20 (−2.16 to 2.56)	0.81 (−0.63 to 2.25)
α-Linolenic 18:3 n-3
APOE ε4-positive, No.	65	65	65	65
β (95% CI)	−0.22 (−2.06 to 1.62)	0.00 (−1.57 to 1.58)	−0.18 (−1.99 to 1.64)	0.39 (−0.66 to 1.43)
APOE ε4-negative, No.	221	218	218	218
β (95% CI)	−0.07 (−1.05 to 0.91)	−0.43 (−1.35 to 0.49)	−0.03 (−0.93 to 0.88)	−0.37 (−0.93 to 0.18)

Abbreviations: APOE ε4, apolipoprotein E; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; MAP, Memory and Aging Project.

^aModels were adjusted for age at death, sex, education, and total energy intake (seafood model only).

^bConsortium to Establish a Registry of Alzheimer Disease score range: 1 (no neuritic plaques) to 4 (frequent neuritic plaques).

^cBraak staging score range: 0 (no tangles in any region) to 6 (frequent tangles across multiple regions).

^dNational Institute on Aging and the Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association score range: 1 (low Alzheimer disease pathology) to 4 (high Alzheimer disease pathology).

We reanalyzed the adjusted models after excluding all dietary assessments that were obtained at the time of an Alzheimer disease diagnosis. The β-coefficients increased in magnitude and remained statistically significant between seafood consumption and global Alzheimer disease pathology (β, −0.88 score units [95% CI, −1.74 to −0.02]) and neuritic plaques (β, −0.83 score units [95% CI, −1.56 to −0.10]), were unchanged for Alzheimer disease diagnostic score (β, −0.53 score units [95% CI, −0.99 to −0.07]), and no longer statistically significant for neurofibrillary tangle severity (β, −0.73 score units [95% CI, −1.54 to 0.08]).

Brain Neuropathology and Brain Metal Concentrations

In further analyses, we examined whether brain concentrations of the neurotoxin mercury or the antioxidant trace element selenium were correlated with brain neuropathology in separate models adjusted for age at death, sex, education, and seafood consumption (Table 5; eTable 2 in the Supplement). There was no significant correlation between brain mercury levels and Alzheimer disease neuropathology. Brain mercury levels were significantly correlated with lower odds of macroinfarcts and microinfarcts. Brain selenium levels were not correlated with the neuropathologic markers except for neurofibrillary tangle severity in which higher selenium levels were correlated with higher tangle severity. Tests for statistical interaction by APOE ε4 status on the brain metals and neuropathology did not reach statistical significance (P > .26).

Table 5.

Effect Estimates for Brain Mercury and Selenium Levels With Alzheimer Disease Neuropathology Based on Adjusted Linear Regression Models (Alzheimer Disease Neuropathology) and Logistic Regression (Vascular and Lewy Body Neuropathology) Among 203 Deceased MAP Participants, 2005–2013^a

	Mercury Level by Tertile			P Value for Trendb	Selenium Level by Tertile			P Value for Trendb
	1	2	3		1	2	3
Tissue Concentration
Mean, μg/g	0.02	0.09	1.28		1.03	1.23	1.94
Median (interquartile [1–3] range), μg/g	0.02 (0.01 to 0.02)	0.06 (0.04 to 0.11)	0.56 (0.23 to 1.48)		1.04 (1.00 to 1.09)	1.23 (1.18 to 1.27)	1.51 (1.36 to 2.31)
Multiple Regression Models of Alzheimer Disease Neuropathology, β (95% CI)
Global Alzheimer disease pathology	0.0 [Reference]	−0.31 (−0.72 to 0.11)	−0.40 (−0.82 to 0.02)	.07	0.0 [Reference]	0.02 (−0.39 to 0.43)	0.24 (−0.19 to 0.67)	.29
Neuritic plaque densityc	0.0 [Reference]	−0.41 (−0.79 to −0.04)	−0.38 (−0.76 to 0.00)	.07	0.0 [Reference]	−0.00 (−0.38 to 0.38)	0.17 (−0.22 to 0.56)	.42
Neurofibrillary tangle severityd	0.0 [Reference]	−0.02 (−0.43 to 0.38)	−0.18 (−0.59 to 0.23)	.37	0.0 [Reference]	−0.04 (−0.44 to 0.35)	0.45 (0.05 to 0.86)	.04
Alzheimer disease diagnostic scoree	0.0 [Reference]	−0.12 (−0.36 to 0.12)	−0.22 (−0.47 to 0.02)	.08	0.0 [Reference]	0.02 (−0.22 to 0.26)	0.13 (−0.12 to 0.37)	.33
Logistic Regression Models of Infarcts and Lewy Bodies, Odds Ratio (95% CI)f
Macroinfarcts	1 [Reference]	1.01 (0.49 to 2.07)	0.48 (0.23 to 1.03)	.05	1 [Reference]	0.64 (0.31 to 3.05)	0.66 (0.31 to 1.33)	.25
Microinfarcts	1 [Reference]	0.37 (0.17 to 0.80)	0.37 (0.17 to 0.80)	.01	1 [Reference]	1.17 (0.56 to 2.48)	0.73 (0.33 to 1.63)	.49
Lewy bodies	1 [Reference]	0.71 (0.29 to 1.73)	0.86 (0.36 to 2.03)	.77	1 [Reference]	1.20 (0.47 to 3.03)	1.79 (0.73 to 4.39)	.21

Abbreviation: MAP, Memory and Aging Project.

^aModels were adjusted for age at death, sex, years of education, seafood consumption, total energy, and for Alzheimer disease neuropathology outcomes apolipoprotein E (APOE ε4) and seafood * APOE ε4).

^bLinear trend variable in which values within each tertile were scored at the median intake level for that tertile.

^cConsortium to Establish a Registry of Alzheimer Disease score range: 1 (no neuritic plaques) to 4 (frequent neuritic plaques).

^dBraak staging score range: 0 (no tangles in any region) to 6 (frequent tangles across multiple regions).

^eNational Institute on Aging and the Ronald and Nancy Reagan Research Institute of the Alzheimer’s Association score range: 1 (low Alzheimer disease pathology) to 4 (high Alzheimer disease pathology).

^fLewy body staging range: 0 (no disease), 1 (nigra-predominant disease), 2 (limbic-type disease), 3 (neocortical disease).

We also investigated the possibility of statistical interaction between brain concentrations of mercury and selenium (an antagonist to mercury toxicity) on the neuropathologic markers.³¹ Tests for interaction between these 2 biometals on brain neuropathologies in the adjusted models did not reach statistical significance (P > .14).

Discussion

In this cross-sectional study of older Midwestern residents, weekly consumption of seafood and dietary intake of long-chain n-3 fatty acids were inversely correlated with Alzheimer disease neuropathology but only among APOE ε4 carriers. Dietary intakes of seafood and long-chain n-3 fatty acids were not correlated with brain infarcts or with Lewy bodies. Higher intake of α-linolenic acid, the shorter-chain n-3 fatty acid found in plants, was correlated with decreased risk of cerebral infarcts; however, there was no evidence of effect modification by APOE ε4 status. Although seafood consumption was correlated with higher brain levels of mercury, the higher mercury levels were not significantly correlated with increased brain neuropathology.

In a large epidemiological study of middle-aged Dutch adults, dietary α-linolenic acid was associated with 35% to 50% reduction in incident strokes.³² A large number of epidemiological studies reported protective relations between seafood consumption and n-3 fatty acids with cognitive decline^{22,27,33–37} and incident dementia.^1–5,7 Of those that investigated effect modification by APOE ε4 status, the findings are inconsistent. Some reported protective associations with APOE ε4 carriers,^5,25 others with APOE ε4 noncarriers,^2,27 and still others with no effect modification by APOE ε4.^35,38,39 Of 3 randomized controlled trials of fish oil supplementation, 2 observed positive effects on cognitive change in healthy older adults who were APOE ε4 carriers,^28,29 and the third observed positive effects in Alzheimer disease patients who were APOE ε4 noncarriers.²⁴ We did not observe a correlation between fish oil supplement use and brain neuropathology although use was very limited and not consistent over time. DHA, the primary lipid in the brain, decreases with age as a result of lipid peroxidation.⁴⁰ DHA metabolism may be affected by a host of factors including age, APOE genotype, body mass index, sex, and alcohol consumption, which may explain the inconsistencies across studies.⁴¹

Protection against Alzheimer disease neuropathology from dietary n-3 fatty acids in APOE ε4 carriers was also reported in an animal study that measured the effects of a fish oil diet vs control diet in APOE ε4- and APOE ε3-targeted mice.⁴² The study demonstrated the APOE ε4 phenotype of increased Aβ₄₂ (the abnormal protein characteristic of Alzheimer disease pathology), impaired synaptic function, and decreased learning and memory, and that these were prevented by a fish oil diet.⁴²

To our knowledge, this is the first study to report on the relationship between brain concentrations of mercury and brain neuropathology or diet. The finding of no deleterious correlations of mercury on the brain is supported by a number of case-control studies that found no difference between Alzheimer disease patients and controls in mercury concentrations in the brain,^19,43,44 serum,⁴⁵ or whole blood.⁴⁶ A cross-sectional population study observed no association between blood concentrations of mercury and cognitive test scores.⁴⁷ In the MAP study, brain mercury levels trended in the protective direction with the neuropathologic markers. One cohort study reported a statistically significant inverse association between blood mercury levels and incident Alzheimer disease.³⁸

There is a vast literature on the benefits of n-3 fatty acids on neurocognitive development.⁴⁸ One study found that the cognitive benefits in early childhood associated with maternal fish consumption in utero were offset by higher erythrocyte mercury levels in the mothers.^10,48 We did not observe a similar interaction between mercury and seafood consumption on brain health in the aged brain.

The study findings were from a cohort of individuals initially free of dementia who were observed until death. The design minimizes the chance of selection bias that occurs in case-control studies. It also provides a wide spectrum of disease pathology from which to observe associations with environmental exposures. The comparatively large number of brains analyzed decreases the likelihood that the findings are due to chance and increased our ability to observe modifications in the correlations by APOE ε4 status. Nevertheless, insufficient statistical power may have prevented detection of small correlations of n-3 fatty acids on neuropathology in APOE ε4 noncarriers. Other strengths of the study include the validated and/or standardized measurements of diet, metal concentrations, and neuropathological measures. The measurements were based on averaged assessments over multiple years (diet) and brain regions (metal concentrations and neuropathologies), which help to minimize bias due to measurement error.

A major limitation of the study is the observational study design, which precludes causal interpretation of the data. In addition, the measurement of brain pathologies and mercury levels at the time of death prevents the assessment of temporality in relation to dietary intake levels. This limitation is mitigated by our finding in a previous study that weekly seafood consumption was associated with slower cognitive decline over a mean 4.9 years among 915 MAP participants.⁴⁹ Another limitation of the study is the subjective measure of dietary intake, although the dietary tool was validated in older adults.^20,50 Further testament to the validity of the diet assessment is the positive correlation between brain mercury concentrations and seafood consumption. Another study limitation is that only half of the deceased MAP participants had both autopsy and dietary intake data for analyses. Concerns of sample bias are allayed by the fact that the analyzed sample was comparable with all deceased participants in age, sex, education, and APOE ε4 status. Seafood intake in the MAP study population was moderate, and therefore the findings cannot be generalized to populations with higher seafood consumption or with high mercury exposure. It is likely that types of fish consumed in this study reflect the top 10 consumed species in the United States, which have low-to-moderate levels of mercury.⁵¹ Nevertheless, the levels of mercury in the MAP study population were comparable with levels previously reported for cortical brain regions.¹⁹

The study findings were from a very old, largely non-Hispanic white cohort and may not be generalizable to younger adults or other racial or ethnic groups.¹² Brain concentrations of DHA decrease with older age because of lipid peroxidation,⁵² which is higher among APOE ε4 carriers.⁵³ Thus, fish consumption may be more beneficial with older age.

Conclusions

In cross-sectional analyses, moderate seafood consumption was correlated with lesser burden of brain Alzheimer disease neuropathology in APOE ε4 carriers. Although seafood consumption was correlated with higher brain levels of mercury, these levels were not correlated with brain neuropathology.