Assessment of the MIND diet score

The MIND diet score was computed based on the responses to a semi-quantitative food frequency questionnaire (FFQ), a modified version of the Harvard FFQ that was validated for use in Chicago community residents [25]. MAP participants reported the usual frequency of consumption of 144 food items over the previous 12 months. The MIND diet score has 15 dietary components, including 10 brain-healthy food groups (green leafy vegetables, other vegetables, nuts, berries, beans/legumes, whole grains, fish, poultry, olive oil, and wine) and 5 unhealthy food groups (red meat, fried and fast foods, pastry and sweets, butter, and cheese) [19]. Based on the frequency of intake reported for the healthy and unhealthy food groups, the MIND diet score was computed for each participant summing all 15 of the component scores. A cumulative average of MIND diet score across follow-up was used in analyses to limit measurement error [26]. The average of dietary assessments in the MAP cohort study was 3.5.

Assessment of [postmortem] brain pathology

A standardized protocol was used for brain removal, sectioning and preserving of tissue, and quantifying Alzheimer’s disease pathology and cerebral infarctions, as described in detail previously [27]. Briefly, staff blinded to any clinical data removed and weighed brains, and placed each hemisphere in a Plexiglas jig and cut coronally into 1 cm slabs. After slabs from one hemisphere were fixed in 4% paraformaldehyde for 3–21 days, they were dissected into blocks that were embedded in paraffin and cut into 6-μm sections.

Alzheimer disease (AD) pathology

Bielschowsky silver stain was used to visualize neuritic plaques, diffuse plaques, and neurofibrillary tangles in five cortical regions (frontal, temporal, parietal, the entorhinal cortex, hippocampus). Number of neuritic, diffuse plaques, and neurofibrillary tangles identified in a 1mm² area of highest density in 5 cortical regions were counted and used to develop a composite measure of global AD pathology as previously described [4]. In addition, to the global AD measure we also obtained molecularly specific and more quantitative estimates of the 2 major pathologic proteins. amyloid-$\beta$ protein was identified by molecularly-specific immunohistochemistry and quantified by image analysis in each region [27,28]. A summary measure for amyloid-$\beta$ burden across all regions was calculated. Neuronal neurofibrillary tangles are identified by molecularly specific immunohistochemistry (antibodies to abnormally phosphorylated Tau protein, AT8). Computer-assisted sampling was used to quantify the cortical density (per mm2) of tau immunoreactive tangles, and mean tangle density for all regions was used in these analyses.

Lewy body disease pathology

Lewy bodies were assessed in 6 regions, including substantia nigra, anterior cingulate cortex, entorhinal cortex, midfrontal cortex, superior or middle temporal cortex, inferior parietal cortex using a monoclonal phosphorylated antibody to $\alpha$-synuclein (1:20,000; Wako Chemical, Rich- mond, VA) with alkaline phosphatase as the chromogen. Details for the diagnosis of Lewy body we have previously described [29]. In short, Lewy bodies in the cortex were identified as round intracytoplasmic structures, which often lack any halo and an eccentric nucleus, whereas Nigral Lewy bodies were identified as round, intracytoplasmic structures with a darker halo.

TAR DNA-binding protein 43 (TDP-43)

TDP-43 immunohistochemistry was performed on 8 brain regions using phosphorylated monoclonal TAR5P-1D3 (pS409/410; 1:100, Ascenion, Munich, Germany) TDP-43 antibody. Since 2015, this antibody has been obtained from MilliporeSigma, Burlington, MA [30].

Assessment of cognition

All subjects underwent a uniform structured clinical evaluation with emphasis on neurologic function, and neuropsychological testing. Nineteen neuropsychological tests were used to create a composite measure of global cognitive function and five specific cognitive domains, including episodic memory, semantic memory, working memory, perceptual speed, and visuospatial ability. The raw scores of individual tests were standardized and averaged to z-scores, using the mean and standard deviation from the baseline evaluation of the entire study. Detailed information on the individual tests and the derivation of these scores has been described previously [37].

Assessment of covariates

Covariates included age at death, sex, years of education, APOE ϵ – 4, cognitive activities, and total energy intake. APOE genotype was determined for each participant by the Broad Institute for Population Genetics using the Polymorphic DNA Technologies [38,39]. Participants were classified as APOE $ϵ-4$ carriers (>1 allele) or noncarriers. Participation in cognitively stimulating activities was assessed with a structured questionnaire that assessed usual time spent in the past year on seven activities, including reading, writing letters, visiting a library, and playing games such as chess or checkers [40]. Each of these seven activities was scored and then averaged to yield a composite measure of the frequency of participation in cognitively stimulating activities.

MIND diet, brain pathology, and cognition

A higher MIND diet score was associated with better cognition. One standard deviation increase of the MIND diet score was associated with 0.119 units higher in the cognitive score ($\beta$=0.119, SE=0.040, P-value=0.003; Table 2 A). As expected, a higher measure of global AD pathology was associated with a worse cognitive score ($\beta$=−0.609, SE=0.066, p<0.001; Table 2 B–D). When both MIND diet and AD pathology were included in the same model, the MIND diet score remained independently associated with cognition, and estimates did not change substantially ($\beta$=0.111, SE=0.037, P-value=0.003; Table 3 A). Similar findings were noted for the association of the MIND diet with cognition when amyloid-$\beta$ load and tangles were added individually in the multivariable-adjusted model (Table 3 B–C). There was no evidence that AD pathology modified the relationship of MIND diet with cognition, as evidenced by the non-significant interaction terms of the MIND diet with Alzheimer’s disease pathology (P-value=0.667), amyloid-$\beta$ (P-value=0.871), and tangles (P-value=0.930).

Because other common neuropathologies in aging contribute to cognitive impairment, we developed separate linear regression models to examine whether the association of the MIND diet score with cognitive performance is independent of the associations of the other brain pathologies with cognition. In each of these models, the $\beta$-estimates of the association between MIND diet score and cognition were essentially unchanged when other common neuropathologies were included in separate analyses (Table 4 A–I). The greatest attenuation, albeit small, of the MIND diet estimates across all the pathologies considered, was seen in the model that included hippocampal sclerosis ($\beta$=0.103 versus $\beta$=0.119). There were no significant interactions between the MIND diet and any of the other common neuropathologies of aging (p>0.341 for each interaction term).

As a reflection of the overall burden of brain pathologies, all common neuropathologies of aging, together with AD pathology, were included in a single model, along with the MIND diet (Table 5 A-​-B).B). A higher MIND diet score was associated with a better cognition independently of the overall burden of brain pathologies ($\beta$=0.103, SE=0.037, P-value=0.005; Table 5 B).

MIND diet was not associated with brain pathology (AD pathology, amyloid-$\beta$ and tangles), including indices of vascular pathology such as macroinfarcts, microinfarcts, cerebral atherosclerosis, and arteriolosclerosis (Table 6).

The authors thank all the participants in the Rush Memory and Aging Project. They also thank the staff of the Rush Alzheimer’s Disease Center. More information regarding obtaining (MAP) data for research use can be found at the RADC Research Resource Sharing Hub (radc.rush.edu).

Research reported in this publication was supported by the National Institute On Aging of the National Institutes of Health, Grant/Award (R01AG17917, R01AG054476, R21AG070287, K01AG050823). The content is solely the responsibility of the authors and does not necessary represent the official views of the National Insitutes of Health.