Effect of apolipoprotein E ε4 and its modification by sociodemographic characteristics on cognitive measures in South Asians from LASI‐DAD

Abstract BACKGROUND We investigated the effects of apolipoprotein E (APOE) ε4 and its interactions with sociodemographic characteristics on cognitive measures in South Asians from the Diagnostic Assessment of Dementia for the Longitudinal Aging Study of India (LASI‐DAD). METHODS Linear regression was used to assess the association between APOE ε4 and global‐ and domain‐specific cognitive function in 2563 participants (mean age 69.6 ± 7.3 years; 53% female). Effect modification by age, sex, and education were explored using interaction terms and subgroup analyses. RESULTS APOE ε4 was inversely associated with most cognitive measures (p < 0.05). This association was stronger with advancing age for the Hindi Mental State Examination (HMSE) score (β ε4×age = −0.44, p = 0.03), orientation (β ε4×age = −0.07, p = 0.01), and language/fluency (β ε4×age = −0.07, p = 0.01), as well as in females for memory (β ε4×male = 0.17, p = 0.02) and language/fluency (β ε4×male = 0.12, p = 0.03). DISCUSSION APOE ε4 is associated with lower cognitive function in South Asians from India, with a more pronounced impact observed in females and older individuals. Highlights APOE ε4 carriers had lower global and domain‐specific cognitive performance. Females and older individuals may be more susceptible to ε4 effects. For most cognitive measures, there was no interaction between ε4 and education.


INTRODUCTION
Dementia, a group of neurological disorders characterized by cognitive impairment, is a leading cause of death, disability, and dependency among older adults worldwide. 1,24][5] In European ancestry populations, having one ε4 allele is associated with an increase of 2-3 times in the risk of AD compared to the most common and risk-neutral ε3/ε3 genotype, while having two ε4 alleles is associated with an increase in risk of up to 15 times. 6Conversely, the ε2 allele is generally associated with reduced AD risk.These APOE alleles are also linked, to a lesser extent, with global cognition and specific cognitive domains including episodic memory, executive function, and verbal fluency in individuals without dementia. 7ciodemographic factors such as age, sex, and education are also associated with cognitive function in older adults 8 and contribute to the risk of dementia. 9,10There is increasing evidence that these factors interact with APOE ε4 on cognitive function and dementia.Some studies suggest that the ε4 effect on AD may be more pronounced in young-old adults (e.g., 60-70 years) compared to older individuals (> 70 years), 6,11,12 in females compared to males, 6,[13][14][15] and in those with lower education. 16,179][20][21] In contrast to AD, the ε4 effect on cognitive function appears to increase progressively with age. 22,23[26][27] To date, genomic studies on cognitive function and dementia have predominantly focused on European ancestry populations. 28,29India is the most populous country in the world with over 4.1 million people estimated to have dementia, 2 yet Indian and South Asian populations in general are rarely represented in genomic studies.This underrepresentation is concerning as genetic effects and disease associations could vary across ancestry groups due to differences in allele frequencies and linkage disequilibrium patterns. 30Indeed, some evidence suggests that the effect of APOE ε4 on cognitive function and dementia differs across ancestries and race/ethnicities. 31Previous studies showed that the association between ε4 and AD was weaker among African Americans and Hispanics compared to Caucasians, but stronger in Japanese. 6Other studies investigating the association of ε4 with cognitive function and decline also reported a weaker effect among African Americans and Hispanics. 32,335][36][37] However, these studies have primarily relied on geographically restricted samples within India.Furthermore, unique cultural, social, and environmental characteristics of India, such as dietary practices and pathogen or toxicant exposures, may be associated with cognitive function and dementia and/or modulate the genetic effect of APOE ε4 in various ways.

Study population
The Longitudinal Aging Study of India (LASI) is a nationally representative sample of more than 73,000 adults from India aged 45 years or older. 38LASI-DAD is an add-on study of late-life cognition and dementia, in which a subsample of approximately 4000 community-residing LASI respondents aged 60 or older were administered in-depth cognitive tests and informant interviews.A two-stage stratified random sampling method was used in LASI-DAD to ensure a sample with a broad distribution of cognitive ability including respondents with dementia and mild cognitive impairment (MCI).The respondents were stratified according to their state of residence and their risk of cognitive impairment based on their performance on the cognitive tests and proxy interview (i.e., interview with a family member when the respondent is incapable of participating in the interview) in the main LASI study.Next, within each state, a predetermined sample size proportional to the state's population size was used to randomly select an equal number of respondents at high and low risk of cognitive impairment. 39The baseline (Wave 1) LASI-DAD sample consists of 4096 older adults aged 60 or older, representing > 600 ethnically and geographically diverse areas within 18 states and union territories.

Whole genome sequencing (WGS)
WGS was performed on a total of 2762 individuals at MedGenome, Inc (Bangalore, India) at an average read depth of 30.After removing the control samples, a total of 2736 LASI-DAD participants who consented to blood sample collection had WGS data available for analysis.There is minimal evidence of significant differences in cognitive function between participants who underwent WGS and those who did not (Table S1).Genotype calling and quality control (QC) on the raw LASI-DAD WGS data were performed at the Genome Center for Alzheimer's Disease (GCAD) at the University of Pennsylvania. 40,41iefly, sample-level QC included checks for low coverage, sample contamination, sex discrepancies, concordance with previous genotype data, and duplicates. 40After further excluding samples with low quality and/or unresolved identity, a total of 2680 samples were retained in the analysis.At the genotype level, each genotype was evaluated and set to missing if read depth was less than 10 or genotype quality score was less than 20.At the variant level, a variant was excluded if it was monomorphic, was above the 99.8% Variant Quality Score Recalibration (VQSR) Tranche (the quality score was beyond the range that contains 99.8% of true variants), had a call rate ≤ 80%, or had an average mean depth > 500 reads.We further removed variants that were in low complexity regions identified with mdust. 42After QC and filtering, we retained a total of 71,109,961 autosomal bi-allelic variants that include 66,204,161 single nucleotide polymorphisms (SNPs) and 4,905,800 indels.Genetic principal components (PCs) were calculated using PCAir 43 to take into account the relatedness in the samples.
Specifically, the PCs were first calculated in a set of unrelated individuals (kinship cutoff = 0.044) and then projected to all other individuals.
For this analysis, we included 2607 individuals that are not related at 3rd degree or above.The top 10 PCs were included in all analyses to adjust for population stratification.criteria. 44A study flow chart is presented in Figure S1.In LASI-DAD, appropriate modifications were made to the HCAP to reflect India's unique cultural and contextual constraints.Notably, to accommodate the considerable proportion of older Indian adults who are illiterate and/or innumerate, the selection and administration of specific cognitive tests were adapted to reduce the dependency on literacy and numeracy skills. 39,45Cognitive tests were evaluated through two pretests conducted prior to the first wave of data collection.Further adjustments were made to the protocol based on the observations and feedback received from these two pretests.widely used cognitive screening instrument. 47The HMSE score ranges from 0 to 30, with higher scores indicating better cognitive functioning.
General cognitive function and domain-specific cognitive performance were derived using factor analysis of the LASI-DAD cognitive battery, as described by Gross et al. 46 Briefly, cognitive tests were classified into broad domains and further divided into narrow subdomains based on a priori knowledge and theory, following the structure of the Cattell-Horn-Carroll (CHC) theory of human cognitive abilities.The specific cognitive tests included in each domain are listed in Table S2.Unidimensional factor analyses were conducted separately for each narrow and broad cognitive domain.Once an adequate fit was achieved for each model, all the domains were combined into a hierarchical multiple domain factor analysis that included a general factor.The resulting factor scores were internally standardized to a N(0,1) distribution in the entire sample of 4096 LASI-DAD participants.

Covariates and effect modifiers
We examined several sociodemographic and socioeconomic characteristics that were self-reported by the participants at baseline as Age, sex, and education were also assessed as potential effect modifiers of the association between APOE ε4 and cognitive function.Effect modification by age was tested using interaction terms between ε4 and continuous age.We also examined ε4 effects in stratified analysis by 5year age group, with those aged 85 and older combined into a single group.We examined the interaction between ε4 and education using the three-level categorical education variable described above.Nearly half of our sample (∼48.7%) had never attended school, which aligns with the 56.5% illiteracy rate among older adults in India. 48Thus, we further investigated the interaction between APOE ε4 and a dichotomous education variable, distinguishing between individuals with no formal education and those with any formal education (0 years of education vs. > 0 years of education groups).

Statistical analysis
The We first examined the association between APOE ε4 (carrier vs. noncarrier) and cognitive measures in the full analytic sample using linear regression models.We then tested the interaction between APOE ε4 and age, sex, or education by including the two independent variables and their cross-product term into the same model.We further conducted age-, sex-, and education-stratified analyses to examine the ε4 effects separately within each subgroup, regardless of whether an interaction was detected.We employed two sets of models for all We performed two sensitivity analyses.First, we repeated our analyses to examine the effects of specific APOE genotypes and their modification by age, sex, and education on cognitive measures.In this analysis, we excluded individuals with the ε2/ε4 (protective/risk allele) and ε2/ε2 (low frequency; N = 5) genotypes and focused on the remaining four possible APOE genotypes: ε3/ε3 (reference group), ε2/ε3, ε3/ε4, and ε4/ε4.However, due to small subgroup sample sizes, we did not estimate the effect of APOE genotype within 5-year age groups or assess the APOE genotype interaction with the three-tier education variable and its subgroup effects.Second, there is growing evidence that vascular risk factors, including hypertension, diabetes, hypercholesterolemia, obesity, and smoking, increase the risks of cognitive decline and dementia. 49In India, despite the high prevalence of these risk factors, their impact on cognitive function remains unclear. 50erefore, for any identified ε4 effects and interactions, we addi-

APOE ε4 and sociodemographic characteristic associations with cognitive measures
Table 2 presents the associations between APOE ε4 and cognitive measures in the full sample.All effect estimations were in the expected direction, with ε4 carriers having lower cognitive function.In models adjusting for age, sex, state of residence, and the top 10 genetic PCs (Model 1), ε4 carrier status was associated with all cognitive measures except visuospatial functioning (p < 0.05).Specifically, being an ε4 carrier was associated with a 0.095-0.142SD decrease in cognitive scores and a 0.978 points reduction in HMSE score.The associations were slightly attenuated after further controlling for education, literacy, urban/rural residence, caste, and per capita household consumption in Model 2. Nevertheless, the association between APOE ε4 and HMSE score, general cognitive function, orientation, and memory remained significant (p < 0.05).

Effect estimates and p-values for all other covariates in Models
1 and 2 can be found in Tables S6 and S7, respectively.In Model 1, older age was inversely associated with all cognitive measures, while male sex was positively associated with all cognitive measures.These associations stayed largely consistent in Model 2. Education was also independently associated with all cognitive measures.Compared to individuals with less than lower secondary education, those with upper secondary or vocational training education showed a 0.179-0.458SD increase in cognitive scores and a 1.480 point increase in HMSE score, while those with tertiary education showed a 0.327-0.893SD increase in cognitive scores and a 2.392 points increase in HMSE score.2).In this further adjusted model, the individual contribution of ε4 carrier status to the total variance explained ranged from 0.01% to 0.27%.

Effect modification of age, sex, and education on the associations between APOE ε4 and cognitive measures
We next examined whether the associations of APOE ε4 with cognitive measures were modified by age, sex, and education.Table 3 provides the results from testing the APOE ε4 × age interaction term.We found a significant APOE ε4 by age interaction on HMSE score, orientation, and language/fluency in Model 2, with the associations being more pronounced with advancing age.To illustrate the identified age interactions, Figure 1 shows the predicted cognitive scores for ε4 carriers and noncarriers at the 25th (64 years) and 75th percentile (74 years) of age.Examining the ε4 effect across 5-year age groups demonstrates a similar trend, where there is an overall increase in the magnitude of the ε4 effect as age increases, particularly from the 70-75 age group onward, culminating in the strongest association identified in the ≥ 85 age group.This trend was consistent across most cognitive measures, except for visuospatial function, where the ε4 effect was most pronounced in the 60-65 age group (Figure S3).However, due to the small sample sizes in each 5-year age group, we note that while the trend appears to show increased effects at older ages, most effect estimates are not significantly different from zero.Table 4 presents the results from testing the APOE ε4 by sex interaction term.In Model 1, an APOE ε4 × male interaction was observed for language/fluency and memory, with the effect of ε4 stronger in females.While the interaction terms were not statistically significant for the other cognitive measures, sex-stratified analysis showed that APOE ε4 was associated with all cognitive measures among females except visuospatial functioning, whereas there was no association between ε4 and any cognitive measures among males (Table 5).Sex differences in the ε4 effect attenuated in Model 2, but the APOE ε4 × male sex interaction term remained statistically significant for both language/fluency and memory (Figure 2, Tables 4   and 5).
Table S8 provides the results from testing the APOE ε4 by education interaction terms.Using the three-level education variable, all interaction terms between APOE ε4 and education were not statistically significant except the interaction between ε4 and upper secondary or vocational training on visuospatial function.A closer examination of the education-stratified analysis showed that most significant ε4 associations were predominantly observed in the less than lower secondary education group (Table S9).Given that this group constituted ∼75% of our sample, with more than 64% of them having no formal education, we further explored the APOE ε4 by education interaction by separating our sample into two categories: those with no formal education (0 years of education) and those with any type of formal education (> 0 years of education).We found no interaction between APOE ε4 and the dichotomous education variable on any of the cognitive measures (Table S10).However, when we examined the ε4 associations with cognitive measures separately within these two subgroups, we observed that ε4 was associated with all cognitive measures except visuospatial function among individuals without formal education in Model 1 (Table S11).In contrast, among those who had received some education, ε4 was only associated with HMSE score and orientation domain score.This pattern of association largely persisted in Model 2, except for the loss of significance in the associations between ε4 and executive function and language/fluency among those with no formal education.

Sensitivity analysis
Sensitivity analyses examining APOE genotypes showed that the main effect of ε3/ε4 compared to ε3/ε3, as well as its potential modification by age, sex, and education closely mirrored what we observed for ε4 carriers in the primary analyses (Tables S12-S17).Furthermore, even though not statistically significant, we found that the effect of ε4/ε4 on cognitive function was in the expected direction, with lower scores observed among ε4/ε4 compared to ε3/ε3 in the full sample and within sex and education subgroups.Contrary to our expecta-tion based on previous literature, we did not consistently observe a clear dose-response relationship between the number of ε4 alleles and worse cognitive function, as indicated by the smaller magnitude of effect in ε4/ε4 homozygotes compared to ε3/ε4 heterozygotes in the full sample, as well as within subgroups of male sex and 0 years of education (Tables S12, S15, and S17).However, we caution that our sample included only a small number of individuals carrying the ε4/ε4 genotype (N = 36), which likely contributed to our inability to establish statistically significant associations or interactions.Finally, results from the sensitivity analyses also revealed an association between APOE ε2/ε3 (Table S12).We did not detect any significant interactions between the ε2/ε3 genotype and age, sex, or education (Tables S13, S14, and S16).
Nevertheless, an adverse effect of the ε2/ε3 genotype on visuospatial function was found among females but not among males (Table S15).
Results from sensitivity analyses further controlling for vascular risk factors also yielded similar findings to our main analyses (Tables S18 and S19).The main effect of APOE ε4 and its interactions with age were negligibly different from those observed in the main analyses.
Interactions between APOE ε4 and sex on language/fluency and memory were slightly attenuated but remained in the same direction and marginally significant (p = 0.062 for language/fluency and p = 0.052 for memory).

DISCUSSION
In this large, nationally representative sample of Indians aged 60 and older, APOE ε4 carrier status was associated with lower cognitive function, both globally and in multiple specific domains.We also found that females may be more susceptible to ε4 effects on memory and language/fluency, while older individuals may be more susceptible to ε4 TA B L E 4 Two-way interaction between APOE ε4 carrier status and sex on cognitive measures.for South Asians 51 and reaffirms the relatively lower ε4 prevalence among South Asians compared to the global frequency of 14%. 3 We found modest adverse effects of APOE ε4 on a broad range of cognitive domains among older Indians.This finding contributes to the growing body of literature indicating that ε4 effects on cognitive function vary across racial and ethnic groups.For example, a previous study found that ε4 was associated with cognitive decline in middle-aged Whites but not African Americans. 32In Hispanic and Asian popula-tions, evidence of the ε4 effect on cognition is inconsistent, with some studies reporting a lack of association. 52,53Direct comparison of ε4 effects on cognition is challenging as methodological factors, including measurement and modeling of cognitive measures, may vary substantially across studies.Nonetheless, the observed domain specificity of ε4 effects on cognition in our study is in line with two meta-analyses of cognitively healthy European ancestry individuals which indicate that ε4 carriers perform worse on global cognition and episodic memory measures, but not on language/fluency (verbal ability) and visuospatial function. 22,54In contrast, we did not find an association between ε4 carrier status and executive function after adjusting for sociodemographic and socioeconomic factors.These results also differ slightly from earlier LASI-DAD studies in 932 participants, 55,56 which found no significant ε4 effects on cognition, potentially due to inadequate power and a more restricted geographic sample.
We also found a relatively consistent, though not always statistically significant, pattern across cognitive measures whereby the magnitudes of the ε4 effects on cognition appeared stronger among older individuals.This is in line with our hypothesis and several studies indicating that the negative effect of ε4 on cognitive function increases with advancing age. 22,57,58The molecular mechanisms underlying the potential age-dependent ε4 effect on cognitive function remain unknown.One possibility is the resource modulation hypothesis, 58 which posits that effects of common genetic variants, such as APOE, may become stronger as neurochemical and anatomical brain resources decline with age. 59,60Conversely, the observed stronger association between ε4 and cognition with advancing age might reflect the increased presence of dementia cases in older individuals, as ε4 is more robustly associated with dementia.However, not all studies have observed this interaction. 54,61In fact, one study showed that ε4 effects on global cognitive functioning and episodic memory decreased with age, although these findings were not statistically significant. 54These inconsistencies across studies may be attributed to methodologic differences, such as sample size, age range of the participants, and measurements of cognitive function.
We found that the ε4 association on cognition was stronger in females than males for language/fluency and memory, which is consistent with our hypothesis and a growing body of literature.
Earlier research indicates that the ε4 allele confers a greater risk for AD in women than men, 13,62 with recent studies confirming its more pronounced effect on AD biomarkers and cognitive function in females. 18,63,64Although the precise biological mechanisms underlying these observed sex differences remain largely unclear, there is compelling biological plausibility.For example, APOE may interact with estrogen through various pathways, exacerbating the adverse effects of estrogen fluctuations/loss during perimenopause and postmenopause on cognitive function and AD. 65Alternatively, the observed sex differences in our sample may be partly due to a more pronounced selection bias among males.APOE ε4 has pleiotropic effects on cardiovascular disease and mortality, which are also associated with being male. 19Given the older age of our LASI-DAD sample, it is likely to include healthier male ε4 carriers who demonstrate resilience against the detrimental ε4 effects on cognitive function and age-related diseases.Consequently, our study may underestimate the association between ε4 and cognition in males.Indeed, while our findings align with most studies examining sex differences, a few studies, including a recent analysis of UK Biobank participants with white British ancestry, found no interactions between ε4 and sex on cognitive abilities. 66Further investigation is required to elucidate the observed sex-specific ε4 effects on cognitive function and dementia in diverse populations.
In our study, higher education was associated with better cognitive performance as expected.However, contrary to our hypothesis, the APOE ε4 association with cognitive function did not differ significantly across education groups.Previous studies in European ancestry demonstrated that higher education may counteract the negative effects of ε4 on cognitive function and dementia by providing greater cognitive reserve, which enables individuals to better cope with brain pathologies, thereby mitigating genetic risk associated with ε4. 16,67ile we found no evidence of interaction when including a crossproduct interaction term between APOE ε4 and > 0 years of education, education-stratified analysis showed that the magnitude of the association between ε4 and most cognitive measures trended greater in those who never attended school compared to those who did.The lack of interactions may partially stem from analyzing education as a dichotomous variable based on having ever attended school, as some studies suggest that the interaction between ε4 and education may be at the tertiary level. 67Although we examined and found no interaction at the tertiary level, we may have lacked statistical power to detect such an interaction given that we had only 95 participants with tertiary education.Further, the moderate overall sample size may have limited our power to detect smaller effects.The complex social hierarchy and high social stratification in India, which we could not completely capture, might also influence this interaction.
Sensitivity analyses using APOE genotypes further revealed that ε2/ε3 was associated with lower visuospatial function compared to ε3/ε3.While the ε2 allele is generally considered protective against AD, previous studies of cognition have yielded inconsistent findings.For instance, one study found worse visuospatial attention in mid-adulthood ε2 carriers, 68 whereas another study in older Koreans without dementia reported better visuospatial performance in ε2 carriers. 69Additional research is needed to replicate our findings and elucidate the ε2 effect on cognitive function.
There is a higher prevalence of cardiometabolic risk factors in LASI-DAD compared to nationally representative US-based aging studies. 70wever, most traditional cardiovascular risk factors were not associated with cognitive function in LASI-DAD. 71In our study, adjusting for vascular factors did not alter the main findings, except for the interactions between ε4 and sex on language/fluency and memory, which remained directionally consistent but only marginally significant.While this may be due to reduced power from the smaller number of participants who had data on vascular risk factors, it is possible that the interaction between ε4 and sex on cognition may be confounded or moderated by these vascular factors.Future research is warranted to elucidate the role of vascular risk factors in ε4 associations with cognitive function, particularly in South Asians.
This study has limitations.First, we were unable to reliably estimate the dosage effect of ε4 due to the rarity of ε4 homozygotes (N = 36).Second, LASI-DAD only included participants aged 60 or older, which may have induced selection bias.Our results might underestimate the true association between ε4 and cognition since older adults carrying ε4 and with poorer cognitive function might be underrepresented due to morbidity or death.Contrary to previous studies that reported a decrease in ε4 frequency with increasing age, 72,73 our study showed no significant age differences in ε4 frequency.We did observe a decreasing trend in ε4 frequency up to age 85, followed by an increase, which may be the result of underlying population structure differences or the small sample size of the oldest age groups.
Third, despite adjusting for many potential confounders, there may be residual confounding from lifestyle factors and environmental toxicants.Fourth, all socioeconomic characteristics were self-reported and may be subject to response bias.Fifth, we were unable to examine longitudinal ε4 effects due to the cross-sectional study design.Future studies using longitudinal cognitive measures in LASI-DAD, which are currently being collected, could provide further insights into the interplay among ε4, sociodemographic, and cognitive aging in South Asians.
Finally, our findings need to be replicated in other South Asian/Indian samples.
In conclusion, this is the first study to investigate the effect of APOE administered a detailed battery of cognitive tests to measure a range of key cognitive domains affected by cognitive aging including memory, processing speed, language, executive function, and visuospatial skills, closely following the Harmonized Cognitive Assessment Protocol (HCAP).The HCAP was developed for the assessment of dementia and MCI in the US Health and Retirement Study and its sister studies around the world to enable international comparisons.
covariates.Sociodemographic factors included age, sex, educational level according to a three-tier simplified version of the 1997 International Standard Classification of Education (ISCED-97) codes (less than secondary education, upper secondary education and vocational training, or tertiary education), urban/rural residence, literacy (can read or write), and state of residence.We also included caste and per capita household consumption (in rupees) as measures of socioeconomic status.Respondents self-reported their caste and were classified into four official categories recognized by the Indian government: scheduled caste, scheduled tribe, other backward class (OBC), or no caste/other caste.Per capita household consumption was used to capture the economic status of the respondents and was calculated by taking total household consumption over the previous year for food, household utilities, fees, durable goods, education, healthcare, discretionary spending, transit, and remittances divided by the number of people in the household.In cases of missing data, which were limited to two covariates (caste and per capita household consumption) in our main analyses and affected fewer than 0.6% of the study sample, we conducted a complete case analysis by excluding those individuals from the model.

1
Plots of significant two-way interactions between apolipoprotein E (APOE) ε4 and age on cognitive measures.Predicted HMSE score (A), orientation score (B), and language/fluency score (C) of APOE ε4 carriers and noncarriers at the 25th (64 years) and 75th percentile (74 years) of age are shown with standard error bars by age group.Cognitive measures are predicted using models adjusted for age, sex (male), state of residence, top 10 genetic PCs, education (less than lower secondary, upper secondary or vocational training and tertiary education), urban/rural residence, literacy, caste, quintiles of per capita household consumption, and APOE ε4 × age (Model 2).Color coding is according to age.APOE, apolipoprotein E; HMSE, Hindi-mini-mental state examination; PCs, principal components.F I G U R E 2 Plots of significant two-way interactions between APOE ε4 and sex on cognitive measures.Predicted memory score (A) and language/fluency score (B) of APOE ε4 carriers and noncarriers are shown with standard error bars by sex.Cognitive measures are predicted using models adjusted for age, sex (male), state of residence, top 10 genetic PCs, education (less than lower secondary, upper secondary or vocational training, and tertiary education), urban/rural residence, literacy, caste, quintiles of per capita household consumption, and APOE ε4 × male (Model 2).Color coding is according to sex.APOE, apolipoprotein E; PCs, principal components.genotype and lower visuospatial function when compared to ε3/ε3

ε4
and its potential modification by sociodemographic factors on cognitive function in a nationally representative sample of older Indians.Our study provides important initial insights into the relationship between the strongest known genetic risk factor for AD and cognitive function in the South Asians population.Future research is warranted to provide a better understanding of the genetic influences on cognition and dementia within this genetically and socially diverse population.
We reviewed the literature using traditional sources (e.g., PubMed).There were several studies examining age, sex, and education as potential effect modifiers of the association between apolipopro- Therefore, large-scale epidemiological studies employing nationally representative data are needed RESEARCH IN CONTEXT 1. Systematic review: 3. Future directions: Future large-scale longitudinal research in the Indian/South Asian population is warranted to replicate our findings and further evaluate the age-, sex-, and education-related changes in the effects of APOE ε4 on cognitive aging.to comprehensively explore the effects of APOE ε4 on cognition and dementia and identify potential effect modifiers in Indian/South Asian populations.The Diagnostic Assessment of Dementia for Longitudinal Aging Study in India (LASI-DAD) is a population-based prospective cohort study that has collected nationally representative data on Indians aged 60 and older to better understand the determinants of late-life cognition, cognitive aging, and dementia.In this study, we examined the effect of the APOE ε4 allele and its potential modification by sociodemographic factors including age, sex, and education on cognitive measures in LASI-DAD.Based on prior literature, we hypothesize that the association between APOE ε4 and cognitive function would be more pronounced in older individuals, females, and those with a lower education level.
Descriptive statistics by APOE ε4 carrier status among 2563 participants in LASI-DAD.Note: Individuals with the APOE ε2/ε4 genotype were excluded from the sample.N (%) are reported unless otherwise specified.Abbreviations: APOE, apolipoprotein E; HMSE, Hindi mental state examination; LASI-DAD, diagnostic assessment of dementia for the longitudinal aging study of India; SD, standard deviation.
data for these variables (N = 2351) to address potential confounding from cardiovascular risk factors.For all analyses, statistical significance was established at p-value < 0.05.Given that all our hypotheses were pre-defined and grounded in priori evidence, and considering the inherently correlated nature of cognitive measures, we did not correct for multiple testing in this study.TA B L E 1 a p-value calculated from Wilcoxon rank sum test.b p-value calculated from Chi-square test.c p-value calculated from t-test.

Table S3
APOE ε4 carriers and noncarriers did not differ with respect to age, sex, and state of residence.However, compared to APOE ε4 noncarriers, ε4 carriers were more likely to have less than upper secondary education, reside in rural areas, and score lower on all cognitive mea- provides the distribution of APOE genotypes and alleles in the full study population, as well as by 5-year age intervals ranging from 60 years to ≥ 85 years.The allele frequencies of APOE ε2, ε3, and ε4 were 0.048, 0.844, and 0.108, respectively.There was no significant difference in the frequency of the ε4 allele by age when compared across 5-year age groups (p = 0.813) (TableS3and FigureS2).About 19% (N = 497) of the participants were classified as APOE ε4 carriers (18% ε3/ε4 heterozygotes and 1% ε4/ε4 homozygotes).Genotype frequencies observed in the full study sample were in Hardy-Weinberg equilibrium (p = 0.667).Descriptive statistics for sociodemographic characteristics and cognitive measures by APOE ε4 carrier status are shown in Table1.The mean age of the participants was 69.56 years (SD 7.29) and approximately 53% (N = 1358) were female.Seventy-five percent (N = 1926) had less than lower secondary education, with 48.7% (N = 1249) participants having no formal education.Twenty-one percent (N = 542) had upper secondary or vocational training, and 4% (N = 95) had tertiary education.ers were more likely to be underweight.Distribution of other vascular risk factors did not significantly differ between the two groups.Table S5 provides the correlation among the seven cognitive measures examined in this study.The correlation among the five cognitive domain scores ranged from 0.491 to 0.733.The correlations between the five domain scores and general cognitive function and HMSE ranged from 0.733 to 0.948 and from 0.560 to 0.856, respectively.
Effects of APOE ε4 carrier status on cognitive measures in LASI-DAD.Model 1 adjusted for age, sex, state of residence, and the top 10 genetic PCs.Model 2 adjusted for age, sex, state of residence, top 10 genetic PCs, education (less than lower secondary education, upper secondary or vocational training and tertiary education), literacy, urban/rural residence, caste, and quintiles of per capita household consumption.ΔR 2 represents the change in R 2 when APOE ε4 carrier status was added to the corresponding regression models.Beta coefficient and p-value in bold indicates statistically significant association at p < 0.05.Abbreviations: APOE, apolipoprotein E; HMSE, Hindi mental state examination; LASI-DAD, diagnostic assessment of dementia for the longitudinal aging study of India; PCs, principal components.
, APOE ε4 carrier status and other covariates in Model 1 explained approximately 18.6% (visuospatial function) to 33.9% (general cognitive function) of the total variance in cognitive TA B L E 2 measures.Specifically, ε4 carrier status alone contributed between 0.09% (visuospatial function) and 0.51% (HMSE score) of the explained variance.The total variance explained increased to 37.9%-62.4% when education, literacy, urban/rural residence, caste, and per capita household consumption were added to the model (Model Two-way interaction between APOE ε4 carrier status and age on cognitive measures.ondary education, upper secondary or vocational training and tertiary education), literacy, urban/rural residence, caste, quintiles of per capita household consumption, and APOE ε4 × Age.Beta coefficient and p-value in bold indicates statistically significant association at p < 0.05.Asterisk (*) denotes cognitive measures for which a statistically significant interaction term was observed.Abbreviations: APOE, apolipoprotein E; HMSE, Hindi mental state examination; LASI-DAD, diagnostic assessment of dementia for the longitudinal aging study of India; PCs, principal components.
Note: Model 1 adjusted for age (continuous and centered at the sample mean), sex (male), state of residence, top 10 genetic PCs, and APOE ε4 × Age.Model 2 adjusted for age (continuous and centered at the sample mean), sex (male), state of residence, top 10 genetic PCs, education (less than lower sec- Model 1 adjusted for age, sex (male), state of residence, top 10 genetic PCs, and APOE ε4 × male.Model 2 adjusted for age, sex (male), state of residence, top 10 genetic PCs, education (less than lower secondary education, upper secondary or vocational training and tertiary education), literacy, urban/rural residence, caste, quintiles of per capita household consumption, and APOE ε4 × male.Beta coefficient and p-value in bold indicates statistically significant association at p < 0.05.Asterisk (*) denotes cognitive measures for which a statistically significant interaction term was observed.Abbreviations: APOE, apolipoprotein E; HMSE, Hindi mental state examination; LASI-DAD, diagnostic assessment of dementia for the longitudinal aging study of India; PCs, principal components.
Associations between APOE ε4 carrier status and cognitive measures stratified by sex.Model 1 adjusted for age, state of residence, and top 10 genetic PCs.Model 2 adjusted for age, state of residence, top 10 genetic PCs, education (less than lower secondary education, upper secondary or vocational training and tertiary education), literacy, urban/rural residence, caste, and quintiles of per capita household consumption.Beta coefficient and p-value in bold indicates statistically significant association at p < 0.05.Asterisk (*) denotes cognitive measures for which a statistically significant interaction term was observed.