Abstract
We examined the association between APOE ɛ2/ɛ4 with incident Alzheimer’s disease (AD) and mild cognitive impairment (MCI) among African Americans using the national dataset from the National Alzheimer’s Coordinating Center (NACC) from 2005 to September 2019. Compared to ɛ3/ɛ3 carriers, ɛ2/ɛ4 carriers exhibited a similar risk of incident AD (adjusted hazard ratio [aHR] = 0.85, 95% CI [0.39, 1.84]) among the AD cohort and similar risk of incident MCI (aHR = 0.88, 95% CI [0.51, 1.50]) among the MCI cohort. Our findings suggest that, unlike the increased risk of AD and MCI in non-Latino whites, APOE ɛ2/ɛ4 genotype is not associated with the incidence of AD and MCI among African Americans.
INTRODUCTION
The APOE gene is widely recognized as a major susceptibility factor for Alzheimer’s disease (AD) [1–4]. Its three common alleles (ɛ2, ɛ3, and ɛ4) give rise to six genotypes (ɛ2/ɛ2, ɛ2/ɛ3, ɛ3/ɛ3, ɛ2/ɛ4, ɛ3/ɛ4, and ɛ4/ɛ4). The ɛ3 is the most common allele and considered to play a “neutral” role for the onset of AD. Meanwhile, ɛ4 (ɛ3/ɛ4 or ɛ4/ɛ4) represents the largest genetic risk factor for AD, and ɛ2 (ɛ2/ɛ2 or ɛ2/ɛ3) may confer a protective effect against AD [5–9]. There is a paucity of studies on the effect of the APOE ɛ2/ɛ4 genotype on phenotype due to both the relative rarity of the ɛ2/ɛ4 genotype in the population and the counteractive effects of the ɛ2 and ɛ4 alleles. Our and others’ recent research linking APOE ɛ2/ɛ4 and AD demonstrates that APOE ɛ2/ɛ4 has a statistically significant effect on increased risk of AD and mild cognitive impairment (MCI) among non-Latino whites [10, 11]. Nevertheless, African Americans remain a genetically understudied group, although the incidence of AD is higher within this group than among whites [4, 12]. To date, only a few studies have investigated the association between APOE ɛ2/ɛ4 and the onset of AD and MCI among African Americans. In this study, we evaluated the effects of APOE ɛ2/ɛ4 on the incident AD and MCI among African Americans, using a large national dataset provided by the National Alzheimer’s Coordinating Center (NACC).
METHODS
The NACC recruited and followed-up participants starting in 2005, and this analysis includes participants recruited through September 2019 (N = 41,459). A consensus diagnosis of AD and MCI was established using standardized clinical criteria [13, 14] in the NACC dataset. The details on NACC patient recruitment, evaluation, and institutional review board approval have been previously published [14, 15]. Based on APOE genotyping, NACC participants were classified into four groups: ɛ2/ɛ4, ɛ2 (ɛ2/ɛ2, ɛ2/ɛ3), ɛ4 (ɛ3/ɛ4, ɛ4/ɛ4), and using ɛ3/ɛ3 as reference. Participants’ demographics, including sex, age, education, cognitive status at initial visit and the calendar year of the initial visit, were collected. Incidence of AD and MCI were defined from NACC-derived variables.
Our analysis was restricted to the African American ethnic group. Additionally, the included participants had to have an APOE genotype available and made at least two visits. The AD cohort was defined as dementia free at their first visit, and the MCI cohort was defined as both dementia and MCI free at their first visit. Based on these criteria, two cohorts were constructed for analyses: an AD cohort (N = 2,007) and an MCI cohort (N = 1,480) (Fig. 1).
Consort diagram displaying derivation of study sample for the analyses with source of exclusion. AD, Alzheimer’s disease; MCI, mild cognitive impairment; APOE, apolipoprotein.
Statistical analysis
Descriptive statistics of participants’ baseline were summarized for both the AD and MCI cohorts. The group differences of demographics across the four APOE groups were compared by using one-way ANOVA and chi-square tests (Table 1). Both unadjusted and adjusted Cox proportional hazards models were performed to examine the association between APOE genotype and incident AD dementia and MCI in each defined cohort. The adjusted model was controlled for demographics including age, sex, education, cognitive status at initial visit, and the calendar year of initial visit (Table 2). Survival time was defined as the length of time in years from the initial visit to the first event of interest (AD diagnosis for the AD cohort; Dementia or MCI diagnosis for the MCI cohort) or death without incident event. Otherwise, survival time was calculated as time from first visit to last follow-up evaluation. The proportional hazard assumption was tested based on the Schoenfeld residuals. In addition, sensitivity analyses with a competing risk subdistribution hazard model were conducted. In this model, we considered death as a competing risk event instead of censoring (Supplementary Table 1). All statistical analyses were made using SAS 9.4 and STATA 16.
Demographic Characteristics of Participants at Initial Visit
Hazard Ratio of APOE Genotype from Cox Proportional Hazards Model
*Models were adjusted for age, sex, education, baseline cognitive status, and year of initial visit.
RESULTS
Demographic characteristics are displayed in Table 1. Participants in the AD (N = 2,007) and MCI (N = 1,480) cohorts were similar: mean age at enrollment was approximately 71 years, participants were predominantly female (over 70%), and average education level was 14 years. We observed statistically significant differences of sex distribution across the four APOE groups for the AD cohort.
APOE genotype and incidence of AD
Among AD cohort participants (N = 2,007), ɛ2/ɛ4 accounted for 4.9%, ɛ2/ɛ2 & ɛ2/ɛ3 15%, ɛ3/ɛ3 & ɛ4/ɛ4 34.3%, and ɛ3/ɛ3 45.8%. The average follow-up time was 4.6 years. Overall, 216 (10.8%) developed AD, with the following percentages for each allele carrier: 7 of 99 (7.1%) ɛ2/ɛ4, 108 of 688 (15.7%) ɛ3/ɛ4 & ɛ4/ɛ4, 19 of 301 (6.31%) ɛ2/ɛ2 & ɛ2/ɛ3, and 82 of 919 (8.92%) ɛ3/ɛ3. Using a multivariable Cox proportional hazards model (Table 2), compared to ɛ3/ɛ3 carriers, we observed that ɛ2/ɛ4 carriers exhibited a similar risk for AD (7.1% versus 8.9%; aHR = 0.85, 95% CI [0.39, 1.84], p = 0.679). Meanwhile, ɛ2/ɛ2 & ɛ2/ɛ3 carriers exhibited a 25% decreased risk of AD (aHR = 0.75, 95% CI [0.45, 1.24], p = 0.262), although the decrease was not statistically significant. In contrast, ɛ3/ɛ4 & ɛ4/ɛ4 carriers revealed double the risk of AD (aHR = 2.04, 95% CI [1.52, 2.73], p < 0.0001).
APOE genotype and incidence of MCI
Among MCI cohort participants (N = 1,480), ɛ2/ɛ4 accounted for 4.8%, ɛ2/ɛ2 & ɛ2/ɛ3 16.1%, ɛ3/ɛ4 & ɛ4/ɛ4 32.4%, and ɛ3/ɛ3 46.7%. The average follow-up time was 4.5 years. In total, 15 of 71 (21.1%) ɛ2/ɛ4, 127 of 480 (26.5%) ɛ3/ɛ4 & ɛ4/ɛ4, 50 of 238 (21.0%) ɛ2/ɛ2 & ɛ2/ɛ3, and 144 of 691 (20.8%) ɛ3/ɛ3 carriers developed MCI. Overall, 336 (22.7%) participants in the MCI cohort developed AD or MCI.
Following a multivariable Cox proportional hazards model (Table 2), compared to ɛ3/ɛ3 carriers, we observed that ɛ2/ɛ4 carriers exhibited a similar risk for MCI (aHR = 0.88, 95% CI [0.51, 1.50], p = 0.64). Meanwhile, individuals with ɛ2/ɛ2 & ɛ2/ɛ3 exhibited a 12% decreased risk of MCI (aHR = 0.88, 95% CI [0.63, 1.21], p = 0.43), although the decrease was not statistically significant. In contrast, individuals with ɛ3/ɛ4 & ɛ4/ɛ4 revealed a 40% increased risk of MCI (aHR = 1.40, 95% CI [1.10, 1.78], p = 0.006) compared to ɛ3/ɛ3 carriers.
Our sensitivity analysis, using a competing risk subdistribution hazard model, yielded results similar to the Cox proportional hazards model (Supplementary Table 1).
DISCUSSION
Using the national NACC dataset, which comprised approximately 2,000 African Americans without dementia, we found no associations between APOE ɛ2/ɛ4 and the incidence of AD and MCI. Moreover, consistent with the literature, we confirmed that compared to ɛ3/ɛ3 carriers, individuals carrying ɛ4 (ɛ3/ɛ4 and ɛ4/ɛ4) demonstrated statistically significant elevated risk of AD and MCI [12]. Our prior studies have shown that ɛ2/ɛ4 increases the risk of AD and MCI among non-Latino whites [11]. However, in this study, we observed a “protective” trend of ɛ2/ɛ4 on AD (aHR = 0.85; 95% CI [0.39, 1.84]) and MCI (aHR = 0.88; 95% CI [0.51–1.50]) among African Americans, although this trend was not statistically significant. Furthermore, the magnitude of increased risk of ɛ4 (ɛ3/ɛ4 and ɛ4/ɛ4) was much weaker than what we found among non-Latino whites [11]. This suggests the much lower “risk” effect of the ɛ4 allele on AD and MCI among African Americans. African Americans carrying the ɛ2/ɛ4 genotype manifest a trend of decreased incidence of AD and MCI, which is more like ɛ2 than ɛ4, although this trend is not statistically significant.
Prior studies generally have excluded participants carrying the ɛ2/ɛ4 genotype because of 1) the relative rarity of the ɛ2/ɛ4 genotype in the population and 2) the counteractive effect of ɛ2 and ɛ4 allele. Our and others’ prior research has investigated the association between APOE ɛ2/ɛ4 and incident AD and MCI among non-Latino whites [10, 11]. However, few studies have evaluated such relationships among African Americans. For example, one meta-analysis has examined the association of ɛ2/ɛ4 with AD in populations of various ethnic group and reported no association among African Americans (OR = 1.8; 95% CI [0.4, 8.1]; p = 0.27); however, this study included only 10 ɛ2/ɛ4 carriers [4]. Another population-based longitudinal study reported similar findings (OR = 1.34; 95% CI [0.44, 4.13]; p = 0.61), and it included only slightly more ɛ2/ɛ4 carriers (n = 19) [16]. Indeed, given such small sample size of ɛ2/ɛ4 carriers included in these studies, the parameter estimates of the effect of the ɛ2/ɛ4 may be neither stable nor reliable. In contrast, we took advantage of the large national NACC dataset and examined this association with a much larger sample size (99 ɛ2/ɛ4 carriers in the AD cohort; 71 ɛ2/ɛ4 carriers in the MCI cohort). Similar to previous studies, we found no effect of ɛ2/ɛ4 on AD and MCI. However, the direction of association was toward “protection” rather than “risk”, although not statistically significant. These inconsistent findings warrant further research on the effects of APOE ɛ2/ɛ4 on AD and MCI among African Americans.
Although the large NACC sample size confers strength on our study, we must acknowledge several limitations. First, the individuals assessed at the Alzheimer’s Disease Research Center are not representative of the general population, and this inherent sampling bias may limit the generalizability of our findings. Second, we had 99 and 71 ɛ2/ɛ4 carriers in our analytical sample of the AD and MCI cohorts, respectively, which were still small number relative to other genotype groups. Although our sample was far larger than that published in prior studies, the findings should be replicated with even larger samples. Lastly, given the observational nature of this study, the potential for unmeasured confounding variables may have influenced our findings.
Footnotes
ACKNOWLEDGMENTS
This work was partially supported by US National Institutes of Health grants (R03AG068413). The NACC database is funded by NIA/NIH Grant U01 AG016976. NACC data are contributed by the NIA-funded ADCs: P30 AG019610 (PI Eric Reiman, MD), P30 AG013846 (PI Neil Kowall, MD), P30 AG062428–01 (PI James Leverenz, MD) P50 AG008702 (PI Scott Small, MD), P50 AG025688 (PI Allan Levey, MD, PhD), P50 AG047266 (PI Todd Golde, MD, PhD), P30 AG010133 (PI Andrew Saykin, PsyD), P50 AG005146 (PI Marilyn Albert, PhD), P30 AG062421–01 (PI Bradley Hyman, MD, PhD), P30 AG062422–01 (PI Ronald Petersen, MD, PhD), P50 AG005138 (PI Mary Sano, PhD), P30 AG008051 (PI Thomas Wisniewski, MD), P30 AG013854 (PI Robert Vassar, PhD), P30 AG008017 (PI Jeffrey Kaye, MD), P30 AG010161 (PI David Bennett, MD), P50 AG047366 (PI Victor Henderson, MD, MS), P30 AG010129 (PI Charles DeCarli, MD), P50 AG016573 (PI Frank LaFerla, PhD), P30 AG062429–01(PI James Brewer, MD, PhD), P50 AG023501 (PI Bruce Miller, MD), P30 AG035982 (PI Russell Swerdlow, MD), P30 AG028383 (PI Linda Van Eldik, PhD), P30 AG053760 (PI Henry Paulson, MD, PhD), P30 AG010124 (PI John Trojanowski, MD, PhD), P50 AG005133 (PI Oscar Lopez, MD), P50 AG005142 (PI Helena Chui, MD), P30 AG012300 (PI Roger Rosenberg, MD), P30 AG049638 (PI Suzanne Craft, PhD), P50 AG005136 (PI Thomas Grabowski, MD), P30 AG062715–01 (PI Sanjay Asthana, MD, FRCP), P50 AG005681 (PI John Morris, MD), P50 AG047270 (PI Stephen Strittmatter, MD, PhD).This manuscript was edited for lexicogrammar, mechanics, cohesion, and coherence prior to submission by Brian Greene, EdD, Director for International Affairs at the University of Pittsburgh, School of Nursing.
