ACCOUNTING FOR GENETIC CONTEXT: INCORPORATING INTERACTIONS BETWEEN POLYGENIC RISK SCORES FOR ALZHEIMER’S AND ASSOCIATED RISK FACTORS AND THEIR RELATION TO MILD COGNITIVE IMPAIRMENT

Abstract

Alzheimer's disease (AD) is under considerable genetic influence. Despite widespread interest in AD risk genes, little attention has been paid to the interactive effects of different risk genes. Polygenic risk scores (PRSs) represent a useful approach to summarize multiple small genetic effects across the genome. We previously showed that higher AD-PRSs were associated with significantly increased odds of having mild cognitive impairment (MCI) in adults who were only in their 50s (Logue et al., Mol Psychiatry, in press). Risk factors for AD such as cardiovascular disease and obesity are also under genetic influence, yet it is unknown how genetic risk for these conditions interacts with genetic risk for AD. We included 1118 cognitively normal participants, 89 amnestic MCI, and 46 non-amnestic MCI — determined by extensive neuropsychological testing in the Vietnam Era Twin Study of Aging (VETSA). Mean age=56.75 years; range=51-67. An AD-PRS was developed from the International Genomics of Alzheimer's Project database. PRSs for coronary artery disease (CAD) and BMI—known AD risk factors—were calculated for all subjects. We tested main and interaction effects of the AD-PRS and these risk factor PRSs in both MCI groups. There was a significant main effect of the AD-PRS in both MCI groups. There was an interaction between the AD-PRS and CAD-PRS in amnestic MCI subjects (p<0.05), and between the AD-PRS and BMI-PRS in non-amnestic MCI subjects (p=0.055). In both cases, the association between the AD-PRS and MCI was attenuated as scores on the risk factor PRS increased. These results indicate the importance of considering broader genetic context when examining risk for MCI or AD. When genetic risk for conditions that contribute to AD is low, an AD-PRS may be more predictive of MCI status. However, an AD-PRS may be less predictive in individuals at high risk for other conditions. These findings suggest that incorporating PRSs for other AD risk factors and their interactions may better characterize variability in the genetic etiology of AD. Further work is necessary to determine whether such interactions can identify individuals whose impairment is due to AD versus other conditions.