Assessing the presence of shared genetic architecture between Alzheimer\u2019s disease and major depressive disorder using genome-wide association data

J Gibson,M J Adams,L S Hall,A D Murray,C Hayward,G Davies,I J Deary,B H Smith,T C Russ,T-K Clarke,D M Howard,E M Wigmore,A M Mcintosh,A M Fernandez-Pujals,D J Porteous

doi:10.1038/tp.2017.49

Abstract

Major depressive disorder (MDD) and Alzheimer’s disease (AD) are both common in older age and frequently co-occur. Numerous phenotypic studies based on clinical diagnoses suggest that a history of depression increases risk of subsequent AD, although the basis of this relationship is uncertain. Both illnesses are polygenic, and shared genetic risk factors could explain some of the observed association. We used genotype data to test whether MDD and AD have an overlapping polygenic architecture in two large population-based cohorts, Generation Scotland’s Scottish Family Health Study (GS:SFHS; N=19 889) and UK Biobank (N=25 118), and whether age of depression onset influences any relationship. Using two complementary techniques, we found no evidence that the disorders are influenced by common genetic variants. Using linkage disequilibrium score regression with genome-wide association study (GWAS) summary statistics from the International Genomics of Alzheimer's Project, we report no significant genetic correlation between AD and MDD (rG=−0.103, P=0.59). Polygenic risk scores (PRS) generated using summary data from International Genomics of Alzheimer's Project (IGAP) and the Psychiatric Genomics Consortium were used to assess potential pleiotropy between the disorders. PRS for MDD were nominally associated with participant-recalled AD family history in GS:SFHS, although this association did not survive multiple comparison testing. AD PRS were not associated with depression status or late-onset depression, and a survival analysis showed no association between age of depression onset and genetic risk for AD. This study found no evidence to support a common polygenic structure for AD and MDD, suggesting that the comorbidity of these disorders is not explained by common genetic variants.

Highlights

Major depressive disorder (MDD) and Alzheimer’s disease (AD) are among the leading causes of disability worldwide, affecting an estimated 350 million and 44 million people respectively.[1,2] They are both common conditions in older age and are frequently comorbid;[3] their inter-relationship is complex and not well understood
The high prevalence of MDD in this sample is due to the ensure that any genetic correlation between AD and MDD Polygenic risk scores (PRS) was not driven by a single locus, the analysis was repeated using MDD PRS
This study provides little evidence for overlap in the polygenic architecture of lifetime MDD and AD, based on either cross-trait linkage disequilibrium (LD) score regression or a polygenic profile score approach in two large, independent population cohorts

Summary

INTRODUCTION

Major depressive disorder (MDD) and Alzheimer’s disease (AD) are among the leading causes of disability worldwide, affecting an estimated 350 million and 44 million people respectively.[1,2] They are both common conditions in older age and are frequently comorbid;[3] their inter-relationship is complex and not well understood. This study uses GWAS summary data and two complementary techniques, linkage disequilibrium (LD) score regression and polygenic profile scoring, to assess whether MDD and AD have an overlapping polygenic architecture in two large, independent UK-based population cohorts These techniques are reliant on different metrics for assessing pleiotropy between traits, baseline medical assessments carried out, and biological samples were taken for DNA extraction and genotyping. Further information on sample collection, DNA extraction, genotyping, quality control, and assessment of depression phenotype, for both samples, are provided in the Supplementary Materials Both depend on the analyzed trait being polygenic in nature, with many genetic variants of individually small effect contributing to the overall phenotypic variation. To test the extent to which shared genetic architecture is predictive of phenotypic variation in the samples on an individual subject basis, GWAS summary data can be used to calculate polygenic risk scores (PRS) for the traits under investigation. Power calculations were implemented in AVENGEME, following the method outlined by Palla and Dudbridge.[43,44] The proportions of trait

MATERIALS AND METHODS

RESULTS

DISCUSSION

CONFLICT OF INTEREST