Characterisation and Genetic Architecture of Major Depressive Disorder Subgroups Defined by Weight and Sleep Changes

Abstract

AbstractMajor depressive disorder is highly heterogeneous and is postulated to contain subgroups with different underlying aetiologies. The aim of this work is to further characterise depression subgroups defined using sleep and weight changes and investigate the effect of adjustment for BMI on results. Probable lifetime MDD cases (N = 26,662) from the UK Biobank were stratified into three subgroups defined by self-reported weight and sleep changes during worst episode: (i) increased weight and sleep (↑WS), (ii) decreased weight and sleep (↓WS) and (iii) the remaining individuals (non-↑↓WS). Analyses compared the depression characteristics, mental health scores, genetic architecture, and neurological and inflammatory comorbidities between subgroups and with 50,147 controls from UK Biobank. Sensitivity analyses compared results to those of subgroups defined by weight change only and analyses unadjusted for BMI. In contrast to ↑WS depression, ↓WS depression had a higher age of onset and lower proportion reporting countless or continuous episodes. The ↓WS depression subgroup also showed a higher wellbeing score compared to non-↑↓WS. Subgroup-specific GWAS identified three genome-wide significant loci associated with ↑WS in genes previously associated with obesity and response to anticonvulsants. In case-case subgroup comparisons, the ↑WS depression subgroup was significantly enriched for comorbid epilepsy, migraine and asthma, and there was significant variability in the odds ratios for dementia between subgroups. These findings provide further evidence for differences in the characteristics and genetic architecture of depression subgroups defined by sleep and weight change and highlight the importance of dividing non-↑WS individuals into ↓WS and non-↑↓ WS subgroups in analyses.