Serum proteomic correlates of mental health symptoms in a representative UK population sample

Poor mental health constitutes a public health crisis due to its high prevalence, unmet need and its mechanistic heterogeneity. A comprehensive understanding of the biological correlates of poor mental health in the population could enhance epidemiological research and eventually help guide treatment strategies. The human bloodstream contains many proteins, several of which have been linked to diagnosed mental health conditions but not to population mental health symptoms, however recent technological advances have made this possible. Here we perform exploratory factor analyses of 184 proteins from two panels (cardiometabolic and neurology-related) measured using proximity extension assays from Understanding Society (the UK Household Longitudinal Study; UKHLS). Data reduction results in 28 factors that explain 55-59% of the variance per panel. We perform multiple linear regressions in up to 5304 participants using two mental health symptom-based outcomes: psychological distress assessed with the general health questionnaire (GHQ-12) and mental health functioning assessed with the 12-Item Short Form Survey, Mental Component Summary (SF12-MCS) using the proteomic factors as explanatory variables and adjusting for demographic covariates. We use backward selection to discard non-significant proteomic factors from the models. Ten factors are independently associated with population mental health symptoms, three of which are immune-related (immunometabolism, immune cell-mediated processes, acute phase processes), three brain-related (neurodevelopment, synaptic processes, neuroprotective processes), two proteolysis-related (proteolysis & the kynurenine pathway, haemostasis & proteolysis), growth factors & muscle, and oxidative stress & the cytoskeleton. Associations partially overlap across the two outcomes, and a sensitivity analysis excluding people taking antidepressants or other central nervous system medications suggestively implicates some of the factors in treatment-resistant poor mental health. Our findings replicate those of case-control studies and expand these to underlie mental health symptomatology in the adult population. More work is needed to understand the direction of causality in these associations.