Evaluating metabolome-wide causal effects on risk for psychiatric and neurodegenerative disorders

Abstract

Evidence indicates phenotypic and biological overlap between psychiatric and neurodegenerative disorders. Further identification of underlying mutual and unique biological mechanisms may yield novel multi-disorder and disorder-specific therapeutic targets. The metabolome represents an important domain for target identification as metabolites play critical roles in modulating a diverse range of biological processes. Here, we used Mendelian randomisation (MR) to test the causal effects of [~]1000 plasma metabolites and [~]300 metabolite ratios on anxiety, bipolar disorder, depression, schizophrenia, amyotrophic lateral sclerosis, Alzheimers disease, Parkinsons disease and multiple sclerosis. In total, 85 causal effects involving 77 unique metabolites passed FDR correction and robust sensitivity analyses (IVW-MR OR range: 0.73-1.48; pFDR < 0.05). No evidence of reverse causality was identified. Multivariate analyses implicated sphingolipid metabolism in psychiatric disorder risk and carnitine derivatives in risk for amyotrophic lateral sclerosis and multiple sclerosis. However, polygenic risk scores for prioritised metabolites showed limited prediction in the UK Biobank. Downstream colocalisation in regions containing influential variants identified greater than suggestive evidence (PP.H4 [≥] 0.6) for a shared causal variant for 29 metabolite/psychiatric disorder trait-pairs on chromosome 11 at the FADS gene cluster. Most of these metabolites were lipids containing linoleic or arachidonic acid. Additional colocalisation was identified between the ratio of histidine-to-glutamine, glutamine, Alzheimers disease and SPRYD4 gene expression on chromosome 12. Although no single metabolite had a causal effect on a psychiatric and a neurodegenerative disease, results suggest a broad effect of lipids across brain disorders. Metabolites identified here may help inform future targeted interventions.