Associations of 2,922 genetically predicted plasma protein levels with mental illness and response to treatment

Abstract:

Abstract There is growing evidence that mental disorders are associated with altered levels of plasma proteins, potentially indicating molecular targets for drug development or repurposing, but determining which associations are due to confounding and which are causal for illness onset or course is challenging. We used two-sample Mendelian randomisation, which can examine associations free from the effects of confounding between genetically predicted levels of the blood proteome and multiple psychiatric traits. We used the UK Biobank’s largest genome-wide association study (GWAS) of 2,922 plasma proteins, selecting lead cis-quantitative trait loci variants as genetic instruments. We report 337 false discovery rate (FDR) corrected associations of plasma proteins involved in immunity, cytoskeletal integrity, and vesicular trafficking with psychiatric traits: 188 with schizophrenia, 86 with bipolar disorder, 32 with opioid use disorder, 30 with major depressive disorder, and one with treatment resistance in depression. Immune-metabolic protein BTN2A1 associated with depression, schizophrenia, and bipolar disorder. We did not find strong evidence for obsessive compulsive disorder, panic disorder, treatment resistance in schizophrenia, or the depression trait antidepressant response rate of improvement. Colocalisation analyses were applied to the above results and provided evidence for 33 associations that are less likely to be due to horizontal pleiotropy. Our results can help to inform efforts toward identifying potential aetiological pathways of mental disorders and assist in drug development or repurposing.