Proteome-wide Mendelian randomization and colocalization analyses identify potential biomarkers for schizophrenia

We performed proteome-wide Mendelian randomization (MR) and colocalization analyses to explore the causal relationships between proteins and schizophrenia (SCZ).

In the primary analysis, genetic instruments of 4,907 plasma protein from 35,559 Icelanders served as the exposure, summary statistics for SCZ (35,476 cases, 46,839 controls) Working Group of the Psychiatric Genomics Consortium (PGC) served as the outcome. The initial findings underwent sensitivity analyses and were externally validated using cis-pQTLs from the Fenland study (4,979 proteins, 10,708 participants) and UK Biobank Pharma Proteomics Project (UKB-PPP, 2923 proteins, 33,043 participants), and brain cis-eQTLs from Genotype-Tissue Expression (GTEx). Bayesian colocalization assessed shared causal variants. Protein-protein interactions with antipsychotic drug targets were explored.

In the primary analysis, genetically predicted levels of seven plasma proteins were significantly associated with SCZ risk: ADAM22 (OR = 0.85, P = 8.97×10^-7), LIMA1 (OR = 0.43, P = 1.28×10^-5), CTSS (OR = 1.27, P = 9.18×10^-7), FOXO3 (OR = 2.80, P = 6.05×10^-6), IRF3 (OR = 2.09, P = 1.15×10^-6), KLC1 (OR = 2.17, P = 3.38×10^-¹¹), and MMP16 (OR = 2.00, P = 1.60×10^-5). External validation partially confirmed these associations: LIMA1, CTSS, FOXO3, KLC1, and MMP16 replicated in the Fenland study; ADAM22 and CTSS replicated in UKB-PPP; MMP16 and CTSS expression in specific brain tissues replicated using GTEx brain eQTLs. Colocalization strongly supported shared causal variants for FOXO3 (PPH4 = 0.966), IRF3 (PPH4 = 0.932), and LIMA1 (PPH4 = 0.985) with SCZ. CTSS, FOXO3, IRF3, and MMP16 showed interactions with known antipsychotic drug targets.

This large-scale MR study provides robust evidence supporting causal roles for specific plasma proteins in SCZ pathogenesis, highlighting promising candidates for mechanistic studies and therapeutic development.