Correlation networks of blood proteins in the neuroimmunology of schizophrenia—replication and extension

The Pearson sample correlation between two biomarkers across a group of individuals can sometimes be much stronger than expected by chance. In the context of psychosis risk, we previously analyzed blood plasma protein data from initial presentations as collected in the North American Prodrome Longitudinal Study 2 (NAPLS2). We found enhanced correlation between proteins SERPINE1 and TIMP1, both promoters of coagulation and inhibitors of remodeling of extracellular matrix (ECM). Participants were unaffected community controls vs. others of clinical high risk. The SERPINE1-TIMP1 correlation was consistently higher in individuals at clinical high risk for psychosis who later converted to a psychotic disorder vs. participants who were nonconverters or unaffected community controls. Here, we extend those findings using data from a larger cohort, the North American Prodrome Longitudinal Study 3 (NAPLS3). Again, the correlation between SERPINE1 and TIMP1 remained higher in psychosis high-risk converters vs. the other groups. In NAPLS3 we added an assay for PLAT (anti-coagulation plasminogen activator strongly inhibited by SERPINE1). Comparing the three NAPLS3 groups we found a decreased correlation between SERPINE1 and PLAT in converters. In summary, the increased correlation of SERPINE1 and TIMP1 in converters is consistent with restricted brain circuit remodeling and increased tendency to coagulation. Rigorous application of permutation testing yielded NAPLS2 vs. NAPLS3 consistency of SERPINE1-TIMP1 correlation patterns with empirical p-value 0.03.