W73. POLYGENIC RISK SCORES PROVIDE STRONG EVIDENCE FOR TWO MOLECULAR SUBTYPES OF SCHIZOPHRENIA

We have previously described how the DLPFC transcriptomes of schizophrenia patients can be used to define two patient groups: “Type 1” patients with a relatively normal DLPFC transcriptome and “Type 2” patients with hundreds of genes differentially expressed in the DLPFC at a statistical significance which survives Bonferroni correction. The biological significance of the distinction between Type 1 and Type 2 patients remains an open question. In the present report we examine the polygenic risk scores (PRS) in those two groups of schizophrenic patients and observe Type1 and Type 2 schizophrenia have dramatically different genetic architectures. This study is a re-analysis of gene expression data from the dorsolateral prefrontal cortex of schizophrenic patients and neurotypical controls of postmortem brains in the Human Brain Collection Core of the NIMH. The expression array data is publicly available from the NIMH Data Archive. GWAS summary statistics on which this study is based - Psychiatric Genetics Consortium schizophrenia dataset (Trubetskoy, 2022) - Cannabis Use Disorder GWAS based on multiple very large cohorts (Levey, 2023) - PRS for treatment resistant schizophrenia (TRS) was based on GWAS summary statistics estimated by (Pardinas, 2022). Calculation of PRS statistics PRS statistics were calculated using the clumping and thresholding approach as implemented in the PRSice software package (Choi, 2019). The variance in risk attributable to PRS was estimated based on Nagelkerke's R squared calculated using the R function Nagelkirke() in the package {fsmb} There were three important results in this study. 1> Strong, statically significant differences in PRS for risk of schizophrenia in Type 1 vs Type 2 schizophrenic patients were observed. 2> There was a strong schizophrenia type x self-identified ethnic background interaction which persisted in spite of the fact that the PRS scores were adjusted for the genetically-defined percent of African ancestry in this cohort containing only Caucasians and African-Americans. 3 > PRS for the genetic factors for the risk of treatment resistance in schizophrenic patients account for variance in the risk of schizophrenia in Type 2, but not Type 1 patients. This work shows that Type 1 and Type 2 schizophrenia have dramatically different genetic architectures. That result supports the hypothesis that the DLPFC transcriptome-defined schizophrenia subtypes are biologically meaningful. An important unexpected finding was the dramatic difference in the schizophrenia type-specific genetic architecture of schizophrenia in the Caucasians and African-Americans. We speculate that in this particular cohort African-American ancestry may be a surrogate marker for poorer prenatal obstetric care and/or adverse environmental influences during early post-natal life. Further study of this cohort may represent an opportunity to study genetic vs environmental schizophrenia risk factors. Finally, we have made the serendipitous observation that PRS for the genetic factors for the risk of treatment resistance in schizophrenic patients account for variance in the risk of schizophrenia in Type 2, but not Type 1 patients. The implications of this result for efforts to use gene expression data to identify novel pharmacologic therapies for schizophrenia are obvious.