Comparison of pathway-based polygenic scores between schizophrenia carriers and non-carriers of high impact de novo variants

Both common and rare variants contribute to schizophrenia liability, and within individuals with schizophrenia, a lower genome-wide burden of schizophrenia common alleles is observed among carriers of high impact de novo variants, which is expected under a liability threshold model (PMID:31932766). However, studies have not yet tested whether particular sets of genes are enriched or depleted for common alleles in schizophrenia carriers of high impact de novo variants. Here, we used array and exome-sequencing data from 1148 schizophrenia proband-parent trios to compare pathway-based polygenic scores between carriers and non-carriers of high impact de novo variants, focussing on pathways previously implicated in schizophrenia. Array and exome-sequencing data were analysed from two published schizophrenia trio cohorts: 665 Bulgarian trios (PMID:24463507) and 483 trios collected from multiple European sites (PMID: 31932766). Pathway-based polygenic scores were generated using PGC3 GWAS summary statistics and the PRSice-2 PRSet function for 21 gene sets previously associated with schizophrenia in the SCHEMA study (PMID: 35396579). We then generated pathway-based polygenic Transmission Disequilibrium Test (pTDT) scores for each trio, and compared these scores between 61 probands carrying a high impact de novo variant, defined as loss-of-function mutations or deletions affecting a loss-of-function intolerant gene, and 1087 proband non-carriers using a logistic regression model that controlled for five principal components and sex. Each trio cohort was analysed separately and then meta-analysed using inverse variance weighting with fixed effects. Genome-wide schizophrenia pTDT scores were significantly lower in carriers of high impact de novo variants compared with non-carriers (beta = -0.36, p-value = 0.019). Carriers of high impact de novo variants had lower pathway-based pTDT scores for common alleles that map to 690 genes involved in the post-synaptic density (beta = -0.39, p-value = 0.0047); however, this association does not survive correction for 21 gene sets tested. Genome-wide schizophrenia pTDT scores remained lower in carriers of high impact de novo variants after excluding alleles from the post-synaptic density set (beta = -0.30, p-value = 0.045). Additional pathways with nominally significant lower pTDT scores in carriers of high impact de novo variants include cerebellum-expressed genes (beta = -0.34, p-value = 0.02), cortex-expressed genes (beta = -0.32, p-value = 0.027), and CDH8 targets in neuronal stem cells (beta = -0.29, p-value = 0.032). Using the largest schizophrenia GWAS summary statistics currently available, our study reproduces previous findings showing schizophrenia carriers of high-impact de novo variants have a lower genome-wide burden of schizophrenia common alleles compared with non-carriers. We provide suggestive evidence that this effect might be stronger for common alleles mapping to genes involved in the post-synaptic density; however, further work is required to replicate this finding in additional samples, and also to test whether our pathway-based findings simply reflect the strength to which these sets are enriched for common alleles in schizophrenia.