Abstract
Background: The objective of this study was to examine how common genetic variants associated with cognition at baseline predict antipsychotic treatment response. Methods: The sample includes 418 individuals with schizophrenia of European ancestry from the Clinical Antipsychotic Trial of Intervention Effectiveness (CATIE) study. Genome-wide genotype and clinical data were acquired through the publically available CATIE dataset, and pharmacokinetic data were acquired through an ancillary project. We performed the genome-wide association analysis of 5 cognitive domains (working memory, processing speed, verbal memory, reasoning and problem solving, vigilance/attention) at baseline. Polygenic cognitive score (PCS) of single nucleotide polymorphisms (SNPs) showing association signals with cognition were created for association analysis with treatment response outcomes based on the number of minor alleles and regression coefficients in each of the 5 cognitive domains. Cox proportional hazard regression was used for analysis of the time to discontinuation; and Positive and Negative Syndrome Scale (PANSS) was analyzed using general linear mixed-model that incorporated measures at baseline and end of study. Results: Whole genome analysis identified 15 858 autosomal SNPs showing association signals (P < .01) with at least 1 of the 5 cognitive domains. Strong signals were observed at rs17575566 in SLC4A3, rs17142641 in MACC1||ITGB8, rs531269 in RFX3, rs4740196 in FUBP3, rs6589208 in RP11-89C3.4, and rs10492093 in RP11-1038A11.3 (P < 10 × 10-07); and several others at MACC1, FHIT, CLINT1, 1PCBD1, KALRN, HEPHL1, ADAMTS8, and DIAPH3 (P < 5 × 10-06). PCS explained the variation in all domains (R(2) = 42% to 72%) except reasoning and problem solving (R(2) = 24%) and in cognitive composite score (R(2) = 76%). Sixty-one SNPs also showed nominal association with both time to discontinuation and improvement of symptoms in PANSS. PCS of 61 SNPs significantly explained 7.6% to 12.7% of variation in all 5 cognitive domains and affected the time to discontinuation mainly in individuals treated with olanzapine and quetiapine. Conclusion: The SNPs associated with cognition significantly predict the treatment response in those treated with olanzapine and quetiapine. This may have an implication for prescribing antipsychotics as those individual who have higher PCS are expected to have better efficacy when treated with these 2 antipsychotics.