Genetic variations reducing dysbindin-1 are associated with better executive functions in antipsychotic-treated subjects. a Reduction of dysbindin-1 expression in the DLPFC of Dys Hap+/− and −/− human subjects (light and dark blue circles, respectively) compared to Dys Hap+/+ (white circles) (n = 594; one-way ANOVA, F(2, 591) = 5.70, p < 0.005). Analysis with the R package haplo.stats v1.7.7 yielded a global p value of 3.5 × 10−7 for DTNBP1 at the gene level, which was in part driven by the specific Dys haplotype (p = 2.54 × 10−5), associated with lower expression (i.e., Dys Hap+/+ > Dys Hap+/− > Dys Hap−/−). b–d Patients with schizophrenia with reduced dysbindin-1 expression (Dys Hap−/−) made b fewer perseverative (n = 259; one-way ANOVA, F(2, 251) = 4.64, p < 0.01), and c fewer non-perseverative errors (one-way ANOVA, F(2, 251) = 4.21, p < 0.01) on the WCST, and d passed more categories (one-way ANOVA, F(2, 251) = 4.74, p < 0.01). No Dys-haplotype-dependent differences were present in demographical characteristics or in the duration or dosage of antipsychotic treatments (see Supplementary Table 1). All single-nucleotide polymorphisms (SNPs) were in Hardy–Weinberg equilibrium (HWE) for healthy subjects and patients with schizophrenia (see Supplementary Table 2). e Patients with schizophrenia with reduced dysbindin-1 (Dys Hap+/− and −/−, blue circles), who were enrolled in the NIH CATIE trial showed better WCST performance than those with the Dys Hap+/+ (white circles) genotype. The WCST score was calculated by averaging z-scores for preservative errors and categories achieved (n = 359; two-way ANOVA, time effect, F(3, 1046) = 5.03, p < 0.005; genotype effect, F(1, 1046) = 4.38, p < 0.04). Error bars represent S.E.M. *p < 0.05, **p < 0.005