Shared genetic architecture links schizophrenia with hemispheric white matter microstructure

Schizophrenia is a highly heterogeneous psychiatric disorder affecting ~1% of the global population, typically emerging in adolescence or early adulthood and characterized by hallucinations, delusions, disorganized thinking, and cognitive impairments; while its etiology remains elusive, genetic factors are widely recognized as fundamental, and GWAS have identified over 200 genome-wide significant loci. White matter microstructure is essential for neural communication and is strongly heritable; diffusion tensor imaging (DTI) quantifies white matter integrity via FA, MD, and the three tensor eigenvalues (λ1–λ3), and widespread abnormalities in these metrics have been reported in schizophrenia. Although prior work suggests genetic overlap between schizophrenia and white matter microstructure, approaches such as LDSC and PRS mainly capture genome-wide correlation without pinpointing shared variants and may underestimate overlap due to cancellation across loci with heterogeneous effect directions.

Moreover, most studies emphasize FA/MD while overlooking eigenvalue-based metrics and often rely on bilateral tract averages without systematically testing hemisphere-specific differences. Therefore, we identify and contrast shared genetic variants linking schizophrenia with left- vs right-hemisphere white-matter tracts to test whether the overlap exhibits hemispheric asymmetry.”

Yujie Zhang, author, researcher at Tianjin Medical University

This study leveraged GWAS summary statistics to dissect cross-trait genetic sharing. Schizophrenia data were obtained from the PGC (European-ancestry subset: 53,386 cases and 77,258 controls), and white-matter microstructure data were from the UK Biobank (European ancestry: N=33,224), covering five DTI metrics (FA, MD, and eigenvalues λ1–λ3) across 48 JHU ICBM-DTI-81 atlas–defined tracts. Genetic overlap was interrogated using an empirical-Bayes condFDR/conjFDR framework with bidirectional conditioning (swapping primary/secondary phenotypes) and conjFDR (maximum of the two) to identify shared variants; conditional Q–Q plots were used to visualize enrichment, while the MHC and 8p23.1 regions were excluded due to complex LD; thresholds were set at condFDR<0.01 and conjFDR<0.05. Independent loci were defined via LD-based clumping (independent significant SNPs, lead SNPs, locus boundaries, and merging loci within 250 kb), and novelty was assessed by distance from prior GWAS signals and absence from the GWAS Catalog. Downstream interpretation included functional annotation (CADD, RegulomeDB, ChromHMM) and gene mapping (positional, eQTL using GTEx v8, and chromatin interaction), followed by GO and WebCSEA enrichment plus PsychENCODE-based lifespan spatiotemporal expression trajectories. Additional validation/biological triangulation involved eQTL colocalization (coloc; PP.H4>0.8), TWAS (S-PrediXcan/S-MultiXcan), and drug–gene interaction mining (DGIdb). Hemispheric asymmetry was evaluated in 21 paired left–right tracts by classifying shared loci as left-specific, right-specific, or bilateral. Robustness was assessed using LDSC for genome-wide correlation, LAVA for local correlation replication, an independent East Asian schizophrenia GWAS for consistency of allelic effect directions, and comparison with prior FA-overlap findings.

At the genome-wide level, conditional Q–Q plots indicated clear SNP enrichment between schizophrenia and white-matter microstructure, consistent with polygenic overlap; under condFDR<0.01, conditioning on DTI metrics yielded 391–433 distinct schizophrenia loci, while conditioning on schizophrenia identified 405–496 loci for DTI phenotypes. Using conjFDR<0.05 to pinpoint shared signals across 48 tracts, the aggregated numbers of shared loci were 198 (FA), 188 (MD), 233 (λ1), 200 (λ2), and 182 (λ3). Combining across all five DTI parameters produced 435 distinct shared loci, including 49 shared across all parameters and 154 shared only with eigenvalue-based metrics (λ1–λ3) but not FA/MD, suggesting eigenvalues capture additional shared architecture. The strongest shared region was chr4:102,638,777–103,388,441 (lead SNP rs13107325), jointly associated with schizophrenia and λ1 in the bilateral cerebral peduncles. Among the 435 shared loci, 112 were novel for schizophrenia, 130 novel for white-matter microstructure, and 27 novel for both.

Hemispheric analyses across 21 bilateral tract pairs showed substantial hemisphere specificity: left-specific loci accounted for 25.5%–34.4% and right-specific loci for 23.9%–33.7%; notably, the left posterior thalamic radiation (PTR) and tapetum (TAP) consistently showed higher proportions of shared loci than their right counterparts. Functionally, multiple shared lead SNPs were annotated as potentially pathogenic (CADD>12.37), showed regulatory evidence (RegulomeDB<3), and largely lay in open chromatin regions; gene mapping identified 1,171–1,336 protein-coding genes per DTI metric, with 499 genes shared across all five parameters, and enrichment highlighted neurodevelopmental processes and synapse organization (for FA/MD/λ1). Multi-omic integration further supported transcriptomic convergence: colocalization implicated 577 loci for schizophrenia and 422 loci across white-matter phenotypes, with shared loci colocalizing to the same genes for schizophrenia and 200 white-matter phenotypes; TWAS identified 7,348 significant gene–trait associations (880 genes) with broad sharing across phenotypes, and spatiotemporal trajectories showed pronounced prenatal expression activity stabilizing after ~0.5–2.6 postnatal years.

This study provides compelling evidence of a shared genetic basis between schizophrenia and white matter microstructure, highlighting the critical role of white matter integrity in the pathophysiology of schizophrenia. By utilizing the cond/conjFDR approach, authors identified substantial genetic overlap, revealing potential biological pathways involved in neurodevelopment and central nervous system function. Notably, the inclusion of additional DTI metrics and the examination of hemisphere-specific loci offer a more comprehensive understanding of the genetic influences on white matter, capturing more specific alterations in both the left and right hemispheres. “These findings underscore the complexity of the genetic underpinnings of schizophrenia and suggest that targeting these shared loci may offer new avenues for understanding the structural brain abnormalities and cognitive impairments associated with this disorder.” said Yujie Zhang.

Authors of the paper include Yujie Zhang, Mengge Liu, Shaoying Wang, Wanwan Zhang, Haoyang Dong, Qian Qian, Yue Wu, Qian Wu, Jinglei Xu, Ying Zhai, Haolin Wang, Jingchun Liu, Yuxuan Tian, Qi Luo, Xinxing Li, Lining Guo, Fengtan Li, and Feng Liu.

This research was supported by the Natural Science Foundation of China (82072001, 82572306, and 82572169), the Beijing-Tianjin-Hebei Basic Research Collaboration Project (J230040), the Tianjin Natural Science Foundation (24ZXGQSY00060, 23JCZXJC00120, and 24JCYBJC00970), and the Tianjin Key Medical Discipline Construction Project (TJYXZDXK-3-008C).

The paper, “Hemispheric Asymmetry in the Genetic Overlap between Schizophrenia and White Matter Microstructure” was published in the journal Cyborg and Bionic Systems on Jan. 9, 2026, at DOI: 10.34133/cbsystems.0451.