Skip to main content
NCBI home page
Schizophrenia logoLink to Schizophrenia
. 2026 Jul 7;12(1):82. doi: 10.1038/s41537-026-00782-7

Shared genetic architecture between schizophrenia and subcortical brain susceptibility phenotypes

Yingying Xie 1,2,#, Jiaojiao Du 3,#, Yao Zhao 4, Jie Sun 5, Shiqi Lin 1,2, Ningnannan Zhang 5, Zhang Zhang 5, Feng Liu 5,✉, Dairong Cao 1,2,6,7,✉
PMCID: PMC13720105  PMID: 42414318

Abstract

Schizophrenia (SCZ) is a highly heritable and complex neuropsychiatric disorder. Emerging evidence implicates dysregulated brain iron, particularly in subcortical regions, in SCZ pathophysiology. Here, we systematically dissected the shared genetic architecture between SCZ and subcortical brain susceptibility phenotypes by integrating large-scale genome-wide association study (GWAS) data with quantitative susceptibility mapping (QSM) phenotypes across 16 subcortical regions. Using the MiXeR framework, we quantified the extent and pattern of shared genetic overlap between SCZ and each QSM phenotype, revealing the strongest overlap in the left and right nucleus accumbens. Conditional and conjunction false discovery rate analyses identified 666 unique shared SNPs. Functional annotation highlighted enrichment in pathways related to synaptic function and neuronal development, with pronounced expression in neurons. Additionally, we characterized their spatio-temporal expression patterns of shared genes and identified 89 significant expression-trait associations linked to SCZ-related phenotypes. Furthermore, the virtual drug screen identified 691 potential protein-drug pairs that may contribute to both iron regulation and SCZ. Our findings provide new insights into the complex interaction between subcortical brain susceptibility phenotypes and SCZ, highlighting pathways that may offer novel therapeutic strategies for SCZ and related brain susceptibility-associated conditions.

Subject terms: Schizophrenia, Neuroscience, Genetics of the nervous system

Introduction

Schizophrenia (SCZ) is a complex and debilitating psychiatric disorder characterized by profound cognitive, emotional, and behavioral disturbances, affecting approximately 1% of the global population. Brain iron is central to dopaminergic neurotransmission, which plays a key role in SCZ pathology1,2, and is essential for various neural processes, including myelination, neurotransmitter synthesis, and healthy brain function and development3. Dysregulation of iron homeostasis has been linked to oxidative stress and neuroinflammation, both of which are hypothesized to contribute to SCZ1,4,5. Quantitative Susceptibility Mapping (QSM), an advanced MRI technique, is highly sensitive to iron, particularly in subcortical nuclei, but its signal also reflects myelin and other microstructural properties, making it a composite in vivo measure of tissue susceptibility. Although both SCZ and QSM phenotypes exhibit substantial heritability, the genetic and biological mechanisms underlying SCZ and subcortical susceptibility variation remain largely unknown.

Previous studies using magnetic susceptibility-sensitive MRI techniques have demonstrated that alterations in subcortical susceptibility, primarily driven by iron-related microstructural properties, may be associated with the pathophysiology of SCZ and psychotic disorders1,6–8. For example, increased QSM values have been reported in the bilateral putamen in patients with SCZ1, whereas patients with first-episode SCZ exhibit significantly reduced QSM values in the bilateral substantia nigra, left red nucleus, and left thalamus6. Importantly, altered susceptibility in subcortical regions has been implicated in psychotic spectrum disorders, with reduced or dysregulated iron-related homeostasis associated with increased risk and more severe clinical manifestations9. In parallel, subcortical dopamine dysfunction is a core pathophysiological feature of SCZ10. Given the critical role of iron in dopamine synthesis, metabolism, and neuronal signaling, as well as the sensitivity of QSM to iron-related microstructural properties, investigating QSM-derived susceptibility in subcortical regions may provide a more comprehensive window into the neurobiological substrates underlying SCZ.

Recent advances in molecular genetics have opened new avenues for understanding the structural abnormalities associated with SCZ, particularly those involving subcortical brain regions11. Efforts have been made to explore the genetic contributions to the iron-induced oxidative stress disturbance and cognitive impairment in SCZ in the single nucleotide polymorphism (SNP) polymorphism level12. However, a comprehensive genome-wide analysis to investigate the shared genetic architecture between SCZ and subcortical brain susceptibility phenotypes is critically lacking. Addressing this gap is important for improving our understanding of susceptibility-related imaging alterations in SCZ, which may provide new insights into potential biological mechanisms and therapeutic targets for this debilitating condition.

In this study, we aimed to explore the genetic architecture between SCZ and subcortical brain susceptibility using QSM phenotypes. Utilizing large-scale genome-wide association study (GWAS) summary statistics of SCZ and QSM-derived susceptibility measures from 16 subcortical regions in individuals of European (EUR) ancestry, we identified polygenic overlap, shared SNPs and local genetic correlations between SCZ and regional susceptibility phenotypes. To further explore the biological underpinnings, we conducted four gene-level analyses on the identified shared genes. These analyses included (1) elucidating critical biological processes, (2) enriching cell types associated with these genes, (3) examining expression-trait associations, and (4) characterizing spatio-temporal expression patterns within each subcortical brain region. Additionally, we performed protein–protein interaction (PPI) and virtual drug screening analyses to characterize the interaction networks among the identified shared genes and to identify potential therapeutic avenues. This comprehensive approach offers new insights into the role of subcortical brain susceptibility in SCZ. Understanding these shared genetic mechanisms may reveal novel therapeutic targets for correcting susceptibility imbalances and mitigating their potential contribution to the onset and progression of SCZ. The flowchart of the study design is shown in Fig. 1.

Fig. 1. Study design flowchart.

Fig. 1

Open in a new tab

This study incorporates GWAS summary statistics for SCZ and 16 QSM phenotypes to assess shared genetic architectures between SCZ and subcortical brain susceptibility phenotypes. Univariate analyses and pleiotropic genetic overlap was determined using MiXeR analysis, and shared SNPs were identified through conditional/conjunction FDR analysis. Local genetic correlations and colocalization analyses were then evaluated based on these shared SNPs. Functional annotations were performed on the shared SNPs. Furthermore, four analyses were conducted on corresponding shared genes: (1) exploration of critical biological pathways, (2) identification of key cell types, (3) investigation of gene expression-trait associations, and (4) clarification of spatial-temporal expression patterns. Lastly, a PPI network was constructed for the key shared genes, and a virtual drug screen was performed to identify potential drug candidates. CADD Combined Annotation Dependent depletion, PPI protein–protein interaction, QSM quantitative susceptibility mapping, RDB RegulomeDB, SCZ schizophrenia, SNP single nucleotide polymorphisms. Created in part with Servier Medical Art (https://smart.servier.com/), licensed under a Creative Commons Attribution 4.0 International License. The top-left graphic was further modified with segmentation overlays.

Methods

GWAS datasets

The SCZ GWAS summary dataset was obtained from the Psychiatric Genomics Consortium (PGC)13 including 53,386 cases and 77,258 controls in EUR ancestry. GWAS summary datasets for 16 QSM phenotypes were obtained from UK Biobank discovery (n = 19,720) and replication (n = 9859) cohorts14, including bilateral accumbens, amygdala, caudate, hippocampus, pallidum, putamen, substantia nigra and thalamus. QSM phenotypes were calculated as median χ values, representing magnetic susceptibility, which quantifies the degree to which a material becomes magnetized in response to an external magnetic field. Meta-analysis was performed to obtain the finally used GWAS datasets for QSM phenotypes by using METAL (https://github.com/statgen/METAL)15. Detailed QSM processing steps are provided in the previous study14. All methods were conducted in accordance with relevant guidelines and regulations. Ethical approval for the original GWAS was obtained from the respective institutional review boards or ethics committees, and informed consent was obtained from participants in the original studies.

MiXeR analysis

We applied the MiXeR framework (https://github.com/precimed/mixer)16,17 to quantify the polygenic architecture and shared genetic basis between SCZ and each QSM phenotype. MiXeR is a causal mixture model that leverages GWAS summary statistics to estimate the number of trait-influencing variants and the proportion of phenotypic variance explained by common SNPs. First, univariate MiXeR models were fitted separately for each QSM-derived phenotype to characterize their genetic architecture. This analysis provides three key parameters: polygenicity, discoverability, and SNP-based heritability. Building upon the univariate framework, bivariate MiXeR analyses were conducted to estimate the extent of shared and trait-specific causal variants between SCZ and each QSM phenotype. The bivariate model partitions variants into four categories: those influencing neither trait, those specific to SCZ, those specific to the QSM phenotype, and those jointly influencing both traits. To further quantify polygenic overlap, the Dice coefficient was calculated, which measures the proportion of shared causal variants relative to the total number of trait-associated variants across both phenotypes.

Identification of pleiotropic SNPs

Conditional/conjunction FDR (cond/conjFDR) analyses were used to identify the shared SNPs between SCZ and each of the QSM phenotypes18–22. To enhance the discovery of genetic variants for each trait, the primary phenotype’s test statistics were re-ranked taking into account the strength of association with a secondary phenotype. Conditional Q-Q plots were also generated to visually represent enrichment patterns between traits based on the different significance in the secondary trait. We excluded both the major histocompatibility complex (MHC) region and 8p23.1 regions in consideration of the complex LD pattern in the regions. In accordance with previous researches23–27, significance thresholds were set at 0.01 for condFDR and 0.05 for conjFDR. Independent SNPs were identified using linkage disequilibrium (LD) clumping with a 250 kb window and an r² threshold of 0.1, referencing the EUR-LD data from the 1000 Genomes Project28.

Local genetic correlations

Local genetic correlations between SCZ and QSM phenotypes were estimated using the Local Analysis of [co]Variant Association (LAVA, https://github.com/josefin-werme/LAVA)29 framework. In contrast to global approaches, LAVA enables the quantification of local genetic correlation within specific genomic regions, thereby providing higher-resolution insights into shared genetic architecture. Genomic loci were defined as 250 kb regions centered on each identified independent shared SNP. Subsequently, we conducted bivariate local genetic correlation analyses between SCZ and QSM phenotypes within the selected loci. To ensure sufficient genetic signal for reliable estimation, we first performed univariate LAVA analyses and retained loci showing evidence of genetic signal in both traits (univariate P < 0.05). Bivariate local genetic correlation analyses were then conducted between SCZ and QSM phenotypes within the selected loci. Multiple testing correction was performed using the false discovery rate (FDR) approach, and loci with q < 0.05 were considered to show significant local genetic correlations.

To account for the potential influence of genetically correlated traits, we further performed conditional analyses using a partial correlation framework implemented in LAVA11. Specifically, for each pairwise SCZ-QSM analysis, we separately conditioned on type 2 diabetes30 and multiple sclerosis31, which may act as potential confounders due to shared genetic architecture with the included phenotypes, allowing the estimation of local genetic correlations while accounting for their potential confounding genetic effects independently.

Functional annotations for shared SNPs

FUMA (https://fuma.ctglab.nl)32 was employed to annotate the independent shared SNPs. The Combined Annotation Dependent Depletion (CADD) score integrates multiple annotations to predict the deleteriousness of variants across the human genome33. A higher CADD score suggests that a variant is more likely to have a damaging effect on gene function or be associated with disease, and SNPs with a CADD score greater than 12.37 have been considered potentially deleterious. RegulomeDB (RDB) annotates variants based on their regulatory potential34, and a lower score indicates strong evidence of regulatory function. The chromatin state indicates whether a region is active in gene regulation, with scores from 1 to 7 considered indicative of open chromatin state32,35.

Colocalization analyses

We performed colocalization analyses by integrating brain cis-eQTL data from the GTEx Consortium (GTEx v7) across multiple relevant brain tissues36. For loci defined around shared SNPs identified between SCZ and QSM traits, colocalization analyses were conducted using coloc (https://chr1swallace.github.io/coloc/index.html)37 to estimate the posterior probability for hypothesis 4 (PP.H4), representing a shared causal variant between association signals. For each QSM phenotype, analyses were restricted to anatomically matched subcortical brain regions available in GTEx. Within each matched region, separate colocalization analyses were performed between SCZ and regional eQTL signals and between the corresponding QSM phenotype and the same regional eQTL signals, aiming to identify shared loci potentially driven by common gene regulatory mechanisms. A genomic locus was defined as the region extending ±250 kb around each index shared SNP, and colocalization analyses were performed accordingly. Default prior settings were applied in coloc, and a PP.H4 greater than 0.8 was considered as strong evidence for a shared causal variant between GWAS and eQTL signals.

Gene-level analysis based on shared genes

Enrichment analyses

Using FUMA, the shared genes were derived from the unique nearest genes based on the shared SNPs between SCZ and all QSM phenotypes. We used ToppGene (http://toppgene.cchmc.org)38 to conduct gene functional annotations based on the shared genes (q < 0.05, FDR Benjamini-Hochberg correction [FDR-BH]). Besides, the Web-based Cell-type-Specific Enrichment Analysis (WebCSEA, https://bioinfo.uth.edu/webcsea/)39 was utilized to recognize cell-type-specific gene expression patterns from transcriptomic data.

Spatio-temporal gene expression trajectory analysis

A spatio-temporal analysis of brain gene expression was performed to explore how genes shared between SCZ and QSM phenotypes are expressed at different lifespan stages. The mRNA-seq data from the PsychENCODE project were utilized40,41, covering a developmental range from 8 post-conception weeks (PCW) to 40 years of age. A total of nine developmental windows were constructed, including 8–9 PCW, 12–13 PCW, 16–17 PCW, 19–22 PCW, 35 PCW to 4 months, 0.5–2.6 postnatal years (PY), 2.8–10.7 PY, 13–19 PY, and 21–40 PY40. The expression levels were measured in RPKM (reads per kilobase of transcript per million mapped reads), and the values were log2-transformed and normalized. For the shared genes associated with the corresponding subcortical region QSM phenotype, the averaged gene expressions in this region were calculated. Finally, LOESS curves were generated to illustrate their temporal expression trajectories.

Expression-trait association

The gene-to-phenome association resource that integrates GWAS results with gene expression and regulation data was obtained from PhenomeXcan (https://github.com/hakyimlab/phenomexcan)42. The association statistics (P values) between predicted gene-expression variation and traits were obtained. In this study, we included a total of 23 SCZ-related traits, organized into coherent categories—cognitive ability, alcohol intake, adiposity, dietary intake (fish, sweets, and iron), and smoking exposure (Supplementary Table 1)—all of which have been previously implicated in SCZ or related phenotypes.

PPI network analysis

A PPI network was constructed to further explore the functional relationships among the shared genes. Specifically, we utilized the STRING database (https://string-db.org/)43 to evaluate potential interactions among the identified shared genes. To ensure high-confidence results, we applied a stringent interaction threshold, retaining only interactions with a confidence score ≥0.900.

Molecular docking

Molecular docking has been widely adopted in the process of drug discovery44. In this study, DrugRep (http://cao.labshare.cn:10180/DrugRep/php/index.php)45 was utilized for the virtual screening of drugs. The possible binding pockets were automatically identified by the curvature-based cavity detection approach (CurPocket)46. The biggest binding pocket was submitted to a docking-based virtual screening using AutoDock Vina47,48. Specifically, we selected top proteins (genes) to explore candidate drugs, which were shared between SCZ and more than five QSM phenotypes. The top 100 results were included and a vina score less than −7 was considered as promising docking.

Results

The polygenic overlap between SCZ and subcortical QSM phenotypes

Univariate MiXeR analysis characterized the heritability, polygenicity, and discoverability for QSM phenotypes. The QSM phenotypes exhibited SNP heritability ranging from 0.024 to 0.272, polygenicity ranging from 1.74 × 10−5 to 1.12 × 10−3, and discoverability ranging from 5.11 × 10−5 to 5.22 × 10−3 (Fig. 2a). For Bivariate MiXeR analysis, SCZ exhibited the highest number of shared variants with QSM phenotypes in the left (1.4 K) and right (2.5 K) accumbens, with Dice coefficients of 0.251 and 0.373, respectively (Fig. 2b, c). The conditional Q-Q plot demonstrated a consistent leftward deflection for subsets of SNPs with higher significance, both for SCZ conditioned on accumbens and for accumbens conditioned on SCZ (Fig. 2d, e). Additionally, the left hippocampus and right thalamus showed 0.4 K (Dice = 0.080) and 0.3 K (Dice = 0.050) shared variants with SCZ. The detailed results for SCZ and other QSM phenotypes are shown in Supplementary Figs. 1–8.

Fig. 2. Pleiotropic genetic overlap between SCZ and bilateral accumbens QSM phenotypes.

Fig. 2

Open in a new tab

a Univariate MiXeR estimates of polygenicity, discoverability, and SNP-based heritability for SCZ and subcortical QSM phenotypes, stratified by left/right hemisphere. Colored symbols represent distinct subcortical brain regions, and symbol shape indicates laterality, with diamonds depicting left hemisphere measures and circles depicting right hemisphere measures. b, c Venn diagrams display the numbers of shared (gray) causal SNPs, counted in thousands, for SCZ (blue) and accumbens QSM phenotypes (orange). The Dice coefficient indicates a substantial genetic overlap between the two traits. d, e Conditional Q-Q plots show expected versus observed −log10(P) values for SCZ, based on the significance of association with accumbens QSM phenotypes, and vice versa. L Acc left accumbens, R Acc right accumbens, SCZ schizophrenia.

Additionally, we extended the MiXeR analyses to include multiple sclerosis and diabetes, in addition to SCZ, across all 16 QSM phenotypes. In total, 32 pairwise MiXeR analyses were performed, enabling a systematic comparison of the polygenic overlap and pleiotropic effects between SCZ and these other complex disorders. The results showed that the overall pattern of genetic overlap observed for SCZ exhibited both shared and disorder-specific features compared with multiple sclerosis and diabetes. Several common pleiotropic signals were consistently identified across disorders, alongside distinct regional differences in the magnitude and spatial distribution of genetic overlap (Supplementary Fig. 9). For instance, the bilateral accumbens continued to exhibit relatively higher polygenic overlap across all three disorders, suggesting partially shared susceptibility mechanisms. Nevertheless, clear disease-specific patterns were also observed. Compared with SCZ, diabetes exhibited relatively strong genetic pleiotropy with the left hippocampus, whereas multiple sclerosis demonstrated broader pleiotropic associations across multiple brain regions, particularly involving the bilateral amygdala and caudate.

Pleiotropic SNPs shared by SCZ and subcortical QSM phenotypes

The cond/conjFDR analyses were used to identify pleiotropic SNPs between SCZ and each of the QSM phenotypes. We identified 62 independent shared SNPs for the left accumbens, 124 SNPs for left caudate, 2 SNPs for left hippocampus, 114 SNPs for left pallidum, 121 SNPs for left putamen, 117 SNPs for left substantia nigra, 6 SNPs for left thalamus, 3 SNPs for right accumbens, 118 SNPs for right caudate, 3 SNPs for right hippocampus, 38 SNPs for right pallidum, 116 SNPs for right putamen, 96 SNPs for right substantia nigra, and 3 SNPs for right thalamus. For example, rs6855246 exhibited a pleiotropic effect between SCZ and multiple subcortical QSM phenotypes, including the left substantia nigra (conjFDR = 6.98 × 10−8), right pallidum (conjFDR = 1.18 × 10−7), left pallidum (conjFDR= 1.20 × 10−7), right substantia nigra (conjFDR = 1.20 × 10−7), left caudate (conjFDR = 1.51 × 10−7), right caudate (conjFDR = 1.70 × 10−7), left hippocampus (conjFDR = 4.50 × 10−7), and right hippocampus (conjFDR = 4.31 × 10−3). Among all the subcortical QSM phenotypes analyzed, the left caudate and left putamen revealed the highest number of pleiotropic SNPs, with 124 and 121 shared SNPs, respectively. Together, we identified a total of 666 unique shared independent SNPs between SCZ and 14 subcortical QSM phenotypes (Fig. 3 and Supplementary Table 2).

Fig. 3. The pleiotropic SNPs shared by SCZ and QSM phenotypes.

Fig. 3

Open in a new tab

a Manhattan plots display the conjunctional FDR analysis results for SCZ and each left (left panel) and right (right panel) subcortical QSM phenotype. Each point represents a SNP according to its genomic position and −log10(conjFDR) value. Different colors represent various brain regions. b Genomic locations of SNPs shared by SCZ and each subcortical QSM phenotype are shown, with different colors indicating distinct QSM phenotypes. QSM quantitative susceptibility mapping, SCZ schizophrenia, SNP single nucleotide polymorphism.

Local genetic correlations based on shared SNPs

A total of 188 loci defined by shared SNPs between SCZ and QSM phenotypes were identified for significant local genetic correlation analyses (q < 0.05, Supplementary Table 3). To evaluate the robustness of these associations, we compared local genetic correlations before and after conditioning on diabetes and multiple sclerosis within these loci. No significant differences were observed between the unadjusted and conditioned estimates (all P > 0.05; two-tailed Wilcoxon signed-rank test for paired samples), including conditioned diabetes (P = 0.50) and multiple sclerosis (P = 0.11), suggesting that these traits do not materially influence the local genetic correlations between SCZ and QSM phenotypes.

Functional annotations for identified shared SNPs and genes

FUMA (https://fuma.ctglab.nl)32 was utilized to conduct SNP-level annotations for the identified shared SNPs, including CADD scores, RDB scores, and chromatin states. For the identified 666 unique SNPs shared by SCZ and subcortical QSM phenotypes, the majority of candidate SNPs (49.4%) were located within intronic regions, while 14 SNPs were located within exonic regions, and the rest were distributed across other regions. Furthermore, 39 SNPs had CADD scores above 12.37, with rs11557154 having the highest score at 25.3. Additionally, 21 SNPs with RDB scores of 1b to 1 f, and 594 SNPs exhibited a chromatin state value < 8 (Supplementary Table 4 and Supplementary Fig. 10).

Additionally, shared genes were derived based on the nearest genes of shared SNPs. ToppGene (http://toppgene.cchmc.org)38 was used to explore the significant biological processes (q < 0.05, FDR-BH). For the unique pleiotropic genes shared by SCZ and all subcortical QSM phenotypes, 135 significant biological processes were identified (q < 0.05, Supplementary Table 5 and Fig. 4a). These biological pathways are mainly involved in the development and function of the nervous system, cell structure and morphogenesis, gene transcription and regulation. Notable examples include the synapse organization (GO:0050808, FDR-BH q = 1.30 × 10−6), regulation of trans-synaptic signaling (GO:0099177, FDR-BH q = 8.57 × 10−4), and neuron projection development (GO:0031175, FDR-BH q = 1.22 × 10−5). Furthermore, the cell-type-specificity enriched in unique genes shared by SCZ and all subcortical QSM phenotypes was evaluated using the bioinformatics tool WebCSEA (https://bioinfo.uth.edu/webcsea/submit.php)39. The findings indicated that the neuron was the most enriched cell type (P = 1.74 × 10−5), followed by the excitatory neuron (P = 1.02 × 10−4) and the Purkinje cell (P = 2.99 × 10−4) (Fig. 4b).

Fig. 4. Enrichment and association analyses based on shared genes.

Fig. 4

Open in a new tab

a The top 10 significantly enriched biological processes based on unique shared genes between SCZ and QSM phenotypes, with the x-axis and color indicating the −log10(q) for each biological process. b The top 20 enriched cell types identified based on shared genes, with the red dashed line marking the Bonferroni-corrected significance threshold across 1355 tissue-cell types, and the gray solid line marking the nominal significance level (P = 1 × 10−3). c The significant expression-trait associations from PhenomeXcan based on shared genes, with the y-axis representing −log10(P). d The spatio-temporal gene expression trajectories of shared genes in corresponding subcortical brain regions. The x-axis displays nine distinct developmental stages, while the y-axis represents the mean log2-transformed expression value. CAEM cellular anatomical entity morphogenesis, PCW post-conceptional weeks, PMBCPM plasma membrane bounded cell projection morphogenesis, PY postnatal years.

Colocalization analyses

We identified a total of 81 locus-tissue pairs showing strong evidence of colocalization (PP.H4 > 0.8). The largest number of colocalized signals was observed for putamen-related QSM phenotypes (left: 23; right: 23), followed by caudate nucleus (left: 9; right: 9) and nucleus accumbens (left: 9) phenotypes. In comparison, fewer colocalized signals were detected for substantia nigra-related phenotypes (left: 4; right: 4) (Supplementary Table 6).

Expression-trait association analysis

The gene-to-phenome association resource that integrates GWAS results with gene expression and regulation data was obtained from PhenomeXcan (https://github.com/hakyimlab/phenomexcan)42. The analysis incorporated 23 SCZ-related phenome traits (Supplementary Table 1), and a total of 1078 significant results were obtained with nominal P < 0.05, and 89 significant results were obtained with Bonferroni-corrected threshold of P < 4.40 × 10−6 (corrected for 494 unique shared genes and 23 traits, Supplementary Table 7 and Fig. 4c). The body mass index (BMI) and fluid intelligence score (FIQ) showed the most associations with the expression of shared genes (nominal P < 0.05). The most significant association was between the expression of RBM6 and BMI (P = 6.61 × 10−32). In particular, 19 genes showed significant associations with iron consumption based on food and beverage (nominal P < 0.05).

Spatio-temporal gene expression trajectory analysis

The spatio-temporal patterns of gene expression in the corresponding subcortical brain region for shared genes were analyzed using mRNA-seq data obtained from PsychENCODE40,41. The shared genes between SCZ and bilateral thalamus QSM phenotypes showed a higher level of averaged expression in the thalamus throughout the lifespan. In contrast, the averaged expression of shared genes between SCZ and bilateral accumbens QSM phenotypes remained consistently the lowest in the accumbens. Furthermore, the expression trajectory of SCZ-thalamus shared genes in the thalamus showed an initial increase, followed by a decrease, with peak expression occurring during the 19–22 PCW stage. Similarly, the average expression of SCZ-hippocampus shared genes in the hippocampus and SCZ-accumbens shared genes in the accumbens both demonstrated an initial increase followed by a decrease, with peak expression occurring during the 0.5–2.6 PY stage. In contrast, the average expression of the shared genes between SCZ and bilateral pallidum, putamen, and caudate exhibited a relatively stable trend over time in the corresponding brain regions (Fig. 4d).

PPI network analysis

The resulting PPI network based on shared genes exhibited a non-random interaction pattern, suggesting that the shared genes are functionally interconnected rather than acting independently (Supplementary Fig. 11). In addition, based on network topology measures, we identified several hub genes, including EP300, MAPK3, and PPP3CA, which may play central roles in mediating the biological overlap between SCZ and QSM-related traits.

Virtual drug screen

We performed molecular docking experiments of the drugs with the target proteins corresponding to the significant candidate proteins derived from shared genes using DrugRep (http://cao.labshare.cn:10180/DrugRep/php/index.php)45. A total of 12 available proteins were included which were shared between SCZ and more than five QSM phenotypes, including MAPT, BANK1, CRHR1, SLC39A8, FAM120A, CANX, PPP3CA, ANKRD11, PLEKHO1, PROX2, NRIP1, and MAP2K5. The findings revealed a total of 691 interactions, with 11 unique proteins showing binding potential with 355 unique drugs (Supplementary Table 8). Notably, the proteins corresponding to BANK1 and FAM120A showed the highest number of unique drugs (both n = 100), followed by PLEKHO1 (n = 99), PPP3CA (n = 98), SLC39A8 (n = 98), and CANX (n = 93). Among these drugs, Paliperidone demonstrated binding to multiple receptors, including FAM120A (vina score = −9.8), BANK1 (vina score = −9.3), PPP3CA (vina score = −8.5), SLC39A8 (vina score = −8.2), CANX (vina score = −7.6), MAP2K5 (vina score = −7.4), and CRHR1 (vina score = −7.1). Furthermore, Pimozide showed a particularly high vina score of −10.2 with FAM120A, while Ziprasidone exhibited a vina score of −9.9 with BANK1. Additionally, Deferasirox interacted with both PPP3CA and PLEKHO1, with vina scores of −8.3 and −8.2, respectively (Fig. 5).

Fig. 5. Virtual drug screen based on critical shared genes.

Fig. 5

Open in a new tab

Examples of virtual drug screens for FAM120A with Pimozide and Paliperidone, BANK1 with Ziprasidone, and PPP3CA with Deferasirox. The corresponding vina scores are shown.

Discussion

Our study investigated the shared genetic architecture between SCZ and subcortical brain susceptibility phenotypes, leveraging QSM phenotypes across 16 subcortical regions. By integrating GWAS summary statistics of SCZ and brain QSM measures, we identified polygenic overlap between SCZ and subcortical QSM phenotypes. Furthermore, based on the shared genes, the critical biological processes and cell types were revealed, and the correlations between gene expression and related traits were identified. Finally, the virtual drug screen found 691 protein-drug pairs which showed docking potentials. These findings provide additional insights into the complex etiology of SCZ and its association with brain susceptibility-related processes.

SCZ and QSM phenotypes exhibited polygenic overlap, highlighting shared genetic factors that influence both the disorder and brain iron regulation. A total of 666 unique shared SNPs were identified between SCZ and QSM phenotypes. For example, rs12814239 was shared between SCZ and bilateral caudate and putamen, located in the exonic region of LRP1 with a CADD score of 18. The de novo mutations of LRP1 were reported in individuals with SCZ49, and LRP1 is involved in the process by which macrophages uptake iron and iron-containing molecules50. As another example, rs11557154 was shared by SCZ and caudate, located in the exonic region of DCAF12 with a CADD score of 25. DCAF12 is critical for normal synaptic function and plasticity at larval neuromuscular junctions51. Notably, accumulating evidence has demonstrated that iron levels in subcortical brain regions may exert a causal influence on psychiatric disorders, particularly in structures such as the substantia nigra52. In line with these findings, our results support the hypothesis that genetically regulated variation in tissue magnetic susceptibility contributes to region-specific vulnerability in SCZ, potentially linking underlying microstructural alterations to disease-relevant neuroanatomical changes.

Moreover, the shared genes identified in this study are enriched in the biological processes that have been implicated in both SCZ pathology and iron regulation. Notably, genes are involved in the synapse and morphogenesis of cells and neurons. Dysregulation of synaptic dynamics has been linked to SCZ53. Besides, we observed that the shared genetic variants are predominantly expressed in neurons, particularly excitatory neurons, supporting the notion that disruptions in excitatory signaling may play a central role in the relationship between SCZ and brain susceptibility homeostasis. From a translational perspective, recent data-driven studies have increasingly leveraged neuroimaging and multi-modal biomarkers to identify biologically meaningful subtypes of SCZ54. In particular, the integration of neuroimaging features with peripheral biological markers, such as immune profiles, has enabled the delineation of distinct biological subgroups with potential therapeutic relevance55. Within this framework, our findings provide genetic evidence supporting the role of susceptibility-based imaging phenotypes derived from QSM as candidate biomarkers, which may facilitate stratification of SCZ and inform precision medicine strategies.

Additionally, the investigation into spatio-temporal gene expression patterns further clarified how shared genes are dynamically regulated in corresponding subcortical regions implicated in SCZ and iron homeostasis. We observed that shared genes are highly expressed during critical neurodevelopmental periods, which could mean that early-life disturbances in iron homeostasis might contribute to the developmental pathophysiology of SCZ. While genes linked to SCZ and iron deposition show peak expression during critical neurodevelopmental windows, our spatio-temporal trajectory analysis reveals pronounced brain region-specific heterogeneity in these expression dynamics. For example, the thalamus maintains persistently high mean expression across prenatal and postnatal developmental stages, whereas the hippocampus exhibits a distinct, transient expression peak in early postnatal life. Furthermore, the expression-trait associations indicated that these shared genes show significant correlations with personal lifestyle factors and neurocognitive performance. This suggests that genetic variation in these genes may be associated not only with neural function but also with environmental factors related to brain health and resilience.

Virtual drug screen yielded several potential therapeutic targets. For example, Paliperidone is an atypical antipsychotic that is chemically the primary active metabolite of the older antipsychotic risperidone, specifically known as 9-hydroxyrisperidone56–58; Pimozide is an orally active antipsychotic drug product which shares with other antipsychotics the ability to blockade dopaminergic receptors on neurons in the central nervous system59,60; Ziprasidone is classified as a second generation or atypical antipsychotic and is a dopamine and 5HT2A receptor antagonist with a unique receptor binding profile61,62; Deferasirox is an FDA-approved orally active iron chelator used in the treatment of iron toxicity63. Repurposing approved drugs targeting these pathways may offer potential value for addressing abnormalities in iron metabolism and magnetic susceptibility in SCZ and related pathological processes.

Several limitations should be acknowledged. First, while the study focused on a European cohort, it remains unclear whether these findings generalize across other populations with different genetic backgrounds. Expanding future analyses to more diverse populations could provide a more comprehensive understanding of the genetic architecture linking SCZ and brain iron homeostasis. Second, although the conjFDR method prioritizes SNPs with joint associations, residual confounding cannot be completely excluded, and certain SCZ-QSM associations may arise from pleiotropic influences or overlapping genetic architecture. Third, the limited sample size of the QSM GWAS may reduce the statistical power and precision of MiXeR estimates, and the results should be interpreted with appropriate caution. Fourth, although the phenome traits included in our analysis were selected based on prior evidence linking them to SCZ or related domains, they represent only a curated subset rather than a comprehensive coverage of all SCZ-associated phenotypes. Fifth, the molecular docking results are based on in silico predictions and include many compounds not directly relevant to SCZ, and therefore should be interpreted as hypothesis-generating, requiring further experimental validation to confirm their relevance and potential therapeutic value. Sixth, although QSM can to some extent reflect brain iron content, it is not specific to iron and instead captures a combination of magnetic susceptibility sources, which may confound the interpretation of the observed associations in terms of iron-related biological mechanisms. Finally, the number of sampling time points varies across brain regions and developmental stages, and some region-stage combinations have small effective sample sizes, which may introduce bias and reduce the precision of fitted gene expression trajectories.

This study provides insights into the shared genetic architecture of SCZ and subcortical QSM-derived imaging phenotypes, suggesting a potential link between SCZ-related genetic factors and brain susceptibility measures related to iron-sensitive MRI signals. Our findings highlight specific SNPs and genes, as well as enriched biological processes and cell types, that may contribute to the observed genetic overlap between SCZ and QSM phenotypes. By integrating genetic associations with imaging-derived traits, this work may help improve the understanding of SCZ’s complex etiology and support further investigation into biologically informed hypotheses relevant to SCZ and brain susceptibility-related alterations.

Supplementary information

Supplementary Figures (1.2MB, pdf)
Supplementary Tables (205.8KB, xlsx)

Author contributions

Dairong Cao and Yingying Xie designed the study. Yingying Xie, Jiaojiao Du, Yao Zhao, and Jie Sun analyzed the data. Yingying Xie, Jiaojiao Du, and Dairong Cao wrote the manuscript. Shiqi Lin, Zhang Zhang, Ningnannan Zhang, Feng Liu, and Dairong Cao supervised this work. All authors critically reviewed the manuscript.

Funding

We acknowledge funding from the National Natural Science Foundation of China (82502309); Fujian Provincial Health Technology Project (2025GGA042); Natural Science Foundation of Fujian Province (2025J08168).

Data availability

The GWAS summary statistics for SCZ are publicly available from the Psychiatric Genomics Consortium. The QSM GWAS summary statistics were obtained from the UK Biobank brain imaging GWAS resources.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Yingying Xie, Jiaojiao Du.

Contributor Information

Feng Liu, Email: fengliu@tmu.edu.cn.

Dairong Cao, Email: dairongcao@fjmu.edu.cn.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41537-026-00782-7.

References

  • 1.Ravanfar, P. et al. In vivo 7-Tesla MRI investigation of brain iron and its metabolic correlates in chronic schizophrenia. Schizophrenia8, 86 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Howes, O. D. & Kapur, S. The dopamine hypothesis of schizophrenia: version III—the final common pathway. Schizophr. Bull.35, 549–562 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Carpenter, K. L. H. et al. Magnetic susceptibility of brain iron is associated with childhood spatial IQ. NeuroImage132, 167–174 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bitanihirwe, B. K. & Woo, T. U. Oxidative stress in schizophrenia: an integrated approach. Neurosci. Biobehav. Rev.35, 878–893 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Marques, T. R. et al. Neuroinflammation in schizophrenia: meta-analysis of in vivo microglial imaging studies. Psychol. Med.49, 2186–2196 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Xu, M. et al. Brain iron assessment in patients with First-episode schizophrenia using quantitative susceptibility mapping. Neuroimage Clin.31, 102736 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Sonnenschein, S. F. et al. Subcortical brain iron deposition in individuals with schizophrenia. J. Psychiatr. Res.151, 272–278 (2022). [DOI] [PubMed] [Google Scholar]
  • 8.Cuesta, M. J. et al. Motor abnormalities and basal ganglia in first-episode psychosis (FEP). Psychol. Med.51, 1625–1636 (2021). [DOI] [PubMed] [Google Scholar]
  • 9.Sui, Y. V. et al. Decreased basal ganglia and thalamic iron in early psychotic spectrum disorders are associated with increased psychotic and schizotypal symptoms. Mol. Psychiatry27, 5144–5153 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kesby, J. P., Eyles, D. W., McGrath, J. J. & Scott, J. G. Dopamine, psychosis and schizophrenia: the widening gap between basic and clinical neuroscience. Transl. Psychiatry8, 30 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Xie, Y. et al. Shared genetic architecture among gastrointestinal diseases, schizophrenia, and brain subcortical volumes. Schizophr. Bull.50, 1243–1254 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Chen, P. et al. Polymorphism of transferrin gene impacts the mediating effects of psychotic symptoms on the relationship between oxidative stress and cognition in patients with chronic schizophrenia. Antioxidants 10.3390/antiox11010125 (2022). [DOI] [PMC free article] [PubMed]
  • 13.Trubetskoy, V. et al. Mapping genomic loci implicates genes and synaptic biology in schizophrenia. Nature604, 502–508 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Wang, C. et al. Phenotypic and genetic associations of quantitative magnetic susceptibility in UK Biobank brain imaging. Nat. Neurosci.25, 818–831 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Willer, C. J., Li, Y. & Abecasis, G. R. METAL: fast and efficient meta-analysis of genomewide association scans. Bioinformatics26, 2190–2191 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Frei, O. et al. Bivariate causal mixture model quantifies polygenic overlap between complex traits beyond genetic correlation. Nat. Commun.10, 2417 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Holland, D. et al. Beyond SNP heritability: polygenicity and discoverability of phenotypes estimated with a univariate Gaussian mixture model. PLoS Genet16, e1008612 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Andreassen, O. A., Thompson, W. K. & Dale, A. M. Boosting the power of schizophrenia genetics by leveraging new statistical tools. Schizophr. Bull.40, 13–17 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Andreassen, O. A. et al. Improved detection of common variants associated with schizophrenia by leveraging pleiotropy with cardiovascular-disease risk factors. Am. J. Hum. Genet.92, 197–209 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Andreassen, O. A. et al. Improved detection of common variants associated with schizophrenia and bipolar disorder using pleiotropy-informed conditional false discovery rate. PLoS Genet.9, e1003455 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Ray, D. & Chatterjee, N. A powerful method for pleiotropic analysis under composite null hypothesis identifies novel shared loci between Type 2 Diabetes and Prostate Cancer. PLoS Genet.16, e1009218 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zhang, Y. et al. Hemispheric asymmetry in the genetic overlap between schizophrenia and white matter microstructure. Cyborg Bionic Syst.7, 0451 (2026). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Bahrami, S. et al. Genetic loci shared between major depression and intelligence with mixed directions of effect. Nat. Hum. Behav.5, 795–801 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.O’Connell, K. S. et al. Characterizing the genetic overlap between psychiatric disorders and sleep-related phenotypes. Biol. Psychiatry90, 621–631 (2021). [DOI] [PubMed] [Google Scholar]
  • 25.Xie, Y. et al. Genetic and neural mechanisms shared by schizophrenia and depression. Mol. Psychiatry30, 3975–3987 (2025). [DOI] [PubMed] [Google Scholar]
  • 26.Zhao, Q. et al. Genome-wide analysis identifies novel shared loci between depression and white matter microstructure. Mol. Psychiatry30, 3455–3465 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Liu, M. et al. Investigating the shared genetic architecture between depression and subcortical volumes. Nat. Commun.15, 7647 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Genomes Project, C. et al. A global reference for human genetic variation. Nature526, 68–74 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Werme, J., van der Sluis, S., Posthuma, D. & de Leeuw, C. A. An integrated framework for local genetic correlation analysis. Nat. Genet.54, 274–282 (2022). [DOI] [PubMed] [Google Scholar]
  • 30.Suzuki, K. et al. Genetic drivers of heterogeneity in type 2 diabetes pathophysiology. Nature627, 347–357 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.International Multiple Sclerosis Genetics, C. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science 10.1126/science.aav7188 (2019). [DOI] [PMC free article] [PubMed]
  • 32.Watanabe, K., Taskesen, E., van Bochoven, A. & Posthuma, D. Functional mapping and annotation of genetic associations with FUMA. Nat. Commun.8, 1826 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Rentzsch, P., Witten, D., Cooper, G. M., Shendure, J. & Kircher, M. CADD: predicting the deleteriousness of variants throughout the human genome. Nucleic Acids Res.47, D886–d894 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Boyle, A. P. et al. Annotation of functional variation in personal genomes using RegulomeDB. Genome Res.22, 1790–1797 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Ernst, J. & Kellis, M. ChromHMM: automating chromatin-state discovery and characterization. Nat. Methods9, 215–216 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Consortium, G. T. et al. Genetic effects on gene expression across human tissues. Nature550, 204–213 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Giambartolomei, C. et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet10, e1004383 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Chen, J., Bardes, E. E., Aronow, B. J. & Jegga, A. G. ToppGene Suite for gene list enrichment analysis and candidate gene prioritization. Nucleic Acids Res.37, W305–W311 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Dai, Y. et al. WebCSEA: web-based cell-type-specific enrichment analysis of genes. Nucleic Acids Res.50, W782–w790 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Li, M. et al. Integrative functional genomic analysis of human brain development and neuropsychiatric risks. Science 10.1126/science.aat7615 (2018). [DOI] [PMC free article] [PubMed]
  • 41.Akbarian, S. et al. The PsychENCODE project. Nat. Neurosci.18, 1707–1712 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Pividori, M. et al. PhenomeXcan: mapping the genome to the phenome through the transcriptome. Sci. Adv. 10.1126/sciadv.aba2083 (2020). [DOI] [PMC free article] [PubMed]
  • 43.Szklarczyk, D. et al. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res.51, D638–D646 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Pinzi, L. & Rastelli, G. Molecular docking: shifting paradigms in drug discovery. Int. J. Mol. Sci. 10.3390/ijms20184331 (2019). [DOI] [PMC free article] [PubMed]
  • 45.Gan, J. H. et al. DrugRep: an automatic virtual screening server for drug repurposing. Acta Pharmacol. Sin.44, 888–896 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Liu, Y. et al. CB-Dock: a web server for cavity detection-guided protein-ligand blind docking. Acta Pharm. Sin.41, 138–144 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Eberhardt, J., Santos-Martins, D., Tillack, A. F. & Forli, S. AutoDock Vina 1.2.0: new docking methods, expanded force field, and Python bindings. J. Chem. Inf. Model61, 3891–3898 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Trott, O. & Olson, A. J. AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J. Comput Chem.31, 455–461 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Jouan, L. et al. Investigation of rare variants in LRP1, KPNA1, ALS2CL and ZNF480 genes in schizophrenia patients reflects genetic heterogeneity of the disease. Behav. Brain Funct.9, 9 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Winn, N. C., Volk, K. M. & Hasty, A. H. Regulation of tissue iron homeostasis: the macrophage “ferrostat”. JCI Insight 10.1172/jci.insight.132964 (2020). [DOI] [PMC free article] [PubMed]
  • 51.Patron, L. A. et al. Cul4 ubiquitin ligase cofactor DCAF12 promotes neurotransmitter release and homeostatic plasticity. J. Cell Biol.218, 993–1010 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Du, W., Tang, B., Liu, S., Zhang, W. & Lui, S. Causal associations between iron levels in subcortical brain regions and psychiatric disorders: a Mendelian randomization study. Transl. Psychiatry15, 19 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Keshavan, M., Lizano, P. & Prasad, K. The synaptic pruning hypothesis of schizophrenia: promises and challenges. World Psychiatry19, 110–111 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Zhang, W., Qiu, C. & Lui, S. Imaging biomarker studies of antipsychotic-naive first-episode schizophrenia in China: progress and future directions. Schizophr. Bull.51, 379–391 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Zhang, W., Sweeney, J. A., Bishop, J. R., Gong, Q. & Lui, S. Biological subtyping of psychiatric syndromes as a pathway for advances in drug discovery and personalized medicine. Nat. Ment. Health1, 88–99 (2023). [Google Scholar]
  • 56.Awouters, F. H. & Lewi, P. J. Forty years of antipsychotic Drug research—from haloperidol to paliperidone—with Dr. Paul Janssen. Arzneim.-Forsch.57, 625–632 (2007). [DOI] [PubMed] [Google Scholar]
  • 57.Dolder, C., Nelson, M. & Deyo, Z. Paliperidone for schizophrenia. Am. J. Health-Syst. Pharm.65, 403–413 (2008). [DOI] [PubMed] [Google Scholar]
  • 58.Chue, P. S., MacKenzie, E. M., Chue, J. A. & Baker, G. B. The pharmacology and formulation of paliperidone extended release. Expert Rev. Neurother.12, 1399–1410 (2012). [DOI] [PubMed] [Google Scholar]
  • 59.Tabatabaei Dakhili, S. A. et al. The antipsychotic dopamine 2 receptor antagonist diphenylbutylpiperidines improve glycemia in experimental obesity by inhibiting succinyl-CoA:3-ketoacid CoA transferase. Diabetes72, 126–134 (2023). [DOI] [PubMed] [Google Scholar]
  • 60.Mothi, M. & Sampson, S. Pimozide for schizophrenia or related psychoses. Cochrane Database Syst. Rev. 10.1002/14651858.CD001949.pub3 (2013). [DOI] [PMC free article] [PubMed]
  • 61.Gunasekara, N. S., Spencer, C. M. & Keating, G. M. Ziprasidone: a review of its use in schizophrenia and schizoaffective disorder. Drugs62, 1217–1251 (2002). [DOI] [PubMed] [Google Scholar]
  • 62.Stahl, S. M. & Shayegan, D. K. The psychopharmacology of ziprasidone: receptor-binding properties and real-world psychiatric practice. J. Clin. Psychiatry64, 6–12 (2003). [PubMed] [Google Scholar]
  • 63.Imai, T. et al. Deferasirox, a trivalent iron chelator, ameliorates neuronal damage in hemorrhagic stroke models. Naunyn-Schmiedeb. Arch. Pharmacol.394, 73–84 (2021). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Figures (1.2MB, pdf)
Supplementary Tables (205.8KB, xlsx)

Data Availability Statement

The GWAS summary statistics for SCZ are publicly available from the Psychiatric Genomics Consortium. The QSM GWAS summary statistics were obtained from the UK Biobank brain imaging GWAS resources.

Articles from Schizophrenia are provided here courtesy of Nature Publishing Group

RESOURCES