Evidence for Shared Genetic Aetiology Between Schizophrenia, Cardiometabolic, and Inflammation-Related Traits: Genetic Correlation and Colocalization Analyses

Abstract

Background: Schizophrenia commonly co-occurs with cardiometabolic and inflammation-related traits. It is unclear to what extent the comorbidity could be explained by shared genetic aetiology.

Methods: We used GWAS data to estimate shared genetic aetiology between schizophrenia, cardiometabolic, and inflammation-related traits: fasting insulin (FI), fasting glucose, glycated haemoglobin, glucose tolerance, type 2 diabetes (T2D), lipids, body mass index (BMI), coronary artery disease (CAD), and C-reactive protein (CRP). We examined genome-wide correlation using linkage disequilibrium score regression (LDSC); stratified by minor-allele frequency using genetic covariance analyzer (GNOVA); then refined to locus-level using heritability estimation from summary statistics (ρ-HESS). Regions with local correlation were used in hypothesis prioritization multi-trait colocalization to examine for colocalisation, implying common genetic aetiology.

Results: We found evidence for weak genome-wide negative correlation of schizophrenia with T2D (r_g = -0.07; 95% C.I., -0.03,0.12; P = .002) and BMI (r_g = -0.09; 95% C.I., -0.06, -0.12; P = 1.83 × 10^-5). We found a trend of evidence for positive genetic correlation between schizophrenia and cardiometabolic traits confined to lower-frequency variants. This was underpinned by 85 regions of locus-level correlation with evidence of opposing mechanisms. Ten loci showed strong evidence of colocalization. Four of those (rs6265 (BDNF); rs8192675 (SLC2A2); rs3800229 (FOXO3); rs17514846 (FURIN)) are implicated in brain-derived neurotrophic factor (BDNF)-related pathways.

Conclusions: LDSC may lead to downwardly-biased genetic correlation estimates between schizophrenia, cardiometabolic, and inflammation-related traits. Common genetic aetiology for these traits could be confined to lower-frequency common variants and involve opposing mechanisms. Genes related to BDNF and glucose transport amongst others may partly explain the comorbidity between schizophrenia and cardiometabolic disorders.

Keywords: cardiometabolic disorders; colocalization; common aetiology; correlation; genetic; schizophrenia.

Fig. 1.

Manhattan plot showing regions of local genetic correlation between schizophrenia and BMI. Manhattan plot showing local genetic correlation estimates (top panel); local covariance estimates (second panel); and local SNP heritability estimates (bottom two panels), at LD blocks across chromosomes 1–22. Areas coloured red/blue in top two panels correspond to LD-blocks surpassing Bonferroni significance threshold. Loci within the major histocompatibility complex (MHC) region were not considered for further analysis due to complex LD patterns in the region. SCZ = schizophrenia; BMI = body mass index.

Fig. 2.

Examples of regional genetic association plots for four loci returning strong evidence for colocalization between schizophrenia, cardiometabolic and inflammatory traits. A.rs8192675 – SLC2A2, B.rs13107325 – SLC39A8, C.rs6265 – BDNF, D.rs17514846 - FURIN. Regional association plots denote chromosomal location (x axis) and strength of association with listed trait (−log_10(p)) (y axis) alongside genomic position. SNP r² estimated from the EPIC-Norfolk cohort. See Supplementary Figure 3 for regional association plots of the remaining colocalized variants described in Table 2. Scz = schizophrenia; bmi = body mass index; tg = triglycerides; hdl = high-density lipoprotein; t2ds = type 2 diabetes; crp = c-reactive protein.

Fig. 2.

Examples of regional genetic association plots for four loci returning strong evidence for colocalization between schizophrenia, cardiometabolic and inflammatory traits. A.rs8192675 – SLC2A2, B.rs13107325 – SLC39A8, C.rs6265 – BDNF, D.rs17514846 - FURIN. Regional association plots denote chromosomal location (x axis) and strength of association with listed trait (−log_10(p)) (y axis) alongside genomic position. SNP r² estimated from the EPIC-Norfolk cohort. See Supplementary Figure 3 for regional association plots of the remaining colocalized variants described in Table 2. Scz = schizophrenia; bmi = body mass index; tg = triglycerides; hdl = high-density lipoprotein; t2ds = type 2 diabetes; crp = c-reactive protein.

Fig. 2.

Examples of regional genetic association plots for four loci returning strong evidence for colocalization between schizophrenia, cardiometabolic and inflammatory traits. A.rs8192675 – SLC2A2, B.rs13107325 – SLC39A8, C.rs6265 – BDNF, D.rs17514846 - FURIN. Regional association plots denote chromosomal location (x axis) and strength of association with listed trait (−log_10(p)) (y axis) alongside genomic position. SNP r² estimated from the EPIC-Norfolk cohort. See Supplementary Figure 3 for regional association plots of the remaining colocalized variants described in Table 2. Scz = schizophrenia; bmi = body mass index; tg = triglycerides; hdl = high-density lipoprotein; t2ds = type 2 diabetes; crp = c-reactive protein.

Fig. 2.

Examples of regional genetic association plots for four loci returning strong evidence for colocalization between schizophrenia, cardiometabolic and inflammatory traits. A.rs8192675 – SLC2A2, B.rs13107325 – SLC39A8, C.rs6265 – BDNF, D.rs17514846 - FURIN. Regional association plots denote chromosomal location (x axis) and strength of association with listed trait (−log_10(p)) (y axis) alongside genomic position. SNP r² estimated from the EPIC-Norfolk cohort. See Supplementary Figure 3 for regional association plots of the remaining colocalized variants described in Table 2. Scz = schizophrenia; bmi = body mass index; tg = triglycerides; hdl = high-density lipoprotein; t2ds = type 2 diabetes; crp = c-reactive protein.