Skip to main content
Download PDF

Exploring Lead loci shared between schizophrenia and Cardiometabolic traits

BMC Genomics volume 23, Article number: 617 (2022) Cite this article

Abstract

Individuals with schizophrenia (SCZ) have, on average, a 10- to 20-year shorter expected life span than the rest of the population, primarily due to cardiovascular disease comorbidity. Genome-wide association studies (GWAS) have previously been used to separately identify common variants in SCZ and cardiometabolic traits. However, genetic variants jointly influencing both traits remain to be fully characterised. To assess overlaps (if any) between the genetic architecture of SCZ and cardiometabolic traits, we used conditional false discovery rate (FDR) and local genetic correlation statistical framework analyses. A conjunctional FDR was used to identify shared genetic traits between SCZ and cardiometabolic risk factors. We identified 144 genetic variants which were shared between SCZ and body mass index (BMI), and 15 variants shared between SCZ and triglycerides (TG). Furthermore, we discovered four novel single nucleotide polymorphisms (SNPs) (rs3865350, rs9860913, rs13307 and rs9614186) and four proximate genes (DERL2, SNX4, LY75 and EFCAB6) which were shared by SCZ and BMI. We observed that the novel genetic variant rs13307 and the most proximate gene LY75 exerted potential effects on SCZ and BMI comorbidity. Also, we observed a mixture of concordant and opposite direction associations with shared genetic variants. We demonstrated a moderate to high genetic overlap between SCZ and cardiometabolic traits associated with a pattern of bidirectional associations. Our data suggested a complex interplay between metabolism-related gene pathways in SCZ pathophysiology.

Peer Review reports

Background

Individuals with schizophrenia (SCZ) have a 10- to 20-year shorter life span when compared with healthy individuals in the same population [1, 2]. Previous studies have indicated that cardiovascular disease could be a major cause of this shorter life expectancy in SCZ patients [1]. The link between the increased incidence of cardiovascular and metabolic disorder was previously established in SCZ patients when compared with the general population [2, 3]. For example, the risk of obesity and type 2 diabetes (T2D) are approximately 3.5- and 2-fold higher, respectively, in individuals with SCZ [4, 5]. Historically, the increased risk and prevalence of cardiometabolic disease (CMD) has been attributed to social determinants and lifestyle factors (including poor diet, sedentary behaviour and alcohol and substance use) and the effects of psychotropic medication [6, 7]. Furthermore, several psychopharmacological agents, in particular antipsychotics, are obesogenic and contribute to adverse events due to metabolic disorders [7] These factors suggest that CMD risks are both key risk factors and long-term health concerns in patients with SCZ [8]. However, for decades, cardiometabolic comorbidity and associated mortality have remained high in these patients and have suggested that most patients with SCZ have not benefited from clinical advancements [8, 9]. Therefore, comorbidity may be a result of other factors hitherto not considered. Therefore, a systems biology approach could provide new pathophysiological knowledge as indicated by recent genetics studies [10, 11].

Recently, the polygenic nature of SCZ and cardiometabolic traits have become increasingly clear [12]. These traits are reported with substantial heritability, estimated at 79% for SCZ [13], 24–90% for BMI [14], 36–61% for waist-hip ratio (WHR) [15], 11% for triglycerides (TG) [16], approximately 89% for total cholesterol (TC), 22–93% for high-density lipoprotein (HDL), 22–91% for low-density lipoprotein (LDL), 38–66% for fasting glucose (FG) [17], 47% for fasting insulin (FIN) [18] and 25–80% for T2D [19]. Several genetic studies have established links between CMD and SCZ, and an increased CMD prevalence has been associated with treatment responses in SCZ [20, 21]. Due to strong associations between BMI and SCZ, and also between TG and SCZ [22, 23], several neurobiological hypotheses related to potential underlying mechanisms have been proposed. However, associations are complex, as weight loss [24, 25] and weight gain [26] are associated with SCZ, and inconsistent associations between TG and SCZ have been reported [27, 28].

Even though abundant genetic variants are associated with SCZ comorbidity and cardiometabolic traits, understanding the functional consequences of genetic variations and identifying pleiotropic genes and pathways for both phenotypes remains challenging. Powerful statistical approaches, specifically designed to analyse the polygenic architectures of complex traits, could improve gene or loci discovery and replication rates [25,26,27]. The conditional FDR and conjunctional FDR methods, which specifically analyse the polygenic architecture of multiple disorders, allow for the identification of shared genetic variants, and in turn, elucidate common pathobiology and molecular mechanisms across different disorders [10]. Using this strategy, common associations between two phenotypes can be identified by evaluating the contribution from all SNPs from two independent GWAS [28,29,30]. The discovery of shared genetic variants could facilitate the development of risk prediction models for CMD traits and enabling targeted CMD interventions for SCZ patients.

In this study, we analysed GWAS summary statistics of SCZ and cardiometabolic traits, including WHR, BMI, TG, TC, HDL, LDL, FG, FIN and T2D, using pleiotropic-based conditional and conjunctional FDR statistics to estimate shared genetic characteristics between SCZ and cardiometabolic traits. We hypothesised that these methods could help identify shared genetic variants, shared polygenic architecture, and potential pleiotropic genes and biological pathways shared between SCZ and cardiometabolic traits.

Results

Polygenetic overlap and genetic correlations between SCZ and Cardiometabolic traits

As shown (Fig. 1), we performed conditional FDR and local genetic covariance analyses to identify pleiotropic effects between SCZ and cardiometabolic traits in a European ancestry background. Briefly, the conditional FDR approach was based on an empirical Bayesian statistical framework and used GWAS summary statistics as a primary trait (e.g. SCZ), together with a conditional trait (e.g. BMI) to estimate the posterior probability that an SNP had no association with the primary trait, given that P-values for that SNP in both primary and conditional traits were as small as, or smaller than the observed P-value.

Fig. 1
figure 1

Work flow of the study design

Full size image

Our fold-enrichment plots demonstrated that SNPs were highly enriched (4.8–25.0 fold) in SCZ, across increasingly stringent significance levels, for a BMI association (SCZ|BMI) (Fig. 2a), while SNPs were moderately enriched (1.3–4.3 fold) in SCZ, across increasing stringent significance levels, for the SCZ|TG association (Fig. 2c). The reverse conditional association (BMI|SCZ and TG|SCZ) showed a ~ 2.5–9.0 fold and ~ 2.5–34.0 fold-enrichment, respectively (Fig. 2b, d). These results supported a moderate to high level of polygenic overlap between SCZ and BMI or TG.

Fig. 2
figure 2

Pleiotropy analysis. a Plot of fold enrichment vs. nominal -log10 P for SCZ below the standard GWAS threshold of P value < 5 × 10− 8 as a function significant of the association with BMI. b Plot of fold enrichment versus nominal -log10 P values for BMI below the standard GWAS threshold of P value < 5 × 10− 8 as a function of significance of the association with SCZ. c Plot of fold enrichment vs. nominal -log10 P for SCZ below the standard GWAS threshold of P value < 5 × 10− 8 as a function of significance of the association with TG. d Plot of fold enrichment of TG versus nominal -log10 P values for below the standard GWAS threshold of P value < 5 × 10− 8 as a function of significance of the association with SCZ. e Local genetic correlation and local SNP-heritability between SCZ and BMI. f Local genetic correlation and local SNP-heritability between SCZ and TG. For each panel (e&f), the top section represents local genetic correlation, the middle section represents local genetic covariance, where significant local genetic correlation and covariance after multiple testing correction are highlighted in blue; and the bottom part represents local SNP-heritability for individual trait

Full size image

We also used the Heritability Estimation from Summary Statistics (HESS) package to estimate and visualise local SNP-heritability and genetic covariance, to examine if a specific genomic region was genetically linked to SCZ and cardiometabolic traits. We estimated local genetic covariance and correlations in 104 regions between SCZ and BMI, and 32 regions between SCZ and TG. In analysis between two trait pairs (SCZ|BMI and SCZ|TG), we identified four genomic regions on different chromosomes: chr11:27020461–28,481,593, chr12:122007651–124,977,980, chr16:29036613–31,382,943 and chr16:63691589–65,938,566, which showed strong local genetic associations between SCZ and BMI (Fig. 2e and Additional file 1 Supplementary Table 2). We identified two genomic regions: chr6:31571218–32,682,664 and chr7:71874885–73,334,602 which showed strong local genetic correlations between SCZ and TG (Fig. 2f and Additional file 1 Supplementary Table 3).

Genetic variants and genes identified by conjunctional FDR analysis are shared between SCZ and BMI and TG

To identify genetic variants shared between SCZ and BMI and TG, we performed conjunctional FDR analysis. This approach assessed the posterior probability that an SNP was null for either trait or both, given that P-values for both phenotypes were as small as, or smaller than P-values for each trait individually. Conjunctional FDR is an extension of the conditional FDR approach and is defined as the maximum of two conditional FDR statistics for a specific SNP. In total, 144 distinct genetic variants were shared between SCZ and BMI at a conjunctional FDR value < 0.05 (Fig. 3a and Additional file 1 Supplementary Table 4). Of these, 80 variants (56%) were not found in the original BMI GWAS [31], while 75 (52%) were not found in the original SCZ GWAS [32] and 79 (55%) were not reported in similar studies [3, 31]. Four genetic variants were identified which were novel for both phenotypes and, after mapping the most proximate genes to associated SNPs, we identified DERL2, SNX4, LY75 and EFCAB6 as novel genes in SCZ and BMI associations (Table 1). By integrating information from local genetic covariance analyses, we identified six SNPs [rs6265 (chr11:27679916), rs7975482 (chr12:124481690), rs4243232 (chr16:30514723), rs7953704 (chr12:122625992), rs10744211 (chr12:122931820) and rs4787491 (chr16:30015337)] and six most proximate genes BDNF, ZNF664, ITGAL, MLXIP, ZCCHC8 and INO80E at three significant genomic regions (chr11:27020461–28,481,593, chr12:122007651–124,977,980 and chr16:29036613–31,382,943). All genes were previously reported as being associated with BMI and SCZ.

Fig. 3
figure 3

Common genetic variants jointly associated with SCZ and BMI (3a) and TG (3b) at conjunctional false discovery rate (conjFDR) less than 0.05. Manhattan plot showing the –log10 transformed conjFDR values for each SNP on the y-axis and chromosomal position along the x-axis. The dotted horizontal line represents the threshold chosen for reporting shared associations (−log (FDR) values of 1.3 corresponds to a cFDR ≤0.05). Independent lead single-nucleotide polymorphisms are highlighted with a black outline. The significant shared signal in the major histocompatibility complex region (chr6:25119106–33,854,733 and chr8:7242715–12,483,982) were deleted in the analysis. Further details are provided in Additional file 1 Supplementary Table 4 and Supplementary table 4_1

Full size image
Table 1 Novel shared genes reaching statistical significance between SCZ and cardiovascular risk traits (BMI and TG) (conjFDR ≤ 0.05)
Full size table

In total, 15 genetic variants were shared between SCZ and TG at a conjunctional FDR value < 0.05 (Fig. 3b and Additional file 1 Supplementary Table 4_1). Of these, 10 (67%) genetic variants were not previously reported in the original TG GWAS [33]; while 10 (67%) were not previously reported in the original SCZ GWAS, and 6 (40%) were not previously reported in a similar study [3]. One genetic variant rs1472584, when the most proximate gene was mapped to the associated SNP, defined HCN1 as a novel gene for TG, and was associated with SCZ in a previous study (Table 1 and Additional file 1 Supplementary Table 4_1). When we integrated information from local genetic covariance analysis, we identified one SNP rs3130544 and one proximate gene C6orf15 in a significant genomic region (chr6:31571218–32,682,664). This novel TG gene was associated with SCZ (Table 1). By comparing the association direction for the top SNPs shared between SCZ and BMI at a conjunctional FDR value < 0.05, we identified mixed association direction patterns, with SNPs having concordant association directions in 72/144 genetic variants (50.0%) shared between BMI and SCZ (Additional file 1 Supplementary Table 5–6), while 8/15 genetic variants (53.3%) had concordant association directions between TG and SCZ (Additional file 1 Supplementary Table 7–8).

Annotating genetic variants shared between SCZ and BMI and TG

The functional annotation of SNPs at a conjunctional FDR value < 0.05 for SCZ and BMI is shown (Additional file 1 Supplementary Table 4 and Additional file 2 Supplementary Fig. 1). Most SNPs were in intronic (54.29%) and intergenic (31.43%) regions, and 12.77% had a RegulomeDB score < 3, predicting potential regulatory functions (Additional file 2 Supplementary Fig. 1a). Details on RegulomeDB scores are shown (Additional file 1 Supplementary Table 9).

After functional annotation using ANNOVAR, we identified four novel SNPs (rs3865350, rs9860913, rs13307 and rs9614186) and four most proximate genes (DERL2, SNX4, LY75 and EFCAB6) shared between SCZ and BMI. Ten candidate SNPs, in strong linkage disequilibrium (LD) (r2 ≥ 0.8) with rs3865350 (DERL2) at 17p13.2, were extracted using the HaploReg v4.2 tool (Additional file 1 Supplementary Table 10). We observed that rs3865350 was predicted to alter the binding of two transcription factor (TF) motifs GCNF and PLZF in HaploReg, which was not confirmed in GVATdb. In total, 42 candidate SNPs in strong LD (r2 ≥ 0.8) with rs9860913 (SNX4) at 3q21.2 were identified (Additional file 1 Supplementary Table 11). Using HaploReg, this locus was located within an activated enhancer and DNAse site in different cell types, but was not confirmed in EnhancerDB. We also identified 67 candidate SNPs in strong LD (r2 ≥ 0.8) with rs13307 (LY75) at 2q24.2 (Additional file 1 Supplementary Table 12). The binding site of three TF motifs (Evi-1, RPEB-1 and RXR) were affected by this variant in HaploReg, which was confirmed in GVATdb. In total, eight candidate SNPs in strong LD with rs9614186 (EFCAB6) at 22q13.2 were identified (Additional file 1 Supplementary Table 13).

The functional annotation of SNPs at conjunctional FDR < 0.05 for SCZ and TG are shown (Additional file 2 Supplementary Fig. 1). Most SNPs were within intronic (35.71%) or intergenic (42.86%) regions, and 6.67% had a RegulomeDB score < 3 (Additional file 2 Supplementary Fig. 1b).

We identified two novel SNPs (rs1472584 and rs3130544) and two most proximate genes (HCN1 and C6orf15) which were not previously associated with TG, but with SCZ. Ninety candidate SNPs, in strong LD (r2 ≥ 0.8) with rs1472584 (HCN1) at 5p12, were extracted (Additional file 1 Supplementary Table 14). Additionally, the binding site of the TF motif, Sox, was affected by this variant in HaploReg, but not confirmed in GVATdb. Thirteen candidate SNPs, in strong LD (r2 ≥ 0.8) with rs3130544 (C6orf15) at 6p21.33, were extracted (Additional file 1 Supplementary Table 15). The binding site of the TF motifs CEBPα, ERα-a and RORα1 were affected by this variant in HaploReg, but this was not confirmed in GVATdb.

Pathway analysis of genetic variants shared between SCZ and BMI and TG

We performed pathway enrichment analyses for shared genetic variants between SCZ and BMI, and SCZ and TG, to separately identify overrepresented pathways among most proximate genes nearest identified genetic variants. For SCZ and BMI, 297 pathways were significantly overrepresented and related to central nervous system (CNS) neuron differentiation, brain-derived neurotrophic factor (BDNF) signalling, positive regulation of growth, post synapse and modulation of chemical synaptic transmission (Fig. 4a and Additional file 1 Supplementary Table 16). Gene analyses were consistent and opposite association directions were separately identified between BMI and SCZ and indicated only a minor overlap in overrepresented pathways. Concordant genes were enriched in heterocycle catabolic processes, BDNF signalling, neuronal cell bodies, the PIP3 activation of AKT signalling, intracellular signalling by second messengers and stem cell differentiation. Opposite genes were enriched in cellular responses to fluid shear stress, CNS neuro differentiation, excitatory postsynaptic potential, chemical synaptic transmission, postsynaptic transmission and the regulation of postsynaptic membrane potential. More results from genetic analyses are shown (Additional file 1 Supplementary Tables 17 and 18).

Fig. 4
figure 4

Functional enrichment analysis of shared genes between SCZ and BMI. a Pathway enrichment analysis. b Tissue enrichment analysis using 53 tissues from the GTEx database (version 7). Significantly enriched differential expressed gene (DEG) sets (Bonferroni corrected P < 0.05) are highlighted in red. c PPI network of the shared genes between SCZ and BMI. d Significant cluster related to the PPI network (Module1)

Full size image

Four pathways were significantly enriched in genes nearest identified genetic variants which were shared by SCZ and TG, including small molecule catabolic processes, cellular responses to organo-nitrogen compounds, cell responses to nitrogen compounds and inorganic cation transmembrane transport (Fig. 5a and Additional file 1 Supplementary Table 19).

Fig. 5
figure 5

Functional enrichment analysis of shared genes between SCZ and TG. a Pathway enrichment analysis. b Tissue enrichment analysis using 53 tissues from the GTEx database (version 7). Significantly enriched differential expressed gene (DEG) sets (Bonferroni corrected P < 0.05) are highlighted in red. c PPI network for the shared genes between SCZ and TG. d and e Significant clusters related to the PPI network

Full size image

GTEx enrichment analysis identified four independent tissues which were significantly enriched after Benjamini-Hochberg corrections for shared gene expression between SCZ and BMI: these included brain, liver, heart and kidney (Fig. 4b). The most strongly enriched tissue was the brain-anterior-cingulate-cortex. No significant tissue enrichment results were identified for shared genes between SCZ and TG (Fig. 5b).

eQTL analysis of novel shared SNPs

Our eQTL analysis of novel shared SNPs was performed using the GTEx V7 database. DERL2, SNX4, LY75 and EFCAB6 expression levels were evaluated in brain, adipose and whole blood tissue in GTEx datasets. The risk allele rs13307-A was correlated with higher LY75 expression levels in adipose, brain and whole blood tissue (Additional file 2 Supplementary Fig. 2 and Additional file 1 Supplementary Table 20). To identify the posterior probability of the LY75 causal gene and phenotype at the same genetic variant rs13307, we performed tissue-eQTL colocalisation analyses and showed that eQTLs for LY75 in the brain strongly colocalised with the genetic variant rs12469374, a perfect proxy of rs13307 (R2 = 0.90, D′ = 0.98) in BMI (H4-Posterior Probability: 81.2%) and SCZ (H4-Posterior Probability: 82.1%) (Additional file 1 Supplementary Table 21).

Both rs1472584 and rs3130544 were not identified as HCN1 and C6orf15 eQTLs in brain, adipose, whole blood and liver tissue (Additional file 1 Supplementary Table 22).

Protein-protein interaction (PPI) network analysis

To identify potential interactions between shared genes, the Search Tool for the Retrieval of Interacting Genes (STRING) database was used to perform PPI network analysis. The PPI network of genes shared between SCZ and BMI consisted of 219 nodes (genes) and 535 edges (interactions) (Fig. 4c). After using the Molecular Complex Detection (MCODE) plug-in to identify modules from the PPI network for shared genes between SCZ and BMI, the top central modules with MCODE scores > 10 were selected. Module 1, with scores of 12, consisted of 13 nodes and 72 edges (Fig. 4d). From Gene Ontology (GO) and Kyoto Encyclopaedia of Gene and Genomes (KEGG) pathway enrichment analysis, Module 1 was mainly enriched for cyclin-dependent protein serine/threonine kinase activity, transcription regulation of G1/S transition in the mitotic cell cycle, mitotic G1 DNA damage checkpoint signalling and cell cycle G1/S phase transition. PPI network analysis of genetic variants with concordant and opposite association directions identified one module (enriched for axon guidance, cellular responses to growth factors and cellular responses to oxidative stress) and one module (enriched for CNS neuron differentiation and positive regulation of peptidyl-tyrosine phosphorylation) for highly interconnected nodes, respectively (Additional file 2 Supplementary Fig. 3).

The PPI network for SCZ and TG shared genes consisted of 114 nodes (genes) and 699 edges (interactions) (Fig. 5c). After using the MCODE plug-in to identify modules from this PPI network, the top two central modules, with MCODE scores > 10, were selected. Module 1 had a score of 11.79, consisted of 30 nodes and 171 edges and was mainly enriched for genes involved in DNA-template transcription initiation, regulation of pri-miRNA transcription by RNA polymerase II, nuclear receptor activity, Huntington disease and RNA polymerase II-specific DNA-binding TF binding (Fig. 5d). Module 2, with a score of 10.4, consisted of 11 nodes and 52 edges, and was enriched for cell responses to fatty acids and adipocytokine signalling (Fig. 5e). PPI network analysis of concordant and opposite genes identified one module (enriched for cholesterol transport regulation) and two modules (module 1 was enriched for macroautophagy, the autophagosome, protein localisation to phagophore assembly sites and autophagosome assembly, and module 2 was enriched for the positive regulation of the mitotic cell cycle, top signalling and TOR signalling regulation), respectively (Additional file 2 Supplementary Fig. 4).

Mendelian randomisation (MR) analysis

In MR analysis, when cardiometabolic traits were considered the exposure, no causal associations were observed between traits associated with SNPs for SCZ risk (Additional file 1 Supplementary Table 23). The test for horizontal pleiotropy, estimated using the MR-Egger intercept between BMI and SCZ, was significant (P = 0.049, Additional file 1 Supplementary Table 24) and indicated shared genetic variants between exposure and outcome. When cardiometabolic traits were considered an outcome, no causal relationship for SCZ was identified (Additional file 1 Supplementary Table 23).

Discussion

In this study, we investigated the polygenic overlap between SCZ and two cardiometabolic traits (BMI and TG). The MR analysis indicated that genetic liability, the heritability of a disease, to SCZ exerted no influence on BMI|TG and vice versa, suggesting no evidence identified for a causal relationship between SCZ and BMI|TG. We identify 144 genetic variants between SCZ and BMI, and 15 genetic variants between SCZ and TG. Genetic variants were mainly enriched for neuronal system functions, including CNS neuron differentiation, BDNF signalling and the positive regulation of growth. Also, shared genetic variants demonstrated a mixture of concordant associations and associations with opposite directions between trait pairs. In total, 50% of shared SCZ SNPs had positive associations with BMI, while 46.7% had positive associations with TG. Additionally, we identified four novel genetic variants (rs3865350, rs9860913, rs13307 and rs9614186) shared by SCZ and BMI, and two novel genetic variants (rs1472584 and rs3130544) shared by SCZ and TG.

Shared genetic variants demonstrated a mixture of consistent and opposite association directions between trait pairs. This polygenic overlap between SCZ and BMI agreed with previous epidemiological association evidence [34, 35]. Also, 50% of genetic variants shared between BMI and SCZ had negative associations with BMI, and furthermore, 57% of genetic variants between BMI and SCZ were negatively associated with BMI when conjunctional FDR analysis was conducted at the 0.1 threshold (Additional file 1 Supplementary Table 25). These results agreed with genetic correlation analyses; we identified a negative correlation between SCZ and BMI (rg = − 0.081, P < 0.01, Additional file 1 Supplementary Table 26). Previous studies [5, 11] on SCZ and cardiometabolic traits indicated similar results with similar conclusions, however, in our study, we provided further evidence using different analytical approaches (HESS, partional LDSC and MR), thereby confirming associations may not be causal, but more likely pleiotropic in nature [11]. We also identified two chromosomal regions (chr11:27020461–28,481,593 and chr12:122007651–124,977,980; Fig. 2e and Additional file 1 Supplementary Table 2) with positive local genetic correlations and two genetic regions (chr16:29036613–31,382,943 and chr16:63691589–65,938,566; Fig. 2f and Additional file 1 Supplementary Table 3) with negative local genetic correlations, together with negative genetic correlations for DNase I hypersensitivity sites and TF binding sites (Additional file 1 Supplementary Table 27). MR analyses indicated no causal relationships between BMI and SCZ, but the test for horizontal pleiotropy, estimated using the MR-Egger intercept between BMI and SCZ, was significant (P = 0.049) and indicated shared genetic variants between the exposure (BMI) and outcome (SCZ). While underlying mechanisms remain unclear, one hypothesis suggests that poor nutrition, though subtle, may exert negative effects on neural development, leading to a increased incidence of mental health disorders, such as SCZ [11]. This observation suggested that variables such as antipsychotic treatments, dietary habits, or lifestyle may be primary factors contributing to weight gain in patients with long-term disease. Moreover, low BMI is viewed as a risk factor for SCZ [36], while a recent study reported an increased underweight frequency in patients with SCZ [37]. Our findings suggest variations in weight gain can occur during antipsychotic medication administration and may be partly mediated by genetics [38].

We performed several analyses, including eQTL, colocalisation, and functional annotation to identify a novel shared genetic overlap between SCZ and BMI|TG. Our eQTL analyses showed that the risk allele rs13307-A was associated with LY75 expression in adipose and brain tissue. The genetic variant rs13307 for LY75 in the brain strongly colocalised with BMI (H4-Posterior Probability: 81.2%) and SCZ (H4-Posterior Probability: 82.1%). Also, rs13307 may exert effects on the binding sites of TF motifs (Evi-1, RPEB-1 and RXR). This evidence suggested that the novel genetic variant rs13307 and the most proximate gene LY75 exerted potential effects on SCZ and BMI comorbidity. A previous mouse study that used quantitative complementation, qualitative phenotypic and causal analysis, showed that an allele of the LY75 locus potentially exerted pleiotropic effects on the weight of total and inguinal fat pads [39]. LY75 is predominantly expressed by dendritic cells [40] and plays critical roles in endocytosis and T cell antigen presentation via major histocompatibility complex molecules, thereby contributing to immune function (e.g. antigen processing and complement pathways) [41, 42]. Recent studies suggested that chronic inflammation may be an important mediator linking metabolic abnormalities and severe mental illness [42, 43]. For example, elevated pro-inflammatory cytokines, including tumour necrosis factor-α and interleukin-6, were observed in patients with psychosis and CMD [43]. These findings support a potential role for LY75, which is involved in inflammation and immune pathways, in the shared genetic architecture of SCZ with cardiometabolic traits [44]. The exploration of potential mechanisms underlying combined TF motifs and LY75 functions in SCZ and BMI is warranted.

We noticed that the minor allele frequency (MAF) of rs13307-A is 0.27 with the highest population frequency (1000 Genomes Phase 3, ESP and gnomAD) could be 0.49. The MAF of rs3865350-C is 0.26, with the highest population frequency (1000 Genomes Phase 3, ESP and gnomAD) could be 0.50. This indicates these alleles are more common in populations not merely in SCZ patients. In the QC process, numerous germline mutations were removed immediately after being generated, either by selection or randomly [45]. The retained variants may expand in the population; however, some mutations may cause disease and disorders owning to environmental changes over time, known as risk alleles [46]. Studies indicated that the deviation from 0.5 in the proportion of SNPs in which minor alleles were the risk alleles was relatively small (0.591–0.631) when the MAF was relatively high (> 0.1), indicating that most of the SNPs with those high MAFs were associated with diseases resulting from the changes of environment [47, 48]. Gorlov et al. suggested that environment or lifestyle-dependent diseases tend to have a higher frequency of risk-associated variants [49]. Besides, the analysis of the NHGRI-EBI Catalog data demonstrated that complex diseases, such as Alzheimer’s disease (late-onset), Parkinson’s disease, multiple sclerosis, metabolic syndrome, and schizophrenia, were more likely to have a high average risk allele frequencies [48]. In our study, the eQTL analyses indicated that the minor allele rs13307-A was associated with the expression of LY75 in adipose and brain tissue, which may be a risk allele for the comorbidity of SCZ and BMI. However, since the rs13307-A with a high MAF, we speculate that recent environmental changes (including epigenetic changes or other factors) may play a crucial role, and these environmental changes should be considered for the study of the pathogenesis of comorbidity in future research.

Functional annotation of shared genetic variants showed that some genes were associated with gene expression in the brain and several biological and molecular processes, including CNS neuron differentiation, BDNF signalling, positive regulation of growth and the modulation of chemical synaptic transmission. A large proportion (approximately 67%) of shared genetic variants were brain-related and suggested that BMI regulation involved brain-related mechanisms [50]. Brain functions determine our behaviours as they determine lifestyle choices such as diet and exercise, which in turn affect BMI [51]. For example, BDNF signalling affects neural circuit structure and function and also modulates multiple neurotransmitter systems - functions closely related to SCZ [52]. Extensive evidence now suggests that BDNF is an essential contributor to food intake and body weight control [53]. Heterozygous BDNF knockout mice had reduced BDNF expression which led to age-dependent obesity and an insulin-resistant phenotype, and were characterised by elevated circulating levels of insulin, leptin and glucose [54, 55]. Critically, similar brain mechanisms may be involved in behaviours related to mental health disorders and BMI.

We failed to identify significant genetic correlations (rg = − 0.029, P = 0.187, Additional file 1 Supplementary Table 26) between SCZ and TG. However, from local genetic covariance analysis, two specific chromosomal regions were positively genetically correlated with SCZ and TG. These data were generally consistent with a previous meta-analysis on lipid profiles in first-episode psychosis patients, which reported higher TG levels [27]. In longitudinal studies [56, 57], dyslipidemia in patients with SCZ was typically studied as a side effect of antipsychotic medication. However, researchers reported that dyslipidemia and other metabolic risk factors may be present in early disease stages, before treatment initiation [58]. Thus, sustained elevated TG levels could mean dyslipidemia is associated with SCZ and not only due to medication [59]. Of note, a previous MR analysis supported the notion that SCZ was casually associated with increased TG levels [3], however, we did not identify a causal association between SCZ and TG.

Functional annotation analysis of shared genetic variants showed that variants contributed to both SCZ and TG, and were associated with small molecule catabolic processes, cell responses to organo-nitrogen compounds, cell responses to nitrogen compounds and inorganic cation transmembrane transport. A previous rat study reported that an inorganic cation transmembrane transport pathway was enriched in the cortex of rats treated with antipsychotic drugs [60]. In mice, cell responses to organo-nitrogen compounds and nitrogen compounds were involved in liver peroxisomal and mitochondrial roles to maintain TG balance and oxidative stress [61]. While we investigated pathways putatively involved in shared pathophysiology between SCZ and lipid traits, further experimental studies are warranted to elucidate the exact mechanisms.

Data from other phenotype pairs (TC|SCZ, HDL|SCZ, LDL|SCZ, T2D|SCZ, FG|SCZ and FIN|SCZ) (Additional file 4 Additional Tables and Additional file 5 Additional Figures) go beyond standard genetic association as shown by the findings in the conditional FDR and conjunctional FDR analyses that can evaluate the directions of association of the shared genetic variant. For example, despite the lack of genetic correlations for TC and SCZ (rg = − 0.0294), we identified 160 shared genetic variants (Additional file 3 Additional datasets). The shared genetic variant discovery could facilitate risk predictions for comorbid SCZ, thereby generating targeted interventions for cardiometabolic symptoms or similar diseases in patients with SCZ.

We failed to identify shared genetic variants between WHR, a measure of abdominal or visceral adiposity and SCZ, despite BMI being moderately correlated with WHR in adults (rg = 0.3) [62]. Some studies [12, 62] reported that SNPs related to body composition and fat distribution (WHR = waist and hip circumference) were also associated with mental disorders, and simultaneously associated with BMI, inconsistent with our study data. A possible explanation could be that our GWAS summary statistics for WHR were adjusted for BMI, and there is no or little isolated effect of body fat distribution on SCZ, as corroborated by several clinical and epidemiological studies [62,63,64].

Our study had some limitations. Firstly, our analyses were based on large-scale GWAS studies, which primarily consisted of Caucasian participants, thus in future work, we will extend our research remit to other ethnic groups. Secondly, in analyses using paired summary results, we could not control for other clinical factors, e.g., BMI when studying lipid traits. Thirdly, it was challenging to assess small effect sizes and speculate on molecular mechanisms underlying effective variants when examining potentially overlapping phenotypes. Overall, our methods enhanced the discovery of additional shared polygenic architectures and identified potential pleiotropic genes and biological pathways between two complex traits.

Conclusions

We demonstrated a moderate to high genetic overlap between SCZ and BMI|TG, with a pattern of bidirectional associations, indicating a complex interplay of metabolism-related gene pathways in SCZ pathophysiology. Furthermore, we identified four novel SNPs and four most proximate genes which were shared by SCZ and BMI, and also two novel SNPs and two most proximate genes shared by SCZ and TG. Our findings contribute to a better understanding of the shared biological mechanisms underpinning SCZ and BMI|TG, and may facilitate reduced BMI and TG comorbidities among SCZ patients.

Methods

Study design, data summary and quality control (QC)

The overall study design is shown (Fig. 1). We retrieved summary statistics from publicly available GWAS studies, including SCZ [32] from the Psychiatric Genomics Consortium (N = 105,318 of European ancestry), BMI [65] (N = 236,231 of mixed ancestry) and WHR [66] (N = 348,501 of European ancestry) from the GIANT Consortium, T2D from the DIAGRAM Consortium [67] (N = 159,208 of European ancestry), fasting glucose (FG) [68] (N = 58,047 of European ancestry) and fasting insulin (FIN [68]) (N = 51,750 of European ancestry) from the MAGIC Consortium, and blood lipids (HDL [N = 99,900], LDL [N = 95,454 of European ancestry], TC [N = 100,184 of European ancestry] and TG [N = 96,598 of European ancestry]) [33] from the ENGAGE Consortium [69]. Dataset details are shown (Additional file 1 Supplementary Table 1).

We applied standardised GWAS summary data to minimise potential biases due to different array platforms and QC procedures. Firstly, we compared the md5 code of GWAS summary statistics and reported total SNPs to check data quality. Secondly, we used the LiftOver tool (http://genome.sph.umich.edu/wiki/LiftOver) to convert all GWAS summary data to the GRCh37/hg19 reference genome. Thirdly, we filtered out variants with a minor allele frequency (MAF) < 1% and removed SNPs with duplicates by keeping the first one. Lastly, we deleted SNPs with any missing values of OR/beta, SE, or P-values. Additionally, we restricted our analysis to autosomal chromosomes and European ancestry. The original number of SNPs and those remaining after QC are shown in Additional file 1 Supplementary Table 1.

Pleiotropy analysis

Our pleiotropy analysis strategy was based on conditional FDR. Fold-enrichment plots are described elsewhere [70, 71]. Briefly, the conditional FDR method establishes an empirical Bayesian statistical framework and uses GWAS summary statistics from the trait of interest (e.g. SCZ) together with statistics for a conditional trait (e.g. BMI), to estimate the posterior probability that an SNP has no association with the primary trait, given that the P-value for that SNP in both primary and conditional traits are as small as, or smaller than the observed P-value. Fold-enrichment plots graphically depicted pleiotropy by showing fold-enrichment in terms of SNP numbers on the ordinate, and nominal -log10(P) values for associations with SCZ on the abscissa [7]. Separate curves were established for SNP subsets which reached specific significance levels for associations with SCZ.

Our analyses followed two directions; firstly, with SCZ as the primary phenotype A and cardiometabolic traits as the conditional phenotype B, then vice versa for the second direction. We also generated a fold-enrichment plot to assess polygenic overlap between SCZ and cardiometabolic risk traits.

Local genetic covariance analysis

To investigate if a local genetic correlation existed between SCZ and cardiometabolic traits, the HESS [72] package was used to estimate local genetic correlations between a pair of traits at each LD-independent region in the genome. Approximate independent LD blocks, averaging 1.5 Mb in length, were used to calculate each local genetic heritability trait and genetic covariance. A total of 1685 approximate LD-independent genomic regions (excluding the major histocompatibility complex region) were used for analysis. Genomic regions were also excluded if the estimated local single-trait heritability was negative due to insufficient study power.

Conjunctional FDR analysis

To determine if genetic variants were likely to be shared by two phenotypes, we computed conjunctional FDR statistics. The conjunctional FDR is an extension of the conditional FDR and is defined as the maximum of two conditional FDR statistics for a specific SNP. The conjunctional FDR estimates the posterior probability that an SNP is null for either trait or both, given that the P-values for both phenotypes are as small as, or smaller than the P-values for each trait individually. More details are provided elsewhere [73,74,75]. In our study, we included shared SNPs with conjunctional FDR < 0.05. Manhattan plots were constructed based on the conjunctional FDR to show the genomic location of shared genetic variants. Traits were selected based on two criteria: 1) genetic pleiotropy existed in two phenotypes and 2) local genetic covariance analysis was significant in at least one region. Two trait pairs met these criteria: SCZ vs. BMI and SCZ vs. TG.

Functional annotations

We mapped SNPs identified by conjunctional FDR to promising genes using ANOVAR software [76] and described the distribution of shared SNPs. We extracted novel shared SNPs with strong LD (r2 ≥ 0.8) with the index variant based on the 1000 Genomes Phase 1 European individuals from the online HaploReg v4.2 tool [77]. Using data from ENCODE [78] and Roadmap [79] databases, we predicted regulatory elements (promoters and enhancers, etc.) using histone modification markers (H3K4me3, H3K4me1 and H3K27ac), chromatin state segmentation and DNase I hypersensitivity sites (DHS) in 125 cell types.

Functional Mapping and Annotation of GWAS (FUMA https://fuma.ctglab.nl/) was used to annotate significantly shared lead SNPs with functional categories using combined annotation dependent depletion scores (CADD) [80], RegulomeDB scores and chromatin states. A CADD score > 12.37 indicated a deleterious protein association with outcomes. The RegulomeDB score indicated the regulatory functionality of SNPs, based on the expression of quantitative trait loci (eQTL) and chromatin markers. The chromatin state indicated the accessibility of genomic regions using 15 categories as predicted by ChromHMM and based on five chromatin markers for 127 epigenomes. GTEx tissue enrichment analysis was based on 53 general tissue types and conducted using FUMA [81]. The Genetic Variants Allelic TF Binding Database (GVATdb) was used to search for SNPs with differential TF binding (http://renlab.sdsc.edu/GVATdb/search.html) capabilities and was also used to characterise the allelic binding of common human SNPs (MAF > 1% in European and Asian populations) to distinct TFs. The transcriptional regulation in the context of enhancers (EnhancerDB) resource was used to define tissue-specific enhancers by setting threshold scores for tissue-specific enhancers (http://lcbb.swjtu.edu.cn/EnhancerDB/).

Bioinformatics analysis

To understand the biological role of genes nearest the shared genetic variants between SCZ and cardiometabolic traits, we performed multiple post-GWAS functional analyses in shared genes identified by conjunctional FDR. We used the Metascape tool [82] (http://metascape.org), with default parameters, to assess the overrepresented enrichment of shared gene sets between SCZ and cardiometabolic traits (BMI and TG) in KEGG pathway analyses (www.kegg.jp/kegg/kegg1.html), GO Biological Processes, GO Cellular Components, GO Molecular Functions, WikiPathways, Hallmark and Reactome gene sets.

To evaluate the possible effects of genetic variants on transcriptional activity, we performed an eQTL analysis using the GTEx V7 database [83]. Previous studies indicated that SCZ and BMI-associated genetic variants showed strong gene expression enrichment in brain tissues [84, 85]. As for other cardiometabolic traits, including obesity and blood lipids, these were reportedly stored in subcutaneous adipose tissue [84]. The liver is related to fat metabolism and secretes TG [86]. Considering that genetic variants may affect gene expression in a tissue-specific manner, eQTL analyses were performed on brain, adipose and whole blood tissue to identify shared SNPs between SCZ and BMI, while the GTEx database was used for the same tissue and liver to identify shared SNPs between SCZ and TG.

Colocalisation between GWAS and eQTL signals

Colocalisation analysis was performed to test the probability of genetic variant the same one of GWAS and tissue-specific eQTL using GTEx dataset. Summary statistics for SNPs (regardless of GWAS P-value) within 200 kb of significant lead SNPs and common to both GWAS and eQTL studies, were inputted into coloc under default parameter settings [87]. This approach tested the probability of five hypotheses (H0–4), of which H4 tested the hypothesis that the same causal variant was shared between GWAS and tissue-specific eQTLs. Genetic variants with 80% or higher probability for H4 were compared to understand the LD structure and the most prominent variant being shared by GWAS and eQTL [88].

Protein-protein interaction (PPI) networks

PPI networks of shared genes were generated using the STRING 11.0; https://string.embl.de/) database, with a confidence score = 0.7 (high confidence) [89]. The PPI network was visualised using Cytoscape 3.8.2 software, and network modules were screened using the MCODE plug-in with MCODE scores > 10 [90]. The connectivity cut-off degree = 2, node score cut-off = 0.2, k-core = 2 and a maximum depth of 100 was permitted [91]. Pathway enrichment analysis of genes in modules was performed and P < 0.05 was considered a statistically significant difference.

MR analysis

To test for causal relationships between SNPs identified by conditional analysis, we performed MR analysis [92] using cardiometabolic trait GWAS SNPs as instrumental variables in the ‘TwoSampleMR’ package. For the exposure, default parameter settings of a P threshold of 5 × 10− 8, LD R2 = 0.001 and clumping distance = 10 kb were used [93]. We conducted MR using five approaches: inverse variance weighted (IVW), MR-Egger, weighted median, simple mode and weighted mode methods.

In the main analysis, we reported IVW estimates, however, if Egger’s method identified horizontal pleiotropy (e.g., SNPs associate with exposure but influence the outcome through pathways not specific to exposure), then MR-Egger results were used.

Availability of data and materials

The datasets analysed during the crrent study are available in the following public domain resources: https://www.ebi.ac.uk/gwas/publications/25673413 (Accession Number: GCST002783); https://www.ebi.ac.uk/gwas/publications/31453325 (Accession Number: GCST009127); https://www.ebi.ac.uk/gwas/publications/20081858 (Accession Number: GCST000571 and GCST000568); https://www.ebi.ac.uk/gwas/publications/20686565 (Accession Number: GCST000758, GCST000760, GCST000759, and GCST000755); https://figshare.com/articles/dataset/scz2018clozuk/14681220 (Data DOI: https://doi.org/10.6084/m9.figshare.14681220).

References

  1. Laursen TM, Plana-Ripoll O, Andersen PK, McGrath JJ, Toender A, Nordentoft M, et al. Cause-specific life years lost among persons diagnosed with schizophrenia: is it getting better or worse? Schizophr Res. 2019;206:284–90.

    Article  PubMed  Google Scholar 

  2. McElroy SL, Keck PE. Obesity in bipolar disorder: an overview. Curr Psychiat Rep. 2012;14:650–8.

    Article  Google Scholar 

  3. So H-C, Chau K-L, Ao F-K, Mo C-H, Sham P-C. Exploring shared genetic bases and causal relationships of schizophrenia and bipolar disorder with 28 cardiovascular and metabolic traits. Psychol Med. 2019;49:1286–98.

    Article  PubMed  Google Scholar 

  4. Tully A, Smyth S, Conway Y, Geddes J, Devane D, Kelly JP, et al. Interventions for the management of obesity in people with bipolar disorder. Cochrane Db Syst Rev. 2020:CD013006.

  5. Hert MAD, Winkel R van, Eyck DV, Hanssens L, Wampers M, Scheen A, et al. Prevalence of the metabolic syndrome in patients with schizophrenia treated with antipsychotic medication. Schizophr Res 2006;83:87–93.

  6. Zheng J, Erzurumluoglu AM, Elsworth BL, Kemp JP, Howe L, Haycock PC, et al. LD hub: a centralized database and web interface to perform LD score regression that maximizes the potential of summary level GWAS data for SNP heritability and genetic correlation analysis. Bioinformatics. 2017;33:272–9.

    Article  CAS  PubMed  Google Scholar 

  7. Ringen PA, Engh JA, Birkenaes AB, Dieset I, Andreassen OA. Increased mortality in schizophrenia due to cardiovascular disease – a non-systematic review of epidemiology, possible causes, and interventions. Frontiers Psychiatry. 2014;5:137.

    Article  Google Scholar 

  8. Rødevand L, Steen NE, Elvsåshagen T, Quintana DS, Reponen EJ, Mørch RH, et al. Cardiovascular risk remains high in schizophrenia with modest improvements in bipolar disorder during past decade. Acta Psychiatr Scand. 2019;139:348–60.

    Article  PubMed  CAS  Google Scholar 

  9. Correll CU, Robinson DG, Schooler NR, Brunette MF, Mueser KT, Rosenheck RA, et al. Cardiometabolic risk in patients with first-episode schizophrenia Spectrum disorders: baseline results from the RAISE-ETP Study. Jama Psychiat. 2014;71:1350–63.

    Article  Google Scholar 

  10. Andreassen OA, Djurovic S, Thompson WK, Schork AJ, Kendler KS, O’Donovan MC, et al. Improved detection of common variants associated with schizophrenia by leveraging pleiotropy with cardiovascular-disease risk factors. Am J Hum Genet. 2013;92:197–209.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Andreassen OA, Harbo HF, Wang Y, Thompson WK, Schork AJ, Mattingsdal M, et al. Genetic pleiotropy between multiple sclerosis and schizophrenia but not bipolar disorder: differential involvement of immune-related gene loci. Mol Psychiatry. 2015;20:207–14.

    Article  CAS  PubMed  Google Scholar 

  12. Andreassen OA, McEvoy LK, Thompson WK, Wang Y, Reppe S, Schork AJ, et al. Identifying common genetic variants in blood pressure due to polygenic pleiotropy with associated phenotypes. Hypertension. 2014;63:819–26.

    Article  CAS  PubMed  Google Scholar 

  13. Hoge EA, Pollack MH, Kaufman RE, Zak PJ, Simon NM. Oxytocin levels in social anxiety disorder. Cns Neurosci Ther. 2008;14:165–70.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Elks CE, den HM, Zhao JH, Sharp SJ, Wareham NJ, Loos RJF, et al. Variability in the heritability of body mass index: a systematic review and Meta-regression. Front Endocrinol. 2012;3:29.

    Article  Google Scholar 

  15. Rose KM, Newman B, Mayer-Davis EJ, Selby JV. Genetic and behavioral determinants of waist-hip ratio and waist circumference in women twins. Obes Res. 1998;6:383–92.

    Article  CAS  PubMed  Google Scholar 

  16. Williams PT. Gene-environment interactions due to quantile-specific heritability of triglyceride and VLDL concentrations. Sci Report. 2020;10:4486.

    Article  CAS  Google Scholar 

  17. Simonis-Bik AMC, Eekhoff EMW, Diamant M, Boomsma DI, Heine RJ, Dekker JM, et al. The heritability of HbA1c and fasting blood glucose in different measurement settings. Twin Res Hum Genet. 2008;11:597–602.

    Article  PubMed  Google Scholar 

  18. Poulsen P, Levin K, Petersen I, Christensen K, Beck-Nielsen H, Vaag A. Heritability of insulin secretion, peripheral and hepatic insulin action, and intracellular glucose partitioning in young and old Danish twins. Diabetes. 2005;54:275–83.

    Article  CAS  PubMed  Google Scholar 

  19. McCarthy M, Menzel S. The genetics of type 2 diabetes: genetics of type 2 diabetes. Brit J Clin Pharmaco. 2001;51:195–9.

    Article  CAS  Google Scholar 

  20. Verma SK, Subramaniam M, Liew A, Poon LY. Metabolic risk factors in drug-naive patients with first-episode psychosis. J Clin Psychiatry. 2009;70:997–1000.

    Article  PubMed  Google Scholar 

  21. Zammit S, Rasmussen F, Farahmand B, Gunnell D, Lewis G, Tynelius P, et al. Height and body mass index in young adulthood and risk of schizophrenia: a longitudinal study of 1 347 520 Swedish men. Acta Psychiatr Scand. 2007;116:378–85.

    Article  CAS  PubMed  Google Scholar 

  22. Solberg DK, Bentsen H, Refsum H, Andreassen OA. Lipid profiles in schizophrenia associated with clinical traits: a five year follow-up study. Bmc Psychiatry. 2016;16:299.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  23. Consortium R, Consortium PG. 3 GC for AN of the WTCCC, Bulik-Sullivan B, Finucane HK, Anttila V, et al. an atlas of genetic correlations across human diseases and traits. Nat Genet. 2015;47:1236–41.

    Article  CAS  Google Scholar 

  24. Padmavati R, McCreadie RG, Tirupati S. Low prevalence of obesity and metabolic syndrome in never-treated chronic schizophrenia. Schizophr Res. 2010;121:199–202.

    Article  PubMed  Google Scholar 

  25. Inamura Y, Sagae T, Nakamachi K, Murayama N. Body mass index of inpatients with schizophrenia in Japan. Int J Psychiatry Med. 2012;44:171–81.

    Article  PubMed  Google Scholar 

  26. Song X, Pang L, Feng Y, Fan X, Li X, Zhang W, et al. Fat-mass and obesity-associated gene polymorphisms and weight gain after risperidone treatment in first episode schizophrenia. Behav Brain Funct Bbf. 2014;10:35.

    Article  PubMed  CAS  Google Scholar 

  27. Pillinger T, Beck K, Stubbs B, Howes OD. Cholesterol and triglyceride levels in first-episode psychosis: systematic review and meta-analysis. Brit J Psychiat. 2017;211:339–49.

    Article  PubMed  Google Scholar 

  28. Niarchou M, Byrne EM, Trzaskowski M, Sidorenko J, Kemper KE, McGrath JJ, et al. Genome-wide association study of dietary intake in the UK biobank study and its associations with schizophrenia and other traits. Transl Psychiatry. 2020;10:51.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Andreassen OA, Thompson WK, Schork AJ, Ripke S, Mattingsdal M, Kelsoe JR, et al. Improved detection of common variants associated with schizophrenia and bipolar disorder using pleiotropy-informed conditional false discovery rate. PLoS Genet. 2013;9:e1003455.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Perry BI, Burgess S, Jones HJ, Zammit S, Upthegrove R, Mason AM, et al. The potential shared role of inflammation in insulin resistance and schizophrenia: a bidirectional two-sample mendelian randomization study. PLoS Med. 2021;18:e1003455.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Eyre HA, Air T, Proctor S, Rositano S, Baune BT. A critical review of the efficacy of non-steroidal anti-inflammatory drugs in depression. Prog Neuro-psychopharmacology Biological Psychiatry. 2015;57:11–6.

    Article  CAS  Google Scholar 

  32. Pardiñas AF, Holmans P, Pocklington AJ, Escott-Price V, Ripke S, Carrera N, et al. Common schizophrenia alleles are enriched in mutation-intolerant genes and in regions under strong background selection. Nat Genet. 2018;50:381–9.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  33. Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, et al. Biological, clinical and population relevance of 95 loci for blood lipids. Nature. 2010;466:707–13.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Manu P, Dima L, Shulman M, Vancampfort D, Hert MD, Correll CU. Weight gain and obesity in schizophrenia: epidemiology, pathobiology, and management. Acta Psychiatr Scand. 2015;132:97–108.

    Article  CAS  PubMed  Google Scholar 

  35. Yang X, Sun L, Zhao A, Hu X, Qing Y, Jiang J, et al. Serum fatty acid patterns in patients with schizophrenia: a targeted metabonomics study. Transl Psychiatry. 2017;7:e1176.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  36. Sørensen HJ, Mortensen EL, Reinisch JM, Mednick SA. Height, weight and body mass index in early adulthood and risk of schizophrenia. Acta Psychiatr Scand. 2006;114:49–54.

    Article  PubMed  Google Scholar 

  37. Sugawara N, Maruo K, Sugai T, Suzuki Y, Ozeki Y, Shimoda K, et al. Prevalence of underweight in patients with schizophrenia: a meta-analysis. Schizophr Res. 2018;195:67–73.

    Article  PubMed  Google Scholar 

  38. Zhang J-P, Lencz T, Zhang RX, Nitta M, Maayan L, John M, et al. Pharmacogenetic associations of antipsychotic drug-related weight gain: a systematic review and Meta-analysis. Schizophr Bull. 2016;42:1418–37.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Makino K, Ishikawa A. Genetic identification of Ly75 as a novel quantitative trait gene for resistance to obesity in mice. Sci Report. 2018;8:17658.

    Article  CAS  Google Scholar 

  40. East L, Isacke CM. The mannose receptor family. Biochimica Et Biophysica Acta Bba - Gen Subj. 2002;1572:364–86.

    Article  CAS  Google Scholar 

  41. Tel J, Benitez-Ribas D, Hoosemans S, Cambi A, Adema GJ, Figdor CG, et al. DEC-205 mediates antigen uptake and presentation by both resting and activated human plasmacytoid dendritic cells. Eur J Immunol. 2011;41:1014–23.

    Article  CAS  PubMed  Google Scholar 

  42. Henderson DC, Vincenzi B, Andrea NV, Ulloa M, Copeland PM. Pathophysiological mechanisms of increased cardiometabolic risk in people with schizophrenia and other severe mental illnesses. Lancet Psychiatry. 2015;2:452–64.

    Article  PubMed  Google Scholar 

  43. Revelo XS, Luck H, Winer S, Winer DA. Morphological and inflammatory changes in visceral adipose tissue during obesity. Endocr Pathol. 2014;25:93–101.

    Article  CAS  PubMed  Google Scholar 

  44. Byrne P, Cullinan J, Smith SM. Statins for primary prevention of cardiovascular disease. Bmj. 2019;367:l5674.

    Article  PubMed  Google Scholar 

  45. Kido T, Sikora-Wohlfeld W, Kawashima M, Kikuchi S, Kamatani N, Patwardhan A, et al. Are minor alleles more likely to be risk alleles? BMC Med Genet. 2018;110(1):3.

    Google Scholar 

  46. Chan Y, Lim ET, Sandholm N, Wang SR, MK AJ, Ripke S, et al. An excess of risk-increasing low-frequency variants can be a signal of polygenic inheritance in complex diseases. Am J Hum Genet. 2014;94:437–52.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Welter D, MacArthur J, Morales J, Burdett T, Hall P, Junkins H, et al. The NHGRI GWAS catalog, a curated resource of SNP-trait associations. Nucleic Acids Res. 2014;42D:1001–6.

    Article  CAS  Google Scholar 

  48. Neel JV. Diabetes mellitus: a “thrifty” genotype rendered detrimental by “progress”? Am J Hum Genet. 1962;14:353–62.

    CAS  PubMed  PubMed Central  Google Scholar 

  49. Gorlov IP, Gorlova OY, Amos C. Allelic spectra of risk SNPs are different for environment/lifestyle dependent versus independent diseases. PLoS Genet. 2015;11:e1005371.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  50. Makaronidis JM, Batterham RL. Obesity, body weight regulation and the brain: insights from fMRI. Br J Radiol. 2018;91:20170910.

    Article  PubMed  PubMed Central  Google Scholar 

  51. Kishi T, Elmquist JK. Body weight is regulated by the brain: a link between feeding and emotion. Mol Psychiatry. 2005;10:132–46.

    Article  CAS  PubMed  Google Scholar 

  52. Briana DD, Malamitsi-Puchner A. Developmental origins of adult health and disease: the metabolic role of BDNF from early life to adulthood. Metabolis. 2018;81:45–51.

    Article  CAS  Google Scholar 

  53. Vanevski F, Xu B. Molecular and neural bases underlying roles of BDNF in the control of body weight. Front Neurosci. 2013;7:37.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  54. Coppola V, Tessarollo L. Control of hyperphagia prevents obesity in BDNF heterozygous mice. Neuroreport. 2004;15:2665–8.

    Article  PubMed  Google Scholar 

  55. Duan W, Guo Z, Jiang H, Ware M, Mattson MP. Reversal of behavioral and metabolic abnormalities, and insulin resistance syndrome, by dietary restriction in mice deficient in brain-derived neurotrophic factor. Endocrinology. 2003;144:2446–53.

    Article  CAS  PubMed  Google Scholar 

  56. Saari K, Koponen H, Laitinen J, Jokelainen J, Lauren L, Isohanni M, et al. Hyperlipidemia in persons using antipsychotic medication: a general population-based birth cohort study. J Clin Psychiatry. 2004;65:547–50.

    Article  CAS  PubMed  Google Scholar 

  57. MacKenzie NE, Kowalchuk C, Agarwal SM, Costa-Dookhan KA, Caravaggio F, Gerretsen P, et al. Antipsychotics, metabolic adverse effects, and cognitive function in schizophrenia. Front Psychol. 2018;9:622.

    Article  Google Scholar 

  58. Wedervang-Resell K, Friis S, Lonning V, Smelror RE, Johannessen C, Agartz I, et al. Lipid alterations in adolescents with early-onset psychosis may be independent of antipsychotic medication. Schizophr Res. 2020;216:295–301.

    Article  PubMed  Google Scholar 

  59. Lett TAP, Wallace TJM, Chowdhury NI, Tiwari AK, Kennedy JL, Müller DJ. Pharmacogenetics of antipsychotic-induced weight gain: review and clinical implications. Mol Psychiatry. 2012;17:242–66.

    Article  CAS  PubMed  Google Scholar 

  60. Petralia MC, Ciurleo R, Saraceno A, Pennisi M, Basile MS, Fagone P, et al. Meta-analysis of transcriptomic data of dorsolateral prefrontal cortex and of peripheral blood mononuclear cells identifies altered pathways in schizophrenia. Genes-basel. 2020;11:390.

    Article  CAS  PubMed Central  Google Scholar 

  61. Asimakopoulou A, Fülöp A, Borkham-Kamphorst E, Leur EV de, Gassler N, Berger T, et al. Altered mitochondrial and peroxisomal integrity in lipocalin-2-deficient mice with hepatic steatosis. Biochimica Et Biophysica Acta Bba - Mol Basis Dis 2017;1863:2093–2110.

  62. Peters T, Nüllig L, Antel J, Naaresh R, Laabs B-H, Tegeler L, et al. The role of genetic variation of BMI, body composition, and fat distribution for mental traits and disorders: a look-up and Mendelian randomization Study. Front Genet. 2020;11:373.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. Cortese S, Moreira-Maia CR, Fleur DSt, Morcillo-Peñalver C, Rohde LA, Faraone SV. Association between ADHD and obesity. Syst Rev Meta-Analysis Am J Psychiat. 2016;173:34–43.

    Google Scholar 

  64. Nigg JT, Johnstone JM, Musser ED, Long HG, Willoughby M, Shannon J. Attention-deficit/hyperactivity disorder (ADHD) and being overweight/obesity: new data and meta-analysis. Clin Psychol Rev. 2016;43:67–79.

    Article  PubMed  Google Scholar 

  65. Study TLC, Consortium TAdipog, Group TA-BW, Consortium TCardiogram, Consortium TCkdg. GLGC T, et al. genetic studies of body mass index yield new insights for obesity biology. Nature. 2015;518:197–206.

    Article  CAS  Google Scholar 

  66. Wang H, Zhang F, Zeng J, Wu Y, Kemper KE, Xue A, et al. Genotype-by-environment interactions inferred from genetic effects on phenotypic variability in the UK biobank. Sci Adv. 2019;5:eaaw3538.

    Article  PubMed  PubMed Central  Google Scholar 

  67. Scott RA, Scott LJ, Mägi R, Marullo L, Gaulton KJ, Kaakinen M, et al. An expanded genome-wide association Study of type 2 diabetes in Europeans. Diabetes. 2017;66:2888–902.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  68. Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, Jackson AU, et al. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet. 2010;42:105–16.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  69. Consortium E, Surakka I, Horikoshi M, Mägi R, Sarin A-P, Mahajan A, et al. The impact of low-frequency and rare variants on lipid levels. Nat Genet. 2015;47:589–97.

    Article  CAS  Google Scholar 

  70. Desikan RS, Schork AJ, Wang Y, Thompson WK, Dehghan A, Ridker PM, et al. Polygenic overlap between C-reactive protein, plasma lipids, and Alzheimer disease. Circulation. 2015;131:2061–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  71. Wang X-F, Lin X, Li D-Y, Zhou R, Greenbaum J, Chen Y-C, et al. Linking Alzheimer’s disease and type 2 diabetes: novel shared susceptibility genes detected by cFDR approach. J Neurol Sci. 2017;380:262–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  72. Shi H, Mancuso N, Spendlove S, Pasaniuc B. Local genetic correlation gives insights into the shared genetic architecture of complex traits. Am J Hum Genet. 2017;101:737–51.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  73. Hellard SL, Wang Y, Witoelar A, Zuber V, Bettella F, Hugdahl K, et al. Identification of gene loci that overlap between schizophrenia and educational attainment. Schizophr Bull. 2017;43:654–64.

    PubMed  Google Scholar 

  74. Smeland OB, Frei O, Kauppi K, Hill WD, Li W, Wang Y, et al. Identification of genetic loci jointly influencing schizophrenia risk and the cognitive traits of verbal-numerical reasoning, reaction time, and general cognitive function. Jama Psychiat. 2017;74:1065.

    Article  Google Scholar 

  75. Lutz MW, Sprague D, Barrera J, Chiba-Falek O. Shared genetic etiology underlying Alzheimer’s disease and major depressive disorder. Transl Psychiatry. 2020;10:88.

    Article  PubMed  PubMed Central  Google Scholar 

  76. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38:e164.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  77. Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 2012;40:D930–4.

    Article  CAS  PubMed  Google Scholar 

  78. Consortium TEP, Moore JE, Purcaro MJ, Pratt HE, Epstein CB, Shoresh N, et al. Expanded encyclopaedias of DNA elements in the human and mouse genomes. Nature. 2020;583:699–710.

    Article  CAS  Google Scholar 

  79. Bernstein BE, Stamatoyannopoulos JA, Costello JF, Ren B, Milosavljevic A, Meissner A, et al. The NIH roadmap Epigenomics mapping Consortium. Nat Biotechnol. 2010;28:1045–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  80. Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J. A general framework for estimating the relative pathogenicity of human genetic variants. Nat Genet. 2014;46:310–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  81. Watanabe K, Taskesen E, van BA, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;8:1826.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  82. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019;10:1523.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  83. Aguet F, Brown AA, Castel SE, Davis JR, He Y, Jo B, et al. Genetic effects on gene expression across human tissues. Nature. 2017;550:204–13.

    Article  Google Scholar 

  84. Pan DZ, Garske KM, Alvarez M, Bhagat YV, Boocock J, Nikkola E, et al. Integration of human adipocyte chromosomal interactions with adipose gene expression prioritizes obesity-related genes from GWAS. Nat Commun. 2018;9:1512.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  85. Glastonbury CA, Viñuela A, Buil A, Halldorsson GH, Thorleifsson G, Helgason H, et al. Adiposity-dependent regulatory effects on multi-tissue transcriptomes. Am J Hum Genet. 2016;99:567–79.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  86. Liu H, Sun Y, Zhang X, Li S, Hu D, Xiao L, et al. Integrated analysis of summary statistics to identify pleiotropic genes and pathways for the comorbidity of schizophrenia and Cardiometabolic disease. Front Psychol. 2020;11:256.

    Article  Google Scholar 

  87. Giambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, et al. Bayesian test for Colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet. 2014;10:e1004383.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  88. Guo H, Fortune MD, Burren OS, Schofield E, Todd JA, Wallace C. Integration of disease association and eQTL data using a Bayesian colocalisation approach highlights six candidate causal genes in immune-mediated diseases. Hum Mol Genet. 2015;24:3305–13.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  89. Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, et al. STRING v11: protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 2019;47(Database issue):D607–13.

    Article  CAS  PubMed  Google Scholar 

  90. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498–504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  91. Bader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. Bmc Bioinformatics. 2003;4:2.

    Article  PubMed  PubMed Central  Google Scholar 

  92. North T-L, Davies NM, Harrison S, Carter AR, Hemani G, Sanderson E, et al. Using genetic instruments to estimate interactions in Mendelian randomization studies. Biorxiv. 2019;544734.

  93. Pierce BL, Burgess S. Efficient Design for Mendelian Randomization Studies: subsample and 2-sample instrumental variable estimators. Am J Epidemiol. 2013;178:1177–84.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank the Psychiatric Genomics Consortium, the GIANT Consortium, the DIAGRAM Consortium, the MAGIC Consortium, and the ENGAGE Consortium for providing GWAS summary statistics data and are grateful to all the investigators and participants contributed to the studies.

Funding

This work is supported by City University of Hong Kong New Research Initiatives/Infrastructure Support from Central (APRC) (Grant No.9610401).

Author information

Authors and Affiliations

  1. Department of Biomedical Sciences, City University of Hong Kong, Hong Kong, SAR, China

    Qian He, Adam N. Bennett, Jundong Liu & Kei Hang Katie Chan

  2. Department of Mental Health, Shenzhen Nanshan Center for Chronic Disease Control, ShenZhen, China

    Beifang Fan, Xue Han, Lu Cheng & Yan Chen

  3. Department of Integrative Biology & Physiology, University of California, Los Angeles, USA

    Xia Yang

  4. Brain Research Institute, University of California, Los Angeles, USA

    Xia Yang

  5. Institute for Quantitative and Computational Biosciences, University of California, Los Angeles, USA

    Xia Yang

  6. Department of Electrical Engineering, City University of Hong Kong, Hong Kong, SAR, China

    Kei Hang Katie Chan

  7. Department of Epidemiology, Centre for Global Cardiometabolic Health, Brown University, Providence, RI, USA

    Kei Hang Katie Chan

Authors
  1. Qian He
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Adam N. Bennett
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Jundong Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Beifang Fan
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Xue Han
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Lu Cheng
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Yan Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Xia Yang
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Kei Hang Katie Chan
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Q.H. participated in the study design, analysed the data and wrote the paper. A.N.B. and J.D.L. participated in the discussion of the study design, data analysis, and edited the manuscript. B.F.F., X.H., L.C., Y.C. and X.Y. revised the manuscript and provided comments. K.H.C. conceived the study, designed, coordinated the study, and revised the manuscript. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Kei Hang Katie Chan.

Ethics declarations

Ethics approval and consent to participate

Data used in the preparation of this article are the existing data available on the corresponding consortium website, and all the data are not identifiable. Data were downloaded and analysed according to the ethical guidelines of the Helsinki Declarations and under the approval of the Ethics Review of Human Research, the City University of Hong Kong.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

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

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

He, Q., Bennett, A.N., Liu, J. et al. Exploring Lead loci shared between schizophrenia and Cardiometabolic traits. BMC Genomics 23, 617 (2022). https://doi.org/10.1186/s12864-022-08766-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12864-022-08766-4

Keywords

BMC Genomics

ISSN: 1471-2164

Contact us