Multitrait Genetic Analysis Identifies Novel Pleiotropic Loci for Depression and Schizophrenia in East Asians

Abstract

Background and Hypothesis

While genetic correlations, pleiotropic loci, and shared genetic mechanisms of psychiatric disorders have been extensively studied in European populations, the investigation of these factors in East Asian populations has been relatively limited.

Study Design

To identify novel pleiotropic risk loci for depression and schizophrenia (SCZ) in East Asians. We utilized the most comprehensive dataset available for East Asians and quantified the genetic overlap between depression, SCZ, and their related traits via a multitrait genome-wide association study. Global and local genetic correlations were estimated by LDSC and ρ-HESS. Pleiotropic loci were identified by the multitrait analysis of GWAS (MTAG).

Study Results

Besides the significant correlation between depression and SCZ, our analysis revealed genetic correlations between depression and obesity-related traits, such as weight, BMI, T2D, and HDL. In SCZ, significant correlations were detected with HDL, heart diseases and use of various medications. Conventional meta-analysis of depression and SCZ identified a novel locus at 1q25.2 in East Asians. Further multitrait analysis of depression, SCZ and related traits identified ten novel pleiotropic loci for depression, and four for SCZ.

Conclusions

Our findings demonstrate shared genetic underpinnings between depression and SCZ in East Asians, as well as their associated traits, providing novel candidate genes for the identification and prioritization of therapeutic targets specific to this population.

Introduction

Psychiatric disorders comprise a range of severe illnesses that result in emotional, cognitive, or behavioral disturbances in individuals, making them a significant contributor to the global burden of disability.¹ Empirical studies indicate that psychiatric disorders result from the intricate interplay of genetic and environmental risk factors.² Heritability estimates, as determined by twin-based and genome-wide association studies (GWAS), emphasize the substantial role of genetic factors within the spectrum of psychiatric disorders.^3–6 Specifically, the heritability estimates for major depression are approximately 30%–40%,⁷ while for schizophrenia (SCZ), these estimates range from 65% to 80%.^8–10

Depression and SCZ both exhibit a substantial degree of polygenicity, with numerous risk loci have been identified through GWAS in individuals of European ancestry.^6,11 In a recent study, the largest GWAS of depression, involving more than 1.3 million individuals (including 371 184 with depression), identified a total of 243 risk loci. This study also unveiled a critical role of neurodevelopmental pathways associated with prenatal GABAergic neurons, astrocytes and oligodendrocyte lineages.¹² For SCZ, a total of 287 significant loci have been identified in the largest GWAS encompassing 76 755 cases and 243 649 controls. These identified associations predominantly cluster within genes that are actively expressed in both excitatory and inhibitory neurons within the central nervous system.¹³ However, the majority of previous genetic studies have been performed in European ancestry cohorts, leaving a conspicuous research gap in Asian populations. The largest GWAS of depression in East Asians (15 771 cases and 178 777 controls) only identified two genetic loci.¹⁴ Similarly, the largest GWAS conducted on SCZ among individuals of East Asian origin (with 22 778 cases and 35 362 controls) identified 19 genetic loci,¹⁵ which is significantly fewer than those observed in European studies.

Recently, the emergence of multitrait analysis of GWAS (MTAG), a tool for conducting cross-trait meta-analysis, has brought forth a powerful approach to explore the genetic overlap and shared causality among complex traits.¹⁶ By conducting meta-analysis of GWAS summary statistics for different genetically correlated traits, multitrait analysis can enhance the statistical power and pinpoint pleiotropic genetic variants that influence interrelated traits. Moreover, it reveals the intricate genetic interplay existing between these phenotypes and unveils the common biological pathways that underpin their development. Such cross-trait analysis methods have been successfully applied to the study of psychiatric disorders. For example, a meta-analysis using association analysis based on subsets (ASSET)¹⁷ across eight psychiatric disorders identified 109 pleiotropic loci linked to at least two disorders, revealing the common genetic structures shared among different psychiatric disorders.⁶ Besides, a cross-trait meta-analysis revealed a substantial number of pleiotropic genetic loci between psychiatric disorder and gastrointestinal tract diseases, emphasizing the shared biological mechanism concerning immune response, synaptic structure and function, and potential gut microbiome in these conditions.¹⁸ To the best of our knowledge, MTAG has predominantly centered on European populations, and, no such analysis has been undertaken for psychiatric disorders within East Asian populations so far.

In this study, we conducted a comprehensive multitrait analysis using GWAS data of depression and SCZ in cohorts of East Asian descent. We also investigated the deep-phenotype GWAS data encompassing a range of health aspects such as diseases, biomarkers and medication usage sourced from BioBank Japan (BBJ).¹⁹ Moreover, we performed analyses to quantify both overall and local genetic correlations, further identifying novel pleiotropic loci for depression and SCZ in East Asians.

Methods

GWAS Data

Our study is a secondary analysis of existing GWASs. GWAS summary statistics of depression were retrieved from the hitherto largest GWAS meta-analysis among individuals of East Asia (15 771 cases and 178 777 controls), based on combined data from the China, Oxford, and Virginia Commonwealth University Experimental Research on Genetic Epidemiology (CONVERGE) consortium,²⁰ China Kadoorie Biobank (CKB), and the Taiwan-Major Depressive Disorder (MDD) study, as well as US- and UK-based cohorts with DNA samples of individuals of East Asian ancestry.¹⁴ And GWAS summary statistics of SCZ were retrieved from the largest GWAS among individuals across East Asia, compiled 22 778 SCZ cases and 35 362 controls from 20 sample collections from East Asia.¹⁵

GWAS summary statistics of 96 phenotypes from East Asians including diseases, biomarkers and medication usage were downloaded from GWAS Catalog,²¹ information of all included GWAS data from East Asians is summarized in Supplementary table 1.

Meta-analysis Between Depression and SCZ

Inverse-variance weighted meta-analysis was performed on depression and SCZ using the basic meta-analysis function in PLINK (v1.90b7) and the fixed-effect meta-analysis P value and fixed-effect ORs were estimated. We prioritized significant SNPs that reached genome-wide significance (P < 5 × 10⁻⁸) in the meta-analysis and suggestive significance (P < .01) in the original single-trait GWAS.

Global Genetic Correlation Analysis

Genetic correlation r_g between depression and SCZ was estimated by LD (Linkage Disequilibrium) score regression (LDSC) using GWAS summary statistic.²² Pre-computed linkage disequilibrium scores for HapMap3 SNPs calculated based on East Asian-ancestry data from the 1000 Genomes Project were used in the analysis.

Local Genetic Correlation Analysis

Considering that genetic correlation estimated by LDSC aggregates information across all variants in the genome, we further estimated the pairwise local genetic correlation using ρ-HESS (heritability estimation from summary statistics).²³ ρ-HESS are designed to quantify the local genetic correlation between pairs of traits at each of the 1703 prespecified LD independent segments with an average length of 1.6 Mb. A Bonferroni corrected P value less than .05/1703 was considered as statistically significant.

Multitrait GWAS Analysis

In addition to the above traditional meta-analysis method, the MTAG framework, a generalized meta-analysis method that outputs trait-specific SNP associations,¹⁶ was furthered conducted. MTAG can increase the power to detect loci from correlated traits by analyzing GWAS summary statistics jointly. It also accounts for sample overlap and incomplete genetic correlation, when comparing with the conventional inverse-variance weighted meta-analysis. The first step of MTAG is to filter variants by removing non common SNPs, duplicated SNPs, or SNPs with strand ambiguity. MTAG then estimates the pairwise genetic correlation between traits using LDSC²⁴ and uses these estimates to calibrate the variance-covariance matrix of the random effect component. MTAG next performs a random effect meta-analysis to generate the SNP-level summary statistics. Prior to the MTAG analysis, the summary statistics underwent a standard quality control process, which included: aligning A1 and A2 alleles across different GWAS datasets and retaining the intersection of SNPs; removing SNPs with missing values; excluding SNPs with minor allele frequency (MAF) ≤ 0.01; discarding SNPs with standard errors (SE) < 0 or NaN values; filtering out SNPs with out-of-bounds P values; removing SNPs with duplicated rs numbers; and excluding SNPs with N < 0. We prioritized significant pleiotropic SNPs that reached genome-wide significance (P < 5 × 10⁻⁸) in the multitrait analysis and suggestive significance (P < .01) in the original single-trait GWAS. Pleiotropic loci exhibiting an LD r² greater than 0.1 with previously reported loci associated with SCZ or depression in East Asians were categorized as novel, while others were considered known.

Genetic Correlation Analysis Between Psychiatric Disorders and Deep Phenotypes

To investigate the potential risk factors for psychiatric disorders in East Asian populations, we included 96 GWAS summary statistics of deep phenotypes from the BioBank Japan (BBJ), including diseases, biomarkers and medication usage. The heritability z-score for each phenotype and genetic correlation r_g between depression/SCZ and each phenotype were estimated by LD (Linkage Disequilibrium) score regression (LDSC) using GWAS summary statistic.²²

Multitrait GWAS Analysis of Psychiatric Disorders and Risk Factors

Multitrait GWAS meta-analysis for psychiatric disorders and the corresponding genetically correlated deep phenotypes were performed by the MTAG framework. In MTAG analysis, we consistently designated psychiatric disorders as trait 1 (schizophrenia/depression) and other risk factors as trait 2. The MTAG results reported are specific to trait 1 (schizophrenia/depression).

We prioritized significant pleiotropic SNPs that reached genome-wide significance (P < 5 × 10⁻⁸) in the multitrait analysis and suggestive significance (P < .01) in the original single-trait GWAS. Pleiotropic loci exhibiting an LD r² greater than 0.1 with previously reported loci associated with SCZ or depression in East Asians were categorized as novel, while others were considered known. Based on this definition, the significant loci identified through MTAG can be categorized as follows: (1) “Novel for Trait 1”: This refers to loci not previously identified in GWAS of SCZ or depression. Such loci are considered novel discoveries for Trait 1. (2) “Novel for Trait 2”: This refers to loci not previously identified in GWAS of the other risk factors analyzed. Such loci are considered novel discoveries for Trait 2. (3) “Known for Either Trait but New Pleiotropic Effect”: This refers to loci previously identified in GWAS of either SCZ or depression or the other risk factors, but not previously known to have a pleiotropic effect between the two traits. Identifying such loci enhances our understanding of the shared genetic architecture between the traits. A potential issue with MTAG is that SNPs may be null for one trait but non-null for another, potentially causing MTAG’s effect size estimates for these SNPs in the first trait to deviate from 0. This could lead to an increase in the false positive rate (FDR). Hence, we estimated the maximum FDR for each trait by invoking the “--fdr” option when running MTAG.

Association Analysis Based on Subsets for Pleiotropic Loci Identified in MTAG

To further validate the pleiotropic loci identified in the aforementioned MTAG analysis of psychiatric disorders and risk factors, Association analysis based on subsets (ASSET) analysis was conducted for each locus. ASSET explored all possible subsets of the given traits to identify association signals, resulting in the optimal combination of traits to maximize the test statistic.¹⁷ In the ASSET analysis of pleiotropic loci for depression, all depression-associated risk factors were incorporated (LDSC P < .05). The same approach was applied in the ASSET analysis of pleiotropic loci for SCZ.

Results

Meta-analysis Between Depression and SCZ

We conducted a comprehensive meta-analysis by combing summary statistics derived from the largest GWAS of depression and SCZ within East Asian populations. Our analysis yielded three pleiotropic loci surpassing the significance threshold of P < 5 × 10⁻⁸. Among the above three pleiotropic loci, 2p16.1 (rs13016665) and 18q23 (rs58736086) were known in the study of SCZ among East Asian populations,¹⁵ while none of these three pleiotropic loci was previously reported in the GWAS of depression (table 1, Supplementary figure 1). The single-tissue eQTLs for novel pleiotropic loci were presented in Supplementary table 2, originate from the Genotype-Tissue Expression (GTEx) Analysis Release V8. The lead variant rs12031894 of the only novel locus is located in the intergenic region near gene SOAT1. It is worth mentioning that this locus reached genome-wide significance in the meta-analysis of SCZ data including both European and East Asian samples suggesting its potential relevance and impact in psychiatric disorders across diverse populations.¹⁵ Additionally, the 1q25.2 locus (rs12031894) was identified as an expression quantitative trait locus (eQTL) for neighboring genes across multiple tissues (Supplementary table 2), particularly in brain tissues for genes ABL2, FAM20B, and TDRD5 (Supplementary figure 2).

Table 1.

Meta-analysis between depression and SCZ in East Asians using PLINK

SNP	CHR	BP	A1	A2	Depression		Schizophrenia		Meta		Genes	Annotation
					OR	P Value	OR	P Value	OR	P Value
rs12031894	1	179270314	T	C	0.948	0.001	0.928	1.24E-07	0.937	8.09E-10	SOAT1	New
rs13016665	2	57995348	A	C	1.058	4.11E-04	1.077	3.25E-06	1.067	7.03E-09	VRK2	Known
rs58736086	18	77645575	T	C	1.066	0.003	1.097	2.30E-07	1.084	1.28E-08	SLC66A2	Known

Genetic Correlation Between Depression and SCZ

To further investigate the shared genetic mechanism between depression and SCZ in East Asian populations, we estimated the genetic correlation between the two psychiatric disorders using LDSC. In consistent with findings in European populations,^5,15,25 we observed a significant positive genetic correlation between depression and SCZ in East Asian populations (r_g = 0.46, P = 1.48 × 10⁻⁶). Importantly, the LDSC intercept was close to zero (intercept = 0.003, se = 0.0048), suggesting minimal population stratification and sample overlap. In addition, heritability estimates on the observed scale using GWAS summary statistics of East Asians were 47.05% for SCZ which is comparable to the estimates calculated from Europeans. However, in the case of depression, heritability estimates among East Asians were notably lower at 0.78%, likely attributed to the relatively limited sample size of cases as compared to European studies.

Multitrait Analysis Between Depression and SCZ

We next applied MTAG to detect the potential pleiotropic loci shared between depression and SCZ in East Asians. MTAG differs from conventional meta-analysis in its capacity to enhance the statistical power of GWAS by incorporating information from effect estimates across genetically correlated traits, and generate trait-specific associations of each single nucleotide polymorphism (SNP). In assessing the augmented detection power, we compared the average χ² test statistic for depression or SCZ derived from the multitrait GWAS with that originating from the initial GWAS conducted in East Asian populations. The mean χ² statistics for the initial GWAS results are: χ²_GWAS-DEP = 1.027 and χ²_GWAS-SCZ = 1.265 and the mean χ² statistics for the MTAG-GWAS results are: χ²_MTAG-DEP = 1.052 and χ²_MTAG-SCZ = 1.267. We observed varying degrees of increase in the effective sample size for both depression and SCZ. Specifically, the effective sample size for SCZ (MTAG-SCZ) increased slightly from 58 140 to 58 759, while the effective sample size for depression (MTAG-DEP) experienced a substantial estimated increase of 94.61%, expanding from 194 548 to 378 607. In alignment with the conventional meta-analysis conducted using PLINK, the newly identified signal at 1q25.2 and the known signal at 18q23 were observed in both the results of the MTAG-DEP and MTAG-SCZ analyses (Supplementary table 3, Supplementary figure 1). We also detected one additional locus at 2q33.1 (rs17590956), which was previously reported in GWAS studies of SCZ in East Asians (Supplementary table 3, Supplementary figure 1). Additionally, we checked the significance of locus 2p16.1 (VRK2, table 1) in the MTAG results (rs13016665, P_MTAG-DEP = 6.42 × 10⁻⁸, P_MTAG-SCZ = 6.38 × 10⁻⁷) and locus 2q33.1 (FTCDNL1, Supplementary table 3) in traditional meta-analysis results (rs17590956, P_meta = 6.67 × 10⁻⁵). Although the P values of these results are greater than 5 × 10⁻⁸, they still show a trend towards significance. This suggests that the results from both methods are strongly correlated and provide complementary insights, rather than producing random significant findings.

Genetic Correlations Between Psychiatric Disorders and Deep Phenotypes

To further elucidate the genetic correlations between psychiatric disorders and potential risk factors in East Asians, we analyzed GWAS data of 96 phenotypes sourced from BBJ, ranging from diseases, biomarkers and medication usage. The heritability z-score of each phenotype was shown in Supplementary table 1, phenotypes with z-score below 3 were deemed unreliable, leading to the exclusion of 19 phenotypes from subsequent analyses. Our analysis uncovered significant positive genetic correlations between depression and obesity-related traits. This encompassed variables like weight (r_g = 0.16, P = .009), body mass index (BMI, r_g = 0.24, P = 8.0 × 10⁻⁴), type 2 diabetes (T2D, r_g = 0.24, P = .003), and the use of medications associated with diabetes (r_g = 0.19, P = .022). Besides, serum alkaline phosphatase levels also exhibited a significant positive genetic correlation with depression. Conversely, our analysis unveiled significant negative genetic correlations between depression and two blood biomarkers: high-density lipoprotein (HDL) cholesterol and total bilirubin levels (Supplementary table 4, figure 1A). Likewise, we detected significant negative correlations between SCZ and certain blood lipid markers, particularly HDL (r_g = −0.13, P = .006) and total cholesterol levels (TCL, r_g = −0.12, P = .018). In addition, significant negative genetic correlations were also detected between SCZ and Chronic hepatitis C infection. In contrast, SCZ was significantly correlated with heart diseases such as angina pectoris. We also found significant genetic correlations between SCZ and the use of various medications, including those for Beta blocking agent use measurement (r_g = 0.19, P = .003), Antithrombotic agent use measurement (r_g = 0.16, P = .001), among others (Supplementary table 4, figure 1B).

Fig. 1.

Genetic correlation network between psychiatric disorders and risk factors. (A) Genetic correlation network of depression. (B) Genetic correlation network of SCZ. Genetically correlated traits and diseases are clustered close together. Each circle represents a trait, and each edge represents a significant genetic correlation (P < .05). Positive and negative genetic correlations are indicated by color according to the key. Note: SCZ, schizophrenia; SAP (biomarker), serum alkaline phosphatase levels; TBL, total bilirubin levels; T2D, type 2 diabetes; DD, drugs used in diabetes use measurement; TCL, total cholesterol levels; AP, angina pectoris; CHC, chronic hepatitis C infection; SAP (disease), stable angina pectoris; AARS, agents acting on the renin-angiotensin system use measurement; AnUM, antihypertensive use measurement; AAUM, antithrombotic agent use measurement; BBAU, beta blocking agent use measurement; CCBU, calcium channel blocker use measurement; HCRI, HMG CoA reductase inhibitor use measurement; Asum, aspirin use measurement; VUCD, vasodilators used in cardiac diseases use measurement.

Local Genetic Correlations Between Psychiatric Disorders and Corresponding Risk Factors

We next conducted a genome-wide scan to identify specific genomic regions that contribute to the shared heritability of genetically correlated traits. Despite the absence of statistically significant regions following Bonferroni correction, we observed that 12q23-q24 (P_local = 0.0019) and 16q21(P_local = 0.0024) displayed suggestive correlations between depression and HDL (Supplementary figure 3A). Additionally, 12q23 (P_local = 0.0012) exhibited a suggestive correlation between SCZ and HDL (Supplementary figure 3B).

Multitrait Analysis of Psychiatric Disorders and Corresponding Risk Factors

The insights garnered from our genetic correlation analysis served as a catalyst for a more in-depth exploration, aiming to pinpoint shared pleiotropic genetic loci that influence both psychiatric disorders and associated risk factors. Here, we conducted MTAG analyses for each genetically correlated trait pair, and prioritized SNPs that reached genome-wide significance (P < 5 × 10⁻⁸). We have provided details of quality control process for on the summary statistics in Supplementary table 5, which includes the number of SNPs in the original GWAS data, the number of SNPs included in the MTAG analysis, and the proportion of SNPs analyzed compared to the total number of SNPs in the original psychiatric disorder GWAS dataset. It can be observed that 70%–80% of the loci participated in the MTAG analysis, which falls within a reasonable range. Additionally, the maxFDR for all phenotypes included in the MTAG analysis was shown in Supplementary table 8. While a few phenotypes exhibit higher maxFDR values, the majority fall within acceptable ranges (maxFDR < 0.05). Additionally, we further excluded the signals in MTAG with P > .01 in the original GWAS to avoid false positives.

We identified two significant signals in the analysis of depression and BMI, neither of which had been previously reported in GWAS studies of depression within East Asian populations. Among them, locus at 19q13.32 near GIPR was also detected in the analysis of depression and weight. The lead SNP rs4802269 serves as an eQTL for the nearby gene EML2 in tissues from cerebellar hemispheres. Furthermore, it functions as an eQTL for the nearby genes DMPK, DMWD, and GIPR in adipose tissues (Supplementary table 2), implying the potential relevance of this locus in regulating factors associated with both depression and weight. Additionally, in our exploration of the connection between depression and type 2 diabetes (T2D), we uncovered three shared loci, none of which had been previously identified in studies of depression within East Asian populations. Among these, the variant rs10974438 at 9p24.2, located within an intronic region of GLIS3, has been linked to mania, hypomania, bipolar disorder, or manic-depression in an analysis of UK Biobank data (UKB Neale v2, Supplementary table 6). Furthermore, rs11074451 at 16p12.3 serves as a splicing quantitative trait locus (sQTL, P = 1 × 10⁻⁹) for the nearby gene GP2 in the pancreas according to GTEx, suggesting its potential role in the interplay between depression and T2D. In addition to the above three loci, a novel locus at 18p11.22 was identified when analyzing depression and the measurement of diabetes medication usage. Besides, we identified two novel loci at 12q24.12 (rs11066015) and 16q12.2 (rs35626903) during the analysis of depression and HDL. The locus at 12q24.12 (rs11066015) is specific to East Asian populations and encompasses the pivotal ALDH2 gene, which has been implicated in various metabolic quantitative traits.^19,26–28 The rs35626903 variant, located at 16q12.2 within the intron of the AMFR gene, has been previously related to psychological or psychiatric problem in the study of Europeans.²⁹ We also identified two novel pleiotropic signals between depression and total bilirubin levels (table 2, Supplementary figure 4).

Table 2.

Multitrait meta-analysis between depression and risk factors in East Asians

Risk Factors	SNP	CHR	BP	A1	A2	Trait1		Trait2		MTAG		Genes	Annotation
						Beta	P Value	Beta	P Value	Beta	P Value
BMI	rs13029479	2	655222	A	G	0.101	0.007	−0.052	5.40E-17	0.035	1.31E-09	TMEM18	New
	rs4802269	19	46167469	G	A	−0.07	4.94E-04	0.042	4.60E-16	−0.021	5.01E-11	GIPR	New
Weight	rs4802269	19	46167469	G	A	−0.07	4.94E-04	0.037	4.30E-17	−0.021	1.25E-08	GIPR	New
HDL	rs11066015	12	112168009	A	G	−0.073	0.001	−0.122	1.60E-109	−0.033	8.23E-21	ALDH2	New
	rs35626903	16	56431613	G	A	−0.071	6.24E-06	−0.028	2.50E-09	−0.018	2.52E-09	AMFR	New
TBL	rs12993249	2	234479377	A	G	−0.055	1.40E-03	−0.123	1.60E-208	−0.044	2.55E-46	USP40	New
	rs10770745	12	20986617	T	G	−0.067	8.73E-04	−0.031	2.60E-13	−0.018	9.50E-09	SLCO1B3-SLCO1B7	New
	rs11066015	12	112168009	A	G	−0.073	1.47E-03	−0.035	1.30E-15	−0.02	4.66E-09	ALDH2	New
Type 2 diabetes	rs10013800	4	71731236	T	C	−0.092	4.21E-05	0.081	2.21E-10	−0.023	1.19E-11	GRSF1	New
	rs10974438	9	4291928	A	C	0.045	0.005	0.078	2.35E-16	0.017	4.95E-12	GLIS3	New
	rs11074451	16	20336257	A	G	0.1	2.27E-04	0.103	9.12E-11	0.026	5.61E-11	GP2	New
Drugs used in diabetes use measurement	rs6850072	4	71725868	A	G	0.074	2.37E-04	−0.086	1.14E-12	0.022	1.25E-11	GRSF1	New
	rs10974438	9	4291928	A	C	0.045	0.005	0.072	1.91E-12	0.016	1.64E-09	GLIS3	New
	rs11074451	16	20336257	A	G	0.1	2.27E-04	0.114	2.00E-11	0.028	5.46E-11	GP2	New
	rs662872	18	8881068	T	C	−0.084	1.80E-06	0.038	7.27E-04	−0.016	2.02E-08	MTCL1	New

Note: The sections labeled “Trait 1” and “Trait 2” display the statistical results from the original GWAS for the traits involved in the MTAG analysis, MTAG section of the table presents the MTAG results specific to Trait 1. Trait1, depression; Trait2, risk factors.

We next conducted a MTAG analysis to uncover pleiotropic variants impacting both SCZ and corresponding risk factors. We identified a total of seven significant loci shared between SCZ and HDL including six located in loci reported in SCZ GWAS in East Asian populations. The novel locus at 5q31.3 was also identified in the analysis between SCZ and TCL, as well as the analysis between SCZ and stable angina pectoris (table 3, Supplementary figure 5). Notably, this locus is sited close to the protocadherin gamma gene cluster, which may play a critical role in the establishment and function of specific cell-cell connections in the brain.^30–33 Particularly, the lead SNP rs11956411 is an eQTL of PCDHGA3 in brain cortex (Supplementary table 2). Additionally, we detected four significant loci shared between SCZ and chronic diseases (Chronic hepatitis C infection and Chronic sinusitis), but all of them were reported in GWAS of SCZ conducted on East Asians (Supplementary table 7, Supplementary figures 7 and 8). In the analysis of SCZ and the use of medications, we uncovered three novel loci at 1q25.2 (rs61824368), 4q13.1 (rs62297721), and 8p21.2 (rs117325001) (table 3, Supplementary figure 5). The 1q25.2 and 8p21.2 loci had been reported in the meta-analysis of SCZ, including both European and East Asian populations.¹⁵

Table 3.

Novel loci in the multitrait meta-analysis between SCZ and risk factors in East Asians

Risk Factors	SNP	CHR	BP	A1	A2	Trait1		Trait2		MTAG		Genes	Annotation
						Beta	P Value	Beta	P Value	Beta	P Value
HDL	rs11956411	5	140853856	C	T	−0.097	1.08E-07	−0.02	1.40E-03	−0.035	2.77E-08	PCDHGC3	New
TCL	rs11956411	5	140853856	T	C	0.097	1.08E-07	0.005	1.00E-03	0.036	1.54E-08	PCDHGC3	New
Stable angina pectoris	rs2237079	5	140877905	A	G	−0.085	1.24E-07	0.013	1.94E-03	−0.033	4.65E-08	PCDHGC3	New
Antihypertensive use measurement	rs62297721	4	66504803	A	G	0.074	6.44E-08	0.019	4.81E-03	0.033	1.78E-08	EPHA5	New
Antithrombotic agent use measurement	rs117325001	8	26242272	T	G	0.074	6.62E-08	0.008	8.98E-03	0.033	2.49E-08	BNIP3L	New
Aspirin use measurement	rs117325001	8	26242272	T	G	0.074	6.62E-08	0.008	9.52E-04	0.033	2.44E-08	BNIP3L	New
Vasodilators used in cardiac diseases use measurement	rs61824368	1	179298809	T	C	−0.075	1.19E-07	0.013	2.51E-03	−0.033	3.26E-08	SOAT1	New

Note: The sections labeled “Trait 1” and “Trait 2” display the statistical results from the original GWAS for the traits involved in the MTAG analysis, MTAG section of the table presents the MTAG results specific to Trait 1. Trait1, schizophrenia; Trait2, risk factors.

Association Analysis Based on Subsets for Pleiotropic Loci Identified in MTAG

We further conducted two-sided ASSET for each pleiotropic locus identified in MTAG of psychiatric disorders and risk factors. As shown in Supplementary tables 9 and 10, each row corresponds to a SNP, and the corresponding P values for overall association (including component-wise P values for two-sided search) and names of phenotypes included in the best-subset among the positively or negatively associated subsets respectively. In the ASSET results for pleiotropic loci in MTAG between depression and corresponding risk factors, the best-subset of three pleiotropic loci comprises depression, with the specific loci being rs10770745, rs35626903, and rs4802269 (Supplementary table 9). Interestingly, the best-subset of rs4802269 including depression, type 2 diabetes and drugs used in diabetes use measurement, which was also detected in the MTAG between depression and BMI or the MTAG between depression and weight. For SCZ, all the pleiotropic loci identified in MTAG relating to SCZ and its associated risk factors have been validated in ASSET, with the optimal subsets consistently including SCZ (Supplementary table 10).

Discussion

To our current understanding, this is the first genome-wide multitrait analysis investigating genetic correlations, pleiotropic loci and shared genetic mechanisms of depression and SCZ within East Asian populations. Leveraging the known genetic correlation between these two psychiatric disorders, we pinpointed a novel locus via the inverse-variance weighted meta-analysis, and further validated it using MTAG. Compared to traditional meta-analysis and other multitrait methods, MTAG employs robust techniques to account for population stratification and sample overlap, minimizing the risk of false positives. Additionally, MTAG integrates smoothly with standard GWAS summary data formats and has been widely adopted in genetic studies.^34–38 To explore the potential risk factors for psychiatric disorders among East Asian populations, we conducted an in-depth examination of 96 distinct phenotypes from BBJ. This extensive exploration identified several new risk loci for SCZ and depression, revealing distinctive genetic underpinnings in East Asian populations.

In previous research by the Psychiatric Genomics Consortium (PGC), conventional meta-analysis identified pleiotropic loci associated with psychiatric disorders, with a primary focus on European populations.^6,39 However, recent progress in GWAS focused on SCZ and depression has led to a substantial increase in data concerning East Asian populations, motivating us to initiate research endeavors focused exclusively on East Asian populations. Our meta-analysis of depression and SCZ identified a novel locus at 1q25.2 shared between depression and SCZ in East Asians. The lead SNP, rs12031894, has been identified as an eQTL influencing multiple nearby genes across various tissues. Notably, in a majority of brain tissues, rs12031894 functions as an eQTL for ABL2. ABL2 encodes a nonreceptor tyrosine protein kinase from the Abelson family, crucial for neurulation and neurite outgrowth. These kinases act as intermediaries, transmitting signals from axon guidance cues and growth factor receptors to facilitate cytoskeletal rearrangements. In mature neurons, Abl kinases are situated in both pre- and postsynaptic compartments, contributing to the regulation of synaptic stability and plasticity. Given that this locus exhibits an association with SCZ in European populations as well (figure 2A), it suggests that the candidate gene ABL2 could exerts a widespread influence on SCZ, affecting both East Asians and Europeans. In contrast, the evidence from regional plot (figure 2B) suggests that this locus might be exclusively linked to depression in East Asians but not in Europeans, implying the existence of a complex regulatory mechanism influencing depression across different populations. We then utilized MTAG to identify shared pleiotropic loci between depression and SCZ in East Asians. As expected, the aforementioned 1q25.2 locus appeared in both MTAG-DEP and MTAG-SCZ results, demonstrating the reliability of this novel method.

Fig. 2.

Co-localization for psychiatric disorders in East Asians and Europeans at 1q25.2. Left panel showed the merged association plot for psychiatric disorder in East Asians and Europeans, and right panel showed the regional plots for East Asians and Europeans, respectively. (A) The regional plots for SCZ and points are colored by LD with respect to rs12031894, which is labeled with a diamond. (B) The regional plots for depression and points are colored by LD with respect to rs12031894, which is labeled with a diamond.

Besides correlations among psychiatric disorders,⁴⁰ LDSC analysis has consistently reported genetic correlations between psychiatric disorders and other diseases or risk factors in European populations.^41,42 Therefore, we extended our investigations to East Asian populations by analyzing GWAS data encompassing 96 phenotypes from the BBJ dataset. Our analysis revealed positive genetic correlations between depression and obesity-related traits like BMI and T2D, matching epidemiological studies.^43–46 Intriguingly, serum alkaline phosphatase levels also displayed a positive genetic correlation with depression, consistent with findings by Li et al of higher alkaline phosphatase levels in individuals with depression compared to controls.⁴⁷ Furthermore, recent research on US adults has indicated a significant association between higher alkaline phosphatase levels and an increased risk of depression.⁴⁸ Conversely, we identified negative genetic correlations between depression and two blood biomarkers: HDL and bilirubin. Low levels of HDL have been extensively studied in connection with depression.^49–51 Notably, Liang et al reported that low HDL levels may be linked to more severe depressive symptoms in East Asians.⁵² Additionally, lower HDL levels have been associated with SCZ and psychosis in various studies.^53–55 Consistent with this, we observed negative genetic correlations between HDL and SCZ. In contrast, positive genetic correlations were noted with heart diseases such as angina pectoris, aligning with the observation that cardiovascular diseases are prevalent among SCZ patients.^56–58 These genetic correlations provide valuable insights into the intricate interplay among depression, SCZ, metabolic factors, and various health conditions in East Asian populations, guiding further research and facilitating the identification of shared pleiotropic loci influencing both psychiatric disorders and related risk factors.

We next performed MTAG analysis on depression, SCZ, and their associated traits in East Asian populations. We uncovered ten novel pleiotropic loci for depression through MTAG analysis, linked to traits such as BMI, weight, HDL, TBL, T2D, and diabetes medications. Seven of these loci exhibited associations with psychiatric traits in Europeans, including four acting as eQTLs, implying their potential regulatory role in gene expression. For instance, the lead SNP rs4802269 of the 19q13.32 locus served as an eQTL for the nearby gene EML2 in cerebellar hemispheres and for genes DMPK, DMWD, and GIPR in adipose tissues, hinting at its role in depression and weight regulation. Additionally, 18p11.22 was identified in the analysis of depression and diabetes medication usage. This locus is near the MTCL1 gene, which plays a vital role in maintaining Purkinje neuron axon initial segments.⁵⁹ Furthermore, among the remaining genetic loci, the 12q24.12 locus is specifically presented in East Asian populations. Notably, the lead variant rs11066015 within this region exhibits a high level of LD with a missense variant of gene ALDH2 (rs671, R² = 0.95). Our analysis reveals that the MTAG P value for rs671 (P = 4.46 × 10⁻²⁰) is slightly greater than that of rs11066015 (P = 8.23 × 10⁻²¹), suggesting that the 12q24.12 locus is likely influenced by rs671 in East Asians. Subsequently, we performed GCTA-COJO analyses⁶⁰ for rs11066015, while adjusting for rs671. The conditional analysis reveal that the MTAG P value for rs11066015 became insignificant (P_Conditional = 0.812176), implying that its initial association with depression and HDL levels in the MTAG analysis could potentially be entirely explained by rs671. This is of particular interest because the rs671 variant leads to a loss of ALDH2 enzymatic activity, resulting in adverse reactions to acetaldehyde, which effectively reduce alcohol consumption.⁶¹

In the investigation of SCZ and its related traits, we uncovered four new genetic loci, one of which was the previously mentioned 1q25.2 locus, shared between depression and SCZ. Notably, the 5q31.3 locus emerged consistently in the analysis of SCZ and its associated traits, which included HDL, TCL, and Stable angina pectoris. This locus is situated in close proximity to the protocadherin gamma gene cluster, which plays a potentially vital role in establishing and maintaining specific cell-cell connections in the brain. Particularly, the lead SNP rs11956411 acts as an eQTL for PCDHGA3 in the brain cortex. Moreover, the 4q13.1 locus is located around the EPHA5 gene, known for its crucial involvement in brain development and synaptic plasticity. This locus may also be linked to brain disorders such as bipolar⁶² and Dravet syndrome.⁶³ Similar to the previously mentioned 1q25.2 locus, the 8p21.2 locus centered around the BNIP3L gene surpassed genome-wide significance in the meta-analysis of SCZ GWAS from both European and East Asian populations. Further investigation of both common and rare genetic variants indicated that BNIP3L might serve as a susceptibility gene.⁶⁴ However, it is important to note that eQTL data from GTEx suggests that genotypes of the lead SNP rs117325001 are significantly associated with the expression level of SDAD1P1 in multiple brain tissues, implying a potential role of SDAD1P1 in SCZ.

Limitations

There are several limitations in our study. Firstly, due to limited data in East Asians, our analysis covered only depression and SCZ. With more comprehensive datasets, we will have the opportunity to investigate a broader spectrum of psychiatric disorders in East Asians. Secondly, in exploring trait associations with psychiatric disorders and identifying pleiotropic loci, we tested a broad spectrum of phenotypes. Applying the Bonferroni correction would eliminate many potentially associated traits,⁶⁵ particularly those with small sample sizes. Therefore, we used a more relaxed threshold of 0.05 without multiple testing correction. Thirdly, the genetic correlations between the traits in our study were modest. Despite this, we used MTAG for follow-up analysis because it effectively leverages these genetic correlations to improve the power of identifying pleiotropic loci. Fourthly, since our analysis relied on publicly accessible GWAS data, eliminating sample overlap among different research cohorts is challenging. In light of this, we used MTAG settings to manage and minimize potential overlaps. Finally, the eQTL data utilized in this study primarily stem from non-East Asian populations, predominantly Europeans. Therefore, the observed cis-eQTL effects should be further validated in East Asians once such data becomes available. This verification will enhance the reliability and generalizability of our findings.

Conclusion

In conclusion, our study advances our understanding of the genetic mechanisms of depression and SCZ within East Asian populations by introducing 14 previously undiscovered pleiotropic risk loci to the existing pool of genetic contributors to psychiatric disorders. It is evident from this study that additional genetic loci associated with psychiatric disorders in East Asian populations can be identified through standard GWAS or MTAG analysis with larger sample sizes. This study warrants further investigation to explore the genetic intricacies of mental health conditions within this specific population.

Supplementary Material

Supplementary material is available at https://academic.oup.com/schizophreniabulletin/.

Conflict of Interest Statement

The authors have declared that there are no conflicts of interest in relation to the subject of this study.

Funding

This work is supported by the Special Funds of Taishan Scholar Project, China (tsqn202211224), the National Natural Science Foundation of China (32270661), Excellent Youth Science Fund Project (Overseas) of Shandong China (2023HWYQ-082), Shandong Postdoctoral Science Foundation (SDCX-ZG-202400042), and Shandong Province Higher Education Institution Youth Innovation and Technology Support Program (2023KJ179).