Skip to main content
NCBI home page
Nature Communications logoLink to Nature Communications
. 2026 Mar 15;17:4006. doi: 10.1038/s41467-026-70774-9

Multitrait GWAS and functional validation reveal genetic loci for gastric cancer

Huanxin Ding 1,2,3,4,#, Chuxuan Liu 1,2,3,4,#, Qing Sun 5,#, Yue Jiang 6, Qian Xu 1,2,3,4, Zhenguo Zhu 1,2,3,4,7, Xin Jin 1,2,3,4,8, Qibo Huang 9, Yue Li 1,2,3,4, Yingchao Song 6, Zenglin Liu 1,2,3,4, Tianming Yu 10, Bowen Shi 10, Delin Tan 11, Linzehao Li 6, Frank Mentch 12, Joseph Glessner 12, Xiao Chang 6,13,✉, Hakon Hakonarson 12,✉, Guangyong Zhang 1,2,3,4,✉
PMCID: PMC13136404  PMID: 41833930

Abstract

Gastric cancer is a leading cause of cancer-related death, particularly in East Asia, yet its genetic basis remains incompletely understood. Using genome-wide association study data from BioBank Japan, we show that gastric cancer shares significant genetic associations with 25 other phenotypes, most notably cardiovascular conditions such as angina pectoris and myocardial infarction. Mendelian randomization supports a causal role of these traits in gastric cancer risk. Multitrait analysis of GWAS reveals a novel pleiotropic locus at 12q22 (rs12814712) that is shared between gastric cancer and cardiovascular traits. Functional experiments demonstrate that the risk allele at this locus reduces transcriptional activity and downregulates the expression of nearby genes, particularly VEZT. In gastric cancer cells, overexpression of VEZT and NR2C1 suppress proliferation and migration, whereas their knockdown promotes malignancy. Our analysis identifies rs12814712 as a novel susceptibility locus for gastric cancer and underscores its regulatory role in disease-related gene networks.

Subject terms: Medical genomics, Genome-wide association studies

Multi-trait genetic analysis in East Asians identifies a novel gastric cancer risk locus shared with cardiovascular traits and shows that this variant modulates regulation of nearby genes, including VEZT, with functional effects on gastric cancer cell behaviour.

Introduction

According to global cancer statistics for 2022, gastric cancer ranks fifth in terms of both incidence and mortality worldwide, accounting for 968,350 new cases and 659,853 deaths, respectively1. The incidence of gastric cancer varies regionally, with higher rates in East Asia and Eastern Europe and lower rates in Northern Europe and North America1, which are likely influenced by environmental, lifestyle and genetic factors. Empirical studies have shown that gastric cancer is the result of a complex interaction of genetic and environmental risk factors2. Risk factors encompass a spectrum ranging from Helicobacter pylori (HP) infection to dietary habits such as the consumption of pickled foods, as well as behaviours such as alcohol consumption and tobacco use. Among these, HP infection emerges as the most influential determinant. Moreover, genetic studies, including twin and genome-wide association studies (GWASs), have revealed a substantial genetic contribution to gastric cancer susceptibility, with heritability estimates of approximately 20.9% in the Chinese population and 22% in Nordic twins3,4.

Gastric cancer displays substantial polygenicity, with East Asian populations showing the identification of at least 15 risk loci through GWAS. For instance, in a two-stage GWAS in Japan, a notable association was found between an intronic SNP (rs2976392) within the PSCA gene and diffuse-type gastric cancer5. Similarly, researchers at Nanjing Medical University reported genetic variations at loci 1p35.2 and 6p22.1 in the Chinese population, which correlated with increased gastric cancer risk6. Moreover, they elucidated the roles of PRKAA1 at the 5p13.1 locus and NOC3L at the 10q23.33 locus in gastric tumour pathogenesis7. In contrast, only the 11q22.3 locus was identified in Europeans8. Furthermore, several genes within these risk loci, such as MUC1 (1q22), SPOCD1 (1p35.2), PSCA (8q24.3), PRKAA1 (5p13.1), BTN3A2 (6p22.1) and NOC3L (10q23.33), have been functionally validated as susceptibility genes for gastric cancer5–7,9.

Recent advances in statistical methodologies for GWAS analyses have made it possible to conduct joint analyses of traits that share genetic architecture10. Through meta-analyses of GWAS summary statistics for genetically correlated traits, multitrait analysis of GWAS (MTAG) not only enhances statistical power but also identifies pleiotropic genetic variants that influence interconnected traits. This cross-trait analysis approach has proven successful in numerous cancer research studies. Additionally, it has been applied to the investigation of seven upper gastrointestinal disorders11. However, to the best of our knowledge, no multitrait GWAS analysis specifically targeting gastric cancer has been conducted to date.

In this study, we identified multiple novel pleiotropic loci associated with gastric cancer using GWAS data from the BioBank Japan (BBJ)12. Notably, we observed that several genes adjacent to these loci might be involved in gastric cancer pathogenesis, as documented in the literature. We collected gastric cancer samples and conducted functional experiments to elucidate the regulatory relationships between the identified loci and their neighbouring genes, as well as to explore the potential roles of candidate genes in the aetiology of gastric cancer.

Results

In this study, we utilized GWAS data from BBJ and identified multiple new pleiotropic loci associated with gastric cancer using MTAG. Notably, we observed that several genes near the novel 12q22 locus might be involved in the pathogenesis of gastric cancer. Consequently, we collected clinical gastric cancer samples and conducted functional experiments to elucidate the potential regulatory relationships between the identified loci and their neighbouring genes and to explore the potential roles of candidate genes in the aetiology of gastric cancer. An overview of our study is provided in Fig. 1.

Fig. 1. Overview of the study design and conduct.

Fig. 1

Open in a new tab

Genetic correlation analysis across 95 phenotypes from BioBank Japan identified traits genetically associated with gastric cancer. Multi-trait GWAS revealed a pleiotropic locus at 12q22 (rs12814712), which was further investigated through eQTL analysis in gastric cancer tissues and functional assays in gastric cancer cell lines to identify VEZT as a candidate target gene.

Genetic correlations between gastric cancer and associated phenotypes

We analysed GWAS data from BBJ to explore the genetic correlation between gastric cancer and 95 phenotypes, including diseases, biomarkers and medication usage, in East Asians. Following multiple testing correction, our analysis revealed a significant positive genetic correlation between gastric cancer and 25 cardiovascular-related traits, including diseases like angina pectoris (AP, rg = 0.4611, PFDR = 0.0001), myocardial infarction (MI, rg = 0.3415, PFDR = 0.0009), blood pressure measurements such as diastolic blood pressure (DBP, rg = 0.2549, PFDR = 0.0028), systolic blood pressure (SBP, rg = 0.2421, PFDR = 0.001) and mean arterial pressure (MAP, rg = 0.2612, PFDR = 0.0009), as well as the use of cardiovascular drugs like calcium channel blockers (CCBUM, rg = 0.3561, PFDR = 4.56 × 10−7), renin-angiotensin system inhibitors (AARSUM, rg = 0.4309, PFDR = 1.41 × 10−8), HMG-CoA reductase inhibitors (HCRIUM, rg = 0.3786, PFDR = 1.82 × 10−6) and salicylic acid and derivatives (SADUM, rg = 0.4115, PFDR = 0.0002). Additionally, positive genetic correlations were identified between gastric cancer and other cancer types, including lung cancer (LC, rg = 0.7442, PFDR = 0.0058) and colorectal cancer (CRC, rg = 0.3043, PFDR = 0.002) (Supplementary Table S1 and Fig. 2).

Fig. 2. Genetic correlation network between gastric cancer and associated phenotypes.

Fig. 2

Open in a new tab

Genetic correlation networks across gastric cancer and associated phenotypes. Each circle within the network represents a disease or trait, and the edges depict significant genetic correlations (PFDR < 0.05), with positive correlations in blue and negative correlations in red. Thicker edges on the network correspond to a lower false discovery rate (FDR). GC gastric cancer, AP angina pectoris, CHF chronic heart failure, CRC colorectal cancer, LC lung cancer, MI myocardial infarction, SAP stable angina pectoris, T2D type 2 diabetes, ASAT aspartate aminotransferase levels, CRP C-reactive protein, DBP Diastolic blood pressure, GGT gamma glutamyl transpeptidase, Glu glucose levels, MAP mean arterial pressure, PP pulse pressure, SBP systolic blood pressure, SUA serum uric acid levels, DDUM drugs used in diabetes use measurement, AAUM antithrombotic agents use measurement, VDCDUM vasodilators used in cardiac diseases use measurement, DUM diuretics use measurement, BBAUM beta blocking agents use measurement, CCBUM calcium channel blocker use measurement, AARSUM agents acting on the renin-angiotensin system use measurement, HCRIUM HMG CoA reductase inhibitor use measurement, SADUM salicylic acid and derivatives use measurement.

Causal impact of cardiovascular-related factors on gastric cancer

To further investigate the relationship between cardiovascular traits and gastric cancer, we conducted Mendelian randomization (MR) analysis. The inverse-variance weighted (IVW) method revealed that after false discovery rate (FDR) correction, eight cardiovascular-related factors were significantly associated with an increased risk of gastric cancer, including AP (OR 1.289, 95% CI 1.060–1.568; PFDR = 0.0192), stable angina pectoris (SAP, OR 1.408, 95% CI 1.243–1.596; PFDR = 5.56 × 10−7) and MI (OR 1.179, 95% CI 1.094–1.270; PFDR = 5.36 × 10−5). Additionally, antithrombotic agents (AAUM, OR 1.829; 95% CI 1.343–2.493; PFDR = 3.61 × 10−4), beta-blocking agents (BBAUM, OR 1.478, 95% CI 1.108–1.972; PFDR = 0.0159), SADUM (OR 1.799, 95% CI 1.412–2.293; PFDR = 9.61 × 10−6), CCBUM (OR 1.719, 95% CI 1.203–2.456; PFDR = 0.00679) and HCRIUM (OR 1.368, 95% CI 1.266–1.477; PFDR = 2.52 × 10−14) were used (Supplementary Data 1). Sensitivity analyses further confirmed the majority of these associations (Supplementary Data 1). Additionally, the MR‒Egger intercept test suggested potential horizontal pleiotropy for HCRIUM (PFDR for intercept <0.05). Conversely, the IVW analysis demonstrated that gastric cancer might increase the risk of 13 cardiovascular-related factors, with the exception of vasodilators used in cardiac disease use measurement. Detailed results from the five MR analysis methods are presented in Supplementary Data 1.

Cell type-specific enrichment of SNP heritability

In our analysis of 96 phenotypes, including gastric cancer, we further partitioned SNP heritability on the basis of six chromatin marks and nine cell types. We found that gastric cancer was enriched across all six chromatin marks, specifically in stomach tissues or cell types. The enrichment patterns of cardiovascular-related traits, such as blood pressure, were similar across all the chromatin markers, predominantly in the kidney, adrenal glands, heart, atria and aorta. Additionally, we observed significant genetic enrichment for diabetes-related traits in gastrointestinal tissues (such as the small intestine, stomach and colon) and the pancreas. Interestingly, angina pectoris and MI displayed strong similarities only in the H3K9AC chromatin mark, with enrichment in cardiovascular-related tissues or cell types such as the right ventricle, right atrium and left ventricle (Supplementary Fig. S1).

Multitrait analysis of gastric cancer and corresponding risk factors

Motivated by the findings from genetic correlation analyses, we conducted a comprehensive investigation into potential genetic factors shared between gastric cancer and its genetically correlated traits, aiming to identify novel susceptibility loci associated with gastric cancer risk. Guided by the 25 traits significantly correlated with gastric cancer (LDSC PFDR < 0.05), we performed a joint analysis using MTAG to identify SNPs that reached genome-wide significance (P < 5 × 10−8). Our analysis revealed 76 signals among gastric cancer and 25 traits in East Asians, including 41 previously reported loci, 26 novel findings and 9 located within the MHC region (Supplementary Data 2). Notably, a novel locus at 12q22 (rs12814712) was identified in the MTAG analyses that linked gastric cancer with AP, MI and SADUM (Fig. 3A–C). Considering that 25 phenotypes were analysed—albeit not fully independent—we additionally applied a Bonferroni-corrected threshold to account for multiple trait testing (P < 2 × 10⁻⁹). Under this stricter criterion, rs12814712 remained significantly associated with gastric cancer and MI. To further investigate this genomic region, we employed LocusZoom13 to visualize the genetic associations, which highlighted a distinct signal surrounding the NDUFA12-VEZT locus across all four traits (Fig. 4A–D). We subsequently utilized LocusCompare14 to assess the colocalization of these associated signals at 12q22, which demonstrated strong colocalization among gastric cancer, angina pectoris (AP), MI and salicylic acid and derivatives use measurement (SADUM) (Fig. 5A–C). To further validate this association, we evaluated rs12814712 in an independent GWAS in a Korean cohort. A similar direction of effect was observed (BETA = 0.066), which was consistent with the results of the GC-BBJ analysis (BETA = 0.079; P = 7.65 × 10⁻⁵) (Supplementary Data 2). However, owing to the limited number of cases (n = 378), the association did not reach statistical significance. Interestingly, a nearby SNP, rs9634181, exhibited a stronger association (BETA = 0.19, P = 0.012) and was in weak linkage disequilibrium with rs12814712 (R² = 0.1; Supplementary Fig. S2A). Similarly, in the European-ancestry Million Veteran Program (MVP) cohort (n = 1152), a nominally significant association was observed with a nearby SNP, rs183275771 (BETA = 0.081, P = 0.005), which was in linkage disequilibrium with rs12814712 (R² = 0.2; Supplementary Fig. S2B). These results suggested that increasing the sample size might yield statistically significant replication. Additionally, we have summarized previously reported susceptibility loci for gastric cancer identified through GWAS in Table 1.

Fig. 3. Manhattan plot of the GWAS meta-analysis.

Fig. 3

Open in a new tab

A Multitrait meta-analysis between gastric cancer and angina pectoris in East Asians. B Multitrait meta-analysis between gastric cancer and myocardial infarction in East Asians. C Multitrait meta-analysis between gastric cancer and medication use (salicylic acid and derivatives) in Europeans. Newly identified loci are depicted in yellow, and previously reported loci are depicted in blue. The association p values displayed are from the original GWAS derived from two‑sided Wald tests; no new statistical tests were applied for this visualization. The dashed horizontal line indicates the genome‑wide significance threshold (P = 5 × 10⁻⁸).

Fig. 4. Regional plots of the 12q22 locus associated with gastric cancer and cardiac traits using LocusZoom.

Fig. 4

Open in a new tab

A Associations in gastric cancer. B Associations in angina pectoris. C Associations in myocardial infarction. D Associations in medication use (salicylic acid and derivatives). The purple diamonds indicate the lead SNP, and the circles represent the other SNPs in the region, with colouring from the linkage disequilibrium (East Asian populations from the 1000 Genomes Project) between each SNP and the lead SNP. The y‑axis shows the −log₁₀(P value) from the original GWAS summary statistics (derived from two‑sided Wald tests). LD (r²) is shown in colour according to the scale on the left.

Fig. 5. Colocalization between gastric cancer and cardiac traits in East Asians at the 12q22 locus.

Fig. 5

Open in a new tab

A The left panel shows the merged association plot for gastric cancer and angina pectoris, while the right panel presents regional plots for gastric cancer and angina pectoris separately. B The left panel shows the merged association plot for gastric cancer and myocardial infarction, while the right panel presents regional plots for gastric cancer and myocardial infarction separately. C The left panel shows the merged association plot for gastric cancer and medication use (salicylic acid and derivatives), while the right panel presents regional plots for gastric cancer and medication use (salicylic acid and derivatives) separately. In all diagrams, points are coloured according to linkage disequilibrium (LD) with respect to rs12814712, which is marked with a purple diamond. The y‑axis shows the −log₁₀(P value) from two‑sided association tests in the original GWAS. Colocalization analysis was performed using LocusCompare, which evaluates the shared genetic signal on the basis of LD and association strength without formal statistical testing for colocalization probability.

Table. 1.

Susceptibility variants for gastric cancer identified through GWAS

Region Gene SNPa P value (GC)b P value (MTAG)c ORd Population Reference/ Sourcese
1q21 GBAP1 rs117756257-A 3.60E-19 2.31E-12 0.731 Asian MTAG-BBJ
1q21.3 EFNA3 rs3753642-C 1.12E-17 6.49E-23 0.719 Asian MTAG-BBJ
1q22 MUC1 rs4072037-A 6.28E-17 2.86E-33# 1.35 Asian; European 9
1p36.22 CASZ1 rs77487206-T 1.61E-03 6.31E-10 1.088 Asian MTAG-BBJ
2p24.1 APOB rs13306206-A 1.10E-03 1.12E-41 0.852 Asian MTAG-BBJ
3q11.2 NSUN3 rs7624041-G 4.57E-04 3.95E-01# 1.21 Asian 7
3q13.31 ZBTB20 rs9841504-C 2.00E-09 NA# 1.32 Asian 42
3q25.1 TM4SF4 rs62272016-G 6.64E-03 2.62E-09 1.098 Asian MTAG-BBJ
4p16.3 HGFAC rs16844301-G 1.26E-03 3.25E-08 1.056 Asian MTAG-BBJ
4q28.1 LOC285419, ANKRD50 rs10029005-A 4.52E-06 4.59E-05# 1.16 Asian 7
5p13.1 PRKAA1 rs13361707-C 8.00E-29 7.54E-28# 1.41 Asian; European 42
5q14.3 lnc-POLR3G-4 rs7712641-C 1.21E-11 7.53E-02# 1.19 Asian 43
6p21.1 UNC5CL rs2294693-C 2.50E-08 8.79E-05# 1.18 Asian 44
LRFN2 rs2494938-A 5.00E-09 3.17E-02# 1.18 Asian 45
6p22.3 CDKAL1 rs9356744-C 2.06E-07 2.76E-26 0.916 Asian MTAG-BBJ
6q23.2 TCF21 rs75824506-T 7.42E-05 6.49E-09 1.127 Asian MTAG-BBJ
6q25.3 DYNLT1 rs536241152-T 2.33E-03 1.06E-08 1.269 Asian MTAG-BBJ
8q24.3 PSCA rs2294008-T 1.00E-20 1.09E-37# 1.33 Asian; European 5,43
10p14 NUDT5 rs12241367-G 5.87E-03 6.47E-10# 1.05 Asian MTAG-BBJ
10q23.33 PLCE1 rs2274223-G 1.00E-20 5.41E-02# 1.57 Asian 46
NOC3L rs3781266-C 1.03E-10 5.16E-01# 1.31 Asian 7
rs3740365-T 1.00E-10 NA# 1.31 Asian 7
11q13.4 DHCR7 rs104886038-G 8.51E-06 NA# 0.19 Asian 47
P2RY2 rs11235672-C 1.48E-04 1.24E-08 1.12 Asian MTAG-BBJ
11q22.3 ATM rs758081262-T 8.00E-10 NA# 4.74 European 8
rs768362387-A 8.00E-10 NA# 4.74 European 8
rs777499935-G 8.00E-10 NA# 4.74 European 8
12q22 VEZT rs12814712-A 7.65E-05 5.16E-10 1.08 Asian MTAG-BBJ
12q24.11 FAM222A rs113082829-A 9.05E-03 2.13E-10 1.05 Asian MTAG-BBJ
12q24.12 PHETA1 rs10774621-A 4.30E-04 1.29E-19 1.07 Asian MTAG-BBJ
12q24.12 ACAD10 rs11066008-G 1.32E-03 9.42E-63 1.06 Asian MTAG-BBJ
12q24.12 BRAP rs11065734-C 8.10E-04 1.22E-10 1.06 Asian MTAG-BBJ
12q24.13 ERP29 rs11066132-T 9.73E-04 2.07E-43 1.07 Asian MTAG-BBJ
12q24.13 OAS1 rs7139283-C 7.47E-03 8.97E-11 0.939 Asian MTAG-BBJ
19q13.32 APOC1 rs66626994-A 2.21E-03 4.55E-18 0.913 Asian MTAG-BBJ
22q11.23 SNRPD3 rs140762045-A 6.57E-04 2.09E-64 0.893 Asian MTAG-BBJ
22q11.23 SGSM1 rs9612768-C 9.81E-04 1.94E-09 0.945 Asian MTAG-BBJ
rs140992152-A 4.27E-03 1.22E-13 0.859 Asian MTAG-BBJ
Open in a new tab

aThe rsID of the locus and its effect allele.

bThe P value of the locus in the gastric cancer GWAS.

cThe most significant P value for this SNP in the MTAG analysis (incorporating gastric cancer and additional phenotypes). “NA” indicates that the SNP was not present in the MTAG-BBJ analysis; otherwise, the value represents the strongest association observed.

dOdds ratio (OR): An OR > 1 indicates a positive association (risk-increasing effect), while an OR < 1 indicates a negative association (protective effect).

ePreviously reported gastric cancer loci were sourced from the referenced literature, while novel loci identified in this study were derived from the MTAG-BBJ analysis (see Supplementary Data 2 for associations). GC gastric cancer, BBJ BioBank Japan, MTAG multitrait analysis of GWAS. Association P values and odds ratios were derived from two‑sided tests in the original GWAS or MTAG analyses. The genome‑wide significance threshold was defined as P < 5 × 10⁻⁸.

Expression quantitative trait locus (eQTL) analysis of the gastric cancer risk SNP rs12814712 and neighbouring genes in the Chinese population

We next analysed a cohort of 46 gastric cancer specimens using Sanger sequencing. This experimental validation was designed to assess whether the locus of interest could affect gene expression in gastric cancer. Among these samples, the AG variant of rs12814712 was found in 34.8% (16 cases), while the GG variant was observed in 65.2% (30 cases) (Fig. 6A). The detailed clinical characteristics of these 46 samples are described in Supplementary Table S2. The aforementioned signal at 12q22 (rs12814712) in gastric cancer covered four genes, namely, NDUFA12, FGD6, NR2C1 and VEZT. Multiple lines of evidence suggest that these genes may be involved in the pathogenesis of various cancers, including gastric cancer15–20. To identify the genes potentially regulated by rs12814712 in gastric cancer, we used RT‒qPCR to compare the expression levels of four genes near rs12814712—NDUFA12, FGD6, NR2C1 and VEZT—in gastric cancer tissues and adjacent noncancerous tissues between the AG and GG groups. The expression levels of FGD6, VEZT and NR2C1 were significantly lower in both adjacent nontumour tissues and gastric cancer tissues from patients with the AG genotype of rs12814712 than in those from patients with the GG genotype (P < 0.05; Fig. 6B, C). These findings suggested a potential relationship between the rs12814712 genotype and the regulation of these genes in both nontumour and tumour contexts.

Fig. 6. Association of the rs12814712 genotype with the expression of neighbouring genes in gastric cancer tissues.

Fig. 6

Open in a new tab

A Sanger sequencing chromatograms showing the distribution of the rs12814712 genotypes GG (n = 30 patients) and AG (n = 16 patients). B, C The expression levels of four genes near rs12814712 were quantified by RT‑qPCR in paragastric cancer (B) and gastric cancer tissue (C). Data are presented as box‑and‑whisker plots displaying the median (centreline), 25th–75th percentiles (box), and minimum‑maximum range (whiskers); individual data points are overlaid. Statistical significance between genotype groups was assessed by a two‑sided Mann–Whitney U test: *P < 0.05, **P < 0.01; ns not significant. Exact P values are shown in the figure.

Potential regulatory effect of rs12814712(G > A) on the transcriptional activity of neighbouring genes

To investigate whether rs12814712 had a cis-regulatory function, we first determined its genomic location. It was located in the 12q22 region, downstream of NDUFA12, NR2C1 and FGD6, and upstream of VEZT (Fig. 4). On the basis of its distribution between these genes, we designed two different luciferase reporter assay systems to evaluate the potential regulatory role of this region: (1) A 2001-bp genomic fragment containing the rs12814712 locus (wild-type WT‑G or mutant MUT‑A) and its flanking sequences (GRCh38/hg38 coordinates: chr12:94965934–94967934) was cloned upstream of the SV40 promoter to detect whether this fragment possesses enhancer-like activity (Supplementary Fig. S3A), and (2) the same fragment was inserted into the downstream end of the luciferase gene in the pmirGLO vector to examine its potential to regulate reporter gene expression in a downstream context (Supplementary Fig. S3B). Notably, this study aimed to functionally validate the region containing rs12814712 as an independent noncoding fragment to explore whether it could exert cis-regulatory effects at the transcriptional level. In the enhancer reporter system, the MUT (A) allele significantly reduced luciferase activity (P < 0.05), suggesting that this fragment possessed enhancer-like activity and that the A allele impaired this function (Supplementary Fig. S3C). In contrast, no significant difference was observed between the two alleles (WT and MUT) in the pmirGLO system, which indicated that the fragment did not exhibit significant posttranscriptional regulatory effects under these conditions (Supplementary Fig. S3D). Given that our GWAS signal was located upstream of VEZT and downstream of NR2C1, FGD6 and NDUFA12, this finding supported our eQTL results, indicating a regulatory relationship between the rs12814712 A allele and the expression of VEZT. Taken together, these results suggested that the genomic region containing rs12814712 might play a role in cis-regulation with enhancer-like activity.

Association of genes near rs12814712 with proliferation and migration capacity in gastric cancer

To investigate the role of potential target genes near rs12814712 in gastric cancer, we conducted a series of in vitro experiments using AGS and MKN45 gastric cancer cell lines. NDUFA12, FGD6, NR2C1 and VEZT overexpression models were established via plasmid transfection, with successful overexpression confirmed by RT–qPCR (Fig. 7A). Cell viability assays indicated that overexpression of NR2C1 and VEZT significantly reduced cell proliferation, while NDUFA12 and FGD6 had no notable effect (Fig. 7B). This antiproliferative effect was further supported by the results of EdU assays, which revealed a significant reduction in the percentage of proliferating NR2C1- and VEZT-overexpressing cells but not in that of NDUFA12- and FGD6-overexpressing cells (Fig. 7C). Consistently, the results of the colony formation assays revealed that NR2C1 and VEZT markedly inhibited the colony-forming ability of both cell lines (Fig. 7D). Migration assays, including wound healing and Transwell assays, demonstrated that overexpression of NR2C1 and VEZT significantly impaired cell migration, whereas NDUFA12 and FGD6 did not affect migratory capacity (Fig. 8A, B).

Fig. 7. Effects of candidate genes near rs12814712 on gastric cancer cell proliferation.

Fig. 7

Open in a new tab

A mRNA levels were measured by RT‑qPCR in AGS and MKN45 cells transfected with control (CON) or gene‑specific overexpression plasmids for NDUFA12, FGD6, NR2C1 or VEZT at 48 h. Data are presented as the mean ± SD from n = 3 biologically independent experiments. Significance relative to the CON for each gene was assessed using an unpaired two‑tailed t‑test (no adjustment for multiple comparisons) and denoted as *P < 0.05, **P < 0.01 and ***P < 0.001; ns not significant. Exact P values are provided in the figure. B Cell viability was monitored over time. Data are presented as the mean ± SD from n = 3 biologically independent experiments, each performed in triplicate wells. Significance at each time point relative to CON was determined by two‑way ANOVA followed by Tukey’s multiple comparisons test and denoted as ns not significant, *P < 0.05, **P < 0.01 and ***P < 0.001. Exact P values are provided in the figure. C, D EdU staining (C) and colony formation assays (D) were performed in transfected AGS and MKN45 cells. Images are representative of n = 3 biologically independent experiments. Scale bar, 100 μm. CON control, OP overexpression.

Fig. 8. Association of candidate genes near rs12814712 with the migration capacity of patients with gastric cancer.

Fig. 8

Open in a new tab

Wound-healing (A; scale bar, 250 μm) and Transwell migration (B; scale bar, 100 μm) assays were performed using AGS and MKN45 cells transfected with control (CON) or gene‑specific overexpression (OP) plasmids. Images are representative of three biologically independent experiments with consistent results. CON control, OP overexpression.

To further validate the tumour-suppressive roles of VEZT and NR2C1, we performed gene knockdown experiments using two independent short hairpin RNAs (shRNAs) targeting each gene (Supplementary Fig. S4A). Silencing of either VEZT or NR2C1 significantly increased gastric cancer cell proliferation, as demonstrated by cell growth and colony formation assays (Supplementary Fig. S4B–D), and enhanced cell migration capacity (Supplementary Fig. S5A, B), which was consistent with the results observed in the overexpression experiments.

Discussion

In this study, we conducted a comprehensive genome-wide multitrait analysis to systematically investigate the shared genetic architecture between gastric cancer and a broad spectrum of traits in East Asians. Our analysis revealed several novel East Asian-specific pleiotropic loci associated with gastric cancer and its correlated traits. Among these, we selected the 12q22 locus for functional experiments, uncovering regulatory interactions with nearby genes and indicating potential roles for these genes in the aetiology of gastric cancer.

We examined genetic correlations between gastric cancer and 95 phenotypes in East Asian populations and revealed significant associations with cardiovascular diseases and related risk factors, such as angina, MI, hypertension and the use of antihypertensive agents. Consistent with previous epidemiological findings, our MR analysis further revealed that these cardiovascular factors contribute to an increased risk of gastric cancer. These findings align with those of two studies that reported cardiovascular disease prevalence rates of 26.8% and 34.3% among gastric cancer patients in China21,22. Similarly, a South Korean study of 6458 patients with gastric cancer using National Health Insurance data revealed that 32.7% had been diagnosed with cardiovascular diseases within the year prior to their cancer diagnosis23. Collectively, these studies highlight the frequent coexistence of gastric cancer and cardiovascular events. Additionally, observational studies have shown that type 2 diabetes is linked to a higher risk of gastric cancer, which is consistent with our findings showing positive genetic correlations between T2D, blood glucose levels, diabetes medications and gastric cancer24–26. These genetic correlations offer valuable insights into the complex interactions between gastric cancer, metabolic factors and other health conditions in East Asians, helping to identify shared pleiotropic loci that may influence gastric cancer and its related risk factors.

Our MTAG analysis revealed 26 novel pleiotropic loci, with the majority arising from a joint analysis of gastric cancer and cardiovascular traits. Notably, the 12q22 (rs12814712) locus consistently appears in analyses that link gastric cancer with various heart diseases. This GWAS signal may affect the regulation of nearby genes, including NDUFA12, FGD6, NR2C1 and VEZT, which are implicated in the tumorigenesis of various cancers, including gastric cancer. NDUFA12, a mitochondrial electron transport chain component, inhibits hepatocyte proliferation when it is knocked down15,27. VEZT functions as a tumour suppressor in gastric cancer and may serve as an independent prognostic marker19,28. NR2C1, a member of the steroid/thyroid hormone receptor family, negatively regulates androgen receptor activity in prostate cancer and has prognostic value in hepatocellular carcinoma29–31. FGD6, a Ras-like Rho/Rac protein, regulates the actin cytoskeleton and cell shape32. A TCGA analysis revealed that FGD6 upregulation is correlated with poor gastric cancer prognosis33, and GTEx data revealed that rs12814712 is an eQTL for FGD6 in various tissues, including fibroblasts and arteries34. However, while GTEx data suggest potential regulatory mechanisms among Europeans, the actual regulatory relationships of rs12814712 require experimental validation in East Asians due to varying linkage disequilibrium patterns across populations (Supplementary Fig. S6). In addition to rs12814712, we also identified rs13306206, a genome-wide significant missense variant in APOB (MAF ~ 1% in East Asians). However, it is currently annotated as “likely benign” in ClinVar and lacks supporting regulatory evidence. Given these limitations, we did not pursue functional validation in the present study, but we recognize its potential relevance and propose future studies to clarify its role.

To pinpoint the causal genes of gastric cancer within the 12q22 region, we conducted Sanger sequencing on 46 gastric cancer samples from Chinese patients and identified 16 patients carrying the risk allele A of rs12814712. Subsequent RT‒qPCR results indicated that the presence of the risk allele A of rs12814712 was associated with increased expression of FGD6, VEZT and NR2C1 in both cancerous and adjacent noncancerous tissues, suggesting that these three candidate genes might play critical roles in both the initiation and progression of gastric cancer in East Asian populations. Our luciferase reporter assay provided an indirect mechanistic link, demonstrating that the rs12814712-A risk allele specifically downregulated transcriptional activity, potentially explaining its role in reducing VEZT expression. We next conducted functional experiments to investigate the effects of NDUFA12, FGD6, NR2C1 and VEZT on the proliferation and migration of gastric cancer cells. Consistent with our eQTL results, VEZT significantly inhibited the proliferation and migration of gastric cancer cells, while NR2C1 also had a weaker but still notable effect. In contrast, NDUFA12 and FGD6 did not significantly affect these cellular processes.

Intriguingly, our findings are strongly supported by previous research on VEZT. Earlier studies have revealed that VEZT is a transmembrane protein within adherens junctions, is downregulated in gastric cancer tissues, and is linked to cancer invasion and migration in MKN45 and NCI-N87 cell lines, potentially driven by Helicobacter pylori-induced promoter methylation. Here, we not only observed similar results in MKN45 cells but also reported comparable outcomes in AGS cells19,35. On the basis of the evidence from our study, we suggest that VEZT may be the causal gene for gastric cancer at the 12q22 locus (rs12814712). However, our experimental findings also suggest a possible association between NR2C1 and gastric cancer. Interestingly, a study on pancreatic cancer demonstrated that NR2C1 transcription is linked to epithelial‒mesenchymal transition16. Therefore, a potential role of NR2C1 in gastric cancer cannot be ruled out. Furthermore, our knockdown experiments of VEZT and NR2C1 further confirmed their role in suppressing gastric cancer proliferation and migration.

This study has several limitations. First, given the relatively small sample size of the validation cohort, the associations identified in this study require further replication in larger, independent populations. Accordingly, these findings should be interpreted with caution. In particular, the possibility of false positives cannot be ruled out, as the MTAG approach may capture signals influenced by genetically correlated traits in addition to the focal trait. Second, this study is limited by the absence of information concerning HP infection status, which restricts our ability to distinguish genetic effects on infection risk from those on cancer development. Future studies incorporating both genetic data and HP infection status will be essential to clarify this relationship. Third, owing to the low frequency of the AA genotype in the population, only the GG and AG genotypes were observed in our tissue samples, limiting our ability to assess the effects of the AA genotype. Future studies with larger sample sizes may be needed to evaluate the potential dose-dependent effects more comprehensively. Fourth, our functional validation, including luciferase reporter assays, was conducted in cell-based models that cannot replicate the native chromatin environment and long-range genomic interactions in vivo.

In this study, we systematically analysed public BBJ data to explore the genetic correlations between gastric cancer and various phenotypes in East Asians. Our analysis revealed a novel pleiotropic locus, rs12814712 (12q22), which is shared between gastric cancer and cardiovascular diseases. Additionally, functional experiments confirmed VEZT as a target gene for gastric cancer. These findings enhance our understanding of gastric cancer pathogenesis and may inform the development of more effective therapeutic strategies.

Methods

Ethical compliance

All research involving human participants complied with all relevant ethical regulations. The study protocol (YXLL-KY-2022(068)) was approved by the Institutional Review Board of the First Affiliated Hospital of Shandong First Medical University (Jinan, China). Written informed consent was obtained from all participants, and the study was conducted in accordance with the principles of the Declaration of Helsinki.

GWAS data

The GWAS data for gastric cancer in East Asian populations included 167,122 individuals, including 7921 cases and 159,201 controls. Additionally, we obtained summary statistics for 95 phenotypes (including diseases, biomarkers and medication usage) with case numbers greater than 3000 from the 220 phenotypes available in the BBJ database36. In BBJ, genotyping was conducted using Illumina arrays, and genotype imputation was performed using the 1000 Genomes Phase 3 reference panel combined with Japanese whole-genome sequencing data. Variants with an imputation of Rsq ≥ 0.7 were retained. Phenotypes were derived from curated medical records, ATC-coded medication data and biomarker measurements, with appropriate outlier removal and medication-specific adjustments. A GWAS for binary traits employed SAIGE to account for sample relatedness and case‒control imbalance, whereas BOLT-LMM was used for quantitative traits with adjustments for age, sex and principal components. A comprehensive overview of the included GWAS datasets from East Asian populations is provided in Supplementary Data 3. To validate our findings, we utilized two independent cohorts: the Korean Genome and Epidemiology Study (KoGES) cohort, comprising 378 gastric cancer cases and 71,920 controls of East Asian ancestry, and the MVP cohort, comprising 1152 cases and 449,540 controls of predominantly European ancestry.

Global genetic correlation analysis

Genetic correlation (rg) between each pair of diseases or traits was assessed using LD score regression (LDSC) with GWAS summary statistics37. The equation used in the LDSC was as follows: E[βjγj]=N1N2rgMlj+NsrN1N2, where βj and γj represent the effect sizes of SNP j on the two traits being tested, N1 and N2 are the sample sizes for the two traits, Ns is the number of overlapping samples between the two traits, r is the phenotypic correlation in the overlapping samples, and lj is the LD score. In line with the authors’ recommendations, we used the default LDSC regression weights and limited the analysis to SNPs with an MAF > 5% in the reference panel for estimation.

In this analysis, precalculated linkage disequilibrium scores for HapMap3 SNPs were utilized. These scores were derived from individuals of Asian ancestry from the 1000 Genomes Project. SNP markers with an imputation INFO score less than 0.9 were excluded from the analysis. LDSC was also employed to estimate the SNP-based heritability of the examined traits37.

Cell type-specific enrichment of SNP heritability

Stratified LDSC (S-LDSC) was employed in this study to identify specific functional categories or cell types that significantly affect the heritability of the traits under investigation. The baseline v1.1 model, which contains 53 functional annotations, including gene regions, conserved elements and regulatory features, was used as the reference in all S-LDSC analyses to account for known sources of heritability38. The annotation data used in this analysis were obtained from the Roadmap project and included information on six chromatin marks (DHS, H3K27ac, H3K36me3, H3K4me1, H3K4me3 and H3K9ac) across a diverse set of 88 cell types or tissues. These annotations were subsequently used to partition the SNP heritability of each trait. Additionally, the cell type annotations were grouped into nine categories, namely, adipose, central nervous system, digestive system, cardiovascular, musculoskeletal and connective tissue, immune and blood, liver, pancreas and others. We visualized the annotation-specific enrichment values using a colour-scaled heatmap generated by the ComplexHeatmap package (v2.12.1) in R 4.2.3. The hierarchical clustering of annotations was performed on the basis of Euclidean distance metrics with complete linkage agglomeration, implementing the default clustering algorithm (hclust) within the ComplexHeatmap framework.

Multitrait analysis of GWAS (MTAG)

MTAG was employed to explore the genetic associations between gastric cancer and each of the other phenotypes separately by leveraging combined GWAS summary statistics from gastric cancer and one additional trait in each analysis10. This approach enhances the ability to detect genetic loci associated with correlated traits by addressing sample overlap and incomplete genetic correlation. For the MTAG analysis, we used default parameters, including a genetic correlation prior of N (0,1), and applied cross-trait LD score regression intercepts to account for residual sample overlap. In the initial step of MTAG, variants were filtered by excluding noncommon SNPs, duplicated SNPs, or those with strand ambiguity. The pairwise genetic correlation between traits was subsequently estimated using LDSC, and these estimates were then used to calibrate the variance‒covariance matrix of the random effect component. Following calibration, a random-effects meta-analysis was performed by MTAG to generate SNP-level summary statistics. The suggestive significance threshold (P < 0.01) in single-trait GWAS was selected to capture potentially meaningful associations while maintaining stringency, as these signals were further validated by genome-wide significance in multitrait analysis. Pleiotropic SNPs that achieved genome-wide significance (P < 5 × 10–8) in the multitrait analysis and suggestive significance (P < 0.01) in the single-trait GWAS were prioritized. The major histocompatibility complex (MHC) region was treated as one locus because of its complex LD structure.

Two-sample Mendelian randomization

In the two-sample MR analysis, we implemented five widely recognized statistical methods: IVW, MR‒Egger, weighted median, simple mode and weighted mode. The IVW method, which served as the primary analytical approach, was applied under a multiplicative random effects model to estimate the causal association between genetic predisposition to gastric cancer and cardiovascular-related risk factors identified through the LDSC. The IVW method assumes no intercept term and employs the inverse of the variance as weights to optimize model fitting. To address potential pleiotropic effects, sensitivity analyses were conducted using the MR‒Egger method39, which incorporates a weighted linear regression framework to detect and adjust for horizontal pleiotropy among the instrumental variables (IVs). The results are reported as odds ratios (ORs) with corresponding 95% confidence intervals (CIs), and statistical significance was defined as P < 0.05. All MR analyses were executed using the TwoSampleMR package in R software (version 4.0.2), ensuring robust and reproducible causal inference40.

Patients and tissue samples

Human tissue samples for this study were obtained from a prospective cohort of 46 patients who underwent radical gastrectomy at the First Affiliated Hospital of Shandong First Medical University (Jinan, China) in 2023. For each patient, paired samples of primary gastric cancer tissue and matched adjacent non-tumourous gastric mucosa were collected. The cohort included 33 male and 13 female patients. The genotypic distribution of the rs12814712 variant was as follows: AG genotype (9 males, 7 females) and GG genotype (24 males, 6 females). The participant age distribution is provided in Supplementary Table S2. Sex was determined according to hospital medical records. Although sex was not incorporated as a biological variable in the primary genetic association study design, all clinical data were sex-disaggregated in the source material.

Written informed consent was obtained from all participants for the use of their tissue samples and associated clinical data for research purposes. No financial compensation was provided. Following collection, genomic DNA and total RNA were extracted from portions of the tissue samples for downstream analyses. The original tissue samples were not completely consumed during this process. Residual tissue materials (aliquots and tissue blocks) were cryopreserved and permanently stored at −80 °C in the institutional biobank of the First Affiliated Hospital of Shandong First Medical University.

These deidentified biospecimens and associated clinical data remain available for academic validation and collaborative research. Access requests can be submitted with a scientifically justified proposal and completion of a material transfer agreement (MTA). All such requests should be directed to the corresponding author, Guangyong Zhang (guangyongzhang@hotmail.com), for review and facilitation.

PCR-Sanger sequencing of rs12814712

We collected 25 mg of cancer tissue from each of the 46 gastric cancer patients and extracted genomic DNA using the Ezup Column Animal Genomic DNA Purification Kit (B518251; Sangon Biotech, China). The extracted genomic DNA was amplified by PCR, and the PCR amplification primers for the rs12814712 locus were as follows: forward primer (5′ to 3′): TCATACTATGGGCACTTGGCAC; reverse primer (5′ to 3′): CGGCAGCCCAGAACAGTAAA. The PCR reaction mix included the following components: 10× Taq buffer (with MgCl2), 10 µM dNTPs (mix), forward and reverse primers at 10 μM, 1 μl of template genomic DNA, 5 U/μl Taq DNA polymerase, and sterile water added to a final volume of 25 µL. After amplification, the PCR products were purified by agarose gel electrophoresis, and the sequence of the PCR amplification products was detected by Sanger sequencing.

Quantitative real-time PCR

Total RNA was extracted from frozen gastric cancer tissue or cell lines with TRIzol reagent (G3013; Servicebio, Wuhan, China). Single-strand complementary DNA (cDNA) was synthesized with a ReverTra Ace qPCR RT Kit (FSQ-101; TOYOBO Co., Ltd., Osaka, Japan). For quantitative real-time PCR, SYBR Green Real-time PCR Master Mix (QPK-201; TOYOBO Co., Ltd.) was used following the manufacturer’s instructions. The total PCR volume was 20 μl, containing 0.1 μl of each primer. The target-specific primers used were designed by Huada Gene (Shenzhen, China), and the specific sequences are listed in Supplementary Table S3.

Cell culture

The HEK293T, AGS and MKN45 cell lines were obtained from Procell (Wuhan, China) and cultured in DMEM (for HEK293T), Ham’s F-12K (for AGS) or RPMI-1640 (for MKN45) medium (all from Gibco, Carlsbad, CA, USA), each supplemented with 10% foetal bovine serum (FBS; Gibco) and 1% penicillin/streptomycin (Gibco). The cells were incubated at 37 °C in 5% CO₂. The identity of all the cell lines was verified by short tandem repeat analysis, and the cell lines were routinely screened for mycoplasma contamination.

Plasmid constructs

For luciferase assay-related plasmids, the 2001-bp genomic fragment containing the rs12814712 (G > A) site and 1000-bp flanking sequences was cloned and inserted into (1) the pGL3-promoter vector (E1761; Promega, Madison, WI, USA) to assess its potential enhancer-like activity and (2) the pmirGLO vector (E1330; Promega, Madison, WI, USA) to simulate its regulatory ability in a downstream context (fragment sequences are shown in Supplementary Data 4). The SV40 promoter sequence was used in these two plasmids to determine promoter function (Supplementary Fig. S3A, B).

For gene loss- and gain-of-function assays, plasmids were constructed by LST Biotech (Shandong Co., Ltd., Jinan, China). For gene knockdown experiments, short hairpin RNA (shRNA) sequences targeting NR2C1 and VEZT (Supplementary Table S4) were cloned and inserted into the pLKO.1-EGFP vector. For gene overexpression, the full-length coding sequences of NDUFA12, FGD6, NR2C1 and VEZT were cloned and inserted into the corresponding expression vectors. Information on the sequences and vectors is provided in Supplementary Table S5. The newly constructed plasmids generated in this study are available from the corresponding author upon reasonable request for academic research.

Luciferase assays

HEK293T and AGS cells (1 × 10⁵/well) were taken in 96-well plates and were transfected after 18 h of incubation, pGL3 constructs were transfected with 0.1 μg test plasmid plus 0.01 μg pRL-TK plasmid, and pmirGLO constructs were transfected with test plasmid alone. Luciferase activity was measured using a Dual-Luciferase® Reporter Assay System kit (E1910; Promega, Madison, WI, USA): cells were lysed with 20 μL of PLB (15 min of RT shaking), after which firefly (luciferase assay reagent II) and Renilla (Stop & Glo) luminescence was sequentially measured. Relative activity was calculated as the Firefly/Renilla ratio. The experiments were repeated in triplicate.

Cell transfection

Cells were seeded in 6-well plates (5 × 105 cells/well) and incubated for 18 h. Then, overexpression plasmids for the indicated genes (NDUFA12, FGD6, NR2C1 and VEZT) and knockdown plasmids for the indicated genes (NR2C1 and VEZT) were transfected using Lipofectamine 2000 (11668030; Invitrogen, Carlsbad, CA, USA) and Opti-MEM™ I (31985062; Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. After the cells were incubated for 8 h, the medium was replaced with Ham’s F-12K (for AGS) or 1640 medium (for MKN45) supplemented with 10% FBS. The sequences of the shRNAs for NR2C1 and VEZT are listed in Supplementary Table S4.

Cellular viability assay

Approximately 24 h after transfection, 5000 cells were seeded into 96-well plates. Cell viability was then evaluated at 0, 24 and 48 h using the Cell Counting Kit-8 assay (CK04, Dojindo Laboratories, Kumamoto, Japan), following the methods described previously41. After a 2-h incubation, cellular viability was assessed using a microplate reader, and the absorbance was measured at a wavelength of 450 nm.

EdU staining

In accordance with the manufacturer’s instructions, cell proliferation was assessed using an EdU staining kit (C0071S; Beyotime, Shanghai, China). Cells (1 × 104) in each group were incubated with EdU buffer (10 μM) at 37 °C for 2 h. The supernatant was then discarded, and the cells were washed twice with PBS, followed by fixation with paraformaldehyde for 20 min. Afterwards, the mixture was incubated with freshly prepared cocktail buffer. To calculate the proportion of EdU-positive cells, the number of fluorescently stained cells was divided by the number of DAPI-stained cells. The nuclei were then stained with DAPI (C1006; Beyotime, Shanghai, China). The cells were observed and photographed under a fluorescence microscope. EdU was excited by light at a wavelength of 488 nm to emit green fluorescence, while DAPI was excited by light at a wavelength of 340 nm to emit blue fluorescence.

Colony formation assay

Twenty-four hours after transfection, the transfected gastric cancer cells (1 × 103) were seeded into 6-well plates. After an additional 14 days of culture, the cells were fixed and stained with 1% crystal violet (Solarbio, Beijing, China). After being stained with crystal violet for 20 min, the colonies formed were counted.

Transwell assay

Transwell assays were conducted to evaluate the migration and invasion of AGS and MKN45 cells. For the migration assay, cells (4 × 104) were suspended in serum-free medium (120 μL) and added to the upper chamber. Medium containing 10% FBS (750 μL) was added to the lower chamber to induce cell migration. The cells that migrated were washed three times with PBS, fixed with 4% paraformaldehyde for 30 min, and stained with 1% crystal violet. Images were observed at 100× magnification.

Wound healing assay

Cells (6 × 105) were cultured on 6-well plates. A pipette tip was subsequently used to scratch the bottom of the plate. Free cells were washed away with PBS buffer. The migration distance was observed at 0 h and 48 h.

Statistics and reproducibility

The overall study design involved computational analysis of population-level genetic data (GWAS) followed by experimental validation in independent patient tissues and cell line models. No statistical method was used to predetermine sample size for the GWAS analyses, as these utilized all available data from the specified biobanks (BBJ, KoGES, MVP). The sample size for the patient tissue cohort (n = 46) was based on availability during the study period. No data were excluded from the GWAS summary statistics analyses. For cellular and molecular experiments, sample sizes were determined on the basis of common practices in the field to ensure adequate power for detecting effects, and all experiments were performed with at least three biological replicates. The experiments were not randomized. The investigators were not blinded to allocation during the experiments or outcome assessment. Statistical analyses were conducted using SPSS software (version 22.0) and GraphPad Prism (version 8.0). Data from cellular and molecular experiments are presented as the mean ± standard deviation (SD). Group differences were evaluated using one-way ANOVA. Significance levels in the figures are denoted as follows: *P < 0.05, **P < 0.01 and ***P < 0.001. To control for multiple comparisons in the genome-wide analyses (genetic correlation via LDSC, S-LDSC and MR), Benjamini–Hochberg FDR correction was applied, with significance defined as PFDR < 0.05.

Supplementary information

Supplementary Information (3.6MB, pdf)
41467_2026_70774_MOESM2_ESM.pdf (57.2KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (16KB, xlsx)
Supplementary Data 2 (34.3KB, xlsx)
Supplementary Data 3 (16.2KB, xlsx)
Supplementary Data 4 (10KB, xlsx)
Summary (80.3KB, pdf)
Transparent Peer Review file (1.4MB, pdf)

Source data

Source data (1.6MB, zip)

Acknowledgements

We extend our gratitude to the contributors and custodians of public databases, with special acknowledgement to BioBank Japan (BBJ), the Korean Genome and Epidemiology Study (KoGES) and the Million Veteran Program (MVP). We are also deeply appreciative of the patients whose participation in this study has been invaluable to scientific research. This work is supported by the Special Funds of Taishan Scholar Project, China (tstp20250757, G.Z.; tsqn202211224, X.C.), the National Natural Science Foundation of China (32270661, X.C.; 82472366, G.Z.), the Excellent Youth Science Fund Project (Overseas) of Shandong, China (2023HWYQ-082, X.C.), the China Postdoctoral Science Foundation (2024M761870, Y.S.) and the Shandong Postdoctoral Science Foundation (SDCX-ZG-202400042, Y.S.). The elements in Fig. 1 were designed by Freepik and Linector, both sourced from Flaticon [https://www.flaticon.com/].

Author contributions

Conceptualization: X.C., H.H., G.Z. Methodology: H.D., C.L., Q.S., Y.J., Q.X., Z.Z., X.J., Q.H. Investigation: Y.L., Y.S., Z.L., T.Y., J.G. Visualization: H.D., C.L, Y.J. Formal analysis: H.D., C.L., Q.S. Data curation: B.S., D.T., L.L., F.M., D.T. Funding acquisition: X.C., G.Z., Y.S. Project administration: X.C., H.H., G.Z. Supervision: X.C., H.H., G.Z. Writing–original draft: H.D., C.L., Q.S. Writing–review & editing: H.H., G.Z. and X.C. H.D., C.L. and Q.S. contributed equally to this work. H.H., G.Z. and X.C. jointly supervised this study.

Peer review

Peer review information

Nature Communications thanks Koichi Matsuda and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The GWAS summary statistics data used in this study are available in the GWAS Catalogue under the accession codes summarized in Supplementary Data 3 [https://www.ebi.ac.uk/gwas/]. The individual-level genotype and gene expression data generated in this study are available under restricted access to protect patient privacy and comply with ethical regulations; access can be obtained by submitting a formal request to the corresponding author, Dr. Guangyong Zhang (guangyongzhang@hotmail.com), for academic, noncommercial research purposes, which will then be reviewed by the Ethics Committee of The First Affiliated Hospital of Shandong First Medical University. A completed data transfer agreement may be needed. The raw individual-level clinical data are protected and are not available because of data privacy laws. Source data are provided with this paper.

Code availability

We conducted our analyses using a variety of publicly available software tools. The computational environment consisted of Python (v2.7.18), Perl (v5.16.3) and R (v4.2.3). Genetic correlation analyses between gastric cancer and 95 traits were conducted using LDSC v1.0.1, with subsequent S-LDSC to evaluate cell type-specific SNP heritability enrichment. Bidirectional Mendelian randomization was implemented through the TwoSampleMR R package (v0.6.4). For the multitrait GWAS meta-analysis of gastric cancer and its 25 genetically correlated traits, we employed MTAG v1.0.8. The custom scripts used in this study are deposited in Zenodo and are publicly accessible at https://zenodo.org/records/18085131.

Competing interests

The authors declare no competing interests.

Footnotes

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

These authors contributed equally: Huanxin Ding, Chuxuan Liu, Qing Sun.

Contributor Information

Xiao Chang, Email: changxiao@sdfmu.edu.cn.

Hakon Hakonarson, hakonarson@email.chop.edu.

Guangyong Zhang, Email: guangyongzhang@hotmail.com.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70774-9.

References

  • 1.Bray, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin.74, 229–263 (2024). [DOI] [PubMed] [Google Scholar]
  • 2.Thrift, A. P., Wenker, T. N. & El-Serag, H. B. Global burden of gastric cancer: epidemiological trends, risk factors, screening and prevention. Nat. Rev. Clin. Oncol.20, 338–349 (2023). [DOI] [PubMed] [Google Scholar]
  • 3.Dai, J. et al. Estimation of heritability for nine common cancers using data from genome-wide association studies in Chinese population. Int. J. Cancer140, 329–336 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Mucci, L. A. et al. Familial risk and heritability of cancer among twins in Nordic countries. JAMA315, 68–76 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Sakamoto, H. et al. Genetic variation in PSCA is associated with susceptibility to diffuse-type gastric cancer. Nat. Genet.40, 730–740 (2008). [DOI] [PubMed] [Google Scholar]
  • 6.Zhu, M. et al. Exome array analysis identifies variants in SPOCD1 and BTN3A2 that affect risk for gastric cancer. Gastroenterology152, 2011–2021 (2017). [DOI] [PubMed] [Google Scholar]
  • 7.Yan, C. et al. Meta-analysis of genome-wide association studies and functional assays decipher susceptibility genes for gastric cancer in Chinese populations. Gut69, 641–651 (2020). [DOI] [PubMed] [Google Scholar]
  • 8.Helgason, H. et al. Loss-of-function variants in ATM confer risk of gastric cancer. Nat. Genet.47, 906–910 (2015). [DOI] [PubMed] [Google Scholar]
  • 9.Saeki, N. et al. A functional single nucleotide polymorphism in mucin 1, at chromosome 1q22, determines susceptibility to diffuse-type gastric cancer. Gastroenterology140, 892–902 (2011). [DOI] [PubMed] [Google Scholar]
  • 10.Turley, P. et al. Multi-trait analysis of genome-wide association summary statistics using MTAG. Nat. Genet.50, 229–237 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Han, R. et al. Systematic analyses of GWAS summary statistics from UK Biobank identified novel susceptibility loci and genes for upper gastrointestinal diseases. J. Hum. Genet.68, 599–606 (2023). [DOI] [PubMed] [Google Scholar]
  • 12.Sakaue, S. et al. A cross-population atlas of genetic associations for 220 human phenotypes. Nat. Genet.53, 1415–1424 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Pruim, R. J. et al. LocusZoom: regional visualization of genome-wide association scan results. Bioinformatics26, 2336–2337 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Liu, B., Gloudemans, M. J., Rao, A. S., Ingelsson, E. & Montgomery, S. B. Abundant associations with gene expression complicate GWAS follow-up. Nat. Genet.51, 768–769 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Park, S. I. et al. NDUFA12 as a functional target of the anticancer compound ertredin in human hepatoma cells as revealed by label-free chemical proteomics. J. Proteome Res.23, 130–141 (2024). [DOI] [PubMed] [Google Scholar]
  • 16.Liu, S. et al. NamiRNA-enhancer network of miR-492 activates the NR2C1-TGF-β/Smad3 pathway to promote epithelial-mesenchymal transition of pancreatic cancer. Carcinogenesis44, 153–165 (2023). [DOI] [PubMed] [Google Scholar]
  • 17.Zhang, Y. F. et al. A low amino acid environment promotes cell macropinocytosis through the YY1-FGD6 axis in Ras-mutant pancreatic ductal adenocarcinoma. Oncogene41, 1203–1215 (2022). [DOI] [PubMed] [Google Scholar]
  • 18.Hatthakarnkul, P. et al. Protein expression of S100A2 reveals it association with patient prognosis and immune infiltration profile in colorectal cancer. J. Cancer14, 1837–1847 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Miao, R. et al. VEZT, a novel putative tumor suppressor, suppresses the growth and tumorigenicity of gastric cancer. PLoS ONE8, e74409 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Xie, D., Shang, L., Peng, L. & Li, L. Up-regulation of VEZT by small activating RNA inhibits the proliferation, invasion and migration of gastric cancer cells. Biochem. Biophys. Res. Commun.482, 542–548 (2017). [DOI] [PubMed] [Google Scholar]
  • 21.Zhou, B. et al. Long-term outcomes of esophageal and gastric cancer patients with cardiovascular and metabolic diseases: a two-center propensity score-matched cohort study. J. Transl. Int. Med.11, 234–245 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.He, D. et al. Increased incidence risks of cardiovascular disease among cancer patients: evidence from a population-based cohort study in China. Int. J. Cardiol.396, 131362 (2024). [DOI] [PubMed] [Google Scholar]
  • 23.Han, K. T., Kim, D. W. & Kim, W. Impact of cardiovascular diseases on mortality in gastric cancer patients with preexisting chronic disease. Yonsei Med. J.63, 1043–1049 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Tseng, C. H. The relationship between diabetes mellitus and gastric cancer and the potential benefits of metformin: an extensive review of the literature. Biomolecules11, 1022 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Tseng, C. H. Diabetes conveys a higher risk of gastric cancer mortality despite an age-standardised decreasing trend in the general population in Taiwan. Gut60, 774–779 (2011). [DOI] [PubMed] [Google Scholar]
  • 26.Sekikawa, A. et al. Diabetes mellitus increases the risk of early gastric cancer development. Eur. J. Cancer50, 2065–2071 (2014). [DOI] [PubMed] [Google Scholar]
  • 27.Ahn, M. et al. A requirement for PAK1 to support mitochondrial function and maintain cellular redox balance via electron transport chain proteins to prevent β-cell apoptosis. Metabolism115, 154431 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Li, Y. S., Chen, Y. Z., Guo, X. B., Liu, X. & Li, L. P. VEZT as a novel independent prognostic factor in gastric cancer. Cancer Biomark.15, 375–380 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Lin, D. L. & Chang, C. Identification of an essential cis-acting element (TR2-PACE) in the 5’ promoter of human TR2 orphan receptor gene. Endocrine12, 89–97 (2000). [DOI] [PubMed] [Google Scholar]
  • 30.Mu, X. & Chang, C. TR2 orphan receptor functions as negative modulator for androgen receptor in prostate cancer cells PC-3. Prostate57, 129–133 (2003). [DOI] [PubMed] [Google Scholar]
  • 31.Sun, G., Sun, K. & Shen, C. Human nuclear receptors (NRs) genes have prognostic significance in hepatocellular carcinoma patients. World J. Surg. Oncol.19, 137 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Beck, H. C. et al. A mass spectrometry-based proteome study of twin pairs discordant for incident acute myocardial infarction within three years after blood sampling suggests novel biomarkers. Int. J. Mol. Sci.25, 2638 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Zeng, J., Li, M., Shi, H. & Guo, J. Upregulation of FGD6 Predicts Poor Prognosis in Gastric Cancer. Front Med. (Lausanne)8, 672595 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Human Genomics. The Genotype-Tissue Expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science348, 648–660 (2015). [DOI] [PMC free article] [PubMed]
  • 35.Guo, X. et al. Down-regulation of VEZT gene expression in human gastric cancer involves promoter methylation and miR-43c. Biochem. Biophys. Res. Commun.404, 622–627 (2011). [DOI] [PubMed] [Google Scholar]
  • 36.Buniello, A. et al. The NHGRI-EBI GWAS catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res.47, D1005–d1012 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Bulik-Sullivan, B. et al. An atlas of genetic correlations across human diseases and traits. Nat. Genet.47, 1236–1241 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Finucane, H. K. et al. Partitioning heritability by functional annotation using genome-wide association summary statistics. Nat. Genet.47, 1228–1235 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Bowden, J., Davey Smith, G. & Burgess, S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int. J. Epidemiol.44, 512–525 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Burgess, S., Butterworth, A. & Thompson, S. G. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol.37, 658–665 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Jiang, Y., Zhang, Y., Chu, F., Xu, L. & Wu, H. Circ_0032821 acts as an oncogene in cell proliferation, metastasis and autophagy in human gastric cancer cells in vitro and in vivo through activating MEK1/ERK1/2 signaling pathway. Cancer Cell Int.20, 74 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Shi, Y. et al. A genome-wide association study identifies new susceptibility loci for non-cardia gastric cancer at 3q13.31 and 5p13.1. Nat. Genet.43, 1215–1218 (2011). [DOI] [PubMed] [Google Scholar]
  • 43.Wang, Z. et al. Identification of new susceptibility loci for gastric non-cardia adenocarcinoma: pooled results from two Chinese genome-wide association studies. Gut66, 581–587 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Hu, N. et al. Genome-wide association study of gastric adenocarcinoma in Asia: a comparison of associations between cardia and non-cardia tumours. Gut65, 1611–1618 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Jin, G. et al. Genetic variants at 6p21.1 and 7p15.3 are associated with risk of multiple cancers in Han Chinese. Am. J. Hum. Genet.91, 928–934 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Abnet, C. C. et al. A shared susceptibility locus in PLCE1 at 10q23 for gastric adenocarcinoma and esophageal squamous cell carcinoma. Nat. Genet.42, 764–767 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Chen, Y. et al. Integrated multi-dimensional analysis highlights DHCR7 mutations involving in cholesterol biosynthesis and contributing therapy of gastric cancer. J. Exp. Clin. Cancer Res.42, 36 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

Supplementary Information (3.6MB, pdf)
41467_2026_70774_MOESM2_ESM.pdf (57.2KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (16KB, xlsx)
Supplementary Data 2 (34.3KB, xlsx)
Supplementary Data 3 (16.2KB, xlsx)
Supplementary Data 4 (10KB, xlsx)
Summary (80.3KB, pdf)
Transparent Peer Review file (1.4MB, pdf)
Source data (1.6MB, zip)

Data Availability Statement

The GWAS summary statistics data used in this study are available in the GWAS Catalogue under the accession codes summarized in Supplementary Data 3 [https://www.ebi.ac.uk/gwas/]. The individual-level genotype and gene expression data generated in this study are available under restricted access to protect patient privacy and comply with ethical regulations; access can be obtained by submitting a formal request to the corresponding author, Dr. Guangyong Zhang (guangyongzhang@hotmail.com), for academic, noncommercial research purposes, which will then be reviewed by the Ethics Committee of The First Affiliated Hospital of Shandong First Medical University. A completed data transfer agreement may be needed. The raw individual-level clinical data are protected and are not available because of data privacy laws. Source data are provided with this paper.

We conducted our analyses using a variety of publicly available software tools. The computational environment consisted of Python (v2.7.18), Perl (v5.16.3) and R (v4.2.3). Genetic correlation analyses between gastric cancer and 95 traits were conducted using LDSC v1.0.1, with subsequent S-LDSC to evaluate cell type-specific SNP heritability enrichment. Bidirectional Mendelian randomization was implemented through the TwoSampleMR R package (v0.6.4). For the multitrait GWAS meta-analysis of gastric cancer and its 25 genetically correlated traits, we employed MTAG v1.0.8. The custom scripts used in this study are deposited in Zenodo and are publicly accessible at https://zenodo.org/records/18085131.

Articles from Nature Communications are provided here courtesy of Nature Publishing Group

RESOURCES