Statistics from Altmetric.com
We read with great interest the article by Mocellin et al,1 who conducted a comprehensive meta-analysis that nominated 11 germline variants at nine loci significantly associated with gastric cancer (GC) with high level of summary evidence. Moreover, they also identified 38 single nucleotide polymorphisms (SNPs) with intermediate quality significant associations. Most of these loci were resulted from hypothesis-driven studies based on biological relevance, but most of these studies were small sample size and might lead to publication bias. In order to further evaluate their relevance with GC in individual large studies, we analysed these variants directly or their strong linkage disequilibrium SNPs using existing genome-wide association study (GWAS) datasets (including 2631 cases and 4373 controls) in Chinese populations, including those from Nanjing and Beijing populations conducted by our group2 and from Henan and Shanxi populations conducted by the USA National Cancer Institute.3
After exclusion of the variants with minor allele frequency <0.01 in Chinese population, a total of 31 different SNPs were included in further analysis (table 1). The GC risk-related loci reported in the previous GWAS, including those on 1q22 (MUC1, rs2070803, OR=0.75, p=5.46×10−09), 5p13.1 (PRKAA1, rs13361707, OR=1.26, p=1.41×10−10), 8q24.3 (PSCA, rs2294008, OR=1.14, p=1.65×10−3) and 10q23.33 (PLCE1, rs2274223, OR=1.26, p=5.23×10−8), were all confirmed with consistent directions. In addition, the other eight genetic variants residing in the above four regions were also confirmed. However, in the remaining 19 SNPs that were not reported by GWAS, only one variant was significantly associated with GC risk in the pooling results of GWAS datasets (IL10, rs1800871, OR=0.89, p=3.60×10−3).
We further derived a weighted genetic risk score (wGRS) comprising 22 independent variants (defined as r2<0.3 based on 1000 Genomes Project Phase 3 data) using the following formula4: , where βi is the effect estimate of the ith SNP reported in the meta-analysis1 and SNPi is the dosage of the effect allele (0, 1 or 2 for the risk allele associated with GC). As shown in table 2, logistic regression analysis indicated that the wGRS of total 22 SNPs was significantly associated with increased risk of GC (OR=1.23, 95% CI 1.16 to 1.30, p=2.37×10−12). However, this effect was mainly resulted from the reported four GWAS SNPs (OR=1.79, 95% CI 1.60 to 2.00, p=2.79×10−24), and the effect disappeared after excluding these four loci (OR=1.06, 95% CI 0.99 to 1.14, p=0.078).
Based on these independent variants, we then conducted heritability analysis (see online supplementary table S1). The 5-year prevalence per 100 000 people of GC in China was 53.7/100 000 based on global cancer 2012 (http://globocan.iarc.fr/Pages/fact_sheets_population.aspx). In total, all these 22 independent variants associated with GC risk could explain approximately 1.07% of the phenotypic variance at the relative prevalence, whereas four of the previous GWAS reported variants accounted for approximately 0.87% of the total phenotypic variance in GC. These four GWAS reported variants alone could explain ~81.31% of the phenotypic variance owing to known genetic variations. However, the remaining 18 variants only explained 0.38% of the phenotypic variance.
Supplementary file 1
In conclusion, our findings validated the four loci reported by GWAS and indicated that the loci reported on the basis of candidate gene approach may contribute little to GC susceptibility. These four GWAS reported loci may be useful to identify at-risk individuals and enable implementation of preventive measures for GC. Nevertheless, further studies are warranted to explore more stable and powerful germline variants accounting for GC susceptibility using GWAS approach or whole-genome sequencing-based analysis with larger sample size.
CY and MZ contributed equally.
Contributors GJ designed the study and critically revised the manuscript for important intellectual content. MZ provided technical and data support. TH and FY performed data analysis. CY analysed data and wrote the manuscript.
Funding This work was supported by the national key research and development program of China (grant no. 2016YFC1302703); the national natural science foundation of China (grant nos. 81422042, 81373090); the key grant of natural science foundation of Jiangsu higher education institutions (grant no 15KJA330002); the Top-notch academic programs project of Jiangsu higher education institution (grant no. PPZY2015A067); and the priority academic program for the development of Jiangsu higher education institutions (public health and preventive medicine).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Institutional Review Board of Nanjing Medical University.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.