Article Text
PDF
Statistics from Altmetric.com
We read with great interest the comprehensive study by Mocellin and colleagues1 who developed a quantitative summary of candidate genetic variants for gastric cancer susceptibility. Based on different tumour sites, the authors conducted a subgroup analysis for cardia cancer (CC) and non-cardia cancer (NCC). It is well established that CC has distinctive characteristics compared with NCC, but shares some common risk factors and pathogenesis with oesophageal cancer (OSC). In addition, there are some different genetic factors such as genetic variants between patients with CC and NCC. There is an argument that CC, located at oesophago-gastric junction, belongs to oesophageal more than gastric cancer.
According to the results of subgroup analysis, Mocellin and colleagues1 found several variants showing different results with CC and NCC risk. However, only PLCE1 rs2274223 was reported to be significantly associated with CC but not NCC risk with high credibility in Asian population.1 Unlike other variants such as those in interleukin genes modifying susceptibility of various types of cancers, this variant was de facto identified by genome wide association study (GWAS) from both patients with OSC and CC,2 and thus far has not been reported to be significantly associated with other malignancies, highlighting the unique role of this variant for OSC and CC. Currently published meta-analyses on rs2274223 variant and OSC risk are limited and not comparable due to diverse calculating models. Thus, we conducted a meta-analysis to detect association of this variant with OSC risk using additive model and the strategy described by Mocellin et al 1 With 13 studies included (until May 2015), we found the G allele was associated with significantly increased risk of OSC (OR=1.29, 95% CI 1.18 to 1.40, p<0.001, figure 1), which is consistent with the association of CC reported by Mocellin et al 1 This result was replicated in the Asian group (OR=1.36, 95% CI 1.25 to 1.47, p<0.001) which has a high incidence of CC and OSC. Notably, we observed significant heterogeneity both in the overall analysis (I2=82.0%) and in the Asian subgroup (I2=79.1%).
Forest plot for the overall association of PLCE1 rs2274223 with oesophageal cancer risk under additive model. Weights are from random effects analysis.
Human papillomavirus 16 (HPV16) infection has been recently identified to be related to OSC and CC risk.3 In 2014, a case–control study investigating the PLCE1 rs2274223 and HPV16 genotypes indicated that PLCE1 heterozygote developed higher susceptibility of HPV infection in patients with OSC, which implied that PLCE1 variant carriers, especially heterozygotes, could involve in HPV-related carcinogenesis.4 Therefore, we applied the genetic model GA versus AA to detect the potential heterozygote effect. For meta-analysis of this model, we found that GA genotype was associated with significantly increased risk of OSC in overall (OR=1.30, 95% CI 1.15 to 1.47, p<0.001) and Asian population (OR=1.40, 95% CI 1.26 to 1.54, p<0.001, figure 2), with no significant heterogeneity detected (I2=0.0%) in Asian subgroup (figure 2), suggesting high level of credibility of this finding. However, only two data sets of Asian and one data set of Caucasian were available for this model of CC; thus, no data synthesis was performed.
Forest plot for the association of PLCE1 rs2274223 with oesophageal cancer risk among Asian population (GA vs AA). Weights are from random effects analysis.
In summary, we thought that the rs2274223 variant might be involved with OSC and CC carcinogenesis rather than NCC. Moreover, we proposed a possibility that heterozygotes of this variant might potentially reflect the function better, and additive model might not be informative enough for that effect, although limited by insufficient data of CC. Taking rs2274223 for example, we think it is important to consider related evidence on molecular mechanism or interaction when setting subgroups or choosing genetic models. However, it is a major challenge to balance each factor; just as described by Mocellin et al, 1 the additive model shows a huge advantage, including GWAS, which greatly enlarges the sample size. We look forward to more well-designed studies for rs2274223 variant accumulating and helping us figure out the key role of this variant for carcinogenesis.
Footnotes
YH and CW contributed equally.
Contributors YH and CW: contributed equally to conduct this study and wrote the manuscript. ZW: designed this letter and reviewed the manuscript. ZZ: reviewed the manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.