Relationship between the rs2596542 polymorphism in the MICA gene promoter and HBV/HCV infection-induced hepatocellular carcinoma: a meta-analysis

A comprehensive search of three databases (MEDLINE, Web of Science, and Embase) was performed and relevant publications were retrieved by literature traceability, from inception to April 2019. The search phrase used was: (“liver cell carcinoma” or “carcinoma, hepatocellular” or “Hepatocellular carcinoma” or HCC or hepatoma) AND (MICA or “MHC class I polypeptide-related sequence A” or “Human Major Histocompatibility Complex class I polypeptide-related sequence A”) AND (“polymorphism” or “SNP” or “variation” or “variants” or “locus” or “mutation”).

All retrieved publications were screened by stepwise assessment of the title, abstract, and full text, according to the preset inclusion and exclusion criteria described below. Two independent investigators (Luo and Wang) conducted this work simultaneously, and any disagreement was resolved by discussion with a third investigator.

Inclusion criteria: (1) Studies including rs2596542 genotype and allele frequency data from patients with virus-induced live cirrhosis (LC), chronic hepatitis C (CHC)/chronic hepatitis B (CHB), and HCC; (2) Case-control studies; (3) Sufficient data to calculate an odds ratio (OR) and corresponding 95% confidence interval (CI); (4) In repeatedly published studies, the report with higher quality data and more comprehensive outcomes data, was selected; (5) English language.

Exclusion criteria: (1) Letters, notes of meetings, reviews, etc.; (2) Data could not be extracted.

Quality assessment of the studies and data extraction were performed independently by two investigators (Shen and Deng), based on a set of predetermined criteria (Table 1) extracted and modified from previous studies [19‐23]. Any disagreement was resolved by discussion with a third investigator. Quality scores ranged from 0 to 10, with higher scores representing better quality. The following information was extracted: first author, year of publication, country, the characteristics of participants in each study, sample size, number of cases, number of controls, sample source, detection method, and Hardy-Weinberg equilibrium (HWE) in controls.

The chi-square test was used to assess HWE in control populations, with P > 0.05 considered to indicate consistency with HWE. Meta-analysis was performed using SAS 9.2 (SAS Institute Inc., Cary, NC, USA) and STATA 12.0 (STATA Corp, College Station, TX, USA) software. Pooled OR and 95% CI values were calculated to evaluate the association of MICA rs2596542 with susceptibility to HBV/HCV-induced HCC, based on five genetic models: 1) allelic, C vs. T; 2) heterozygous, CT vs. CC; 3) homozygote, TT vs. CC; 4) dominant, TT/CT vs. CC; and 5) recessive, TT vs. CC/CT. When heterogeneity existed among different studies (P value of Q test < 0.10 and/or I² > 50%), pooled OR values were calculated using a random effect model, while a fixed effect model was used when heterogeneity analysis indicated P > 0.10 and/or I² < 50%. Publication bias was checked using Egger’s test, and Begg’s linear regression with funnel plots, and was considered present when P < 0.05, or the funnel plot was asymmetrical [24].

A total of 148 documents were retrieved from PubMed (n = 36), Embase (n = 61), and Web of Science (n = 51) databases, of which 70 were eliminated, due to duplicate publication. The remaining 78 articles were initially screened, based on the title and abstract, and 59 articles excluded for other reasons, including that: they were reviews, abstract compilations, or meeting reports (n = 30); they investigated other genes (n = 4); they only reported the relationship between sMICA expression and HCC (n = 9); or they investigated other diseases (n = 15). Finally, the full text documents of the remaining 20 articles were read, following which, nine were eliminated because they contained unsuitable data, such as corrigendum (n = 1), no extractable data (n = 4), no controls (n = 2), duplicate data (n = 1), and unclear virus infection (n = 1). Finally, 11 eligible publications [15‐18, 25‐31] were included in the meta-analysis (Fig. 1).

The 11 articles included in the study involved 25,677 blood samples from 4582 patients with HCC and 21,095 non-HCC patients. The subjects included Chinese, Japanese, Egyptian, Swiss, Vietnamese, and Italian individuals infected with HCV or HBV. Quality scores were > 6, indicating that this meta-analysis included high quality studies. SNP data from the control groups of the 11 studies were consistent with HWE. The general characteristics of the included studies are presented in Table 2.

Our analyses demonstrated heterogeneity among the results generated using the five genetic models [I² > 50%, P (heterogeneity) < 0.10] (Table 3). Therefore, the results were pooled under a random effects model, and the results of the analysis showed that, under the recessive model, rs2569542 TT was a risk factor for the development of HBV/HCV-induced HCC. Individuals carrying the TT genotype were more genetically susceptible to HCC compared with other individuals [TT vs. CC/CT: OR = 1.248 (95% CI: 1.040–1.499), P = 0.017] (Table 3, Fig. 2).

Subgroup analyses were conducted under the recessive model, according to control type, virus infection, and ethnicity. Patients with the TT genotype had higher susceptibility to HCC, regardless of whether they were HCV-infected (OR = 1.326, 95% CI: 1.101–1.599, P = 0.003), Asian (OR = 1.273, 95% CI: 1.002–1.618, P = 0.048), or were compared to a CHC control population (OR = 1.506, 95% CI: 1.172–1.936, P = 0.001) (Table 4, Additional file 1: Figure S1).

Publication bias was analyzed under the different genetic models, using Begg’s test to plot funnel diagrams. Ten studies among the 11 included articles are presented in the plots, and their data points were scattered and distributed, with the pooled OR value at the center; that is, they basically formed a symmetrical, inverted funnel-shape (Fig. 3). Analysis using Egger’s test demonstrated that all P values were > 0.05, suggesting no publication bias (Table 3).