Basic Characteristics of the Included Literature
A total of 75 relevant papers were retrieved for this study. After de-duplication, reading the titles and abstracts, 40 duplicate and irrelevant literatures were excluded. After reading the full text, 8 papers were finally included, and the basic characteristics of the included papers are shown in Table 1, including 3 case-control studies of esophageal cancer[17–19] and 5 case-control studies of gastric cancer[18, 20–23]; one[18] of the eight papers simultaneously analyzed the correlation between the risk of esophageal cancer, gastric cancer and CTLA-4 + 49A > G gene polymorphism.
Table 1
Basic Characteristics of the Included Literature
Study | Time | Country | Tumor type | Control group source | Staging method | Case group | Control group | Control group HEW/P-value | NOS scale |
GG | GA | AA | GG | GA | AA |
Liu[17] | 2015 | China | ESCC | hospital | PCR-RFLP | 307 | 254 | 43 | 310 | 296 | 58 | 0.28 | 8 |
Sun[18] | 2008 | China | ESCC | hospital | RFLP | 448 | 434 | 128 | 529 | 406 | 73 | 0.68 | 8 |
Cheng[19] | 2011 | China | ESCC | hospital | PCR-RFLP | 54 | 105 | 46 | 90 | 79 | 36 | 0.01 | 7 |
Hadinia[20] | 2007 | Iran | GC | hospital | PCR-RFLP | 6 | 13 | 24 | 14 | 59 | 117 | 0.09 | 8 |
Yu-qin Y[21] | 2012 | China | GC | hospital | PCR-RFLP | 65 | 45 | 8 | 30 | 45 | 21 | 0.59 | 8 |
Hou[22] | 2010 | China | GC | hospital | PCR-ARMS | 94 | 70 | 41 | 107 | 55 | 100 | ༜0.05 | 7 |
Sun[18] | 2008 | China | GC | Hospital | PCR-RFLP | 235 | 235 | 60 | 282 | 209 | 39 | 0.97 | 8 |
Tang[23] | 2016 | China | GC | hospital | PCR-RFLP | 155 | 153 | 22 | 278 | 264 | 48 | 0.179 | 7 |
ESCC: esophageal squamous cell carcinoma; GC: gastric cancer; HWE: Hardy-Weinberg equilibrium in control groups; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism. |
Meta-Analysis of CTLA-4 + 49A > G Single Nucleotide Polymorphisms and Susceptibility to Upper Gastrointestinal Tract Tumors
In the dominant genetic model: GG + GA: AA, the differences in heterogeneity tests between studies were statistically significant, and using a random-effects model, the combined results showed that the correlation between upper gastrointestinal tumor susceptibility and CTLA-4 + 49 locus polymorphism was not statistically significant (OR = 1.15, 95%C1: 0.74–1.81, see Fig. 1.). The funnel plots were basically symmetrical (see Fig. 2.), and the results of Begg's test had a P value of 0.216 and Egger's test had a P value of 0.111, showing no significant publication bias. The combined effect size of each genetic model did not change significantly after sequentially excluding 1 literature individually, suggesting that the correlation between the risk of upper gastrointestinal tumor development and CTLA-4 + 49 locus polymorphism was not statistically significant and the combined results were stable.
In the recessive genetic model GG: GA + AA, the co-dominant model GG: AA, GA: AA, and the allelic genetic model G: A, the differences in heterogeneity tests between studies were statistically significant, all using the random effects model, and the combined results were: recessive genetic model GG: GA + AA (OR = 0.99, 95% CI: 0.74–1.31), see Fig. 3. Co-dominant model GG: AA (OR = 1.14, 95% CI: 0.67–1.95), see Fig. 4. GA:AA (OR = 1.20, 95% CI: 0.80–1.79), see Fig. 5. Allelic model G: A (OR = 1.06, 95% CI: 0.82–1.38), indicating that the development of upper gastrointestinal tumors was not correlated with CTLA-4 + 49 gene polymorphism was not correlated. In the allelic model G:A, the P value of the result of Begg's test was 0.216 and the P value of the result of Egger's test was 0.114 (see Fig. 6.). The combined effect size of each genetic model did not change significantly when 1 literature was excluded individually in turn. When 2 literatures[19, 22] that did not conform to H-W equilibrium were excluded, the combined results GG: GA + AA (OR = 1.07, 95% CI: 0.78–1.43), GG: AA (OR = 1.18, 95% CI: 0.64–2.18), GA: AA (OR = 1.00, 95% CI: 0.70–1.44), and G: A (OR = 1.08, 95% CI: 0.82–1.42), suggesting that the correlation between the risk of upper gastrointestinal tumor development and the polymorphism of the CTLA-4 + 49 locus was not statistically significant.
Association of the CTLA-4 + 49 Locus with Susceptibility to Gastric Cancer
In the dominant genetic model: GG + GA: AA, the recessive genetic model GG: GA + AA, the co-dominant models GG: AA, GA: AA, and the allelic genetic model G: A, the differences in the test of heterogeneity between studies were statistically significant, and using the random effects model, the combined results were: dominant genetic model (OR = 1.50, 95% CI: 0.80–2.81), see Fig. 7. Recessive genetic model GG: GA + AA (OR = 1.23, 95% CI: 0.81–1.88), see Fig. 8. Co-dominant models GG: AA (OR = 1.65, 95% CI: 0.78–3.50), GA: AA (OR = 1.48, 95% CI: 0.82–2.69), allelic genetic model G: A (OR = 1.29, 95% CI: 0.87–1.91), showed no correlation between the development of gastric cancer and CTLA-4 + 49 gene polymorphism, and the Begg and Egger tests in each genetic model also showed no significant publication bias, and the reliability of the conclusions was good.