HTML
-
A total of 52 articles were identified after an initial search, 20 of which were duplicate articles (Figure 1). Twenty-three articles were further excluded because of duplicate sample use (5), animal or function studies (4), studies on other diseases (8), and review or comments (6) (Figure 1). From the above, we identified nine articles, which were included in the SUMO4 M55V analysis (Table 1). One study was further excluded because of significant disequilibrium in the Hardy-Weinberg equilibrium test. Finally, eight articles including 10 case-control studies, with a total of 2932 cases and 2679 controls, were included in this meta-analysis. The study quality is summarized in Table 2. The quality of the studies included in this meta-analysis was acceptable (at least 5 points).
Study Year Ethnicity/Region Population Group Subject Size Diagnosis Genotyping Method P-value for HWE Noso S[15] 2007 Western area Japanese T2DM 355 ADA Taqman SNP genotyping assay 0.690 Control 398 0.291 Shimada T[20] 2009 Wakayama Japanese T2DM 423 WHO Taqman SNP genotyping assay 0.716 Control 436 0.094 Shimada T[20] 2009 Tokyo Japanese T2DM 451 WHO 0.408 Control 469 0.378 Ji Z[17] 2010 Hunan Chinese T2DM 427 WHO PCR-RFLP 0.830 Control 281 0.079 Li B[21] 2011 Yunan Chinese T2DM 232 WHO PCR-RFLP 0.260 Control 102 0.126 Lin HY[16] 2007 Taiwan Chinese T2DM 574 NA PCR-RFLP 0.193 Control 323 0.466 Fallah S[22] 2010 Tehran Irani T2DM 50 NA PCR-RFLP 0.493 Control 50 0.998 Hu RT[23] 2009 Va Chinese T2DM 96 WHO PCR-RFLP 0.487 Control 104 0.673 Hu RT[23] 2009 Lalu Chinese T2DM 54 WHO PCR-RFLP 0.088 Control 234 0.817 Pu LM[18] 2012 Beijing Chinese T2DM 270 WHO PCR-HRM 1.92 × 10-5 Control 282 0.086 Note. ADA: American Diabetes Association. WHO: World Health Organization. PCR-RFLP: Polymerase Chain Reaction-Restriction Fragment Length Polymorphism restriction fragment length polymorphism. NA: not reported in the paper. Table 1. Characteristics of Case-control Studies Included in Meta-analysis
Study Case Definition Representativeness of the Case Selection of Controls Definition of Controls Comparability Ascertainment of Exposure Same Method of Ascertainment for Cases and Controls Non-Response Rate Noso S 2007[15] √ √ √ √ √ √ √ √ Shimada T 2009[20] √ √ √ √ √ √ √ Shimada T 2009[20] √ √ √ √ √ √ √ Lin HY 2007[16] √ √ √ √ √ √ √ Ji Z 2010[17] √ √ √ √ √ √ Li B 2011[21] √ √ √ √ √ √ √ Hu RT 2009[23] √ √ √ √ √ √ √ Hu RT 2009[23] √ √ √ √ √ √ √ Pu LM 2012[18] √ √ √ √ √ √ √ √ Fallah S 2010[12] √ √ √ √ √ √ Table 2. Assessment of Case-control Studies Using Newcastle-Ottawa Scale for Evaluating Methodological Quality
-
Figure 2 presents the forest plot of the association between the SUMO4 M55V polymorphism (also named G163A in SUMO4, or rs237025) and T2DM in each studies and the meta analysis. The significant association between the SUMO4 M55V polymorphism and susceptibility to T2DM was observed in the dominant model (GG + GA versus AA: OR = 1.21, 95% CI = 1.05-1.40, P = 0.009), recessive model (GG versus GA + AA: OR = 1.29, 95% CI = 1.06-1.56, P = 0.010), homozygous model (GG versus AA: OR = 1.41, 95% CI = 1.15-1.71, P = 0.001), and additive model (G versus A: OR = 1.18, 95% CI = 1.08-1.29, P= 0.001), and marginally significant in the heterozygous model (GA versus AA: OR = 1.16, 95%CI = 0.98-1.36, P = 0.080).
Figure 2. Forest plots of meta-analysis of the association between SUMO4 M55V polymorphisms and T2DM in a dominant model (A), recessive model (B), homozygous model (C), heterozygous model (D), and additive model (E).
No significant subgroup differences were observed (P ranged from 0.52 to 0.56 in all five genetic models). Subgroup analysis showed significant associations between the SUMO4 M55V polymorphism and susceptibility to T2DM in the Chinese population under all but the heterozygous model (GA versus AA: OR = 1.12, 95% CI = 0.86-1.44, P = 0.090) and marginally significant associations in the Japanese population (P ranged from 0.13 to 0.40 in all five genetic models).
-
Funnel plot asymmetry was evaluated by Egger's regression test. If the line passed through the origin, it would indicate the absence of publication bias. The funnel plots and Egger's linear regression test are shown in Figure 3. No publication biases were detected under the five genetic models (all P > 0.05).
Figure 3. Publication biases indicated by the funnel plots. Figure A, dominant model, P = 0.655 and 0.550 in Begg's and Egger's test, respectively. Figure B, recessive model, P = 0.245 and 0.063 in Begg's and Egger's test, respectively. Figure C, homozygous model, P = 0.074 and 0.131 in Begg's and Egger's test, respectively. Figure D, heterozygous model, P = 0.655 and 0.237 in Begg's and Egger's test, respectively. Figure E, additive model, P = 0.788 and 0.866 in Begg's and Egger's test, respectively.
A sensitivity analysis was conducted to explore the sources of heterogeneity, and the results showed that no single study affected the pooled OR and CIs in a qualitative manner.