Literature search results and basic characteristics of inclusion studies
As was shown in Figure. 1, at first there were 93 studies that meet the search strategy, after elaborately screening, 29 papers were excluded and 64 articles were preserved, then 45 articles were excluded for all kinds of reasons and there were 19 trials preserved in the end11,12,16-32. Table 1 showed the basic characteristics and quality evaluation of the included studies.17 articles were written in English, while 2 were in Chinese, all of which were from 10 different countries. We included 2663 patients and 3252 controls. Table 1 revealed that the quality of these studies were moderate. Table 2 presented the number of various genotypes in both groups in each study.
The correlation between VDR BsmI polymorphism and SLE vulnerability
The forest plot in Figure 2 revealed the link between VDR BsmI polymorphism and SLE susceptibility in 15 trials11,12,17-20,22-26,28-31. We found that there was no correlation between VDR BsmI polymorphism and SLE susceptibility in general populations (BB vs Bb+bb: OR=0.895,95% CI=0.656-1.222,p=0.486; B vs b : OR = 1.222, 95% CI: 0.893-1.672, P=0.211; BB vs bb : OR=1.229,95% CI=0.736-2.053,p=0.431; BB+Bb vs bb: OR=1.566,95% CI=0.995-2.465,P=0.052, Figure 2 and Table 3). When stratified by race, higher link was found in Caucasian populations (BB vs Bb+bb : OR=0.734, 95% CI=0.593-0.909,p=0.005; B vs b : OR=0.865,95%CI=0.760-0.983,P=0.026, Figure.2 and Table 3) and African populations (BB+Bb vs bb : OR=2.935, 95%CI=1.944-4.430; B vs b : OR=1.898, 95% CI=1.458-2.470,P=0.000; P=0.000, Figure 2 and Table 3), However, no link was detected in Asian populations. When stratified by gender,no link was detected in female populations (Table 3).
The link between VDR FokI Polymorphism and SLE vulnerability
As shown in Figure 311,12,16-19,21,26-28,30,32. We found that there was no link between VDR FokI polymorphism and SLE susceptibility in general populations (FF vs Ff +ff : OR=1.084,95%CI=0.722-1.629,P =0.697; F vs f : OR=1.043,95%CI=0.777-1.401,P =0.777; FF vs ff : OR=1.171,95%CI=0.682-2.010,P=0.567; FF+Ff vs ff : OR=1.119,95%CI=0.782-1.600, P =0.539, Figure 3 and Table 3). When stratified by race, higher link was detected in African populations (FF vs Ff +ff : OR= 2.424,95%CI=1.673-3.512, P=0.000; F vs f : OR=1.720, 95%CI=1.417-2.087, P=0.000; FF vs ff : OR=3.154, 95%CI=2.083-4.774, P=0.000;FF+Ff vs ff : OR=1.803,95%CI=1.363-2.384, P=0.000; Figure 3 and Table 3), while no link was found in Asian populations and in Caucasian populations. When stratified by gender, No link was found in female populations.
The correlation between VDR ApaI Polymorphism and SLE vulnerability
As shown in Figure 4 11,12,18-20,26, we found that there was a link between VDR ApaI and SLE susceptibility in general populations (AA vs aa : OR=1.374, 95%CI=1.115-1.692,P=0.003; AA+Aa vs aa : OR=1.342, 95%CI=1.139-1.583, P=0.000, Figure.4 and Table 3).When stratified by race, higher link was detected in Caucasian populations (AA vs aa: OR=1.329, 95%CI=1.016-1.740, P =0.038, Figure 4 and Table 3) and in Asian populations ( AA+Aa vs aa: OR=1.351, 95%CI=1.043-1.749, P =0.023, Figure 4 and Table 3). When stratified by gender, higher risk link was also detected in female populations ( AA vs aa: OR=1.392,95%CI=1.049-1.849,P =0.022, Table 3).
The correlation between VDR TaqI polymorphism and SLE vulnerability
As shown in Figure 511,12,18,19. We found that there was no link between VDR TaqI polymorphism and SLE susceptibility in general populations (TT vs Tt + tt: OR=0.943, 95%CI=0.563-1.581, P=0.824; T vs t: OR=0.989, 95%CI=0.725-1.350, P=0.946; TT vs tt genotype: OR=1.104, 95%CI=0.693-1.758, P=0.678;TT+Tt vs tt: OR=1.151, 95%CI=0.917-1.446, P=0.226; Figure.5 and Table 3). No higher link was detected in any ethnicity or gender subgroup analysis.
Publication bias and sensitivity analysis
When the number of studies included was more than 6, we tested its publication bias. And it was found that VDR ApaI among overall populations had publication bias( AA+Aa vs aa genotype: P=0.029; AA vs aa genotype: P=0.017, Figure. 6). The results before and after adjustment by the clipping method were similar, indicating that this meta-analysis results are stable (Table 4). The sensitivity analysis was performed after removing the study with poor quality and non-satisfaction of the HWE test. The difference between the OR value and the 95% CI change was not statistically significant, indicating a stable analytical result (Table 5).