Characteristics of studies selected
The present study describes the role of Cdx2 and ApaI VDR gene polymorphism and PCOS susceptibility. For the metananlysis, a total of 15 case-control studies were identified and included, which comprises 2114 PCOS patients and 1552 control subjects.
Association between VDR gene Cdx2 (rs11568820) and PCOS susceptibility
A total of five studies relevant to VDR rs11568820 polymorphism and risk of PCOS development have been included. The data extracted from these studies have been summarized in (Table 1) (Wehr et al., 2011); (Dasgupta et al., 2015); (Ali et al., 2018); (Malik et al., 2018); (Szafarowska et al., 2019), which include first author, publication year, methodology, ethnicity, genotypic data. Most of them had a Newcastle-Ottawa scale (NOS) score of more than 5. Statistical data synthesized from four different genetic models: Dominant model (AA + AG vs GG): OR=0.51, 95%CI=0.24-1.09, p=0.08, Recessive model (AA vs. AG+GG): OR=1.59, 95% CI=0.91-2.78, p=0.11, Additive model (AA vs. AA + GG): OR=1.30, 95%CI=0.73-2.31, p=0.37 and Allelic model (A vs. G): OR=0.67, 95%CI=0.41-1.10, p= 0.12 has not shown any significant results (Supplementary Fig 1). However, the correlation becomes significant (OR= 0.40, 95% CI= 0.22-0.72, p= 0.002) after a sensitivity analysis test with the exclusion of one major study by Wehr et al. 2011 (Fig 2).
Assoociation between VDR gene ApaI (rs7975232) and PCOS susceptibility
Data has been summarized from thirteen relevant studies describing VDR ApaI (rs7975232) polymorphism and PCOS susceptibility (Table 2) ; (Al-baidani & Abdulhassan, 2017; Dasgupta et al., 2015; El-Shal et al., 2013; Jedrzejuk et al., 2015; Mahmoudi, 2009; Mahmoudi et al., 2015; Santos et al., 2018; Siddamalla et al., 2018; Sur & Chakravorty, 2015; Szafarowska et al., 2019; Wehr et al., 2011; Xavier et al., 2019); Meta-analysis results synthesized from four different genetic models: Dominant Model (AA + AC vs CC): OR= 0.75, 95% CI= 0.62-0.91,p=0.003, Recessive model (AA vs. AC+CC): OR= 0.78, 95% CI= 0.63-0.98, p= 0.03; Additive model (AAvs.AA+CC): OR=0.70, 95% CI= 0.57-0.87, p=0.001 and Allele model (A vs. C): OR= 0.83, 95%CI= 0.74-0.91, p=0.0002 and there is significant association between ApaI polymorphism and PCOS susceptibility under all genetic models (Fig. 3). Subgroup meta-analys when stratified by Asian and Non-Asian population revealed that VDR ApaI polymorphism significantly associated under Dominant model (AA + AC vs CC): OR= 0.69, 95% CI=0.53-0.90, p=0.006 among Asian population while no significant results were observed for Non-Asian population (Fig. 4).
Sensitivity analysis and publication bias
After performing the odd one-out sensitivity analysis for the VDR Cdx2 variant by the removal of one major study by Wehr et al. 2011, the gene variant is found to be significantly associated (OR= 0.40, 95% CI= 0.22-0.72, p= 0.002) with the PCOS risk. Funnel plot verified no publication bias and forest plot show the significant results as mentioned previously. For ApaI variant analysis, after removing one study by Jedrzejuk et al. 2015, heterogeneity changed from moderate (χ2= 21.91, df=12 (p=0.04), I2 =45%) to low (χ2 = 14.67, df=11 (p=0.20), I2 =25%) indicating this study contribute significantly to heterogeneity. Funnel plot and Forest plot (Supplementary Fig. 2) represented a symmetrical funnel plot indicating no publication bias.
The polymorphisms linked to PCOS in the meta-analysis were further investigated in-silico to determine their likely role and biological mechanism.
In silico analysis of the intronic variants
The variations discovered to be related with PCOS in the preceding meta-analysis were then tested in silico to determine their likely function and biological mechanism underlying such statistical relationships. Regulome DB analysis revealed and Cdx2 and ApaI as a transcription factor binding altering polymorphic variants with a score 5 and 3a respectively. The score 3a implies that there is transcription factor binding at ApaI and this overlap with some motifs. The SNP (rs7975232) is also within DNase peak region which implies it is within open chromatin and is less likely to affect binding of transcription factor, while Cdx2 (rs11568820) with score 5 has minimal binding evidence with the transcription factor. The FATHMM-MKL non-coding score for Cdx2 is found to be 0.92453 which predicts the SNP to be deleterious, while for ApaI this tool has not yielded any significant results.
By analysis of HaploReg v4.1 results it is found that the SNP Cdx2 overlaps with enhancer histone marks in two tissues namely, Gastrointestinal tract and Liver, and is found to change the motif of ERalpha-a and Hoxd10 gene which may be further analysed to know the effects of the given intronic variants. ApaI overlaps with an HMM predicted enhancers in 9 major tissue types. There is a cluster of enhancer activity is shown in mesenchymal, epithelial, digestive and other tissues and that is classified as a genetic enhancer by 25 state model of chromatin state and core 15 state model. Modification at H3K4me1, H2K4me3, H3K27ac all contribute to the chromatin state assignment in epithelial tissue while H3K4me1and H3K9ac are found to affect chromatin state in other tissues. The presence of black cells in the table indicates that DNase was not detected by Roadmap in many tissues but DNase 1 hypersensitivity is found in certain tissues like iPSC (induced Pluripotent Stem Cells). ChipSeq and motif data suggests SNP (rs7975232) is found to change the motif of zinc finger protein CTCF and HEN1 gene. (Supplementary Table 2).