To obtain comprehensive data on the FokI (rs2228570), TaqI (rs731236), BsmI (rs1544410), and ApaI (rs7975232) polymorphisms in T1DM, T2DM, and GDM, we performed a systematic search across multiple electronic databases, including PubMed/MEDLINE, Google Scholar, and SCOPUS. The search was conducted from June 15 through the end of September 2024, utilizing predefined keywords and MeSH terms (refer to S1, Table 1). To ensure comprehensive coverage, we also examined the reference lists of relevant studies and prior meta-analyses to identify any potentially overlooked significant studies.

We carried out a thorough search of relevant studies, giving priority to peer-reviewed articles published in English. To broaden the scope of data collection, we also included grey literature and research papers published in Chinese and Arabic. Eligible studies included those with an observational design (either case-control or cohort) that incorporated control groups and examined at least one polymorphism of the vitamin D receptor (VDR) gene (FokI [rs2228570], TaqI [rs731236], BsmI [rs1544410], or ApaI [rs7975232]) in individuals with T1DM, T2DM, or GDM, who were diagnosed based on the World Health Organization (WHO) criteria (1999)33. Studies also needed to provide sufficient data to calculate odds ratios (OR) and 95% confidence intervals (CIs) and report detailed genotype frequencies, along with the total number of cases and controls. Studies were excluded if they were reviews, book chapters, duplicates, lacked sufficient data, or were otherwise deemed irrelevant.

The initial phase of this meta-analysis involved two researchers (HA and AQ) who independently screened study titles and abstracts against the predefined inclusion criteria. After this preliminary review, the full texts of studies deemed potentially relevant were retrieved for further evaluation, while those not meeting the eligibility criteria were discarded. HA, AQ, and AA then systematically extracted key information into a custom Excel spreadsheet, which included details such as the authors, publication dates, SNP frequency data, the number of participants in both patient and control groups, and the overall sample size for each study. Additional recorded variables included the study design, types of biological samples used, key biomarkers for DM (such as HbA1c and fasting or random blood glucose), as well as demographic data like participant age, gender, ethnicity, latitude, and geographical location. Any inconsistencies in data extraction or interpretation were resolved by consultation with the senior author (AA). The included studies were assessed for methodological quality using the Newcastle-Ottawa Scale (NOS) following the guidelines of Stang34. This tool evaluates studies in three categories (Selection, Comparability, and Exposure) with a total of nine criteria. Based on their scores, studies were classified as low (0–3 stars), moderate (4–6 stars), or high quality (7–9 stars), as shown in S1 Table 2 of Supplementary File 1.

In this meta-analysis, we utilized SPSS version 30, adhering to the PRISMA guidelines (see ESF-1, Table 5). For each single nucleotide polymorphism (SNP) examined, at least two studies were required for inclusion. The Hardy-Weinberg equilibrium (HWE) in the control groups was assessed using the chi-squared (χ²) test. The associations between VDR gene polymorphisms and various types of DM (T1DM, T2DM, and GDM) were analyzed using ORs and their respective 95% CIs.

The relationships between VDR gene polymorphisms (FokI, TaqI, BsmI, and ApaI) and the three types of diabetes (T1DM, T2DM, and GDM) were explored through various genetic models. The specified models for the FokI, TaqI, BsmI, and ApaI SNPs included the allelic, recessive, dominant, overdominant, homozygous, and heterozygous models. The symbols for each model were as follows: for FokI, the allelic (T vs. C), recessive model (TT vs. TC + CC), dominant model (TT+TC vs. CC), overdominant model (TC vs. TT+CC), homozygous (TT vs. CC), and heterozygous (TC vs. CC) models, in addition to the (TT vs. TC) model; for TaqI, the allelic (C vs. T), recessive model (CC vs. CT+TT), dominant model (CC+CT vs. TT), overdominant model (CT vs. CC+TT), homozygous (CC vs. TT), and heterozygous (CT vs. TT) models, in addition to the (CC vs. CT) model; for BsmI, the allelic (G vs. A), recessive model (GG vs. GA+AA), dominant model (GG+GA vs. AA), overdominant model (GA vs. GG+AA), homozygous (GG vs. AA), and heterozygous (GA vs. AA) models, in addition to the (GG vs. GA) model; and for ApaI, the allelic (G vs. T), recessive model (GG vs. GT+TT), dominant model (GG+GT vs. TT), overdominant model (GT vs. GG+TT), homozygous (GG vs. TT), and heterozygous (GT vs. TT) models, in addition to the (GG vs. GT) model.

We used Cochran’s Q-statistic (with p < 0.10 indicating statistical significance) and the I-squared (I²) test to assess heterogeneity across studies35. A fixed-effects model (FEM) was applied when the Q-statistic p-value exceeded 0.10 and I² was below 50%, indicating low heterogeneity. Otherwise, a random-effects model (REM) was used36,37. Subgroup analyses and meta-regressions were conducted based on variables such as study year, ethnicity, age, HbA1c, duration of illness, BMI, and sample size to explore sources of heterogeneity.

Furthermore, we performed sensitivity analyses by sequentially excluding individual studies to assess the stability of the pooled estimates. This allowed us to determine whether any specific study disproportionately affected the results. To examine publication bias and the impact of small-study effects, we employed Begg’s funnel plots and Egger’s regression test, considering p-values less than 0.05 as statistically significant38,39.

In the present study, we utilized specific keywords and MeSH terms (refer to Supplementary File 1, Table 1) to conduct a systematic search across several literature databases, including PubMed, Google Scholar, and SciFinder. This initial search retrieved a total of 5,331 studies, as illustrated in the PRISMA flow diagram (Figure 1), which outlines the process of study selection and exclusion. After an exhaustive screening to remove irrelevant and duplicate records, we refined our selection to 174 articles. Of these, only 154 studies met the stringent inclusion criteria for our systematic review. Ultimately, a total of 154 studies were included in the final meta-analysis, adhering to predefined inclusion and exclusion criteria. However, two of the included studies featured cohorts of both T1DM and T2DM patients within the same study; thus, each was treated as two separate studies (one for T1DM and one for T2DM). Additionally, one study included two distinct T1DM cohorts from different cities, which were also considered as two separate studies. Therefore, this meta-analysis incorporated data from 157 studies, including 57 on T1DM, 85 on T2DM, and 15 on GDM, all examining various models of VDR gene polymorphisms, such as FokI (rs2228570), TaqI (rs731236), BsmI (rs1544410), and ApaI (rs7975232). Overall, 49,675 participants were included: 26,450 healthy controls and 23,225 patients with different types of diabetes (T1DM: 8,013, T2DM: 12,206, GDM: 2,306).

Participants ranged in age from 5.8 to 80 years (Supplementary File 1). SNP frequencies in the VDR gene were primarily determined using PCR-Restriction Fragment Length Polymorphism (RFLP) and TaqMan techniques, while other methods were less frequently employed, as detailed in Supplementary File 1, Table 5. Egypt led in contributions with 26 studies, followed by Iraq (16) and India (13). China, Turkey, and Saudi Arabia each contributed 10 studies, while Iran (9), Pakistan (8), Brazil (7), and Chile (4) also made notable contributions. Spain, Germany, Japan, and Jordan each conducted three studies, with Romania, Croatia, Italy, the Czech Republic, and Finland contributing two studies each. Various countries, including Tunisia, Guadeloupe, England, the USA, and others, contributed one study each. A full list is available in Supplementary File 2, Tables 1, 2, and 3. The quality of these studies was evaluated using the Newcastle-Ottawa Scale (NOS)40, yielding scores that ranged from 6 to 9 (Median = 8). Study characteristics and genotype frequencies are summarized in Supplementary File 2, Tables 1, 2, and 3.

A subgroup analysis was performed to assess variations in four VDR gene polymorphisms among different diabetes types—T1DM, T2DM, and GDM. Notably, in the BsmI dominant model, a significant association emerged across the diabetes types, with T2DM showing an effect size of ES = 0.317 (p = 0.017), suggesting a positive association with increased susceptibility to T2DM. Likewise, in the GA vs. AA comparison, a significant effect size of ES = 0.330 (p = 0.008) was observed.

We conducted a meta-regression analysis to identify the primary contributors to heterogeneity within the VDR gene polymorphism data related to DM, as detailed in S1, Table 3. The results showed that, in T1DM, the duration of illness (DOI) was a significant moderator (p < 0.05) and the strongest contributor to heterogeneity, explaining 39% of the variance in the FokI (TC vs. CC) model and 87.8% in the ApaI (G vs. T) model. Additionally, both age and ethnicity significantly contributed to heterogeneity in the FokI (TC vs. CC) model, accounting for 39.0% and 36.0% of the variance, respectively. In contrast, no significant factors were identified to account for the heterogeneity observed in T2DM and GDM.

The initial findings of this study indicate that the FokI (rs2228570) polymorphism, when analyzed under the allelic model (T vs. C) and the heterozygous model (TC vs. CC) across mixed ethnicities, demonstrates a significant association with T1DM in comparison to controls, suggesting a potential trend towards reduced susceptibility to T1DM in carriers of the T allele. Subgroup analyses further revealed significant associations across various ethnic groups; for instance, the TT vs. TC and TC vs. CC models showed notable differences between Asian individuals with T1DM and controls, indicating that these genotypic configurations may confer a decreased susceptibility to T1DM within the Asian population. In T2DM, analysis of the FokI polymorphism across mixed ethnicities (overall) revealed no significant differences between T2DM patients and controls. However, subgroup analyses identified significant associations within specific populations. In the African population, significant differences emerged under the allelic model (T vs. C), recessive model (TT vs. TC+CC), dominant model (TT+TC vs. CC), and homozygous model (TT vs. CC), suggesting an increased susceptibility to T2DM associated with the TT genotype. Conversely, in the Asian population, the heterozygous model (TT vs. TC) demonstrated a significant difference between T2DM patients and controls, implying a potential protective effect against T2DM. In GDM, no significant associations were observed across any genetic models when analyzed under mixed ethnicity.

Recently, a meta-analysis by Zhai et al. indicated no significant overall association between VDR polymorphisms and T1DM risk in the general population. However, subgroup analysis based on ethnicity revealed important insights for the FokI polymorphism: it was associated with a decreased risk of T1DM in European populations, while African populations showed an increased risk under all genotype models31. Another small-scale meta-analysis based on seven studies indicated that no significant associations were found between the FokI polymorphism and T1DM risk30. However, these previous studies did not incorporate all available studies examining various genetic variants in their analyses, potentially limiting the comprehensiveness of their findings. In contrast, our study pooled the overall odds ratios from 38 studies, enabling a more extensive and inclusive analysis. Furthermore, we conducted subgroup analyses based on ethnicity, which enhance the robustness and generalizability of our results by accounting for potential ethnic-specific genetic variations.

The FokI (rs2228570) polymorphism within the VDR gene, located on chromosome 12q13, holds significant implications for diabetes pathogenesis through its influence on immune responses and metabolic pathways41,42. This variant modifies the VDR protein structure through nucleotide substitution of the start codon, potentially impacting its interaction with nuclear co-regulators and altering the transcriptional activity of vitamin D-responsive genes43,44. These alterations can affect pathways linked to immune regulation, beta-cell function, and insulin sensitivity, contributing to diabetes susceptibility. Studies have explored the direct impact of FokI on these pathways, particularly in the context of immune cell behavior and glucose metabolism31,45,46.

Despite extensive research, the association between FokI and T1DM presents varied results across populations. While Shahmoradi et al. (2021) found no overall association30, Zhai et al. (2020) reported a protective effect of the T allele in European cohorts and an increased risk in African populations31. These discrepancies may be attributed to differences in study designs, sample sizes, and regional factors such as UV exposure and dietary vitamin D intake. Future studies should standardize methodologies and include broader sample sizes to better understand these associations, controlling for confounding variables such as lifestyle and socioeconomic status.

For T2DM, which is characterized by insulin resistance and beta-cell dysfunction, significant associations with FokI have been reported, particularly in Asian populations where the T allele is linked to increased risk45. Contrary to previous findings, our results indicate a significant association between the presence of the T allele and a reduced risk of T2DM. This discrepancy highlights the need for additional studies to further explore the relationship between specific alleles and T2DM risk. Genetic predispositions may be enhanced by environmental and lifestyle factors, such as dietary habits and physical activity levels. Examining the role of FokI in combination with other polymorphisms and metabolic risk factors would help clarify its contribution to T2DM susceptibility. Longitudinal studies and genome-wide interaction analyses could provide further insights into these complex relationships.

In GDM, Liu (2021) found FokI significantly associated with increased risk, particularly in Caucasian populations47. This may relate to vitamin D's critical role in insulin secretion and beta-cell function during pregnancy, which is potentially modified by hormonal changes. In our study, we observed that, within the African population, the allelic, dominant, homozygous, and heterozygous models are associated with a lower risk of GDM. In contrast, in the Asian population, the allelic model is associated with an increased risk of GDM. Future research should focus on the interplay between FokI and pregnancy-specific factors to better understand its impact on GDM. Concrete examples, such as intervention trials assessing vitamin D supplementation based on FokI genotypes, could inform personalized prevention strategies.

The second key finding in the present study showed that the TaqI (rs731236) polymorphism has no association with the risk of T1DM when all ethnic groups are combined. However, subgroup analysis detected that the recessive model (CC vs. CT+TT) and homozygous model (CC vs. TT) in South American populations are significantly different among T1DM and controls, suggesting an association with increased risk to T1DM. Likewise, in T2DM, these models, namely the dominant model (CC+CT vs. TT) and heterozygous model (CT vs. TT), are associated with increased risk to T2DM in the African population. No significant difference was found in all genetic models of TaqI SNPs between GDM patients and controls when all ethnicities are combined, suggesting no association between these SNPs and the risk of GDM. It should be noted that we couldn’t perform subgroup analyses due to a lack of studies for each ethnicity.

The TaqI (rs731236) polymorphism, located in the 3' UTR of the VDR gene, affects mRNA stability and VDR expression, which can influence immune responses and insulin signaling31,48. While Zhai et al. found no significant association with T1DM31, population-specific data suggest that TaqI may interact with other immune-related genes, such as HLA-DQ49. This supports the need for multi-gene analyses to elucidate its potential role in T1DM risk. Ethnic variability, particularly trends indicating increased risk in South American cohorts, highlights the importance of exploring region-specific environmental interactions, including sun exposure and dietary habits50.

Aravindhan et al. found that in T2DM, TaqI does not show a strong association overall45, though studies in African populations have linked certain TaqI genotypes to increased risk51. Integrating findings from gene-environment interaction studies and epigenetic analyses could reveal how TaqI modulates glucose homeostasis under different environmental conditions. However, no significant associations were found with the TaqI polymorphism for these complications52. Additionally, Liu et al. found no significant overall association between GDM and TaqI, although pregnancy-related interactions between TaqI and maternal vitamin D levels warrant further study41,47. Larger, multi-ethnic cohort studies that include vitamin D status and hormonal profiles are needed53.

The third major finding of the current study indicates that BsmI SNPs under the overdominant model (GA vs. GG+AA) and (GG vs. GA) models, when all ethnicities are combined, are significantly different among T1DM and controls, suggesting an increasing risk for T1DM when (GA vs. GG+AA) is present and a decreasing risk for T1DM when the (GG vs. GA) model is present. However, subgroup analyses based on ethnicity show that the increased frequency of the allelic (G vs. A) and homozygous (GG vs. AA) models in the African population may be associated with increased risk to T1DM. Likewise, the increasing frequency of the dominant model (GG+GA vs. AA) and heterozygous (GA vs. AA) gene models in the European population, and the overdominant (GA vs. GG+AA) model in the Asian population, may increase the risk of T1DM. However, in the South American population, the increasing frequency of the allelic (G vs. A), recessive model (GG vs. GA+AA), dominant model (GG+GA vs. AA), homozygous (GG vs. AA), and heterozygous (GA vs. AA) models may be associated with a decreased risk to T1DM.

The BsmI (rs1544410) polymorphism, also located in the 3' UTR, is believed to impact VDR expression and consequently affect immune and metabolic pathways. While Shahmoradi et al. observed no overall association with T1DM, specific genetic models suggested nuanced risk variations31. The observed lowered susceptibility to T1DM in American cohorts versus inconsistent results in European and Asian groups could stem from genetic linkage differences and varying environmental exposures. Including visual data representation, such as allele frequency distributions and region-specific odds ratios, would enhance the interpretation and applicability of these findings54.

Aravindhan et al. indicated that in T2DM, BsmI has been associated with risk under the heterozygote model45. Future research should focus on how BsmI modulates inflammatory pathways, as chronic inflammation is a known contributor to T2DM. Functional studies exploring BsmI’s role in immune cell regulation and insulin signaling could offer valuable insights. For GDM, regional differences in BsmI's effect have been noted, with increased risk observed in specific populations47. Investigating the interaction between BsmI, maternal vitamin D levels, and pregnancy-related hormonal changes would clarify its role in GDM. Including environmental data, such as sun exposure and dietary vitamin D intake, could further contextualize these findings.

The fourth major finding of the present study shows that the ApaI (rs7975232) polymorphism under the allelic (G vs. T), recessive model (GG vs. GT+TT), and homozygous (GG vs. TT) models are significantly increased in T1DM compared to controls, which is associated with a decreased risk of T1DM in the overall genetic model where the ethnicities are combined. Subgroup analysis showed no significant differences among different ethnicities. In T2DM, there is no significant difference in all models when all ethnicities are combined compared to controls. However, subgroup analysis showed that in the African population, the recessive model (GG vs. GT+TT) is significantly different between T2DM and controls and associated with a decreased risk of T2DM. In GDM, there are no significant differences in all gene models.

The ApaI (rs7975232) polymorphism, located in intron 8, influences VDR expression, impacting immune function and glucose metabolism. While Shahmoradi et al. reported no overall association with T1DM, allelic models showed protective effects, particularly in African cohorts30. Functional genomic studies should explore how ApaI modulates VDR gene expression under different environmental conditions. For T2DM, potential protective effects of ApaI in certain regions have been reported45, possibly due to its role in enhancing VDR expression. Research focusing on cumulative effects of ApaI with other polymorphisms could elucidate how it impacts metabolic pathways. For GDM, Liu et al. highlighted an increased risk linked to ApaI in specific populations47. Longitudinal studies that include pregnancy-related variables and epigenetic factors would help determine how these interactions influence GDM risk.

Clinical and research implications of genotyping for VDR polymorphisms, particularly FokI, TaqI, BsmI, and ApaI, could suggest personalized diabetes prevention and treatment strategies. Integrating genetic data with environmental factors in multi-gene panels could offer a comprehensive view of diabetes risk. Standardizing study designs and including diverse populations with controlled confounding factors will be crucial for advancing this research. Implementing visual tools, such as allele distribution charts and regional odds ratio plots, would improve data interpretation and clinical applicability. The understanding of VDR polymorphisms as part of a complex genetic landscape influencing diabetes highlights the importance of multifactorial approaches. Combining genetic, environmental, and epigenetic data in future studies could bridge research findings with public health strategies and personalized medical practices.

This study provides a simultaneous comparative evaluation of four VDR SNPs across type 1, type 2, and gestational diabetes. Regarding population specificity and heterogeneity, previous meta-analyses have also failed to emphasize the exact role of each SNP because of inconsistency at the subtype level, where some T1DM analyses found no overall association between the FokI and ApaI polymorphisms, in contrast to our study. Similarly, meta-analyses focused on T2DM did not identify a consistent association for BsmI. Initial significant associations (BsmI, FokI, ApaI in subgroups) were all deemed “less credible” after bias analysis, whereas FokI and BsmI were associated with T2DM overall, and TaqI/ApaI were not significant; this supports a moderate genetic effect, especially for FokI in another meta-analysis. Others suggest polymorphisms affect disease severity rather than risk. These study findings are often population-specific and sometimes inconsistent in direction across ethnic groups. Across meta-analyses, findings on VDR polymorphisms in diabetes are inconsistent and often contradictory, making comparison essential.

In the mechanistic view, FokI affects the protein structure and activity of VDR, while BsmI, ApaI, and TaqI are mostly non-coding variants, influencing mRNA stability or gene expression rather than protein structure. Specifically, these polymorphisms influence insulin secretion (β-cell function), insulin sensitivity (peripheral tissues), and inflammation (immune modulation). Thus, we also attempt to clarify that the clinical value of our work lies in refining the strength and consistency of currently available evidence rather than proposing immediate clinical implementation.