Vitamin D is an essential fat-soluble nutrition from cholecalciferol and steroid pro-hormone, which is predominately obtained from sunlight, dietary sources, and supplementation (1, 2). Vitamin D deficiency is common worldwide, with nearly one billion people experiencing vitamin D deficiency in 2019 (3). In this study, it is reported that 69.2% of the Asian population suffer from vitamin D deficiency, with 61% of postmenopausal women deficient in vitamin D , and the prevalence of vitamin D insufficiency among children is 56.2% (4). The deficiency in vitamin D is critical and is a contributor to the risk of metabolic diseases (5–7), cancers (8, 9), and all-cause mortality (10, 11).

Lowering 25-hydroxyvitamin D (25(OH)D) may act as a trigger for the inflammatory response to disturb the composition and function of circulating cytokines and growth factors (12), influencing endothelial cells (13), hence, increasing the risk of neurodegenerative disorders (14), cardiovascular disease (15, 16), musculoskeletal lesions (17), and mortality (11). The large-scale meta-analysis included 50 randomized controlled trials (RCTs) with a total of 74,655 participants and revealed that vitamin D supplementation statistically significantly reduced the risk of cancer death (RR: 0.85, 0.74–0.97) in adults compared with placebo (18). Moreover, vitamin D supplementation significantly reduced total cancer mortality in an updated meta-analysis of RCTs with 1,591 deaths (19). The meta-analysis comparing the highest with the lowest circulating 25(OH)D concentrations showed a 39% lower risk between levels of total 25(OH)D and colorectal cancer (CRC) risk (OR: 0.61, 0.52–0.71) with 11 case–control studies, and a 20% reduced CRC risk (HR: 0.80, 0.66–0.97) with six prospective cohort studies (20). Moreover, the effects of vitamin D supplementation on decreasing the LDL-c level (21) and releasing insulin resistance (22) were uncovered.

Inversely, the results from an RCT of vitamin D3 (cholecalciferol) at a dose of 2,000 IU per day during a median follow-up of 5.3 years found that supplementation with vitamin D3 did not result in a lower incidence of invasive cancer or cardiovascular events than placebo (23). The negative results were observed in another RCT with 5,108 older adult participants (65.9 years) during a median follow-up of 3.3 years, in which monthly high-dose vitamin D3 (200,000 IU) supplementation did not prevent CVD (24). Meanwhile, the studies did not illuminate the benefits of vitamin D3 supplementation for relapse-free survival of digestive tract cancer at 5 years (25, 26). Although numerous studies explored the association of vitamin D with many clinical diseases, inconsistent results were produced, and the potential confounders may contribute to the unobvious benefit of vitamin D.

Recently, single nucleotide polymorphism (SNP) that was identified by genome-wide association studies (GWASs) was recruited as an instrument in Mendelian randomization (MR) analyses to explore the causal effects of exposures on outcomes (6, 27–30). The MR analysis demonstrated that genetically decreased 25(OH)D was associated with the risk of multiple sclerosis (MS) and provided strong evidence for the causal role of vitamin D in MS susceptibility (8). Moreover, a significant association between 25(OH)D levels and T2DM was found in a European-descent MR with SNP instruments of vitamin D synthesis (31). Inversely, no evidence for the effects of 25(OH)D on T2DM was observed in another MR study with four instrumental SNPs of 25(OH)D concentrations (27). Meanwhile, the delicate associations between genetically predicted 25(OH)D and multiple outcomes were observed, such as in inflammatory bowel disease (IBD) (28), bone mineral density (BMD) (32), non-alcoholic fatty liver disease (NAFLD) (29), and hypertension (33). The genetic instruments of 25(OH)D recruited in a previous MR analysis were limited to four or six variants involved in vitamin D synthesis (DHCR7/NADSYN1, rs12785878 and CYP2R1, rs10741657), transportation (GC, rs3755967), and degradation (CYP24A1, rs17216707), as well as two novel vitamin D metabolism pathways, such as SEC23A (Sec23 homolog A, coat protein complex II component, rs8018720) and AMDHD1 (amidohydrolase domain containing 1, rs10745742). The overall estimate of heritability of 25(OH)D attributable to statistically significant GWAS common SNPs was 2.84% (34). The limited efficiency of SNP instruments may lead to inconsistent results.

In the current study, we conducted a systematic literature review of previous MR studies on 25(OH)D and performed a bidirectional MR analysis to examine the causal association between genetic determinants of 25(OH)D and cardiovascular diseases (CAD), fracture, rheumatoid arthritis (RA), Alzheimer's dementia (AD), inflammatory bowel disease (IBD), chronic kidney disease (CKD), and related clinical risk factors using summary data from large-scale genome-wide association studies.

In the present Mendelian randomization (MR) study, our findings did not support the putative causal effects of 25(OH)D on multiple clinical diseases. Genetically decreased 25(OH)D was significantly associated with the estimated bone mineral density (eBMD), plasma cholesterol, pulse pressure, and elevated lymphocyte count. A bidirectional MR study did not reveal the significant effects of CVD, IBD, T2DM, AD, and musculoskeletal disorders on 25(OH)D concentration. However, chronic kidney disease was positively associated with decreased 25(OH)D.

Dietary and skin-derived vitamin D will play biological roles in multiple organs after hydroxylating in the liver and kidney (67). Patients with chronic kidney disease will decrease the biosynthesis of circulating 25(OH)D by reducing the production of hydroxylase (68). Vitamin D exerts a biological role by binding the vitamin D receptor (VDR) (69), which is commonly found in musculoskeletal cells and various extracellular tissues, such as parathyroid tissue, intestinal tissue, and kidneys (70). Vitamin D deficiency not only increased the risk of rickets, osteomalacia, and osteoporotic fractures (71) but also contributed to extra-skeletal disorders (72) in epidemiological studies. However, no significant associations of 25(OH)D with the risk of clinical diseases were found in our study. Meanwhile, the reversed MR analysis did not reveal any impact of multiple chronic diseases on serum 25(OH)D, except for CKD. The relationship between vitamin D and clinical disease may not be linear, and a linear analysis may yield negative results. A non-linear association of 25(OH)D deficiency with the risk of cardiovascular disease was discovered in a non-linear MR analysis (73), and the non-linear dose–response relationships between 25(OH)D concentrations and coronary heart disease, stroke, and mortality outcomes were discovered in a stratified MR analysis (74).

A weak correlation of 25(OH)D with estimated BMD of heel was found, but not with femoral neck BMD (FN-BMD) or lumbar spine BMD (LS-BMD) in our study. Meta-analysis of RCTs revealed that vitamin D supplementation was not associated with a lower risk of fractures in older adults (75). The MR study showed that the SNP instruments (rs2282679, rs117913124, rs10741657, rs12785878, and rs727479) were not associated with femoral neck BMD (FN-BMD), lumbar spine BMD (LS-BMD), or estimated BMD (eBMD), but rs6013897 near CYP24A1 was associated with FN-BMD at borderline statistical significance (p = 0.01). High 25(OH)D concentration was not associated with higher FN-BMD (p = 0.37) or LS-BMD (p = 0.49) in the inverse-variance weighted analysis or in sensitivity analyses but was associated with estimated BMD (p = 0.02) (76). The differences in the biological pathways of 25(OH)D-SNPs may be the critical factor causing the contradictory result. Mendelian randomization analyses that combine all SNP effects may mask the role of individual SNPs. Moreover, the beneficial effects of vitamin D on many diseases may be largely due to undetected confounders in an epidemiologic study.

In the present study, no significant associations of genetic 25(OH)D with the risk of CAD, stroke, and T2DM were uncovered. The putative causal effects of circulating vitamin D on plasma triglyceride and total cholesterol were uncovered. Previous observational studies revealed that circulating 25(OH)D was inversely associated with blood pressure and the risk of type-2 diabetes (77), but positively correlated with blood lipids (triglycerides, HDL-c, LDL-c, and total cholesterol) (16). Moreover, vitamin D supplementation will increase LDL-cholesterol concentrations (78). The genetic SNPs were not only related to circulating 25-hydroxyvitamin D but also plasma LDL-cholesterol and triglyceride levels. Pleiotropy-associated confounding cannot be completely ruled out (79).

The SNPs used in the previous two-stage and two-sample Mendelian randomization studies were mainly obtained from the SUNLIGHT consortium (study of underlying genetic determinants of Vitamin D and highly related traits) GWAS, which consisted of 31 cohorts from Europe, Canada, and the USA with atotal of 79,366 samples. The statistical effects of all SNPs were calculated by fixed-effect inverse-variance weighted meta-analysis. The joint test with multiple cohorts will easily induce an undetected bias due to the disparate measurements for 25(OH)D and the adjusted covariates. Meanwhile, we were unable to estimate the interactions between genetic variants and dietary intake of vitamin D as well as sunlight exposure as a source of vitamin D production in the skin. In the current study, the adequate instrumental SNPs involved in vitamin D synthesis (DHCR7/NADSYN1 and CYP2R1), transportation (GC), and degradation (CYP24A1), as well as novel vitamin D metabolism pathways, such as SEC23A (Sec23 homolog A, coat protein complex II component) and amidohydrolase domain containing 1 (AMDHD1), were recruited from the UK Biobank study with 401,460 white British participants. Moreover, the association between SNPs and 25(OH)D was estimated by linear mixed-model and the interactions of age, sex, and season with 25(OH)D were evaluated. The subjects in the GWAS analysis investigating the relationship between genetic loci and clinical features were of European and Caucasian descent, which is congruent with the population in UK Biobank. A comprehensive network of Mendelian randomization analysis, inverse-variance weighting (IVW), weighted median, weighted mode, and Mendelian randomization (MR)–Egger regression were performed to explore the causal effects and pleiotropy, as well as avoid bias. Mendelian randomization pleiotropy RESidual Sum and Outlier (MR-PRESSO) was conducted to uncover and exclude pleiotropic SNPs. The methods in our study provided a credible result for the large-scale study.

Here, several strengths of our study should be mentioned. First, the panoptic atlas of 25(OH)D in many clinical diseases and healthy traits were explored in this MR analysis, which will help us to have a more comprehensive understanding of the relationship between 25(OH)D and health. In addition, the bidirectional two-sample MR study was conducted to estimate the influence of diseases on 25(OH)D concentrations. Second, the SNP biomarkers and SNP estimates were obtained in mostly European studies, thus minimizing the possibility of population stratification bias. Third, the genetic SNPs for vitamin D were derived from a recent large-scale GWAS study (n = 443,734) rather than an earlier SUNLIGHT study, which may be more representative of the genetic instruments used to explore the genetic correlation of vitamin D.

However, there are some potential limitations worth noting in the current study.

Although we performed tests to prevent pleiotropy, some genetic SNPs are not only related to circulating 25-hydroxyvitamin D but also to traits, such as plasma LDL-cholesterol and triglyceride levels, and the potential pleiotropy cannot be definitively excluded. Moreover, the non-linear association between 25(OH)D and clinical outcomes could not be assessed by our two-sample MR analysis with a summary dataset. A non-linear MR analysis using raw data is needed in future. Meanwhile, 25(OH)D was recruited as the symbol of serum vitamin D concentration. In addition, the biological roles of active metabolite 1,25-dihydroxy vitamin D in clinical outcomes were not explored. Meanwhile, we still failed to study the biological roles of 24, 25-(OH)₂-D₃, 1, 24, 25-(OH)₃-D₃, and 25, 26-(OH)₂-D₃ metabolites and epimer of vitamin D in health (80). In addition, our results were not likely biased by pleiotropy due to the fact that 25OHD-associated genetic variants were not associated with other lifestyles, which influenced clinical outcomes, such as drinking, physical activity, or smoking. However, we cannot exclude the possibility that such an association may have a genetic basis rather than a causal relationship.

Hence, a large-scale genome-wide scan for genetic variants of vitamin D and further investigation to understand the potential role of vitamin D in the development of clinical outcomes are required. Meanwhile, a long-term and multicentric RCT study that can avoid the interference of numerous known and unpredictable confounders on the results, such as diet, exercise, sleep, geographic latitude, and climate, is seriously needed.

Our study suggested that there was no evidence of the causal effect of 25(OH)D on numerous clinical diseases. Genetically decreased serum vitamin D was associated with estimated bone mineral density evaluated by ultrasound of the heel, plasma cholesterol, pulse pressure, and elevated lymphocyte count. Chronic kidney disease was inversely related to serum 25(OH)D concentration. The putative causal effects of vitamin D on multiple clinical diseases was not supported.

K-qL: concept and design. J-jX: data collection, statistics, and writing the manuscript. X-bZ and TY: data collection. W-tT: data interpretation. K-qL and J-jX: study supervision. All authors critically revised the manuscript and approved the submitted version.