Vitamin D Status, Vitamin D Receptor Polymorphisms, and Risk of Microvascular Complications Among Individuals With Type 2 Diabetes: A Prospective Study Free

Type 2 diabetes (T2D) is a burgeoning health issue worldwide, with an estimated 537 million people living with diabetes in 2021 (1). Diabetic microvascular complications, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy, are major comorbidities plaguing patients with diabetes, causing blindness, end-stage renal disease, pain, and paresthesia, bringing decline in overall quality of life and heavy burden to society (2,3). Therefore, early identification of modifiable risk factors to prevent or postpone the development of microvascular complications among patients with T2D is of great public health importance.

Vitamin D is a multifunctional fat-soluble micronutrient, mainly synthesized from 7-dehydrocholesterol in the skin following ultraviolet B exposure (4). Recent human studies focused on the association of vitamin D with macrovascular disease have suggested a potential threshold effect of vitamin D status, below which (e.g., 50 nmol/L) risk of cardiovascular diseases would significantly increase (5–7). However, evidence is limited regarding the associations between vitamin D status and microvascular complications in patients with T2D, among whom vitamin D deficiency or insufficiency is particularly common (8). Some but not all cross-sectional studies have suggested potentially inverse associations of vitamin D status and diabetic microvascular complications (9–11), but the evidence from prospective studies is scarce. In addition, vitamin D receptor (VDR), widespread in human bodies, enables vitamin D to perform a variety of physiological effects (3). Some studies suggested effect modification of VDR polymorphisms on the associations of 25-hydroxyvitamin D [25(OH)D] concentrations and composite clinical outcomes (12). However, whether genetic variants in VDR would modify the association between serum 25(OH)D and risk of microvascular complications among patients with T2D is unclear.

To fill these knowledge gaps, in the current study we aimed to prospectively evaluate the association of serum 25(OH)D concentrations with risk of microvascular complications in individuals with T2D from the UK Biobank. Moreover, we further examined whether the associations of interest would be modified by VDR polymorphisms.

The UK Biobank is a large-scale prospective cohort study that enrolled over 500,000 adults between 2006 and 2010. With use of a validated algorithm (13), 24,021 patients with T2D were identified. After exclusion of those without eligible serum 25(OH)D measurement (n = 2,117) or who had CVD (n = 4,755), cancer (n = 1,589), or prevalent diabetic microvascular complications (n = 850) at baseline, or who had withdrawn from the UK Biobank (n = 1), 14,709 patients with T2D were included in the final analysis. For the genetic analysis, we restricted to participants who were of European descent and unrelated. After further exclusion of those with excessive heterozygosity and having highly missing or mismatched sex, 10,943 patients with T2D were finally included in the VDR analysis (Supplementary Fig. 1). All participants gave written informed consent at baseline.

Measurement of biochemical markers has previously been described in detail (14). In brief, chemiluminescent immunoassay method on DiaSorin Liaison XL Analyze was used to determine serum 25(OH)D levels (nanomoles per liter) during the baseline visit. Coefficients of variation of 25(OH)D for low, medium, and high internal quality control samples were 6.14%, 5.39%, and 5.04%, respectively.

Outcomes of interest were diabetic microvascular complications, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy. The hospital inpatient records were used to ascertain incident diabetic microvascular complications according to ICD-10: H360 for diabetic retinopathy; E112, E142, and N180–N189 for diabetic nephropathy; and E114, E144, G590, G629, G632, and G990 for diabetic neuropathy.

VDR polymorphisms have been well characterized, and the current study included four single nucleotide polymorphisms, i.e., rs7975232 (ApaI), rs1544410 (BsmI), rs2228570 (FokI), and rs731236 (TaqI) (15), for exploration of whether the association of interest would be modified by genetic variants in VDR.

Touch screen, self-completed questionnaires were used to obtain information on sociodemographic status, lifestyle and dietary factors, history of medical conditions, and medication at the baseline assessment. We defined healthy diet using the following items: total fruit and vegetable intake >4.5 pieces or servings/week, total fish intake >2 times/week, and processed meat intake ≤2 times per week and red meat intake ≤5 times per week, with meeting at least two items considered healthy (16). According to the months of assessment center attendance, blood draw seasons were divided into spring (March–May), summer (June–August), autumn (September–November), and winter (December–February). Anthropometric measurements were taken at baseline, and BMI was calculated as weight in kilograms divided by the square of height in meters. Physical activity including self-reported moderate- and vigorous-intensity activities was computed as MET minutes per week.

According to the Endocrine Society clinical practice guidelines (17), vitamin D status was divided into four groups: <25.0, 25.0–49.9, 50.0–74.9, and ≥75.0 nmol/L. Baseline characteristics are reported as mean (SD) or n (%) according to baseline serum 25(OH)D concentrations. Person-time was calculated from the date of attending the assessment center to the date of diagnosis of outcomes, death, or the end of follow-up (30 September 2020 for England, 28 February 2018 for Wales, and 31 August 2020 for Scotland)—whichever came first.

Cox proportional hazards regression models were used to estimate the hazard ratios (HRs) and 95% CIs for diabetic microvascular complications, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy. We fitted three statistical models: model 1 included adjustment for age (continuous) and sex (male or female); model 2 was additionally adjusted for ethnicity (White or non-White [including Black, Asian, and multiethnic]), Townsend deprivation index (continuous), education (college or university degree, A/AS levels or equivalent or O levels/General Certificate of Secondary Education [GCSE] or equivalent, other professional qualifications, or none of the above), employed (yes or no), smoking status (never smokers, former smokers, or current smokers), drinking status (never or special occasions only, one to three times per month or one to two times per week, three to four times per week, or daily or almost daily), healthy diet (yes or no), use of multivitamin/mineral supplement (yes or no), total physical activity level (MET minutes per week, continuous), time spent outdoors in summer (hours per day, continuous), season of blood drawn (spring, summer, autumn, or winter), estimated glomerular filtration rate (eGFR) (≤90 or >90 mL/min/1.73 m²), BMI (weight in kilograms divided by the square of height in meters, continuous), history of hypertension (yes or no), and hypercholesterolemia (yes or no); and model 3 was further adjusted for duration of diabetes (≤3, 3–10, or >10 years), diabetes medication (no insulin or pills, only diabetes pills, or insulin and/or others), and glycated hemoglobin (HbA_1c) (<7% or ≥7% [<53 or ≥53 mmol/mol]). We also examine the multiplicative interaction between serum 25(OH)D levels and genetic variants in VDR with additional adjustment for genotype measurement batch and principal components of ancestry. Moreover, risks of diabetic microvascular complications according to joint categories of serum 25(OH)D and VDR polymorphisms were also examined. Serum 25(OH)D concentrations were natural logarithmic transformed when analyzed as a continuous variable. To compensate sample size reduction, multiple imputation (n = 5) was used for covariates with missing values.

For exploration of the dose-response relationship between serum 25(OH)D and risk of outcomes, restricted cubic spline models with three knots (25th, 50th, and 75th percentiles) were performed. Stratified analyses were also performed according to age (≤60, >60 years), sex (male, female), smoking status (never smokers, ever smokers), BMI (<30, ≥30 kg/m²), physical activity (<150, ≥150 MET min/week), diabetes duration (≤3, >3 years), and HbA_1c (<7%, ≥7% [<53, ≥53 mmol/mol]). A series of sensitivity analyses were conducted to test the robustness of our findings. First, we excluded participants where outcomes occurred within 2 years of follow-up. Second, to test the potential mediation by inflammation and lipids, we additionally adjusted for C-reactive protein (CRP) levels (continues, milligrams per liter) and lipid profiles, including HDL, LDL, and triglycerides (TG) (all continuous, millimoles per liter).

All analyses were performed with SAS (version 9.4; SAS Institute, Cary, NC). Two-sided P values <0.05 were considered statistically significant.

Among 14,709 participants with T2D (mean age 59 years, 38.5% female), median serum 25(OH)D concentration was 40.7 nmol/L (interquartile range 27.5, 56.4) (43.1 nmol/L [interquartile range 29.5, 58.4] for White participants, 33.3 nmol/L [24.2, 44.8] for Black participants, and 24.2 nmol/L [16.5, 35.8] for Asian participants) and 65.5% of participants had vitamin D <50 nmol/L. Baseline characteristics according to serum 25(OH)D concentrations are shown in Table 1. Participants with higher concentrations of 25(OH)D were older; were more likely to be male, White, employed, and physically active; and tended to have a healthier diet, lower BMI and HbA_1c, and lower prevalence of hypercholesterolemia. In addition, higher serum 25(OH)D were significantly associated with lower levels of glucose, HbA_1c, total cholesterol, LDL, triglycerides, and CRP (all P_trend <0.001) (Supplementary Table 1).

During a median of 11.2 years of follow-up, 1,370 people developed diabetic microvascular complications (some individuals developed more than one complication), including 527 cases of incident diabetic retinopathy, 781 incident diabetic nephropathy, and 298 incident diabetic neuropathy. Participants with higher concentrations of serum 25(OH)D had lower risk of diabetic microvascular complications (Table 2). In comparisons with participants with serum 25(OH)D <25.0 nmol/L, participants with serum 25(OH)D ≥75.0 nmol/L had a multivariable-adjusted HR of 0.65 (95% CI 0.51, 0.84) for composite diabetic microvascular complications (P_trend < 0.001), 0.62 (0.40, 0.95) for diabetic retinopathy (P_trend = 0.01), 0.56 (0.40, 0.79) for diabetic nephropathy (P_trend < 0.001), and 0.48 (0.26, 0.89) for diabetic neuropathy (P_trend = 0.03). Linear dose-response relationships of serum 25(OH)D (range 10–106 nmol/L) with composite diabetic microvascular complications, diabetic retinopathy, and diabetic nephropathy were also demonstrated (all P_linearity < 0.05) (Fig. 1). One-unit increment of natural log-transformed 25(OH)D was associated with a 25% lower risk of composite microvascular complications, 25% lower risk of diabetic retinopathy, and 33% lower risk of diabetic nephropathy (Table 2). No linear association was observed between serum 25(OH)D concentration and diabetic neuropathy.

As for genetic analysis, no significant differences of serum 25(OH)D concentrations across VDR genotypes were observed (Supplementary Table 2). However, compared with the reference group, participants carrying TT alleles of rs1544410 or GG alleles of rs731236 both had a 23% increased risk of composite diabetic microvascular diseases (Supplementary Table 3). Figure 2 demonstrates the joint associations of serum 25(OH)D concentrations and genetic variants in VDR with the risk of diabetic microvascular complications. In comparisons with participants with 25(OH)D <25 nmol/L and minor allele homozygotes (TT of rs1544410 and GG of rs731236), the multivariable-adjusted HRs of composite diabetic microvascular complications were 0.54 (95% CI 0.38, 0.78) and 0.55 (0.38, 0.80) for participants with serum 25(OH)D ≥50 nmol/L and major allele homozygotes (CC and AA), respectively. No significant interactions of VDR polymorphisms with serum 25(OH)D were observed (Supplementary Table 4).

When analyses were stratified by age, sex, smoking status, BMI, physical activity, diabetes duration, and HbA_1c, the associations were not significantly changed (Supplementary Table 5). In the sensitive analyses, the associations between 25(OH)D concentrations and diabetic microvascular complications remained consistent after exclusion of participants who developed diabetic microvascular complications within 2 years of follow-up, although the inverse association for diabetic neuropathy did not reach statistical significance, probably due to limited cases (Supplementary Table 6). The results were largely unchanged with further adjustment for CRP (Supplementary Table 7) or lipid profiles (Supplementary Table 8).

In this large prospective study of patients with T2D, we found that higher serum 25(OH)D was significantly associated with favorable cardiometabolic profiles and lower risk of diabetic microvascular complications, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy. Linear dose-response relationships were also observed between serum 25(OH)D (range 10–106 nmol/L) and diabetic retinopathy as well as diabetic nephropathy. In addition, these associations were not modified by genetic variants in VDR. A series of stratified and sensitivity analyses demonstrated the robustness of these results.

Although some recent megatrials did not show statistically significant results for the beneficial effects of vitamin D supplements on cardiometabolic diseases (18,19), these studies were generally conducted in vitamin D–replete subjects, e.g., mean serum 25(OH)D level of the participants was 77 nmol/L in the VITamin D and OmegA-3 TriaL (VITAL), while the association of vitamin D with health outcomes may be more significant in populations with deficient or insufficient levels. Additionally, recent Mendelian randomization studies illustrated the potential threshold effect of vitamin D status (5,20). However, for patients with T2D, among whom vitamin D deficiency or insufficiency is particularly common, evidence is limited and inconsistent regarding the association between vitamin D status and microvascular complications. For instance, in two cross-sectional studies investigators found that lower 25(OH)D concentrations were significantly associated with higher risk of diabetic retinopathy and diabetic nephropathy (9,10), whereas investigators from two other cross-sectional studies reported no association with risk of diabetic retinopathy and diabetic nephropathy (11,21). Of note, most of these studies were cross-sectional design and limited by small sample sizes and unconsidered confounding factors, such as physical activity, sun exposure, and dietary factors.

To our knowledge, only one prospective study of 9,795 patients with T2D included examination of the association of 25(OH)D levels with composite diabetic microvascular complications, with no significant association found after multivariable adjustment (22). Of note, this study was not primarily designed to explore the association of 25(OH)D with diabetic microvascular complications. Moreover, the previous study did not differentiate specific microvascular complications. In addition, although results of several small interventional trials suggested that vitamin D supplement may postpone the progression of diabetic nephropathy or ameliorate the symptoms of peripheral diabetic neuropathy (23,24), the VITamin D and OmegA-3 TriaL to Prevent and Treat Diabetic Kidney Disease (VITAL-DKD) yielded no effect of 2,000 IU vitamin D₃ supplement per day on eGFR change (25). However, most of these trials were focused on symptoms or biomarkers of diabetic microvascular complications rather than clinical outcomes. Based on a relatively large sample of patients with T2D (n = 14,709) and a long follow-up period (11.2 years), our study is among the first to find that higher concentrations of serum 25(OH)D were significantly associated with lower risk of composite diabetes complications, as well as the individual end points, i.e., diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy, independent of important confounders including time spend outdoors in summer, physical activity, blood draw season, diabetes medication use, diabetes duration, and HbA_1c levels. Moreover, linear dose-response relationships of serum 25(OH)D with diabetic microvascular complications were also demonstrated.

Although some cross-sectional studies suggested that VDR polymorphisms were associated with vitamin D status (26), serum 25(OH)D concentrations did not differ across VDR genotypes in our study. The discrepancy may be partially due to different ethnicities and population characteristics. In addition, current evidence regarding the association of VDR polymorphisms with diabetic microvascular complications is limited and inconsistent. For instance, in a meta-analysis of eight case-control studies investigators found no significant association between VDR polymorphisms and diabetic microvascular complications (27), while in more recent cross-sectional studies it was observed that BsmI and FokI polymorphisms appeared more frequently in patients with complications and BsmI and TaqI polymorphisms were associated with increased risk of diabetic nephropathy (28,29). Of note, the cross-sectional study design could not clarify causality. Limited sample size and unaccounted confounding variables may also bias the results. In the current study, participants carrying TT alleles of rs1544410 (BsmI) or GG alleles of rs731236 (TaqI) had an increased risk of composite diabetic microvascular diseases. In addition, compared with the reference group, participants with serum 25(OH)D ≥50 nmol/L and major allele homozygotes (CC) of rs1544410 (BsmI) or major allele homozygotes (AA) of rs731236 (TaqI) had a significantly lower risk of composite diabetic microvascular diseases, although no significant interactions were observed. More prospective studies are needed to confirm these findings.

Several potential biological mechanisms may underlie the observed association of serum 25(OH)D with diabetic microvascular complications. First, it has been suggested that vitamin D may inhibit neovascularization and modulate intraocular pressure (30–32). Second, vitamin D is a potent inhibitor of the renin-angiotensin aldosterone system (31) and may mitigate glomerulosclerosis and fibrosis (33,34), regulate apoptosis and autophagy, ameliorate podocyte injury, and maintain glomerular filtration barrier structure (34,35). Third, low vitamin D may lead to diabetic nerve damage and impaired nociceptor function (36). Moreover, vitamin D could reduce the expression of transforming growth factor-β and inflammation, maintain vascular homeostasis, improve glucose and lipid metabolism, and enhance antioxidant defense and immune function (30,36,37). Nevertheless, the exact mechanisms underlying the inverse associations of vitamin D with diabetic microvascular complications need to be further elucidated.

Our study had several strengths, including the large sample size and the prospective study design. Moreover, we simultaneously evaluated both composite diabetic microvascular complications and three major components, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy. Whether the associations of interest would be modified by VDR polymorphisms was also examined. In addition, a number of potential confounding factors were carefully adjusted, such as season of blood draw, time spend outdoor in summer, diabetes medication, diabetes duration, and HbA_1c levels. Several limitations should also be considered. First, causality cannot be established based on this observational study design. Second, a single baseline measurement of serum 25(OH)D might not represent long-term levels, though some studies suggest that a single measured 25(OH)D could act as a reliable proxy for vitamin D status (38). Third, most participants in our study were White British people, which might limit the generalizability of the findings, as more and more evidence has indicated that White people tend to have higher levels of serum 25(OH)D and vitamin D metabolism differs by race and ethnicity (39). Fourth, due to lack of information, we could not account for frequency of clinic visit, which may influence the diagnoses of diabetes complications. Fifth, given that ∼85%–90% of circulating vitamin D is bound to vitamin D binding proteins and biologically active vitamin D only accounts for a very small amount (40), more studies are warranted to evaluate the associations of different vitamin D metabolites [including bioavailable 25(OH)D] as well as vitamin D binding protein with the risk of diabetic microvascular complications in a multiethnic population. Finally, though we adjusted for potential confounding as much as we can, residual confounding or unmeasured factors cannot be entirely excluded.

In this large prospective study of patients with T2D, we found that higher concentrations of serum 25(OH)D were significantly associated with lower risk of diabetic microvascular complications, including diabetic retinopathy, diabetic nephropathy, and diabetic neuropathy. In addition, linear dose-response relationships were demonstrated between serum 25(OH)D and risks of diabetic microvascular complications. No effect modifying of genetic variants in VDR was observed. These findings suggest that maintaining adequate vitamin D status may aid in the prevention of microvascular complications among individuals with T2D. Well-designed randomized control trials are needed to confirm these findings.

Acknowledgments. The authors are grateful to the participants of UK Biobank. This research was conducted using the UK Biobank Resource under application number 68307.

Funding. G.L. was funded by grants from National Nature Science Foundation of China (82073554), the Hubei Province Science Fund for Distinguished Young Scholars (2021CFA048), and the Fundamental Research Funds for the Central Universities (2021GCRC076). A.P. was supported by grants from National Nature Science Foundation of China (81930124 and 82021005) and the Fundamental Research Funds for the Central Universities (2021GCRC075).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. G.L. conceived the study design. X.C. conducted analyses and wrote the first draft of the manuscript. Z.W., T.G., K.Z., R.L., Q.L., X.L., S.L., L.C., Y.G., Z.S., L.L., A.P., and J.E.M. contributed to interpretation of the results and critical revision of the manuscript. All authors approved the final version of the manuscript. G.L. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.