Combined associations of 25-hydroxivitamin D and parathyroid hormone with diabetes risk and associated comorbidities among U.S. white and black women

Black Americans tend to have lower total 25-hydroxyvitamin D [25(OH)D] levels than white Americans, potentially due to reduced cutaneous biosynthesis of vitamin D, and also have higher cardiometabolic risk. The vitamin D–parathyroid hormone (PTH) endocrine system modulates insulin/glucose metabolism, RAAS, endothelial function, immune responses, and vascular/cardiac cell function. Observational studies have linked low vitamin D with diabetes and cardiometabolic disorders (obesity, hypertension, CKD), and emerging evidence suggests potential synergistic effects of vitamin D and PTH. However, few studies have examined joint associations of circulating 25(OH)D and PTH with diabetes and related comorbidities in a multiethnic cohort. This study aimed to evaluate (1) joint associations of 25(OH)D and PTH with incident diabetes among U.S. white and black postmenopausal women and (2) their joint associations with diabetes-related cardiometabolic comorbidities (obesity, hypertension, CKD) comparing women with versus without diabetes.

Prior observational studies and meta-analyses generally show an inverse association between circulating 25(OH)D and incident diabetes, with dose-response patterns and possible thresholds in women. A U.S. cohort of older Caucasians and a German cohort reported inverse associations between 25(OH)D and diabetes incidence after multivariable adjustment. A Canadian RCT among high-risk individuals with vitamin D insufficiency found vitamin D3 supplementation improved insulin sensitivity and beta-cell function over 6 months. Joint vitamin D–PTH data are limited: a Canadian postpartum cohort found vitamin D deficiency/insufficiency combined with high PTH predicted worsening beta-cell function and higher glucose; a Greek case-control study found combined low 25(OH)D and high PTH associated with higher fasting glucose. Conflicting evidence includes null associations in an elderly Italian cohort (potential competing mortality risk and underascertainment) and a large RCT in U.S. adults with prediabetes and adequate baseline vitamin D status, where 4000 IU/day did not reduce diabetes risk. Meta-analysis of RCTs showed no overall effect on diabetes incidence, though dose-response suggests >4000 IU/day may be needed to reach >90 nmol/L and improve glycemic markers. For comorbidities, U.S. NHANES showed inverse relationships between 25(OH)D and systolic blood pressure among whites; a Dutch study linked higher PTH (but not 25(OH)D) to hypertension. Elevated PTH has been associated with CKD risk independent of 25(OH)D in Taiwanese adults, while lower 25(OH)D predicted incident CKD in a Chinese elderly cohort. Evidence regarding obesity is mixed, though many studies show inverse 25(OH)D and positive PTH associations with adiposity.

Design and population: Ancillary case-cohort study within the Women's Health Initiative-Observational Study (WHI-OS), which enrolled 93,676 postmenopausal women aged 50–79 years across 40 U.S. clinical centers (1994–1998). This analysis included a race-stratified sample of 4,850 women (3,000 non-Hispanic white; 1,850 non-Hispanic black) without baseline stroke, myocardial infarction, or dialysis. For prospective diabetes analyses, after excluding 37 with missing diabetes follow-up, 4,191 women without diabetes at baseline were followed for a mean of ~11–12 years.

Outcomes: Prevalent diabetes at baseline defined by self-reported physician diagnosis (not during pregnancy) or fasting plasma glucose ≥6.99 mmol/L. Incident treated diabetes during follow-up defined by self-report of new physician diagnosis treated with oral drugs or insulin; validity supported by medication inventories. Comorbidities at baseline: obesity (BMI ≥30 kg/m² by measured height/weight), hypertension (self-reported treatment and/or self-reported diagnosis and/or measured SBP ≥140 mmHg or DBP ≥90 mmHg), CKD (eGFR <60 mL/min/1.73 m² calculated using the CKD-EPI equation). A composite endpoint captured presence of any of obesity, hypertension, or CKD.

Biomarkers: 12-hour fasting baseline plasma stored at −80°C. Total 25(OH)D assayed by enzyme immunoassay (Immunodiagnostic Systems). PTH, creatinine, high-sensitivity C-reactive protein (hs-CRP), fasting glucose, fasting insulin measured by electrochemiluminescence immunoassay (Roche E Modular). Intra-assay CVs: 6.95% (25(OH)D), 3.46% (PTH), 1.82% (creatinine), 3.34% (hs-CRP), 3.26% (glucose), 2.49% (insulin). eGFR computed with CKD-EPI algorithm.

Covariates: Age, race (white/black), clinical center latitude band, education, season of blood draw, smoking, alcohol use, physical activity (MET-h/week), family history of diabetes and CVD, history of high cholesterol, postmenopausal hormone therapy, statin use, and BMI (kg/m² by measurement).

Statistical analyses: Race-stratified inverse-probability weighting (Barlow approach) accounted for case-cohort sampling fractions among whites and blacks. Among women without baseline diabetes, weighted Cox proportional hazards models estimated associations of 25(OH)D and PTH (continuous and quartiles) with incident diabetes; interaction terms tested race differences. Joint exposure groups defined by 25(OH)D (<50 vs ≥50 nmol/L) and PTH (≤6.89 vs >6.89 pmol/L [65 pg/mL]) with ≥50 and ≤6.89 as reference. Cross-sectional weighted logistic regression assessed independent and joint associations of 25(OH)D and PTH with prevalent comorbidities and the composite endpoint among all women and stratified by race and diabetes status. Main models adjusted for age, race/ethnicity, clinical center, education, season, smoking, alcohol, physical activity, postmenopausal hormone therapy; BMI additionally adjusted for CKD and hypertension models. Sensitivity analyses further adjusted for eGFR (where applicable), history of high cholesterol, and statin use. Spearman partial correlations (age-adjusted) examined relationships among biomarkers, stratified by race.

• Sample and events: Among 4,191 women without baseline diabetes, 453 incident diabetes cases occurred over ~12.1 years (incidence rate 9.5 per 1,000 person-years). At baseline, 3,322 prevalent cases of any comorbidity were identified: obesity n=1,629; hypertension n=2,759; CKD n=318.
• Correlations (non-diabetic women): 25(OH)D inversely correlated with PTH (overall r≈−0.36), creatinine, hs-CRP, glucose, and insulin; PTH positively correlated with creatinine, hs-CRP, glucose, and insulin. Patterns varied by race, but inverse 25(OH)D–insulin and positive PTH–hs-CRP/glucose/insulin correlations persisted.
• Prospective diabetes risk: Higher 25(OH)D associated with lower incident diabetes risk in a dose-response manner among all participants (HR per SD increase = 0.78; 95% CI: 0.68–0.88; P trend <0.0001) and among white women (HR per SD = 0.73; 95% CI: 0.62–0.86; P trend <0.0001). PTH was not significantly associated with incident diabetes after full adjustment.
• Joint exposures and incident diabetes: Versus 25(OH)D ≥50 nmol/L and PTH ≤6.89 pmol/L, vitamin D deficiency (<50 nmol/L) was associated with higher diabetes risk regardless of PTH: HR = 1.92 (95% CI: 1.42–2.58) with PTH >6.89; HR = 1.32 (95% CI: 1.12–1.57) with PTH ≤6.89. In white women, HR = 2.30 (95% CI: 1.51–3.49) for deficiency plus PTH excess and HR = 1.46 (95% CI: 1.20–1.78) for deficiency alone. No comparable association was evident among black women.
• Cross-sectional comorbidities: Individually, lower 25(OH)D and higher PTH were associated with higher prevalence of obesity and the composite endpoint across races; associations with hypertension were significant in white women only (e.g., OR for hypertension per lower 25(OH)D = 0.79; 95% CI: 0.72–0.87; per higher PTH = 1.55; 95% CI: 1.39–1.73). PTH was positively associated with CKD in both races.
• Joint exposures and comorbidities: Compared with high 25(OH)D/normal PTH, either vitamin D deficiency alone or PTH excess alone was associated with higher prevalence of obesity, hypertension, or CKD regardless of diabetes status. The combination of vitamin D deficiency and PTH excess was strongly associated with ≥1 comorbidity among women without diabetes (OR = 4.23; 95% CI: 2.90–6.18) but not among women with diabetes or who developed diabetes (OR = 1.89; 95% CI: 0.96–3.71).
• Race interactions: Weighted distributions of 25(OH)D and PTH differed by race and by diabetes/comorbidity status (all P < 0.0001). Interactions by race were noted for 25(OH)D with obesity (P interaction = 0.007) and for PTH with hypertension (P interaction = 0.008).

The study addressed whether circulating 25(OH)D and PTH jointly influence diabetes risk and related comorbidities and whether associations differ by race. Findings demonstrate that vitamin D deficiency is associated with higher diabetes risk among white postmenopausal women, independent of PTH, with suggestive additive synergism when combined with PTH excess. Joint vitamin D–PTH associations were also observed for prevalent cardiometabolic comorbidities, particularly among women without diabetes, suggesting combined assessment may better capture vitamin D–PTH axis dysregulation relevant to cardiometabolic health. These results align with prior observational evidence linking low 25(OH)D to incident diabetes and with studies indicating PTH’s relations to hypertension and CKD, while helping reconcile mixed trial results that often included participants with adequate baseline vitamin D. Potential mechanisms include vitamin D’s negative regulation of RAAS, differential aldosterone sensitivity by race, PTH’s stimulation of aldosterone and renin-angiotensin signaling, and effects of vitamin D/PTH on adipocyte metabolism, inflammation, and endothelial function. Racial differences in vitamin D/PTH physiology and thresholds likely contribute to heterogeneity in associations, underscoring the importance of race-specific evaluations.

In a large cohort of U.S. white and black postmenopausal women, vitamin D deficiency was associated with increased risk of incident diabetes among white women regardless of PTH levels. Either vitamin D deficiency or excess PTH was associated with higher prevalence of obesity, hypertension, or CKD irrespective of diabetes status, with the combination of deficiency and PTH excess particularly predictive among women without diabetes. These findings support the utility of joint 25(OH)D and PTH assessment for cardiometabolic risk stratification and motivate larger longitudinal studies to confirm race-specific thresholds and trajectories of these biomarkers. If confirmed, randomized trials among individuals with true vitamin D deficiency are warranted to clarify preventive dosing and address potential racial differences in cardiometabolic risk mediated by the vitamin D/PTH axis.

• Cross-sectional analyses for comorbidities preclude establishing temporal relationships between 25(OH)D/PTH and development of obesity, hypertension, or CKD.
• Single baseline measurements of 25(OH)D and PTH may introduce random measurement error, likely biasing associations toward the null, though assay CVs were <10%.
• Residual/unmeasured confounding cannot be ruled out (e.g., sun exposure, dietary/supplemental vitamin D intake).
• Generalizability is limited to postmenopausal women and may not extend to men or other populations.