Vitamin D status and its association with clinical outcomes in Hong Kong Chinese

Abstract

Vitamin D is a human steroid and important to human health, but its optimal level is still controversial. Although evidence suggests that a low vitamin D level is linked to risk of cardiometabolic diseases, its association with stroke and diabetes remains inconclusive. The aim of this study was to evaluate the optimal level of serum 25-hydroxyvitmain D [25(OH)D] and its association with the risk of stroke and diabetes among Hong Kong Chinese. Association of metabolomic profiling with vitamin D was also performed to discover vitamin D-associated metabolites and possible health-related pathways. The prevalence of vitamin D deficiency among 5,276 Hong Kong Chinese participants of the Hong Kong Osteoporosis Study [HKOS] was 43.8% and the prevalence of insufficiency (<75nmol/L) or deficiency (<50nmol/L) was 90.1%. In fully adjusted three-phase segmented regression, the estimated first and second break-point of 25(OH)D was 27nmol/L (95% CI:24-30) and 47nmol/L (95% CI:37-56) respectively.

In multivariable Cox-proportional hazard regression in 3,458 HKOS participants aged <45 years, the lower quintiles of 25(OH)D were significantly associated with higher risk of incident stroke compared with the highest quintile (Quintile 1: Hazard Radio [HR], 1.78; 95% CI, 1.16-2.74 and quintile 4: HR, 1.61; 95% CI, 1.07-2.43). A similar association was observed irrespective of gender and specifically for ischemic stroke, but not hemorrhagic stroke. In penalized regression spline plot, the relationship between vitamin D and risk of stroke was a reverse J-shape, showing the lowest HR at 25(OH)D 70 to 80nmol/L. For the relationship between 25(OH)D and incident diabetes among 4,342 HKOS participants aged 20 years or older (1,395 men, 2,947 women; mean age ± standard deviation [SD]: 54.3±16.5 years), fully-adjusted Cox-proportional hazard regression showed no significant difference in the risk of incident diabetes between the lowest and highest quintile of 25(OH)D. The interaction term between 25(OH)D and serum calcium did not affect the risk of incident diabetes significantly (p=0.700).

Identification of the metabolomic profiling associated with 25(OH)D was evaluated using multivariable linear regression in discovery stage using the HKOS baseline cohort 1 of 315 participants (aged 20 years or above; 93 men, 224 women; mean age ± SD: 48.1±15.8 years) and in the discovery stage using the HKOS follow-up cohort 2 of 275 participants (aged 20 years or above; 12 men, 263 women; mean age ± SD: 56.2±9.6 years). In meta-analysis of data from these two cohorts, 13 metabolites were highly correlated with 25(OH)D, and most were lipid in nature. Docosahexaenoylcarnitine and eicosapentaenoylcholine had the highest correlation, with effect estimate of 0.2554 (p=9.60 x10-9) and 0.1682 (p=4.94x10-7) respectively.

In conclusion, the optimal level of vitamin D was estimated to be 47nmol/L for general bone health among Hong Kong Chinese. 25(OH)D was associated with incident ischemic stroke but not with incident diabetes. In metabolomics, a high majority of vitamin D-associated metabolites were lipid in nature, suggesting that lipid metabolism may be related to the role of vitamin D in possible health outcomes.