Main

Vitamin D deficiency leads to multiple adverse health consequences, such as growth retardation and skeletal deformities in children, precipitate osteoporosis, muscle weakness and risk of fracture in the later age.1 Recently, studies suggested that vitamin D insufficiency is also a public health concern that has been associated with multiple diseases in addition to bone health, such as osteoarthritis,2 diabetes3 and cardiovascular disease.4 It has been estimated that over 1 billion people worldwide have vitamin D deficiency or insufficiency.5 Therefore, identification of determinants of vitamin D is particularly important.

Genetics contribute substantially in determining vitamin D metabolism,6 with heritability ranging from 53 to 68.9%.7, 8 Using genome-wide association approach, we previously identified four 25-hydroxyvitamin D (25(OH)D) determining loci, 4p12 (rs2282679 in GC), 11p15 (rs10741657 in CYP2R1), 11q12 (rs12785878 in DHCR7/NADSYN1), and 20q13 (rs6013897 in CYP24A1), in 33 996 individuals of European origins.9 In the current study, we aimed to replicate these four loci in Hong Kong southern Chinese.

In this replication study, 712 unrelated Hong Kong southern Chinese women with serum 25(OH)D data were selected from the Hong Kong Osteoporosis Study Database. The recruitment procedure and exclusion criteria have been detailed elsewhere.10, 11 Briefly, subjects with a history of chronic medical illness, premature menopause age below 40 years, malabsorption, previous major gastrointestinal surgery, metabolic bone disease, endocrine disorders including hyper- and hypothyroidism, or prescription of medication that may affect bone and calcium metabolism, hormone replacement therapy, anti-osteoporosis medication and active vitamin D3 metabolites were excluded from the analysis. Therefore, the primary sources of serum 25(OH)D were expected to be from diet and sunlight. Serum 25(OH)D was measured by a radioimmunoassay (RIA) (DiaSorin, Stillwater, MN, USA) after extraction with acetonitrile. The sensitivity of the assay was 1.5 ng ml−1 with an intra-assay coefficients of variability (CV) of 5.59% at 15.9 ng ml−1 and 11.62% at 58.9 ng ml−1. The inter-assay CV at these two levels were 8.99% and 11.94%, respectively.

Sequenom MassARRAY System (Sequenom, San Diego, CA, USA) was used to genotype the single-nucleotide polymorphisms (SNPs; rs2282679, rs10741657, rs12785878 and rs6013897). Genotyping was repeated in 5% of the samples for verification and quality control; genotype data were confirmed to have a concordance rate of 99.9%. Information on the SNPs studied is provided in Supplementary Table 1. SNPs were tested for association (additive model) with the naturally log transformed serum 25(OH)D (as a continuous variable) and vitamin D insufficiency (categorical variable, defined as 25(OH)D <50 nmol l−1) using a multivariable linear regression model and logistic regression model, respectively. All analyses were adjusted for age, body mass index and season on the measurement day as in our previous study.9 We did not adjust sex because only women were analyzed in this study. We modeled season using categorical variables for summer+autumn (May–October), and winter+spring (November–April). The significance level was set to be 0.00625 in the current study, after correction for four SNPs and two phenotypes studied.

Characteristics of the study cohort are summarized in Table 1. Among 712 subjects studied, 41.8% of them were classified as vitamin D insufficiency. Subjects with vitamin D insufficiency were younger and the serum was less likely to be taken during summer or autumn.

Table 1 Demographic data of the 712 subjects
Full size table

Table 2 shows the association results. For the association with serum 25(OH)D levels, minor allele ‘C’ of rs2282679 near the GC showed a significant association with a β-value of −0.066 (P=9 × 10−5), while it was also significantly associated with increased risk of vitamin D insufficiency with an odds ratio (OR) of 1.51 (95% confidence interval (CI) 1.19–1.93; P=8.6 × 10−4). Previous study suggested that the effect of rs2282679 on 25(OH)D levels is stronger in summer,12 however it was not observed in the current study (Pinteraction>0.4; Supplementary Table 2). Although no association was observed between rs12785878 and serum 25(OH)D levels (continuous variable), minor allele ‘T’ of rs12785878 was nominally associated with reduced risk of vitamin D insufficiency with an OR of 0.79 (95% CI 0.63–0.99; P=0.042). We observed that rs2282679 and rs12785878 were independently associated with vitamin D insufficiency (Supplementary Table 3). In addition, likelihood ratio test showed that addition of rs12785878 in the logistic regression model significantly improved the model with rs2282679, age, body mass index and season (P=0.038). Therefore, we derived a genotype risk score (GRS) by summing risk variants (minor allele of rs2282679 and major allele of rs12785878) for each individual. Each score of the GRS was associated with an increased risk of vitamin D insufficiency with an OR of 1.38 (95% CI 1.17–1.64; Ptrend=1.76 × 10−4; Figure 1). No significant association was observed for the other two SNPs.

Table 2 Association between the four top loci, serum 25(OH)D concentration and vitamin D insufficiency
Full size table
Figure 1
figure 1

Prevalence of vitamin D insufficiency with different genotype risk score.

Full size image

We also assessed the predictive ability of rs2282679 and GRS for vitamin D insufficiency, and the area under receiver operating characteristic curve (AUC) was 0.561 (P=0.005) and 0.576 (P=5 × 10−4), respectively. Although GRS had the highest AUC, the difference between AUC of GRS and rs2282679 was not statistically significant (P>0.5).

The SNP rs2282679 showed consistent and robust association with serum 25(OH)D traits. Although there is substantial difference in genetic background in Chinese residing in different administrative regions of China,13 a recent replication study also showed that rs2282679 was significantly associated with serum 25(OH)D in northern and central Chinese.14 We meta-analyzed the β-value of rs2282679 in northern, central and southern Chinese (Supplementary Figure 1) using inverse variance weighted meta-analysis, a stronger association with an overall β-value of −0.07 (Pmeta=2.56 × 10−27) was observed (Supplementary Table 4).

In the current study, we successfully replicated rs2282679 as a determinant of serum 25(OH)D levels in southern Chinese population, whereas the failure to replicate the associations of the other three SNPs could be due to insufficient power of the current study to detect SNPs with small effect size (Supplementary Figure 2). Nevertheless, the meta-analysis of rs2282679 in southern, central and northern Chinese suggested that rs2282679 had a β-value −0.07 in Chinese population. Interestingly, the effect sizes of rs2282679 and rs12785878 were comparable between Europeans and southern Chinese. The OR of vitamin D insufficiency for those carriers of C allele of rs2282679 was 1.49 and 1.51 in Europeans and southern Chinese, respectively. On the other hand, for those carriers of G allele of rs12785878 was 1.21 and 1.27 in Europeans and southern Chinese, respectively.9 We also showed that these two SNPs are significant predictors of vitamin D insufficiency, despite the predictive ability was only modest; whereas combining lifestyle factors and genetic information may improve the predictive ability.15 Future development of predictive formula of vitamin D insufficiency will be important for early identification of subjects who are susceptible, and early intervention may help reducing its associated diseases.