Vitamin D (as the 1,25-dihydroxyvitamin D metabolite) is a steroid hormone that is well known for its role in maintaining Ca homeostasis and normal bone structure. Recent evidence suggests that in addition to Ca homeostasis, the vitamin may also play a role in a variety of other physiological processes such as modulation of inflammatory pathways(Reference Cohen-Lahav, Douvdevani and Chaimovitz1) and susceptibility to diabetes(Reference Boucher2), cancer(3), and infectious(Reference Baeke, Takiishi and Korf4) and cardiovascular(Reference Reddy Vanga, Good and Howard5) diseases. Thus, the nutrient could play a significant role in public health.
In the USA, the Institute of Medicine recently proposed ≥ 50 nmol/l (20 ng/ml) as the definition of vitamin D adequacy based solely on requirements to optimise bone health, due to a lack of data to support recommendations for the prevention of other disease endpoints(Reference Ross, Taylor, Yaktine and Del Valle6). However, many leading vitamin D researchers continue to recommend serum 25-hydroxyvitamin D (25(OH)D) concentrations of ≥ 75 nmol/l (30 ng/ml) to achieve the broader health benefits(Reference Vieth, Bischoff-Ferrari and Boucher7–Reference Schwalfenberg and Whiting10).
Season, UVB exposure, skin pigmentation, age, race, sex, obesity and dietary/supplemental vitamin D intake have all been previously reported to influence serum 25(OH)D concentrations(Reference Holick11). However, the effect of genetic variation in the vitamin D synthesis and metabolism pathway on circulating concentrations is less well understood. Vitamin D enters the circulation through the activation of vitamin D precursors by UV radiation on the skin to produce cholecalciferol, or via absorption of dietary or supplemental ergo- or cholecalciferol from the intestinal tract. It is then converted to 25(OH)D via 25-hydroxylases (cytochrome P450 (CYP) 2R1, 27A1 and 3A4) in the liver(Reference Holick11). Further hydroxylation of 25(OH)D via 1α-hydroxylase (CYP27B1) in the kidney or at the local tissue level produces 1,25-dihydroxycholecalciferol(Reference Holick11). Catabolism of vitamin D metabolites occurs via 24-hydroxylase (CYP24A1)(Reference Holick11). Vitamin D binding protein (also known as group-specific component, GC) is the transport protein for vitamin D metabolites in circulation(Reference Speeckaert, Huang and Delanghe12). Genetic variation in any of these steps has the potential to alter serum 25(OH)D concentrations.
Previous studies have identified SNP in the vitamin D receptor (VDR)(Reference Ma, Stampfer and Gann13–Reference Smolders, Damoiseaux and Menheere20), CYP2R1 (Reference Ramos-Lopez, Bruck and Jansen21–Reference Bu, Armas and Lappe26), CYP27B1 (Reference Orton, Morris and Herrera22, Reference Ramos-Lopez, Kahles and Weber23, Reference Hypponen, Berry and Wjst27) and GC (Reference Engelman, Fingerlin and Langefeld19, Reference Wang, Zhang and Richards24–Reference Bu, Armas and Lappe26, Reference Kurylowicz, Ramos-Lopez and Bednarczuk28–Reference Janssens, Bouillon and Claes35) genes to be associated with alterations in serum 25(OH)D concentrations. These studies were primarily conducted among Caucasian populations living at higher latitudes with significant seasonal variation in UV exposure, and only a small number of studies have considered environmental vitamin D exposures or personal characteristics as potential effect modifiers.
Data on the distribution and determinants of serum 25(OH)D concentrations in Southeast Asian populations are limited. In this cross-sectional observational study, we evaluated the distribution of serum 25(OH)D concentrations, and identified genetic, dietary and lifestyle predictors of serum 25(OH)D concentrations among middle-aged and older Chinese men and women in Singapore. As Singapore is 1° north of the equator, this study population provides a unique opportunity to evaluate the factors associated with vitamin D status in the absence of seasonal variation in UV exposure, which confounds studies conducted at higher latitudes.
Methods
The Singapore Chinese Health Study (SCHS) is a population-based prospective cohort study of 63 257 Singapore Chinese men and women (aged 45–74 years) that was assembled between 1993 and 1998 to elucidate the role of diet and genetic factors in the causation of human cancer. Participants in the study were recruited from among permanent residents or citizens of Singapore who resided in government-built housing estates, and were one of the two major dialect groups of Singapore Chinese, Hokkien or Cantonese. At recruitment, each study subject was interviewed in person by a trained interviewer using a structured questionnaire that included questions on lifestyle, health behaviours and sociodemographic factors, as well as a 165-item FFQ. The FFQ was specifically designed to assess the dietary habits of Chinese in Singapore, and was subsequently validated using multiple 24 h dietary recalls(Reference Hankin, Stram and Arakawa36). However, because vitamin D is technically challenging to measure in food(Reference Byrdwell, Devries and Exler37), data on vitamin D contained in the nutrient databases used to analyse FFQ for epidemiological studies (including the US Department of Agriculture database used for the SCHS vitamin D analysis) are known to be incomplete(Reference Holden, Lemar and Exler38). Thus, our estimates probably underestimate the participants' actual dietary vitamin D intake. The only form of supplemental vitamin D intake to be assessed was cod liver oil; data on frequency of use over the year before the interview were collected and considered for use in this analysis. Participants were not asked specifically about time spent indoors v. outdoors; however, participants were asked about the average number of hours spent ‘sitting at work’ each day, and hours spent doing ‘vigorous work such as moving heavy furniture, loading or unloading trucks, shovelling or equivalent manual labour’. Responses to these questions were included in our analyses as surrogates for time spent indoors (‘sitting at work’) and outdoors (‘vigorous work’).
Beginning in April 1994, a random 3 % sample of cohort participants was also asked to provide blood or buccal cells, and spot urine samples as a pilot study to determine the feasibility of a larger biospecimen collection effort within the cohort. Details of the biospecimen collection, processing and storage procedures have been described(Reference Koh, Yuan and Sun39). The first 504 SCHS participants who provided biospecimens were included in this substudy. This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects were approved by the Institutional Review Boards of the National University of Singapore and the University of Minnesota. Written informed consent was obtained from all participants.
Serum 25-hydroxyvitamin D assay
Quantitative determination of serum 25(OH)D concentrations was performed using a direct, competitive chemiluminescence immunoassay on the DiaSorin LIAISON platform(Reference Ersfeld, Rao and Body40) by Heartland Assays, a laboratory participating in the Vitamin D External Quality Assessment Scheme(Reference Carter, Carter and Gunter41). This assay measures total 25(OH)D (both the ergocalciferol (D2) and cholecalciferol (D3) derived 25(OH)D metabolites; the assay does not discriminate between the two forms). Two types of blinded controls were randomly interspersed among the study samples for quality control purposes: (1) pooled blood samples (n 10), and (2) vitamin D standard (standard reference material 972, Vitamin D in Human Serum) from the US National Institute of Standards and Technology (NIST)(Reference Tai, Bedner and Phinney42) (n 16). Mean inter- and intra-batch CV for the 25(OH)D concentrations in the pooled blood samples was 6·1 and 5·4 %, respectively, which is consistent with those of previously published reports(Reference Giovannucci, Liu and Rimm43–Reference Cauley, Lacroix and Wu47). For the NIST standard level 1 (58·0 nmol/l 25(OH)D3 in unaltered human serum), the mean inter-batch CV was 6·2 %. Similarly, for the NIST standard level 2 (4·18 nmol/l 25(OH)D2 and 30·0 nmol/l 25(OH)D3 in diluted human serum), the mean inter-batch CV was 5·2 %. The CV for the NIST standards were well below those of previously published reports(Reference Gallicchio, Helzlsouer and Chow48). A number of studies have demonstrated that 25(OH)D is extremely stable with long-term storage (as long as 40 years) and under a variety of storage conditions(Reference Zerwekh49–Reference Bodnar, Catov and Wisner52). Thus, our samples, which were stored for an average of 13 (sd 1·2) years, should have experienced little, if any, 25(OH)D degradation, and any degradation would have been consistent across all samples.
SNP selection
Using a candidate pathway approach, we included common genetic variants (minor allele frequency ≥ 20 %) of genes involved in 25(OH)D synthesis (CYP2R1, CYP3A4, CYP27B1), transport (GC), gene transcription (VDR) and catabolism (CYP24A1). Data from the International HapMap Project (Tagger Pairwise method, HapMap Data Release 27 Phase II+III, February 2009, on NCBI B36 assembly, dbSNP b126 for the Han Chinese in Beijing, China (CHB) population) were used to identify haplotype tagging SNP for the Han Chinese. No HapMap tag SNP were identified for CYP27A1 in the CHB population. Also, any SNP that had been documented in the literature as having functional and/or phenotypic relevance were included.
A total of fifty-five SNP and eight proxies met the inclusion criteria (by rs (reference SNP) number, proxies in parentheses): VDR: 731236 also known as TaqI, 1540339, 1544410 also known as BsmI, 2107301 (12717991), 2189480, 2228570 also known as FokI (and previously reported as rs10735810), 2238136, 2239180, 2239182, 2239184, 2239186, 2254210, 2283342, 2525044, 2853564, 2853559, 3782905, 3847987, 4760658, 7305032, 739837, 757343, 7965281, 7975232 also known as ApaI, 10783215, 10875695 (7136534), 11168268 (11168266), 11168275, 11168287, 11574143, 12721364. CYP2R1: 1993116, 12794714 (10500804), 10741657. CYP3A4: 2242480, 2246709, 3735451. CYP27B1: 4646536. CYP24A1: 912505, 2181874, 2209314, 2296241 (2585428), 2762941, 3787555, 4809958 (6068816), 6022999. GC: 4588 (2298850), 7041, 10488854, 12512631, 16847015, 222016, 2298849 (3733359), 705117, 705120.
DNA extraction and genotype determinations
DNA extraction and genotype determinations were performed by the University of Minnesota's BioMedical Genomics Center. DNA was extracted from buffy coats using a Qiagen Kit (Qiagen, Inc.). Genotype determinations were performed on the commercially available high-throughput genotyping SequenomMassARRAY platform (Sequenom, Inc.). Uniquely located negative controls were routinely included in each plate. These wells were used as controls for genotyping assays, and their unique locations serve as a fingerprint to identify the plate and its orientation. For quality control purposes, only SNP with >95 % call rates were included in the analyses. All SNP were found to be in Hardy–Weinberg equilibrium, except GC rs4588. Genotyping for rs4588 was repeated using a TaqMan assay (Invitrogen). The data from the TaqMan assay were found to be in Hardy–Weinberg equilibrium, and were used for all analyses.
Data analysis
The distribution of serum 25(OH)D in the study population was markedly skewed towards low values. Thus, the statistical analyses were performed on logarithmically transformed values, and geometric means are presented. ANOVA methods were used to compare mean serum concentrations of 25(OH)D by potential lifestyle, sociodemographic, dietary and genetic predictors of 25(OH)D concentration. Dialect group, education level, menopausal status (women), BMI, height, weight, body surface area, physical activity, smoking status, hours spent sitting at work, season of blood draw, use of cod liver oil supplements and dietary intake of vitamin D, Ca, fish, dairy products and alcohol were considered as potential predictors. Age, sex and time interval between last meal and blood draw were included as covariates in all analyses. Dietary vitamin D and hours spent sitting at work were also included as covariates in ANOVA for the assessment of genetic predictors. To test for linear trend, the potential predictor was included as an ordinal variable in general linear models. Each potential predictor was examined individually by assessing its effect on the overall model fit (R 2, F-test). The final multivariate model for non-genetic predictors of serum 25(OH)D concentrations included age, sex, dietary vitamin D intake, hours spent sitting at work, and time interval between last meal and blood draw.
Associations between single SNP and serum 25(OH)D concentrations were evaluated individually in the multivariate-adjusted models. Trends in serum 25(OH)D concentrations across each SNP genotype were tested for statistical significance by including the SNP as a three-level variable (homozygous wildtype, heterozygous and homozygous variant).
All P values quoted are two-tailed, and significance was defined as P< 0·05. For the SNP analyses, statistical significance was defined as P≤ 0·01 to minimise the likelihood of reporting false-positive findings due to multiple comparisons. Calculations were performed using the SAS statistical software system (SAS Institute).
Results
The study population was 56 % women, and the mean age of study participants was 55·7 years (Table 1). Most women were postmenopausal at baseline. Compared to men, women were less educated, less likely to be a smoker, spent less time sitting at work or in vigorous work, and consumed less dietary vitamin D. Mean serum 25(OH)D concentration was 68·6 nmol/l overall, and lower in women (mean: 64·2 nmol/l) than in men (74·3 nmol/l, P< 0·001), and a greater percentage of women (18 %) had 25(OH)D concentrations < 50 nmol/l compared to men (9 %).
25(OH)D, 25-hydroxyvitamin D.
* χ2 or t test P-values for differences by sex for categorical and continuous variables, respectively.
Serum 25(OH)D concentrations were statistically significantly associated with dietary vitamin D, Ca and dairy product intake among women, but not men (Table 2). Serum 25(OH)D levels decreased with increasing number of hours spent sitting at work for both men and women, although the linear relationship was not statistically significant for women. There were no associations with 25(OH)D concentrations for cod liver oil supplement use, fish intake or time between last meal and blood draw, regardless of sex (data not shown). Women engaging in vigorous work for at least half an hour per week showed significantly higher serum 25(OH)D level compared to their counterparts with no vigorous work, whereas there was no difference in serum 25(OH)D level between men with and without vigorous work. Age, BMI, alcohol intake and working status were not associated with serum 25(OH)D concentration in this population. When age, sex, dietary vitamin D intake, hours spent sitting at work, and time interval between last meal and blood draw were considered simultaneously in the final multivariate model, 10·2 % of the variation in serum 25(OH)D concentrations was explained (P< 0·01).
* Mean values were adjusted for age (years), sex (among overall) and time interval from last meal to blood draw.
† Asian-specific BMI cut-points were used(Reference Choo54).
‡ Median values of dietary vitamin D intake for tertiles (μg/4·184 kJ) were: 0·8, 1·4 and 2·5, respectively, for all subjects; 0·9, 1·4 and 2·4, respectively, among men; and 0·8, 1·6 and 2·8, respectively, among women.
§ Median values of dietary Ca intake for tertiles (mg/4·184 kJ) were: 177, 214 and 362, respectively, for all subjects; 168, 217 and 309, respectively, among men; and 185, 260 and 395, respectively, among women.
¶ Median values of dairy product intake (g/4·184 kJ) were: 1·3, 14·4 and 118, respectively, for all subjects; 1·1, 10·8 and 77·0, respectively, among men; and 1·5, 17·3 and 134, respectively, among women.
∥ Among women, only geometric means for none or any alcohol and beer intake is presented, as only nine women reported drinking any alcohol.
Of the fifty-five SNP assessed, eight SNP in CYP2R1, CYP3A4 and GC were associated with 25(OH)D concentrations (Table 3). For five of the SNP, the major allele was associated with lower 25(OH)D concentrations (e.g. CYP2R1 rs10741657 and rs1993116; and GC rs7041, rs2298849 and rs1687015), while for the remaining three SNP, the major allele was associated with higher 25(OH)D concentrations (e.g. CYP2R1 rs12794714; CYP3A4 rs2242480; and GC rs4588). The strongest association was with the GC SNP rs4588, where the decrease in copies of the major allele (e.g. from 2 to 0) was associated with a 11·5 nmol/l decrease in serum 25(OH)D concentration (P< 0·001). Including genotype information for individual SNP into the multivariable model explained an additional 0·8–3·7 % of the variation in serum 25(OH)D concentrations in this cohort (Table 3).
CYP2R1, cytochrome P450 2R1; rs, refSNP or reference SNP; CYP3A4, cytochrome P450 3A4, GC, group-specific component (vitamin D binding protein).
* 25(OH)D concentrations are multivariate-adjusted for: age, sex, dietary vitamin D, hours spent sitting at work, and time interval between last meal and blood draw.
† Variance explained is the model R 2× 100. The partial R 2 for each of the covariates in the individual multivariate models was: age ( < 0·001), sex (0·070–0·076), dietary vitamin D (0·005–0·008), hours spent sitting at work (0·020–0·026), and time interval between last meal and blood draw ( < 0·001).
‡ The variance explained by genotype is the partial R 2× 100 for the individual genotype in a linear regression model with variables for age, sex, dietary vitamin D, hours spent sitting at work, and time interval between last meal and blood draw.
In addition to the single-SNP associations with serum 25(OH)D concentrations, we also evaluated the combined effect of two well-described GC SNP, rs4588 and rs7041. These SNP have been previously shown to jointly determine three well-described protein transcripts: Gc1s, Gc1f and Gc2(Reference Speeckaert, Huang and Delanghe12). As shown in Table 4, the mean 25(OH)D concentration was the highest for individuals with two copies of the Gc1s allele (Gc1s-1s: rs4588CC, rs7041GG), the lowest for individuals with two copies of the Gc2 allele (Gc2-2: rs4588AA, rs7041TT), and intermediate for those with any other Gc haplotype (P-trend < 0·001). When stratified by median dietary vitamin D intake, the trends with 25(OH)D by haplotype are similar to the overall pattern currently presented in Table 4 for both high and low dietary vitamin D intake (data not shown). However, when stratified by hours spent sitting at work, the trend of decreasing 25(OH)D concentration from Gc1s-1s to Gc2-2 haplotype was only evident among those who reported no hours spent sitting at work (n 268, P-trend < 0·001, Table 5). The trend was less evident and was not statistically significant among those who reported any sitting hours at work (data not shown), although the interaction between Gc haplotype and hours spent sitting at work was not statistically significant (P-interaction = 0·24).
* Adjusted geometric mean value was significantly different from that of the Gc2-2 haplotype (P< 0·01).
† From ANCOVA with adjustments for age, sex, dietary vitamin D intake, hours spent sitting at work and time interval between the last meal and blood draw.
† From ANCOVA with adjustments for age, sex, dietary vitamin D intake, hours spent sitting at work and time interval between the last meal and blood draw.
Discussion
On average, participants in this study had serum 25(OH)D concentrations that would be considered sufficient for optimal bone health according to the Institute of Medicine recommendations(Reference Ross, Taylor, Yaktine and Del Valle6), but somewhat below the recommendations from leading vitamin D researchers of 75 nmol/l to address a broader range of health concerns(Reference Vieth, Bischoff-Ferrari and Boucher7). However, the serum 25(OH)D concentrations for participants in this study were higher, on average, than has been reported for comparably aged men (58·0 nmol/l) and women (54·8 nmol/l) in the USA(Reference Looker, Johnson and Lacher53), which might be expected of individuals living near the equator.
BMI was not associated with serum 25(OH)D concentrations in this population, mostly probably due to the fact that the mean BMI for the study participants was within the normal range for Asian populations ( < 23 kg/m2)(Reference Choo54), and the range was narrow. Mean dietary vitamin D intake for these middle- and older-aged Singaporean men (2·8 (sd 1·8) μg/d) and women (2·4 (sd 1·7) μg/d) was lower than comparably aged men (5·1 (sd 0·3) μg/d) and women (3·9 (sd 0·4) μg/d) in the USA according to the 2005–6 National Health and Nutrition Examination Survey data(Reference Bailey, Dodd and Goldman55), and somewhat lower than adult men (3·1 (sd 0·1) μg/d) and women (2·7 (sd 0·1) μg/d) in the UK according to the 2008–9 National Diet and Nutrition Survey data(Reference Bates, Lennox and Swan56).
Sun exposure is known to be a major determinant of vitamin D status. Singapore receives 12 h sunlight/d throughout the year, with a midday solar zenith angle that ranges from a minima of 0–3° (March, September) to a peak of 22–25° (June, December)(57). Despite a small amount of variation in solar zenith angle, we did not observe significant seasonal variation in serum 25(OH)D concentrations in this cohort. The UV index ranges from 10 (December) to 13 (March/April), indicating very high ambient UV radiation levels(58). Given the average daily high temperatures of 31°C (88°F)(59), many Singaporeans avoid the heat of the midday sun. In our study, as the reported number of hours spent sitting at work increased, serum 25(OH)D concentrations decreased (P-trend < 0·001).
Three other reports have described serum 25(OH)D concentrations among healthy adults living within 10° of the equator. Rahman et al. (Reference Rahman, Chee and Yassin60) evaluated 276 post-menopausal women living near Kuala Lumpur, Malaysia (2°N). They found that ethnic Chinese women had significantly higher mean 25(OH)D concentrations compared to Malay women (68·8 (sd 15·7) v. 44·4 (10·6) nmol/l, P< 0·05). Dietary vitamin D intake did not differ between the two ethnic groups. The Chinese women had a significantly lower mean BMI, and reported more regular physical activity than the Malay women. The Malay also tend to have more skin pigmentation than the Chinese, and many Malay women follow Muslim dress codes that further limit UV exposure. Serum 25(OH)D concentrations were significantly correlated with BMI, fat mass, parathyroid hormone concentrations and physical activity scores. Green et al. (Reference Green, Skeaff and Rockell61) evaluated 378 younger women living in Kuala Lumpur (mean age: 25·2 years) and 126 women living in Jakarta, Indonesia (6°S, mean age: 30·0 years). Among the women in Malaysia, they also found higher serum 25(OH)D concentrations among the ethnic Chinese (mean: 58·0 nmol/l, 95 % CI 55·0, 61·0) compared to the Malay (mean: 43·0 nmol/l, 95 % CI 40·0, 46·0) or Indian (mean: 45·0 nmol/l, 95 % CI 43·0, 48·0) women (P< 0·01). The Indonesian women had serum 25(OH)D concentrations that were comparable to the Malay and Indian women (mean: 46·0 nmol/l, 95 % CI 43·0, 48·0). Moy & Bulgiba(Reference Moy and Bulgiba62) recently reported on the vitamin D status of 380 Malay study participants (158 men, 222 women) in a voluntary health screening programme in Kuala Lumpur. The women had significantly lower serum 25(OH)D concentrations (mean: 36·2 nmol/l, 95 % CI 34·5, 38·0) compared to the men (mean: 56·2 nmol/l, 95 % CI 53·2, 59·2, P< 0·001), which could be partially explained by differences in religious dress codes. Age, sex, BMI and abdominal obesity were found to be statistically significantly associated with vitamin D insufficiency in this study cohort. The women in our study had serum 25(OH)D concentrations that were similar to the ethnic Chinese participants in both the Rahman et al.(Reference Rahman, Chee and Yassin60) and Green et al. (Reference Green, Skeaff and Rockell61) studies.
Genetic variants in CYP2R1, CYP3A4 and GC were significantly associated with serum 25(OH)D concentrations in our study. The CYP2R1 rs10741657, rs12794714, and rs1993116 and GC rs4588 and rs7041 findings are consistent with several recent reports(Reference Engelman, Fingerlin and Langefeld19, Reference Ramos-Lopez, Bruck and Jansen21, Reference Bu, Armas and Lappe26, Reference Kurylowicz, Ramos-Lopez and Bednarczuk28–Reference Fu, Yun and Oczak32, Reference Fang, van Meurs and Arp34, Reference Janssens, Bouillon and Claes35), including two large genome-wide association studies(Reference Wang, Zhang and Richards24, Reference Ahn, Yu and Stolzenberg-Solomon25). Several CYP enzymes have been shown to have 25-hydroxylase activity, and CYP2R1 has emerged as the predominant 25-hydroxylase, with the highest binding affinity and specificity for vitamin D(Reference Schuster63). CYP2R1 rs10741657 lies in the promoter region, rs12794714 is a synonymous SNP in exon 1, and rs1993116 is in intron 1. To our knowledge, no previous studies have evaluated the association between genetic variation in CYP3A4 and serum 25(OH)D concentrations in human subjects. CYP3A4 rs2242480 is in intron 10 near the exon/intron boundary. Although the functional relevance of this SNP is unclear, a recent pharmacokinetic study suggests that individuals with the homozygous variant rs2242480TT genotype have significantly lower CYP3A4 activity(Reference Zhang, Chang and Kan64).
The GC SNP, rs4588 and rs7041, are both in exon 11. Consistent with the findings of several previous studies(Reference Engelman, Fingerlin and Langefeld19, Reference Kurylowicz, Ramos-Lopez and Bednarczuk28, Reference Daiger, Miller and Chakraborty65–Reference Constans, Arlet and Viau68), we also observed that individuals with two copies of the Gc2 allele (Gc2-2) have significantly lower 25(OH)D concentrations compared to other GC genotypes. In vitro data have shown that the Gc2 protein has a significantly lower affinity constant for 25(OH)D3 compared to the Gc1s or Gc1f proteins(Reference Speeckaert, Huang and Delanghe12). GC rs16847015 and rs2298849 both lie in intron 1, and their functional relevance is unclear.
While several previous reports identified genetic variants in the VDR (Reference Ma, Stampfer and Gann13–Reference Smolders, Damoiseaux and Menheere20) and CYP27B1 (Reference Orton, Morris and Herrera22, Reference Ramos-Lopez, Kahles and Weber23, Reference Hypponen, Berry and Wjst27) genes as being associated with serum 25(OH)D concentrations, we did not observe any statistically significant associations for those genes in our study population. The studies reporting significant findings for VDR and CYP27B1 variants tended to be among smaller study populations, conducted at higher latitudes, and many failed to adjust for factors known to alter serum 25(OH)D concentrations such as season of blood draw, BMI and dietary/supplemental vitamin D intake.
The present study has several strengths including being the largest study of vitamin D status among Southeast Asians to date, lack of seasonal UV variation, and consideration of dietary vitamin D exposures, lifestyle and sociodemographic factors as well as genetic variation as potential factors contributing to serum 25(OH)D concentrations. We also took a comprehensive approach to assessing the effect of genetic variation in the entire vitamin D metabolism pathway, as opposed to the evaluation of single candidate genes or SNP. Limitations of the present study include incomplete assessment of time spent outdoors during daylight hours and degree of skin pigmentation, factors which may contribute to variation in serum 25(OH)D concentrations. Due to technical challenges in accurately measuring the vitamin D content of foods(Reference Byrdwell, Devries and Exler37, Reference Holden, Lemar and Exler38), our assessment of the participants' dietary vitamin D intake is probably underestimated. This underestimation of dietary vitamin D intake should occur to all study subjects non-differentially, which would result in underestimating the association between dietary vitamin D intake and serum 25(OH)D concentrations. Supplemental vitamin D intake, other than cod liver oil, was not specifically assessed. However, only 8 % of our study cohort reported taking any vitamins or minerals at least once a week.
Our findings confirm the growing body of literature documenting an association between GC and CYP2R1 genetic variation and serum 25(OH)D concentrations. Future studies of both predictors of 25(OH)D concentrations and disease outcomes thought to be associated with vitamin D status should include an assessment of GC and CYP2R1 genotype. Further research is needed to confirm our findings related to CYP3A4 rs2242480.
Acknowledgements
The authors' contributions to the present study were as follows: K. R., L. M. B. and J.-M. Y. were responsible for the study concept and design and obtainment of funding; K. R. was responsible for study supervision; K. R., K. B. B., D. W., W.-P. K. and J.-M. Y. for the acquisition of data; L. M. B., R. W., K. R. and J.-M. Y. for the data analysis and interpretation; and K. R. and L. M. B. for drafting of the manuscript. All authors contributed to critical revision of the manuscript. None of the authors reported a conflict of interest. The authors thank Siew-Hong Low of the National University of Singapore for supervising the fieldwork of the SCHS, and Kazuko Arakawa for the development and maintenance of the cohort study database. Further, the authors acknowledge the founding, long-standing Principal Investigator of the SCHS – Mimi C. Yu. This study was funded by the University of Minnesota Masonic Cancer Center and the National Cancer Institute (R01 CA144034).