Genetic polymorphisms of the vitamin D binding protein and plasma concentrations of 25-hydroxyvitamin D in premenopausal women

Abstract

Vitamin D status, determined on the basis of 25-hydroxyvitamin D [25(OH)D] concentrations, is associated with the risk of several diseases. Vitamin D binding protein (DBP) is the major carrier of vitamin D and its metabolites, but the role of DBP single nucleotide polymorphisms (SNPs) on 25(OH)D concentrations is unclear.
The objective was to evaluate the association of 2 DBP gene SNPs with 25(OH)D concentrations and explore whether such association varies according to the amount of vitamin D that needs to be transported.
This cross-sectional study included 741 premenopausal white women, mostly of French descent. Plasma 25(OH)D concentrations were measured by radioimmunoassay. DBP-1 (rs7041) and DBP-2 (rs4588) were genotyped with a Sequenom MassArray platform. Associations and interactions were modeled by using multivariate linear regression.
DBP-1 and DBP-2 SNPs were in strong linkage disequilibrium and were both associated with 25(OH)D concentrations. An additional copy of the rare allele of DBP-1 or DBP-2 was associated with lower 25(OH)D concentrations (beta = -3.29, P for trend = 0.0003; beta = -4.22, P for trend < 0.0001, respectively). These DBP polymorphisms explained as much of the variation in circulating 25(OH)D as did total vitamin D intake (r2 = 1.3% for DBP-1, r2 = 2.0% for DBP-2, and r2 < or = 1.2% for vitamin D intake).
Circulating 25(OH)D concentrations in premenopausal women are strongly related to DBP polymorphisms. Whether DBP rare allele carriers have a different risk of vitamin D-related diseases and whether such carriers can benefit more or less from dietary interventions, vitamin D supplementation, or sun exposure need to be clarified.

Full text

Genetic polymorphisms of the vitamin D binding protein and plasma
concentrations of 25-hydroxyvitamin D in premenopausal women
1–3
Marc Sinotte, Caroline Diorio, Sylvie Be
´rube
´, Michael Pollak, and Jacques Brisson
ABSTRACT
Background: Vitamin D status, determined on the basis of 25-
hydroxyvitamin D [25(OH)D] concentrations, is associated with
the risk of several diseases. Vitamin D binding protein (DBP) is
the major carrier of vitamin D and its metabolites, but the role of
DBP single nucleotide polymorphisms (SNPs) on 25(OH)D concen-
trations is unclear.
Objective: The objective was to evaluate the association of 2 DBP
gene SNPs with 25(OH)D concentrations and explore whether such
association varies according to the amount of vitamin D that needs
to be transported.
Design: This cross-sectional study included 741 premenopausal
white women, mostly of French descent. Plasma 25(OH)D concen-
trations were measured by radioimmunoassay. DBP-1 (rs7041) and
DBP-2 (rs4588) were genotyped with a Sequenom MassArray plat-
form. Associations and interactions were modeled by using multi-
variate linear regression.
Results: DBP-1 and DBP-2 SNPs were in strong linkage disequi-
librium and were both associated with 25(OH)D concentrations. An
additional copy of the rare allele of DBP-1 or DBP-2 was associated
with lower 25(OH)D concentrations (b¼23.29, Pfor trend ¼
0.0003; b¼24.22, Pfor trend ,0.0001, respectively). These
DBP polymorphisms explained as much of the variation in circu-
lating 25(OH)D as did total vitamin D intake (r
2
¼1.3% for DBP-1,
r
2
¼2.0% for DBP-2, and r
2
1.2% for vitamin D intake).
Conclusion: Circulating 25(OH)D concentrations in premenopausal
women are strongly related to DBP polymorphisms. Whether DBP
rare allele carriers have a different risk of vitamin D–related dis-
eases and whether such carriers can benefit more or less from di-
etary interventions, vitamin D supplementation, or sun exposure
need to be clarified. Am J Clin Nutr 2009;89:634–40.
INTRODUCTION
Higher vitamin D intakes or circulating concentrations are
associated with a lower risk of several chronic illnesses, including
common cancers, autoimmune diseases, infectious diseases, and
cardiovascular diseases (reviewed in 1, 2). A better under-
standing of vitamin D biology may prove useful in clarifying
these effects and designing effective interventions.
The principal circulating vitamin D metabolite, 25-
hydroxyvitamin D [25(OH)D], is recognized (3) as the best
short-term biomarker of total exposure to vitamin D (ingested
from food or dietary supplements and produced by the skin after
sun exposure). More than 99% of 25(OH)D is bound to plasma
protein (4), of which ’90% is bound to the vitamin D binding
protein (DBP) (5). DBP, also known as Gc-globulin, is a mem-
ber of the albumin (ALB) and alpha-fetoprotein (AFP) gene
family. It is mainly synthesized in the liver, where 25(OH)D is
also produced. Serum concentrations of DBP range between
4 and 8 lmol/L. DBP has a short half-life of 2.5–3 d compared
with 1 to 2 months for 25(OH)D (reviewed in 6–8). The precise
role of DBP in vitamin D action is still incompletely understood,
but DBP concentrations do not seem to be influenced by vitamin D
sterols or other calciotropic hormones or by seasonal variation
(reviewed in 7). DBP is significantly elevated during pregnancy
and estrogen therapy (9–11) and is low after liver diseases (4, 12),
nephrotic syndrome, and malnutrition, probably because of a di-
minished synthesis rate or excessive protein loss (reviewed in 7).
Until now, little has been known about the influence of DBP
polymorphisms on circulating 25(OH)D concentrations, but dif-
ferences in the coding of amino acids could affect the concen-
tration of the binding protein or its affinity for vitamin D
metabolites. In this study we assessed the association of 2 single
nucleotide polymorphisms (SNPs) located on exons of the vitamin
D binding protein gene (rs7041 and rs4588) with plasma 25(OH)D
concentrations and explored whether these associations vary ac-
cording to the amount of vitamin D that needs to be transported.
1
From the De
´partement de Me
´decine Sociale et Pre
´ventive, Universite
´
Laval, Que
´bec, Canada (MS, CD, and JB); the Unite
´de Recherche en Sante
´
des Populations, Centre Hospitalier Affilie
´Universitaire de Que
´bec, Que
´bec,
Canada (MS, CD, SB, and JB); the Centre des Maladies du sein Desche
ˆnes-
Fabia, Centre Hospitalier Affilie Universitaire de Que
´bec, Que
´bec, Canada
(CD, SB, and JB); and the Cancer Prevention Research Unit, Lady Davis
Institute of the Jewish General Hospital and McGill University, Departments
of Medicine and Oncology, Montre
´al, Canada (MP).
2
Supported in part by grant 4811-82 from the Canadian Breast Cancer
Research Alliance, a grant from the Translation Acceleration Grants Pro-
gram for Breast Cancer Control of the Canadian Breast Cancer Research
Alliance and the Canadian Institutes of Health Research. MS was supported
by studentships from the Canadian Institutes of Health Research and Na-
tional Cancer Institute of Canada. CD was supported by a postdoctoral fel-
lowships from the Cancer Research Society Inc and the Canadian Institutes
of Health Research.
3
Reprints not available. Address correspondence to C Diorio, Unite
´de
Recherche en Sante
´des Populations, Ho
ˆpital du Saint-Sacrement du Centre
Hospitalier Affilie
´Universitaire de Que
´bec, 1050 Chemin Sainte-Foy, Que
´-
bec, Canada G1S 4L8. E-mail: caroline.diorio@uresp.ulaval.ca.
Received May 23, 2008. Accepted for publication November 25, 2008.
First published online December 30, 2008; doi: 10.3945/ajcn.2008.26445.
634 Am J Clin Nutr 2009;89:634–40. Printed in USA. Ó2009 American Society for Nutrition
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

SUBJECTS AND METHODS
Study population and recruitment procedures
The study design and methods were published previously (13,
14). Briefly, 783 premenopausal women who underwent screening
mammography between February and December 2001 were re-
cruited at the Clinique Radiologique Audet (Que
´bec, Canada).
Eligible women had no personal history of cancer or breast surgery,
had no endocrine diseases, never took selective estrogen-receptor
modulators, and had not used hormonal derivatives in the 3 mo
before blood sampling. Of the 783 eligible women, 741 provided
written informed consentto use their blood samples forassays other
than those planned at recruitment (14) and had DNA available (15).
This study was approved by the Research Ethics Review Board–
Ho
ˆpital du Saint-Sacrement du CHA de Que
´b
ˆec.
Data collection
Anthropometric measurements and blood samples were taken at
recruitment. Breast cancer risk factors were documented by tele-
phone interview, including reproductive and menstrual history,
family history of breast cancer, personal history of breast biopsies,
past use of exogenous hormones, smoking status, alcohol intake,
education, and physical activity. Finally, diet was assessed with a
self-administered 161-item semiquantitative food-frequency ques-
tionnaire (97GP copyrighted at Harvard University, Boston, MA).
Assessment of plasma 25(OH)D
At the time of collection, blood constituents were rapidly stored
in aliquots at 280°C until analyzed. Plasma 25(OH)D concen-
trations were measured between November 2005 and January
2006 by radioimmunoassay after acetonitrile extraction (DiaSorin
Inc, Stillwater, MN). The intrabatch and between-batch CVs were
7.3% and 8.8%, respectively (4 blinded duplicates on average for
each of the 24 batches), and the results met the performance target
set by the international 25-hydroxyvitamin D External Quality
Assessment Scheme (DEQAS) Advisory Panel in 2004–2005.
DNA extraction and SNP genotyping
Genotyping procedures were described previously (15).
Briefly, DNA was extracted from the buffy coat by using the
PUREGene DNA extraction kit (Gentra Inc, Minneapolis, MN)
following the manufacturer’s protocol, and DNA samples were
then blindly genotyped. The rs7041 and rs4588 SNPs, resulting
in a T-to-G transversion [an aspartic acid (Asp: GAT) to a glu-
tamic acid (Glu: GAG)] and a C-to-A transversion [a threonine
(Thr: ACG) to a lysine (Lys: AAG)], respectively, in exon 11 of
DBP (16) were assessed by using the Sequenom MassArray
(Sequenom Inc, San Diego, CA) genotyping platform according
to the manufacturer’s protocol. Each 96-well plate included
negative (no DNA) and positive controls to ensure genotyping
accuracy. Genotyping call rates were 98.7% for both DBP-1
(rs7041) and DBP-2 (rs4588). The protocol can be provided on
request. In this study, concordance of genotyping from the new
Sequenom MassArray platform was compared with Fluorescent
Polarization–Single Base Extension platform on 10% of the sam-
ples; concordance was 100%.
Statistical methods
Crude and adjusted associations of plasma 25(OH)D concen-
trations with 6 continuous and 3 categorical potential explanatory
variables (reviewed in 3, 17) were estimated by using generalized
linear models (GLMs). Season of blood collection and leisure-
time physical activity [metabolic equivalent (MET)-h/wk] in the
past year (proxy variables for sun exposure), total vitamin D (IU/d)
and total calcium (mg/d) intakes from food and supplements in the
past year, body mass index (BMI; in kg/m
2
), smoking status, ed-
ucation, and age (y) were included in our analyses. Total energy
intake (kcal/d) in the past year was also included in the models.
Deviation from the Hardy-Weinberg equilibrium was assessed
for each SNP by a chi-square test with one df, and linkage
disequilibrium strength was evaluated with rand Lewontin’s D’
statistic between SNPs. Univariate and multivariate-adjusted
mean circulating plasma 25(OH)D concentrations by category
of genotypes under codominant mode of inheritance were esti-
mated by using GLM models. Trends between the number of
copies of the rare allele, entered as a continuous variable (0, 1,
or 2), and concentrations of 25(OH)D were evaluated by linear
regression models where the bcoefficients represent the per
allele variation in nmol/L of plasma 25(OH)D concentration.
The strength of associations of genotypes to 25(OH)D concen-
trations among women with a high vitamin D load was com-
pared with that among women with a low vitamin D load by
using the above models to which an interaction term was added.
The Pvalue of these interaction terms between vitamin D loads
(low or high) and the genotypes under the codominant mode of
inheritance were used to assess the effect modification of vita-
min D load. Periods of low and high vitamin D loads were
chosen based on the seasonal variation in 25(OH)D seen in this
cohort (13). Data collected throughout year 2001 were di-
chotomized in two 6-mo periods during which the vitamin D
load is expected to be high (May to October; median 25(OH)D
concentration ¼68.6 nmol/L) or low (November to April; me-
dian 25(OH)D ¼54.1 nmol/L).
Partial r
2
values, generated from the adjusted GLM models as
the ratio of type II sum of squares on the total sum of squares,
were mutually adjusted and interpreted as the independent
contribution of each variable in the model to the explanation of
the variation in 25(OH)D concentrations.
Assumptionofnormalityofresidualsfromtheseanalyseswasmet
with untransformed variables. Neither multicollinearity nor in-
fluentialobservationwas detected.All testswere2-sidedandcarried
out by using SAS version 9.1 (SAS Institute Inc, Cary, NC), and
anominalPvalue of 0.05 was considered statistically significant.
RESULTS
Determinants of 25(OH)D concentrations
Characteristics of the study population are described in detail
elsewhere (14). Briefly, 741 premenopausal women in the present
study had a mean (6SD) age of 46.8 64.6 and plasma 25(OH)D
concentration of 64.9 619.6. The mean (6SD) leisure-time
physical activity and BMI were 27.1 622.2 MET-hour/wk
and 25.2 64.6, respectively. The mean (6SD) total daily vitamin
D, total daily calcium, and total daily energy intakes were of
284 6232 IU, 969 6433 mg, and 1907 6515 kcal, respec-
tively. The percentage of university and college compared with
DBP POLYMORPHISMS AND 25(OH)D CONCENTRATIONS 635
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

secondary school degree or less and of nonsmokers compared
with smokers and former smokers were 61.9% and 44.9%,
respectively.
Several variableswere associated with 25(OH)D concentrations
(Table 1). Mean plasma 25(OH)D was much higher among
women recruited in summer than in those recruited in winter
(difference ¼21.7 nmol/L; an increase of 38%). Leisure-time
physical activity was positively related to 25(OH)D concen-
trations (P,0.001), as were total vitamin D (P¼0.0014) and
total calcium (P,0.001) intakes. BMI was negatively associated
with 25(OH)D concentrations (P,0.001), as were total energy
intake and education (both P,0.05). Smokers tended to have
higher 25(OH)D concentrations than nonsmokers, although this
difference was statistically significant only for ex-smokers. Con-
centrations of 25(OH)D tended to decrease as age increased in this
population of premenopausal women, although this association
was not statistically significant (P¼0.12).
Vitamin D binding protein SNPs and 25(OH)D
concentrations
No significant deviation from Hardy-Weinberg expecta-
tions was observed for polymorphisms DBP-1 (P¼0.19) and
DBP-2 (P¼0.48). Linkage disequilibrium between these SNPs
was almost complete (D’ ¼1.00, r¼0.74). In crude or adjusted
models, rare allele carriers had lower circulating 25(OH)D
concentrations than did homozygotes for the common allele
(Table 2). Each additional rare allele was associated with a re-
duction in 25(OH)D concentrations of 3.29 and 4.22 nmol/L for
DBP-1 and DBP-2, respectively, with a gene dosage compatible
with a codominant mode of inheritance (Pfor trend ¼0.0003
and ,0.0001, respectively).
Vitamin D load and strength of DPB SNP effects
The amount of vitamin D needed for transportation (‘‘vitamin D
load’’) tends tovary with season as reflected by seasonal variations
in circulating 25(OH)D. Thus, vitamin D binding protein poly-
morphisms could have different effects according to vitamin D
load. The effect of both SNPs on 25(OH)D concentrations was
more apparent when the amount of vitamin D to be transported was
high (Table 3). Indeed, associations were stronger in May to
October than in November to April for DBP-1 [b¼23.78 (Pfor
trend ¼0.0018) compared with b¼21.74 (Pfor trend ¼0.21)] as
well as for DBP-2 [b¼25.73 (Pfor trend ,0.0001) compared
with b¼23.03 (Pfor trend ¼0.037)], although the interaction
was not statistically significant (Pfor interaction ¼0.27 and 0.16
for DBP-1 and DBP-2, respectively).
Contribution to variation of 25(OH)D concentrations
Relative contributions of DBP-1 or DBP-2 and other variables
to the variation in plasma 25(OH)D concentrations are shown in
Table 4. The adjusted model without any SNP explained 30.3%
of the variation in 25(OH)D. Adding either polymorphism to
this model improved its explanatory capacity to 31.2% for DBP-
1 (partial r
2
¼1.3%) and to 31.9% for DBP-2 (partial r
2
¼
TABLE 1
Relation between plasma 25-hydroxyvitamin D [25(OH)D] concentrations and potentially explanatory variables in
premenopausal women
Crude models Adjusted models
1
Explanatory variables b6SE
2
Pvalue b6SE
2
Pvalue
Season at time of blood collection
Spring (n¼251) 4.72 62.26 0.037 4.51 62.07 0.030
Summer (n¼186) 21.34 62.36 ,0.001 21.74 62.18 ,0.001
Fall (n¼218) 5.26 62.31 0.023 5.06 62.12 0.017
Winter (n¼86) — — — —
Leisure-time physical activity (MET-h/wk) (n¼740) 0.18 60.03 ,0.001 0.17 60.03 ,0.001
Total vitamin D intake (IU/d) (n¼736) 1.82 60.30 ,0.001 1.06 60.33 0.0014
Total calcium intake (mg/d) (n¼736) 1.19 60.16 ,0.001 0.92 60.20 ,0.001
Total energy intake (kcal/d) (n¼736) 0.23 60.14 0.11 20.27 60.14 0.045
BMI (kg/m
2
)(n¼741) 20.67 60.16 ,0.001 20.51 60.14 ,0.001
Education, degree completed
University (n¼262) 22.72 63.10 0.38 28.55 62.64 0.0013
College (n¼197) 23.96 63.18 0.21 26.35 62.71 0.020
Secondary (n¼235) 23.88 63.13 0.22 25.70 62.64 0.031
Less than secondary (n¼47) — — — —
Smoking status
Smoker (n¼111) 1.51 62.14 0.48 2.06 61.87 0.27
Former smoker (n¼296) 3.13 61.56 0.045 3.40 61.33 0.011
Nonsmoker (n¼334) — —
Age (y) (n¼741) 20.22 60.16 0.16 20.21 60.14 0.12
1
Adjusted for all variables in the table. n¼735 because of 6 missing values. MET, metabolic equivalent.
2
Values are derived from linear regression models and represent absolute mean differences in plasma 25(OH)D concen-
trations (nmol/L) for increments of one MET-h/wk of leisure-time physical activity, 100 IU vitamin D intake, 100 mg total
calcium intake per day, 100 kcal of energy intake per day, 1 unit of BMI (kg/m
2
), and 1 y of age. For categorical variables, b
values represent absolute mean difference in plasma 25(OH)D concentrations (nmol/L) in blood collected in the spring,
summer, or fall as compared with in the winter (crude and adjusted means: 56.4nmol/L); for university, college, and secondary
school degrees compared with less than secondary; and for smokers and former smokers compared with nonsmokers.
636 SINOTTE ET AL
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

2.0%), but did not substantially change the relative contribution
of the other variables. DBP-1 explained as much variation in
25(OH)D concentrations as did total vitamin D intake, BMI, and
education. DBP-2 explained as much variation as did total cal-
cium intake and more than any other variable except for those
related to sun exposure (ie, season and leisure-time physical
activity). The importance of DBP-1 and DBP-2 as explanatory
variables of the variation in 25(OH)D concentrations was even
greater in May to October when the amount of vitamin D to be
transported was high (partial r
2
¼1.6% and 3.7% for DBP-1 and
DBP-2, respectively). During this period, both SNPs explained
25(OH)D concentrations as much or even more than any
other variable, except for leisure-time physical activity. In con-
trast, in November-April, when the amount of vitamin D to be
transported was low, both SNPs explained less of the 25(OH)D
variation, whereas total vitamin D intake, total calcium intake,
and leisure-time physical activity became the major explanatory
variables of plasma 25(OH)D.
DISCUSSION
At the DNA level, this is the first study to our knowledge to
show that plasma 25(OH)D concentrations decrease when the
number of rare alleles of DBP-1 or DBP-2 increases in healthy
TABLE 2
Relation between plasma 25-hydroxyvitamin D [25(OH)D] concentrations and vitamin D binding protein (DBP) genotypes in premenopausal women
Plasma 25(OH)D concentration
Single nucleotide polymorphism rs No. Genotype Subjects
1
Crude mean 6SE
1
Pvalue
2
Adjusted mean 6SE
3
Pvalue
2
n (%) nmol/L nmol/L
DBP-1
4,5
rs7041 GG 228 (31.1) 67.3 61.3 — 67.5 61.1 —
GT 377 (51.4) 65.0 61.0 0.16 64.5 60.8 0.034
TT 128 (17.5) 60.2 61.7 0.0010 60.8 61.5 0.0003
DBP-2
4,6
rs4588 CC 370 (50.5) 67.2 61.0 — 67.2 60.9 —
CA 296 (40.4) 63.2 61.1 0.0081 63.2 60.9 0.0018
AA 67 (9.1) 59.0 62.4 0.0016 58.4 62.0 ,0.0001
1
n¼733 women because of 8 missing values for genotype.
2
The Wald test was used to compare mean concentrations with reference genotype (common homozygote) level.
3
Adjusted for all variables presented in Table 1; n¼727 because of 6 missing values.
4
b6SE values from linear regression models represent the absolute mean difference in plasma 25(OH)D concentrations (nmol/L) for increments of one
rare allele. The Wald test was used to derive Pfor trend values between genotypes entered as a continuous variable and plasma 25(OH)D concentrations [ie,
test of the linear decrease in plasma 25(OH)D concentrations for increments of one rare allele].
5
Crude value: b6SE ¼23.37 61.05, Pfor trend ¼0.0014; adjusted value: b6SE ¼23.29 60.89, Pfor trend ¼0.0003.
6
Crude value: b6SE ¼24.06 61.09, Pfor trend ¼0.0002; adjusted value: b6SE ¼24.22 60.93, Pfor trend ¼,0.0001.
TABLE 3
Relation between plasma 25-hydroxyvitamin D [25(OH)D] concentrations and genotypes of vitamin D binding protein
(DBP) by vitamin D load in premenopausal women
Adjusted plasma 25(OH)D concentration
1
Low vitamin D load
2
High vitamin D load
2
Single nucleotide polymorphism rs No. Genotype Subjects Mean 6SE Subjects Mean 6SE
n (%) nmol/L n (%) nmol/L
DBP-1
3,4
rs7041 GG 86 (27.9) 58.4 61.8 140 (33.4) 73.4 61.4
GT 162 (52.6) 55.6 61.3 214 (51.1) 71.4 61.1
TT 60 (19.5) 55.1 62.1 65 (15.5) 64.8 62.1
DBP-2
3,5
rs4588 CC 155 (50.3) 58.2 61.3 212 (50.6) 74.1 61.1
CA 126 (40.9) 54.8 61.5 169 (40.3) 69.3 61.3
AA 27 (8.8) 52.7 63.2 38 (9.1) 61.5 62.7
1
Adjusted for all variables presented in Table 1, except season. n¼308 and 419 in low- and high–vitamin D load
periods, respectively.
2
Corresponding to periods of low (November to April) and high (May to October) vitamin D loads.
3
b6SE values from linear regression models represent the absolute mean difference in plasma 25(OH)D concen-
trations (nmol/L) for increments of one rare allele. The Wald test was used to derive Pfor trend values between genotypes
entered as a continuous variable and plasma 25(OH)D concentrations [ie, test of the linear decrease in plasma 25(OH)D
concentrations for increments of one rare allele]. For the Pfor interaction, the Wald test was used to compare difference in b
values between low and high vitamin D loads.
4
Pfor interaction ¼0.27. Low vitamin D load: b6SE ¼21.74 61.38 (Pfor trend ¼0.21); high vitamin D load:
b6SE ¼23.78 61.20 (Pfor trend ¼0.0018).
5
Pfor interaction ¼0.16. Low vitamin D load: b6SE ¼23.03 61.45 (Pfor trend ¼0.037); high vitamin D load:
b6SE ¼25.73 61.24 (Pfor trend ,0.0001).
DBP POLYMORPHISMS AND 25(OH)D CONCENTRATIONS 637
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

premenopausal women. In a recent study, serum 25(OH)D con-
centrations in population-based postmenopausal control women
also decreased significantly by Gc genotypes (defined by DBP-1
and DBP-2) (18). The DBP-2 SNP has been shown to be asso-
ciated with lower 25(OH)D concentrations, but only in a group
of Polish Graves’ Disease cases (16). Whereas season remains
the most important explanatory variable of 25(OH)D concen-
trations, DBP polymorphisms explain 25(OH)D variation as
much as do vitamin D intake, calcium intake, and BMI. Each
DBP SNP explains even more of the 25(OH)D variation when
the amount of vitamin D to be transported is high.
Our results suggest that both DBP SNPs generate functionally
different proteins and that such differences affect circulating
25(OH)D concentrations. The rare allele of DBP-1 codes for the
aspartic acid residue at amino acid position 416 of the DBP
protein, whereas DBP-2 codes for the lysine residue at position
420, which allows for differentiation of 3 major DBP protein
phenotypes (19). Being rare homozygote for both DBP-1 and
DBP-2 characterize the glycosylation pattern of the protein
phenotype Gc2-2, which has been shown to be associated with
low mean serum DBP protein concentrations (20) and recently
with low mean serum 25(OH)D concentrations in post-
menopausal women (18). These results suggest that rare alleles
of DBP-1 and DBP-2 are associated with lower 25(OH)D con-
centrations, at least in part because of the lowering effect on
DBP protein concentrations. Whether the variation in DBP
protein concentrations stems from different protein production
or degradation rates associated with different DBP genotypes
and phenotypes is unclear, but Lauridsen et al (21) suggest that,
on the basis of glycosylation patterns, DBP phenotypes related
to a low vitamin D concentration should be metabolized faster.
This would in turn decrease the half-life of 25(OH)D, increase
its conversion to inactive metabolites, and consequently re-
duce 25(OH)D concentrations, as shown in DBP KO mice after
tritium-labeled 25(OH)D injection (22).
Few studies have examined DBP polymorphisms and risk of
vitamin D–related diseases. A nonsignificant increased risk of
breast cancer was associated with the DBP-2 rare homozygote
(23). The rare homozygote for both DBP SNPs was associated
with a reduction in postmenopausal breast cancer (18). The
haplotypes of 3 SNPs on the promoter of the DBP gene suggest
a nonsignificant association with prostate cancer in Americans of
European descent (24). The (TAAA)n-Alu repeat polymorphisms
of the DBP gene was associated with fracture risk (25, 26). DBP
SNPs in linkage disequilibrium with DBP-2 were associated with
bone mineral density in Japanese postmenopausal women (27).
DBP-2, but not DBP-1, was associated with Graves’ disease (16)
in Poland but not in nuclear families from Germany and Italy,
although the DBP polymorphism (TAAA)n-Alu repeat was (28).
Associations with type 2 diabetes mellitus and obesity-related
traits have been observed, but the results have been inconsistent
(reviewed in 7). DBP SNPs can possibly influence bioactive
25(OH)D concentrations through changes in the ratio of free to
bound hormones (29), by differential affinity (30), or through
effects on levels of the DBP/25(OH)D complex that can be in-
ternalized by receptor-mediated endocytosis and activate the
vitamin D receptor pathway, as recently shown in mammary cells
(31). In addition DBP SNPs could have effects on carcinogenesis
through activation of tumoricidal macrophages and antiangiogenic
effects of DBP-macrophage activating factor (reviewed in 7, 32).
Overall, evidence suggests that further studies between DBP SNPs
and health outcomes are needed.
The relation of both DBP SNPs to 25(OH)D seems to be more
apparent when the amount of vitamin D in need of transportation is
TABLE 4
Relative contribution of single nucleotide polymorphisms (SNPs) of vitamin D binding protein (DBP) genes DBP-1 or DBP-2 and other variables to the
variation in plasma 25-hydroxyvitamin D [25(OH)D] concentrations globally and stratified by vitamin D load in premenopausal women
Model without DBP-1 or DBP-2
1
Models including DBP-1
1
Models including DBP-2
1
Vitamin D load
2
Vitamin D load
2
Global Global Low
1
High
1
Global Low
1
High
1
Explanatory variables Partial
3
r
2
Partial
3
r
2
Partial
3
r
2
Partial
3
r
2
Partial
3
r
2
Partial
3
r
2
Partial
3
r
2
Season at time of blood collection 15.7
4
15.4
4
— — 15.3
4
——
Leisure-time physical activity (MET-h/wk) 3.7
4
3.9
4
2.8
5
4.2
4
4.0
4
2.9
5
4.4
4
Total vitamin D intake (IU/d) 1.0
5
1.2
4
3.9
4
,0.1 1.0
5
3.4
4
,0.1
Total calcium intake (mg/d) 2.1
4
1.8
4
3.0
4
2.2
5
2.0
4
3.4
4
2.3
4
Total energy intake (kcal/d) 0.4
5
0.4
5
1.9
5
,0.1 0.5
5
2.1
5
,0.1
BMI (kg/m
2
) 1.3
4
1.3
4
0.6 1.6
5
1.3
4
0.7 1.6
5
Education 1.1
5
1.2
5
1.8 0.2 1.1
5
1.6 ,0.1
Smoking status 0.6
5
0.7
5
0.3 1.5
5
0.7
5
0.2 1.7
5
Age 0.2 0.1 ,0.1 0.9
5
0.1 ,0.1 0.9
5
DBP-1 (rs7041) SNP — 1.3
4
0.7 1.6
5
———
DBP-2 (rs4588) SNP — — — — 2.0
4
1.4
5
3.7
4
Total r
2
(%) 30.3 31.2 22.7 17.1 31.9 23.5 19.2
1
Adjusted for all variables presented in each model. n¼735 in the adjusted model without SNP, and n¼727 in the adjusted model with either DBP-1 or
DBP-2. MET, metabolic equivalent.
2
Corresponding to periods of low (November to April) and high (May to October) vitamin D loads. n¼308 and 419 for the low- and high–vitamin D
load periods, respectively.
3
Partial r
2
represents the independent adjusted contribution of each variable in the model and therefore does not add up to total r
2
.
4
P,0.001.
5
P,0.05.
638 SINOTTE ET AL
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

high (ie, May to October compared with November to April).
Differences in 25(OH)D concentrations were also more apparent
under normal rather than deficient diets in experiments with DBP
knockout mice (22). Our observations appear consistent with
those experimental works. Indeed, if we had assessed this asso-
ciation only during winter, when the vitamin D load is the lowest,
no association or trend would have been detected. The effect
modification by vitamin D load was not statistically significant
though and should be ascertained in other studies. Nevertheless,
these results stress the potential importance of considering base-
line concentrations of 25(OH)D or recruitment season when
studying the association of DBP SNPs with circulating 25(OH)D.
A strength of this study was that recruitment took place over
one full calendar year in a relatively small geographic area over
which the population experiences a large seasonal variation in sun
exposure. This design facilitated the assessment of the effects of
DBP genotypes on 25(OH)D concentrations, but other personal
characteristics and variations in their effects according to vitamin
D load could also be investigated. Leisure-time physical activity
predicts greater 25(OH)D concentrations during a high vitamin
D load and putatively shows that outdoor activity can improve
vitamin D status, even when solar radiation and meteorological
characteristics are less favorable. Even though this variable did not
take into account the actual amount of outdoor exercise, level of
clothing, time of day, and use of sunscreen, the results further
support the idea that leisure-time physical activity is a surrogate of
exposure to solar ultraviolet-B, as suggested in a recent prospective
study by Giovannucci et al (33). The explanation capacity of
25(OH)D concentrations by BMI during a high vitamin D load is
no longer significant during a low vitamin D load. This suggests
that, in this population, possible sequestration of vitamin D into the
subcutaneous fat mass reservoir (34) could be more important
from early May to the end of October, when more vitamin D
3
is
synthesized in the skin. Total vitamin D intake only contributes to
the variation in 25(OH)D concentrations under conditions of a low
vitamin D load, whereas total calcium intake remains significant
in both periods, which suggests that calcium’s effect on 25(OH)D
concentrations is independent of vitamin D intake eventhough both
nutrient intakes are strongly correlated. Age is becoming a signifi-
cant negative predictor of 25(OH)D only during high vitamin D
load, which possibly reflects a reduction in 7-dehydrocholesterol
(ie, precursor of vitamin D
3
) in the skin with aging (reviewed in 35).
Our study had some limitations. False-positive results are
common in studies of the association between genetic markers
and outcomes, but because we only analyzed 2 polymorphisms of
the vitamin D binding protein in relation to circulating 25(OH)D
concentrations, we believe that type 1 errors were not likely to
explain our findings. Moreover, if we had used Bonferroni-
corrected Pvalues, the results would have remained statistically
significant. Population stratification can be a concern with this
type of study (36, 37), although this problem was likely not as
important as anticipated in North American white populations
(38–40). In our study, most of the women were from the Quebec
City area, white (99.7%), and of French descent (87.7%) (15),
which suggests that the associations we found were not due to
population stratification. Confounding was considered; however,
we do not think that this was a major concern because most key
variables known to be associated with vitamin D status were
accounted for in the analysis. Estrogen concentrations, which
were not assessed in the present study, are known to be asso-
ciated with DBP; nevertheless, adjustment for past hormone
derivative use did not affect our results.
In conclusion, the number of rare alleles of DBP-1 and DBP-2
polymorphisms is inversely related to 25(OH)D concentrations
in premenopausal women, and their effects on variation in
25(OH)D concentrations are comparable with those of total vi-
tamin D intakes. DBP SNPs are inherited, which suggests that
the reduction in 25(OH)D concentrations found within the rare
allele carriers would persist over a lifetime. Lifelong low vita-
min D concentrations may have an impact on health, and ad-
ditional studies of the association between DBP polymorphisms
and clinical outcomes are needed. Whether rare allele carriers of
DBP rs7041 and rs4588 SNPs could benefit more or less from
dietary intervention, supplementation, or sun exposure warrants
additional attention.
The authors’ responsibilities were as follows—CD, MS, SB, MP, and JB:
involved in the study design; CD, SB, and JB: supervised the data collection;
and MS: performed the statistical analyses and wrote the manuscript. All
authors contributed to the drafts of the manuscript and approved the final ver-
sion. None of the authors had a personal or financial conflict of interest.
REFERENCES
1. Giovannucci E. The epidemiology of vitamin D and cancer incidence
and mortality: a review (United States). Cancer Causes Control 2005;16:
83–95.
2. Holick MF. Vitamin D deficiency. N Engl J Med 2007;357:266–81.
3. Millen AE, Bodnar LM. Vitamin D assessment in population-based
studies: a review of the issues. Am J Clin Nutr 2008;87:1102S–5S.
4. Bikle DD, Siiteri PK, Ryzen E, Haddad JG. Serum protein binding of
1,25-dihydroxyvitamin D: a reevaluation by direct measurement of free
metabolite levels. J Clin Endocrinol Metab 1985;61:969–75.
5. Bikle DD, Gee E, Halloran B, Kowalski MA, Ryzen E, Haddad JG.
Assessment of the free fraction of 25-hydroxyvitamin D in serum and its
regulation by albumin and the vitamin D-binding protein. J Clin En-
docrinol Metab 1986;63:954–9.
6. Andreassen TK. The role of plasma-binding proteins in the cellular
uptake of lipophilic vitamins and steroids. Horm Metab Res 2006;38:
279–90.
7. Speeckaert M, Huang G, Delanghe JR, Taes YE. Biological and clinical
aspects of the vitamin D binding protein (Gc-globulin) and its poly-
morphism. Clin Chim Acta 2006;372:33–42.
8. Vieth R. Vitamin D supplementation, 25-hydroxyvitamin D concen-
trations, and safety. Am J Clin Nutr 1999;69:842–56.
9. Bouillon R, Van Assche FA, Van Baelen H, Heyns W, De Moor P.
Influence of the vitamin D-binding protein on the serum concentra-
tion of 1,25-dihydroxyvitamin D3. Significance of the free 1,25-
dihydroxyvitamin D3 concentration. J Clin Invest 1981;67:589–96.
10. Rejnmark L, Lauridsen AL, Brot C, et al. Vitamin D and its binding
protein Gc: long-term variability in peri- and postmenopausal women
with and without hormone replacement therapy. Scand J Clin Lab Invest
2006;66:227–38.
11. Hagenfeldt Y, Carlstrom K, Berlin T, Stege R. Effects of orchidectomy
and different modes of high dose estrogen treatment on circulating
‘‘free’’ and total 1,25-dihydroxyvitamin D in patients with prostatic
cancer. J Steroid Biochem Mol Biol 1991;39:155–9.
12. Bikle DD, Halloran BP, Gee E, Ryzen E, Haddad JG. Free 25-
hydroxyvitamin D levels are normal in subjects with liver disease and
reduced total 25-hydroxyvitamin D levels. J Clin Invest 1986;78:748–52.
13. Brisson J, Berube S, Diorio C, Sinotte M, Pollak M, Masse B. Syn-
chronized seasonal variations of mammographic breast density and
plasma 25-hydroxyvitamin d. Cancer Epidemiol Biomarkers Prev 2007;
16:929–33.
14. Diorio C, Pollak M, Byrne C, et al. Insulin-like growth factor-I,
IGF-binding protein23, and mammographic breast density. Cancer
Epidemiol Biomarkers Prev 2005;14:1065–73.
15. Diorio C, Brisson J, Berube S, Pollak M. Genetic polymorphisms in-
volved in insulin-like growth factor (IGF) pathway in relation to
DBP POLYMORPHISMS AND 25(OH)D CONCENTRATIONS 639
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from

mammographic breast density and IGF levels. Cancer Epidemiol Bio-
markers Prev 2008;17:880–8.
16. Kurylowicz A, Ramos-Lopez E, Bednarczuk T, Badenhoop K. Vitamin
D-binding protein (DBP) gene polymorphism is associated with Graves’
disease and the vitamin D status in a Polish population study. Exp Clin
Endocrinol Diabetes 2006;114:329–35.
17. Lips P. Vitamin D status and nutrition in Europe and Asia. J Steroid
Biochem Mol Biol 2007;103:620–5.
18. Abbas S, Linseisen J, Slanger T, et al. The Gc2 allele of the vitamin D
binding protein is associated with a decreased postmenopausal breast
cancer risk, independent of the vitamin d status. Cancer Epidemiol Bio-
markers Prev 2008;17:1339–43.
19. Braun A, Bichlmaier R, Cleve H. Molecular analysis of the gene for the
human vitamin-D-binding protein (group-specific component): allelic
differences of the common genetic GC types. Hum Genet 1992;89:
401–6.
20. Lauridsen AL, Vestergaard P, Nexo E. Mean serum concentration of
vitamin D-binding protein (Gc globulin) is related to the Gc phenotype
in women. Clin Chem 2001;47:753–6.
21. Lauridsen AL, Vestergaard P, Hermann AP, et al. Plasma concentrations
of 25-hydroxy-vitamin D and 1,25-dihydroxy-vitamin D are related to
the phenotype of Gc (vitamin D-binding protein): a cross-sectional study
on 595 early postmenopausal women. Calcif Tissue Int 2005;77:15–22.
22. Safadi FF, Thornton P, Magiera H, et al. Osteopathy and resistance to
vitamin D toxicity in mice null for vitamin D binding protein. J Clin
Invest 1999;103:239–51.
23. McCullough ML, Stevens VL, Diver WR, et al. Vitamin D pathway
gene polymorphisms, diet, and risk of postmenopausal breast cancer:
a nested case-control study. Breast Cancer Res 2007;9:R9.
24. Kidd LC, Paltoo DN, Wang S, et al. Sequence variation within the 5#
regulatory regions of the vitamin D binding protein and receptor genes
and prostate cancer risk. Prostate 2005;64:272–82.
25. Al-oanzi ZH, Tuck SP, Mastana SS, et al. Vitamin D-binding protein
gene microsatellite polymorphism influences BMD and risk of fractures
in men. Osteoporos Int 2008;19:951–60.
26. Papiha SS, Allcroft LC, Kanan RM, Francis RM, Datta HK. Vitamin D
binding protein gene in male osteoporosis: association of plasma DBP
and bone mineral density with (TAAA)(n)-Alu polymorphism in DBP.
Calcif Tissue Int 1999;65:262–6.
27. Ezura Y, Nakajima T, Kajita M, et al. Association of molecular variants,
haplotypes, and linkage disequilibrium within the human vitamin
D-binding protein (DBP) gene with postmenopausal bone mineral
density. J Bone Miner Res 2003;18:1642–9.
28. Pani MA, Regulla K, Segni M, et al. A polymorphism within the vitamin
D-binding protein gene is associated with Graves’ disease but not with
Hashimoto’s thyroiditis. J Clin Endocrinol Metab 2002;87:2564–7.
29. Mendel CM. The free hormone hypothesis: a physiologically based
mathematical model. Endocr Rev 1989;10:232–74.
30. Arnaud J, Constans J. Affinity differences for vitamin D metabolites
associated with the genetic isoforms of the human serum carrier protein
(DBP). Hum Genet 1993;92:183–8.
31. Rowling MJ, Kemmis CM, Taffany DA, Welsh J. Megalin-mediated
endocytosis of vitamin D binding protein correlates with 25-
hydroxycholecalciferol actions in human mammary cells. J Nutr 2006;
136:2754–9.
32. Gomme PT, Bertolini J. Therapeutic potential of vitamin D-binding
protein. Trends Biotechnol 2004;22:340–5.
33. Giovannucci E, Liu Y, Rimm EB, et al. Prospective study of predictors
of vitamin D status and cancer incidence and mortality in men. J Natl
Cancer Inst 2006;98:451–9.
34. Wortsman J, Matsuoka LY, Chen TC, Lu Z, Holick MF. Decreased
bioavailability of vitamin D in obesity. Am J Clin Nutr 2000;72:690–3.
35. Holick MF. 25-OH-vitamin D assays. J Clin Endocrinol Metab 2005;90:
3128–9.
36. Khoury MJ, Beaty TH, Cohen BH. Fundamentals of genetic epidemi-
ology. Monographs in epidemiology and biostatistics. Vol 19. New York,
NY: Oxford University Press, 1993.
37. Rothman KJ, Greenland S. Modern epidemiology. Philadelphia, PA:
Lippincott-Raven, 1998.
38. Rothman N, Wacholder S, Caporaso NE, Garcia-Closas M, Buetow K,
Fraumeni JF Jr. The use of common genetic polymorphisms to enhance
the epidemiologic study of environmental carcinogens. Biochim Bio-
phys Acta 2001;1471:C1–10.
39. Wacholder S, Rothman N, Caporaso N. Counterpoint: bias from pop-
ulation stratification is not a major threat to the validity of conclusions
from epidemiological studies of common polymorphisms and cancer.
Cancer Epidemiol Biomarkers Prev 2002;11:513–20.
40. Cardon LR, Palmer LJ. Population stratification and spurious allelic
association. Lancet 2003;361:598–604.
640 SINOTTE ET AL
at McGill University Libraries on April 22, 2010 www.ajcn.orgDownloaded from