See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/47791470

Vitamin D Receptor Genotypes, Ultraviolet Radiation Exposure, and Risk of

Non-Hodgkin Lymphoma

Article  in  American journal of epidemiology · November 2010

DOI: 10.1093/aje/kwq340 · Source: PubMed

CITATIONS READS
14 55

7 authors, including:

Sandra Eloranta Kristina Duvefelt

Karolinska Institutet Karolinska Institutet
92 PUBLICATIONS   1,413 CITATIONS    26 PUBLICATIONS   992 CITATIONS   

SEE PROFILE SEE PROFILE

Ellen Chang
Exponent
262 PUBLICATIONS   8,920 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Sandra Eloranta on 22 May 2014.

The user has requested enhancement of the downloaded file.

 American Journal of Epidemiology Vol. 173, No. 1
ª The Author 2010. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of DOI: 10.1093/aje/kwq340
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. Advance Access publication:
November 12, 2010

Original Contribution

Vitamin D Receptor Genotypes, Ultraviolet Radiation Exposure, and Risk of Non-

Hodgkin Lymphoma

Karin Ekström Smedby*, Sandra Eloranta, Kristina Duvefelt, Mads Melbye, Keith Humphreys,
Henrik Hjalgrim, and Ellen T. Chang
* Correspondence to Dr. Karin Ekström Smedby, Department of Medicine Solna, Clinical Epidemiology Unit, Karolinska Institutet and
Karolinska University Hospital, SE-171 76 Stockholm, Sweden (e-mail: karin.ekstrom.smedby@ki.se).

Initially submitted April 9, 2010; accepted for publication September 2, 2010.

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

Ultraviolet radiation (UVR) exposure may influence risk of non-Hodgkin lymphoma (NHL) through vitamin D, with
antineoplastic effects mediated through the vitamin D receptor (VDR). To explore the role of vitamin D in NHL risk
and the potential interaction with UVR, the authors genotyped 10 VDR polymorphisms in 2,448 NHL patients and
1,981 controls from Denmark and Sweden who were recruited in 1999–2002. Odds ratios and 95% confidence
intervals were computed with logistic regression. P values were 2-sided. Most VDR variants (e.g., rs731236/TaqI,
rs15444410/BsmI) were not associated with overall risk of NHL, but there was some evidence of a positive
association between rs4760655 and follicular lymphoma risk (nominal Ptrend ¼ 0.004, corrected Ptrend ¼ 0.24).
There was no support for an effect of interaction between VDR variants and UVR exposure on risk of overall NHL or
B-cell lymphoma subtypes. However, there was some evidence that rs731236 altered associations between UVR
and T-cell NHL risk; while increasing UVR frequency lowered T-cell NHL risk among rs731236 TT carriers, an
elevated risk was observed among rs731236 CC carriers (nominal Pinteraction
0.008, corrected Pinteraction  0.12).
VDR does not appear to harbor major determinants of NHL risk, except perhaps for follicular lymphoma. Possible
heterogeneity in effects of UVR exposure on T-cell lymphoma risk by VDR rs731236 genotype merits further
investigation.

case-control studies; lymphoma, follicular; lymphoma, non-Hodgkin; lymphoma, T-cell; polymorphism, single
nucleotide; receptors, calcitriol; ultraviolet rays; vitamin D

Abbreviations: NHL, non-Hodgkin lymphoma; SCALE, Scandinavian Lymphoma Etiology; SNP, single nucleotide polymorphism;
UVR, ultraviolet radiation; VDR, vitamin D receptor.

Ultraviolet radiation (UVR) exposure from the sun was
reported to decrease risk of non-Hodgkin lymphoma (NHL)
in several recent epidemiologic studies (1–5). Because sun
exposure is the primary source of vitamin D in humans and
its biologically active forms have antineoplastic effects, the
observed inverse association with risk of NHL may be me-
diated through UVR-induced vitamin D activation (6). Stud-
ies of vitamin D levels in serum or vitamin D intake and
NHL risk provide additional, though not unanimous, sup-
port for this hypothesis (7–9).
The vitamin D receptor (VDR) protein is a transcriptional
activator affecting expression of many genes related to cell
growth, differentiation, and apoptosis (10), and it is ex-
pressed in lymphocytes (11). Polymorphic variation in the
VDR gene may affect intracellular vitamin D response (12);
for example, the rs10735810 (FokI) single nucleotide poly-
morphism (SNP) alters the transcription initiation site and
has been associated with risk of various solid cancers (13).
The rs731236 (TaqI) and rs1544410 (BsmI) SNPs appear to
alter gene expression and mRNA stability (14). The
rs731236 (TaqI) SNP was recently reported to be associated
with risk of diffuse large B-cell lymphoma (15) and to in-
teract with UVR exposure in risk of follicular lymphoma
(16). To further explore the potential role of vitamin D in
lymphomagenesis, we investigated 10 VDR variants and
their interaction with UVR in risk of NHL and major

48 Am J Epidemiol 2011;173:48–54

subtypes in a Danish-Swedish case-control study including
4,429 persons.
MATERIALS AND METHODS
The Scandinavian Lymphoma Etiology (SCALE) Study
is described in detail elsewhere (2). In brief, SCALE was
a population-based case-control study encompassing resi-
dents of Sweden and Denmark aged 18–74 years, recruited
in 1999–2002. Patients with a first incident NHL were iden-
tified through national hospital-based networks of physi-
cians. Controls were randomly sampled from nationwide
population registers and were frequency-matched to the
cases by age (in 10-year intervals), sex, and country. Persons
with previous organ transplantation, human immunodefi-
ciency virus infection, or hematopoietic malignancy or
who could not speak Danish or Swedish sufficiently were
excluded. Following informed consent, 81% of eligible
cases and 71% of eligible controls participated in a tele-
phone interview including questions on UVR exposure
(2), and 85% of participating cases and 65% of participating
controls provided a blood specimen. The present study en-
compasses participants who were interviewed, gave blood,
and were of second-generation Nordic origin (17). Ethical
approval was granted by national scientific ethics
committees.
The genotyping methods used have recently been de-
scribed (17). For most study participants, genomic DNA
was isolated from white blood cells with the QIAGEN Maxi
Kit (VWR International AB, Stockholm, Sweden). For a
proportion of the Swedish controls, DNA was prepared
from dried whole blood spots on filter paper and whole-
genome amplified with the AmpliQ Genomic Amplifier
Kit (Ampliqon A/S, Skovlunde, Denmark). Because the
overall genotype concordance between amplified and geno-
mic DNA was 99% (17), we used amplified DNA when
genomic DNA was not available. Thirty-two samples were
not available for genotyping.
To estimate genetic variation in VDR, we first selected 3
validated SNPs based on previous reports of associations
with risk of NHL (rs731236, rs1544410 (15, 16), and
rs10735810 (15)). Secondly, from genetic databases (US
National Center for Biotechnology Information
(www.ncbi.nih.gov); International HapMap Project, phase
II (http://hapmap.ncbi.nlm.nih.gov/)), we chose supplemen-
tary tagging SNPs with a minor allele frequency greater than
5% and r2  0.8, with the aim of improving gene coverage.
There were 76 available markers identified using the de-
scribed selection criteria, and we tagged 35 of these (46%
coverage) with 11 of 12 selected SNPs (rs731236 was un-
available in HapMap phase II), with a mean r2 of 0.95.
Samples were genotyped with matrix-assisted laser
desorption/ionization time-of-flight mass spectrometry
(Sequenom Inc., San Diego, California) (17). A success rate
of 85% or more was considered satisfactory; 1 SNP
(rs10735810) had an unacceptable success rate among the
Danish participants and was therefore analyzed statistically
only among the Swedes. Regenotyping of 5% of the study
samples resulted in 99% concordance. Calculations of
Am J Epidemiol 2011;173:48–54
VDR Genotypes, Ultraviolet Light, and NHL Risk 49
Hardy-Weinberg equilibrium were conducted among the
controls using the v2 test; genotype frequencies among con-
trols were considered not to deviate from Hardy-Weinberg
equilibrium if P > 0.05/12 (divided by the number of
SNPs); 2 SNPs were not in Hardy-Weinberg equilibrium
and were excluded. Nine SNPs were included in the final
analysis of both countries. Genotyping was validated using
14 trio families (42 persons in total) with HapMap genotype
data. Concordance analyses with the HapMap data, as well
as analysis of parent-offspring compatibility with the
produced genotypes, were performed; there was 100% con-
cordance for all investigated SNPs. Some samples were
excluded because of a low genotyping completion rate
(<50%, n ¼ 228) or labeling errors (n ¼ 2). A low genotyp-
ing completion rate threshold was used because of a differ-
ence between genomic and amplified DNA samples
(rate 80% for over 90% of genomic samples but only for
50% of amplified samples). As a safeguard against poten-
tially lower quality of the amplified DNA, we repeated some
Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012
statistical analyses after restricting the analyses to persons
for whom genomic DNA was used for genotyping (75).
We computed odds ratios and 95% confidence intervals as
estimates of the relative risk of NHL overall and NHL sub-
types using unconditional logistic regression, with major
allele homozygosity used as the referent category. Trend
per allele was tested with the Cochran-Armitage test. In-
teraction between SNPs and 3 measures of UVR exposure
was assessed in models with and without an interaction
term. Genotype analyses were adjusted for country of resi-
dence, and interaction analyses were additionally adjusted
for sex and age (matching variables). We performed multi-
locus tests of association, comparing models with and with-
out investigated genotype covariates. Although it does not
explicitly assign alleles to chromosomes, this is described as
a powerful haplotype test of association in tightly linked
regions (18). P values were 2-sided and were corrected for
multiple comparisons with the Bonferroni method by mul-
tiplying the uncorrected P value by the number of analyses
performed (45 main-effects analyses þ 15 interaction anal-
yses). Statistical analyses were conducted using STATA
10.1 (Stata Corporation, College Station, Texas).
RESULTS
The final analyses encompassed 2,448 patients with
incident NHL and 1,981 controls (Table 1). Minor allele
frequencies of studied VDR variants were similar to those
in previous reports on Caucasian populations (Table 2)
(www.hapmap.org; 14). None of the variants were associ-
ated with risk of overall NHL, diffuse large B-cell
lymphoma, chronic lymphocytic leukemia (Table 3), or
T-cell lymphoma (data not shown). The rs4760655 variant
was associated with a 60% increase in risk of follicular
lymphoma (Table 3; nominal Ptrend ¼ 0.004, corrected
Ptrend ¼ 0.24), and the odds ratio was unaltered upon
restriction to subjects with genomic DNA (data not shown).
The rs2283342 variant was also associated with elevated
risk of follicular lymphoma in the overall analysis (Table 3;
nominal Ptrend ¼ 0.01), but the association disappeared
50 Smedby et al.

Table 1. Characteristics of Study Participants From Denmark and statistically significant interaction with respect to risk of
Sweden, Scandinavian Lymphoma Etiology Study, 1999–2002 overall NHL or follicular lymphoma (Table 4). However,
Controls Cases
the inverse association between UVR and risk of NHL
and follicular lymphoma previously observed in our study
No. % No. %
was restricted to rs731236 TT carriers, while increasing
Total 1,981 100 2,448 100 UVR frequency among CC homozygotes appeared, if any-
Country of residence thing, to increase risk. For T-cell lymphoma, a stronger pat-
Denmark 766 39 784 32 tern of differential associations with UVR across rs731236
Sweden 1,215 61 1,664 68
genotypes was observed (nominal Pinteraction
0.008
(Table 3)); it was no longer significant, however, after
Sex
correction for multiple comparisons (for sunbathing 5–10
Male 1,076 54 1,474 60 years previously, corrected Pinteraction ¼ 0.48; for sunbathing
Female 905 46 974 40 vacations abroad, corrected Pinteraction ¼ 0.12; and for
Age, yearsa sunbathing at age 20 years, corrected Pinteraction ¼ 0.18).
18–24 41 2 21 1 When analyses were restricted to participants with genomic
DNA, similar results were obtained.
25–34 64 3 60 2
35–44 131 7 150 6
45–54 342 17 464 19 DISCUSSION

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

55–64 620 31 816 33
In this large, population-based, and ethnically homoge-
65–75 783 40 937 38
neous case-control study, there was no evidence that the
Nordic background studied VDR variants constitute major determinants of risk
Both parents born in 1,885 95 2,314 95 of NHL overall or common NHL subtypes, with the poten-
Denmark or Sweden
tial exception of follicular lymphoma. Our results for
Either or both parents born 96 5 134 5 rs4760655 and follicular lymphoma could suggest that
in another Nordic countryb
variation in the 5# promoter region of VDR affects suscep-
a
Age at diagnosis for cases and age at inclusion in the study for tibility to this specific NHL subtype, but they could also be
controls. The median age of both cases and controls was 61 years due to chance. Our results further suggest a possible inter-
(range, 18–74 for cases and 18–75 for controls). action between sunbathing history and rs731236 (TaqI)
b
Norway, Finland, or Iceland. genotypes in risk of T-cell lymphoma, with some evidence
of a beneficial effect of frequent UVR among major-allele
when the analysis was restricted to subjects with genomic homozygotes but a harmful effect among minor-allele
DNA (data not shown). Haplotypes based on 4 adjacent homozygotes.
SNPs at a time were not associated with overall NHL risk In our previous study of sunbathing and other UVR expo-
(data not shown). sure measures and risk of NHL in the same population (2), we
In analyses of the combined effects of rs731236 geno- observed a protective effect of UVR on risk of B-cell NHL
types (previously reported to interact with UVR in risk of but not T-cell NHL. If the presently suggested interaction
follicular lymphoma (16)) and UVR exposure, there was no reflects true biologic variation and not just chance, combined

Table 2. Characteristics of Selected Vitamin D Receptor Gene (12q13.1–q13.3) Single

Nucleotide Polymorphisms, Scandinavian Lymphoma Etiology Study, 1999–2002

Nucleotide Minor Allele

SNP Location of SNP P Valuea
Substitution Frequency, %

rs4760655 5# promoter region A/G 33.9 0.421

rs3890734 5# promoter region G/A 39.0 0.966
rs2853559 5# promoter region G/A 44.0 0.461
rs10735810 (FoqI)b Exon 2 G/A 38.8 0.092
rs886441 Intron 3 A/G 17.0 0.297
rs1540339 Intron 4 C/T 35.9 0.870
rs2239181 Intron 4 A/C 11.0 0.993
rs2283342 Intron 4 A/G 15.0 0.838
rs1544410 (BsmI) Intron 8 G/A 41.9 0.912
rs731236 (TaqI) Exon 9 T/C 42.3 0.743

Abbreviation: SNP, single nucleotide polymorphism.

a
Pearson v2 test for Hardy-Weinberg equilibrium among the controls.
b
Data for Swedish participants only.

Am J Epidemiol 2011;173:48–54
 VDR Genotypes, Ultraviolet Light, and NHL Risk 51

Table 3. Frequencies and Odds Ratiosa for Associations Between Vitamin D Receptor Genotypes and Risk of Non-Hodgkin Lymphoma, Overall
and by Subtype, Scandinavian Lymphoma Etiology Study, 1999–2002

Diffuse Large B-Cell Chronic Lymphocytic

Controls All Non-Hodgkin Lymphoma Follicular Lymphoma
VDR Genotype Lymphoma Leukemiab
No. % No. % OR 95% CI No. % OR 95% CI No. % OR 95% CI No. % OR 95% CI

rs4760655
AA 713 45 1,009 43 1 Referent 267 45 1 Referent 169 39 1 Referent 248 43 1 Referent
AG 698 44 1,045 45 1.06 0.93, 1.22 266 45 1.03 0.84, 1.26 194 45 1.19 0.94, 1.50 264 46 1.09 0.89, 1.34
GG 185 12 279 12 1.08 0.87, 1.33 63 11 0.93 0.67, 1.28 70 16 1.64* 1.18, 2.26 65 11 1.02 0.74, 1.40
rs3890734
GG 552 38 839 37 1 Referent 214 38 1 Referent 169 40 1 Referent 207 37 1 Referent
AG 691 47 1,044 47 1.01 0.87, 1.17 254 5 0.97 0.78, 1.21 187 45 0.91 0.72, 1.16 272 49 1.06 0.85, 1.31
AA 216 15 360 16 1.08 0.88, 1.32 98 17 1.16 0.87, 1.55 63 15 0.94 0.67, 1.31 80 14 0.97 0.72, 1.32
rs2853559
GG 518 33 715 31 1 Referent 181 30 1 Referent 139 32 1 Referent 164 29 1 Referent
GA 758 48 1,120 49 1.09 0.94, 1.26 292 49 1.14 0.91, 1.42 199 46 1.01 0.79, 1.30 289 51 1.23 0.98, 1.54

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

AA 298 19 470 20 1.14 0.95, 1.38 123 21 1.19 0.91, 1.57 90 21 1.14 0.84, 1.54 113 20 1.21 0.92, 1.60
rs886441
AA 1,030 69 1,568 69 1 Referent 388 69 1 Referent 295 70 1 Referent 390 69 1 Referent
AG 409 27 625 28 1.04 0.90, 1.21 163 29 1.09 0.88, 1.36 119 28 1.05 0.82, 1.33 155 27 1.03 0.83, 1.28
GG 49 3 69 3 1.03 0.70, 1.51 15 3 0.92 0.50, 1.67 9 2 0.69 0.34, 1.44 22 4 1.29 0.77, 2.17
rs1540339
CC 716 41 948 42 1 Referent 259 44 1 Referent 153 39 1 Referent 246 44 1 Referent
CT 812 46 1,000 44 0.92 0.80, 1.05 266 45 0.89 0.73, 1.09 181 46 1.04 0.82, 1.32 227 41 0.80* 0.65, 0.99
TT 225 13 318 14 1.05 0.86, 1.27 68 11 0.81 0.60, 1.11 60 15 1.24 0.89, 1.74 86 15 1.10 0.82, 1.47
rs2239181
AA 1,446 79 1,834 79 1 Referent 489 81 1 Referent 327 82 1 Referent 451 79 1 Referent
AC 358 20 432 19 0.95 0.81, 1.11 107 18 0.88 0.69, 1.11 63 16 0.78 0.58, 1.04 110 19 0.98 0.77, 1.24
CC 22 1 39 2 1.35 0.80, 2.30 8 1 1.09 0.48, 2.46 7 2 1.40 0.59, 3.32 11 2 1.58 0.76, 3.29
rs2283342
AA 1,090 74 1,589 71 1 Referent 399 72 1 Referent 286 67 1 Referent 412 74 1 Referent
AG 363 24 584 26 1.05 0.90, 1.22 140 25 1.00 0.80, 1.26 127 30 1.28* 1.00, 1.63 135 24 0.94 0.75, 1.19
GG 29 2 55 2 1.30 0.82, 2.06 11 2 1.07 0.52, 2.17 14 3 1.90 0.99, 3.67 9 2 0.88 0.41, 1.89
rs1544410
GG 620 35 783 34 1 Referent 209 34 1 Referent 135 35 1 Referent 206 36 1 Referent
GA 863 48 1,096 48 1.01 0.88, 1.16 275 45 0.95 0.77, 1.17 194 50 1.03 0.81, 1.32 257 45 0.90 0.73, 1.11
AA 306 17 424 18 1.11 0.93, 1.34 123 20 1.23 0.94, 1.60 59 15 0.89 0.64, 1.25 110 19 1.09 0.84, 1.43
rs731236
TT 604 34 789 35 1 Referent 207 35 1 Referent 139 36 1 Referent 207 36 1 Referent
TC 870 48 1,053 46 0.93 0.81, 1.07 263 44 0.89 0.72, 1.09 186 49 0.93 0.73, 1.19 250 44 0.84 0.68, 1.04
CC 325 18 428 19 1.01 0.85, 1.21 124 21 1.12 0.87, 1.46 56 15 0.75 0.53, 1.05 113 20 1.02 0.78, 1.33

Abbreviations: CI, confidence interval; OR, odds ratio; VDR, vitamin D receptor.
* P < 0.05.
a
Odds ratios were adjusted for country.
b
Also included small lymphocytic lymphoma.

effects of UVR and VDR genotype may vary by lymphoma
cell type. Possible explanations for differing interaction pat-
terns by B-/T-cell lineage could include variation in the fre-
quency of VDR expression and transcriptional effects of
vitamin D, in combination with differences in the inherent
immunoregulatory roles of B- and T-lymphocytes (19).
In a US study (16), the rs731236 and rs1544410 (BsmI)
variants were not associated with risk of overall NHL or
major B-cell subtypes. However, an interaction was reported
between UVR and rs731236 in risk of follicular lymphoma.
Among persons with low sun exposure, rs731236 minor-
allele homozygosity was associated with a 6- to 10-fold

Am J Epidemiol 2011;173:48–54
52 Smedby et al.

Table 4. Odds Ratiosa for Associations of Selected Ultraviolet Radiation Exposures and Vitamin D Receptor rs731236 Genotypes With Risk of
Non-Hodgkin Lymphoma, Overall and by Subtype, Scandinavian Lymphoma Etiology Study, 1999–2002

rs731236 Genotype
Ultraviolet Radiation Exposure TT TC CC
Nca/Ncob OR 95% CI Nca/Nco OR 95% CI Nca/Nco OR 95% CI

All Non-Hodgkin Lymphoma (n ¼ 2,448)

Sunbathing frequency 5–10
years previously, times/week
0 245/176 1 Referent 322/211 1.09 0.83, 1.43 132/102 0.93 0.67, 1.29
1 259/183 0.95 0.72, 1.26 324/275 0.76 0.53, 1.10 133/106 0.98 0.62, 1.55
2–3 140/126 0.74 0.54, 1.02 197/177 0.93 0.61, 1.40 59/59 1.01 0.59, 1.75
>3 143/114 0.83 0.60, 1.14 196/201 0.72 0.48, 1.09 101/57 1.54 0.91, 2.60
P for interactionc 0.26
Lifetime no. of sunbathing
vacations abroad
0 260/159 1 Referent 324/228 0.86 0.66, 1.12 113/90 0.79 0.56, 1.11
1–5 260/208 0.73* 0.56, 0.97 347/269 1.21 0.84, 1.74 148/103 1.48 0.93, 2.36

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

6–20 204/169 0.69* 0.52, 0.93 291/265 1.04 0.72, 1.52 125/97 1.43 0.88, 2.32
>20 65/67 0.58* 0.39, 0.87 88/106 0.95 0.57, 1.61 41/36 1.37 0.70, 2.67
P for interaction 0.29
Sunbathing frequency at
age 20 yearsd, times/week
0 140/80 1 Referent 198/127 0.84 0.58, 1.21 79/58 0.74 0.48, 1.16
1 266/195 0.71 0.50, 1.00 311/258 1.03 0.66, 1.61 125/103 1.18 0.68, 2.05
2–3 171/129 0.66* 0.46, 0.96 213/211 0.90 0.56, 1.45 89/67 1.39 0.76, 2.52
>3 164/137 0.57* 0.40, 0.83 230/195 1.18 0.73, 1.89 92/64 1.74 0.95, 3.16
P for interaction 0.09
Follicular Lymphoma (n ¼ 466)
Sunbathing frequency 5–10 years
previously, times/week
0 40/176 1 Referent 63/211 1.37 0.87, 2.15 18/102 0.75 0.40, 1.39
1 51/183 1.08 0.67, 1.75 54/275 0.50* 0.27, 0.94 20/106 0.92 0.40, 2.15
2–3 19/126 0.57 0.31, 1.05 35/177 0.97 0.45, 2.07 6/59 0.89 0.28, 2.82
>3 28/114 1.01 0.58, 1.75 31/201 0.46* 0.22, 0.94 11/57 1.05 0.39, 2.83
P for interaction 0.69
Lifetime no. of sunbathing
vacations abroad
0 45/159 1 Referent 52/228 0.81 0.51, 1.28 15/90 0.60 0.31, 1.14
1–5 47/208 0.74 0.46, 1.18 65/269 1.32 0.71, 2.46 16/103 1.10 0.45, 2.69
6–20 41/169 0.72 0.44, 1.17 52/265 1.01 0.53, 1.93 21/97 1.54 0.64, 3.72
>20 6/67 0.30* 0.12, 0.73 17/106 1.94 0.65, 5.78 4/36 1.94 0.44, 8.56
P for interaction 0.27
Table continues

increased risk of follicular lymphoma compared with major-
allele homozygosity (16). This interaction pattern is different
from what we observed, although follicular lymphoma was
indicated as potentially associated with VDR variation in both
studies. In an Australian case-control study (15), a positive
association was observed between VDR rs731236 and
rs1544410 and risk of diffuse large B-cell lymphoma, but
interaction with UVR exposure was not investigated.
The intronic rs4760655 variant potentially linked to risk
of follicular lymphoma in our study is located in the 5#
promoter region of VDR and is in linkage disequilibrium
with a functional SNP at the Cdx2 binding site
(rs11568820; D# ¼ 1, r2 ¼ 0.1 in a pairwise comparison)
(14), although the 2 variants appear to belong to different
haplotype blocks (www.hapmap.org). In the 3# untranslated
region, rs731236 and rs1544410 are in strong linkage dis-
equilibrium (20). Different sets of alleles in this region have
been associated with VDR mRNA expression level and sta-
bility, with consequences for the number of VDR proteins
expressed in target cells and for intracellular vitamin D

Am J Epidemiol 2011;173:48–54
 VDR Genotypes, Ultraviolet Light, and NHL Risk 53

Table 4. Continued

rs731236 Genotype
Ultraviolet Radiation Exposure TT TC CC
Nca/Ncob OR 95% CI Nca/Nco OR 95% CI Nca/Nco OR 95% CI
Sunbathing frequency at
age 20 yearsd, times/week
0 26/80 1 Referent 45/127 0.97 0.55, 1.72 17/58 0.80 0.39, 1.64
1 41/195 0.52 0.29, 0.93 48/258 0.90 0.43, 1.89 14/103 0.75 0.28, 2.00
2–3 38/129 0.71 0.40, 1.29 40/211 0.70 0.32, 1.50 8/67 0.45 0.15, 1.39
>3 27/137 0.49* 0.26, 0.92 41/195 1.12 0.51, 2.47 12/64 1.22 0.43, 3.44
P for interaction 0.89
T-Cell Lymphoma (n ¼ 155)
Sunbathing frequency 5–10 years
previously, times/week
0 19/176 1 Referent 15/211 0.61 0.30, 1.24 4/102 0.36 0.12, 1.08
1 21/183 0.75 0.38, 1.49 28/275 1.55 0.61, 3.98 10/106 2.62 0.66, 10.3
2–3 10/126 0.58 0.25, 1.31 15/177 1.72 0.57, 5.23 2/59 1.16 0.17, 7.85

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

>3 7/114 0.44 0.18, 1.10 13/201 2.08 0.64, 6.81 10/57 8.84* 1.95, 40.1
P for interaction 0.008
Lifetime no. of sunbathing
vacations abroad
0 26/159 1 Referent 19/228 0.50* 0.26, 0.95 3/90 0.21* 0.06, 0.73
1–5 18/208 0.45 0.23, 0.86 24/269 2.21 0.89, 5.48 9/103 4.95 1.11, 22.0
6–20 9/169 0.27* 0.12, 0.61 24/265 3.44* 1.24, 9.57 11/97 10.70 2.31, 50.1
>20 4/67 0.35 0.12, 1.06 4/106 1.27 0.27, 6.06 3/36 6.45 0.88, 47.2
P for interaction 0.002
Sunbathing frequency at
age 20 yearsd, times/week
0 10/80 1 Referent 3/127 0.19 0.05, 0.74 1/58 0.15 0.02, 1.21
1 23/195 0.71 0.50, 1.00 311/258 4.36 1.00, 19.0 8/103 4.09 0.42, 39.8
2–3 8/129 0.48 0.17, 1.31 213/211 4.06 0.78, 21.2 5/67 8.32* 1.67, 41.4
>3 10/137 0.42 0.15, 1.14 230/195 8.32* 1.67, 41.4 7/64 14.50* 1.38, 152
P for interaction 0.003

Abbreviations: ca, cases; CI, confidence interval; co, controls; OR, odds ratio.
* P < 0.05.
a
Odds ratios were adjusted for country, sex, and age (matching variables).
b
No. of cases/no. of controls.
c
P values were 2-sided and were computed by means of the Wald test in logistic regression models with and without an interaction term.
d
Restricted to persons over age 39 years at lymphoma diagnosis (cases) or interview (controls).

response (14); rs731236 T and rs1544410 G alleles have
been correlated with lower levels of VDR mRNA expres-
sion, and C and A, respectively, have been correlated with
higher activity (20). The present results would thus be con-
sistent with a reduction in T-cell lymphoma risk by high
UVR exposure among persons with low-activity VDR
alleles.
Our study had several strengths, including its large size,
the population-based design, and the ability to examine as-
sociations with NHL subtypes. Importantly, the study par-
ticipants were of homogeneous ethnic background, which
minimized bias due to population stratification. However,
the power to detect modest genetic effects and to robustly
determine gene-environment interaction was still limited.
Therefore, we may have failed to detect some true associa-
tions while observing some nominally statistically signifi-
cant associations due to chance alone. Upon adjustment for
multiple testing, our results were no longer statistically sig-
nificant. The investigated SNPs did not fully characterize
common variation in the VDR gene, and thus an association
between other VDR variants and NHL risk cannot be ruled
out. In addition, a risk association with common variation in
VDR does not necessarily imply an etiologic role of vitamin
D, since VDR has functions that are independent of vitamin
D (21). Our questionnaire aimed to characterize UVR ex-
posure during summer and overall throughout the year, but
a recent investigation suggested that UVR exposure/vitamin
D status during the winter months may be more relevant (8).

Am J Epidemiol 2011;173:48–54
54 Smedby et al.

Hence, more detailed assessment of seasonal UVR exposure
would have been desirable.
In conclusion, this study does not provide support for the 5.
investigated VDR variants as important determinants of risk
of NHL or major subtypes, except perhaps follicular lym-
6.
phoma. Consequently, the study also provides no additional
strong evidence for a role of vitamin D in risk of B-cell
NHL. However, we found some evidence of an interaction 7.
between VDR rs731236 (TaqI) and UVR exposure in risk of
T-cell lymphoma, which merits further study in larger
pooled data sets. Given the increasing recognition of the
importance of vitamin D for both immunity and cancer, 8.
further studies of different measures of vitamin D status,
including seasonal UVR, in relation to NHL risk are
9.
warranted.
10.
ACKNOWLEDGMENTS
multiple myeloma—a multicentre European case-control
study. Int J Epidemiol. 2008;37(5):1080–1094.
Grandin L, Orsi L, Troussard X, et al. UV radiation exposure,
skin type and lymphoid malignancies: results of a French case-
control study. Cancer Causes Control. 2008;19(3):305–315.
Egan KM, Sosman JA, Blot WJ. Sunlight and reduced risk of
cancer: is the real story vitamin D? J Natl Cancer Inst. 2005;
97(3):161–163.
Polesel J, Talamini R, Montella M, et al. Linoleic acid, vitamin
D and other nutrient intakes in the risk of non-Hodgkin lym-
phoma: an Italian case-control study. Ann Oncol. 2006;17(4):
713–718.
Lim U, Freedman DM, Hollis BW, et al. A prospective in-
vestigation of serum 25-hydroxyvitamin D and risk of lym-
phoid cancers. Int J Cancer. 2009;124(4):979–986.
Erber E, Maskarinec G, Lim U, et al. Dietary vitamin D and
risk of non-Hodgkin lymphoma: the Multiethnic Cohort. Br J
Nutr. 2010;103(4):581–584.
Uitterlinden AG, Fang Y, van Meurs JB, et al. Vitamin D re-
ceptor gene polymorphisms in relation to vitamin D related
disease states. J Steroid Biochem Mol Biol. 2004;89-90(1–5):

Downloaded from http://aje.oxfordjournals.org/ by guest on April 6, 2012

Author affiliations: Clinical Epidemiology Unit, Depart-
ment of Medicine, Karolinska University Hospital, Stock-
holm, Sweden (Karin Ekström Smedby); Department
of Medical Epidemiology and Biostatistics, Karolinska
Institutet, Stockholm, Sweden (Sandra Eloranta, Keith
Humphreys); Clinical Research Centre, Karolinska Univer-
sity Hospital, Stockholm, Sweden (Kristina Duvefelt);
Department of Epidemiology Research, Statens Serum
Institut, Copenhagen, Denmark (Mads Melbye, Henrik
Hjalgrim); Northern California Cancer Center, Fremont,
California (Ellen Chang); and Division of Epidemiology,
Department of Health Research and Policy, Stanford Uni-
versity School of Medicine, Stanford, California (Ellen
Chang).
The current investigation was supported by the Swedish
Research Council (grant K2008-64X-20737-01-2), the Dan-
ish Medical Research Foundation (grant 09-066021), the
Swedish Cancer Fund (grant 2007/667), and the Nordic
Cancer Union (grant S-10/07).
Conflict of interest: none declared.
REFERENCES
1. Hughes AM, Armstrong BK, Vajdic CM, et al. Sun exposure
may protect against non-Hodgkin lymphoma: a case-control
study. Int J Cancer. 2004;112(5):865–871.
2. Smedby KE, Hjalgrim H, Melbye M, et al. Ultraviolet radia-
tion exposure and risk of malignant lymphomas. J Natl Cancer
Inst. 2005;97(3):199–209.
3. Weihkopf T, Becker N, Nieters A, et al. Sun exposure and
malignant lymphoma: a population-based case-control study
in Germany. Int J Cancer. 2007;120(11):2445–2451.
4. Boffetta P, van der Hel O, Kricker A, et al. Exposure to
ultraviolet radiation and risk of malignant lymphoma and
187–193.
11. Jeffery LE, Burke F, Mura M, et al. 1,25-Dihydroxyvitamin
D3 and IL-2 combine to inhibit T cell production of inflam-
matory cytokines and promote development of regulatory T
cells expressing CTLA-4 and FoxP3. J Immunol. 2009;183(9):
5458–5467.
12. Jurutka PW, Whitfield GK, Hsieh JC, et al. Molecular nature of
the vitamin D receptor and its role in regulation of gene ex-
pression. Rev Endocr Metab Disord. 2001;2(2):203–216.
13. Köstner K, Denzer N, Müller CS, et al. The relevance of
vitamin D receptor (VDR) gene polymorphisms for cancer:
a review of the literature. Anticancer Res. 2009;29(9):
3511–3536.
14. Fang Y, van Meurs JB, d’Alesio A, et al. Promoter and 3#-
untranslated-region haplotypes in the vitamin D receptor gene
predispose to osteoporotic fracture: the Rotterdam Study. Am J
Hum Genet. 2005;77(5):807–823.
15. Purdue MP, Lan Q, Kricker A, et al. Vitamin D receptor gene
polymorphisms and risk of non-Hodgkin’s lymphoma.
Haematologica. 2007;92(8):1145–1146.
16. Purdue MP, Hartge P, Davis S, et al. Sun exposure, vitamin D
receptor gene polymorphisms and risk of non-Hodgkin lym-
phoma. Cancer Causes Control. 2007;18(9):989–999.
17. Fernberg P, Chang ET, Duvefelt K, et al. Genetic variation in
chromosomal translocation breakpoint and immune function
genes and risk of non-Hodgkin lymphoma. Cancer Causes
Control. 2010;21(5):759–769.
18. Clayton D, Chapman J, Cooper J. Use of unphased multilocus
genotype data in indirect association studies. Genet Epidemiol.
2004;27(4):415–428.
19. Adams JS, Hewison M. Unexpected actions of vitamin D: new
perspectives on the regulation of innate and adaptive immu-
nity. Nat Clin Pract Endocrinol Metab. 2008;4(2):80–90.
20. Uitterlinden AG, Fang Y, Van Meurs JB, et al. Genetics and
biology of vitamin D receptor polymorphisms. Gene. 2004;
338(2):143–156.
21. Cianferotti L, Cox M, Skorija K, et al. Vitamin D receptor is
essential for normal keratinocyte stem cell function. Proc Natl
Acad Sci U S A. 2007;104(22):9428–9433.

Am J Epidemiol 2011;173:48–54

View publication stats