SEARCH

SEARCH BY CITATION

Keywords:

  • 1958 British birth cohort;
  • 25-hydroxyvitamin D;
  • allergy;
  • gene × environment interaction;
  • total IgE

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

Background

The hormonal form of vitamin D affects both adaptive and innate immune functions involved in the development of allergies. Certain genotypes have been seen to alter the association between vitamin D deficiency (VDD) and the risk of food sensitization in children.

Methods

We examined 27 functional single nucleotide polymorphisms (SNPs) in/near selected candidate genes for association with total immunoglobulin E (IgE) and effect modification by 25-hydroxyvitamin D in the 1958 British birth cohort (aged 45 years, n = 4921). A cut-off value of 50 nmol/L was used to define VDD.

Results

Four SNPs (in FCER1A, IL13, and CYP24A1) and three SNPs (in IL4 and CYP24A1) were associated with total IgE and specific IgE, respectively, after correction for multiple testing. As in a previous study, MS4A2 (rs512555, Pinteraction = 0.04) and IL4 (rs2243250, Pinteraction = 0.02), and their composite score (Pinteraction = 0.009) modified the association between VDD and allergy-related outcome. Vitamin D deficiency was associated with higher total IgE only in the carriers of the ‘C’ allele (IL4), which is present in 86% of white Europeans, while only 26% of Chinese and <20% of some African populations are carriers.

Conclusions

Our study on white European adults was consistent with a previous study on children from largely non-white ethnic groups, suggesting that IL4 and MS4A2 genotypes modify the association between VDD and allergy risk. The risk allele in IL4 is present in nearly 90% of white Europeans, while less than a quarter are carriers in some other populations, highlighting the need to consider possible ethnic differences in allergy-related responsiveness to VDD.

Abbreviations
25(OH)D

25-hydroxyvitamin D

GWAS

genome-wide association study

HWE

Hardy–Weinberg equilibrium

IgE

Immunoglobulin E

SNP

Single nucleotide polymorphism

The hormonal form of vitamin D exerts profound effects on inflammatory pathways and cells involved in the development of allergies [1]. Epidemiological studies suggest that 25-hydroxyvitamin D [25(OH)D, an indicator for nutritional vitamin D status] may reduce the risk of immunological diseases such as type 1 diabetes [2] and multiple sclerosis [3]. For allergic disease and asthma, the evidence is less consistent with both low and high 25(OH)D concentrations suggested to be associated with increased risk [1, 4-6]. However, at the molecular level, the active vitamin D metabolite, calcitriol, has been shown to suppress dendritic cell maturation and consecutive Th1 cell development by blocking the IL-12 signal, which is an independent key mechanism for the development of allergy [5, 7-9]. Also, studies in mice have shown that treatment with 1,25-dihydroxyvitamin D results in the reduced secretion of Th1-type cytokines IL-2 and INF-γ and an increase in Th2-type IL-4 [10]. Furthermore, epidemiological studies have provided evidence for a link between early vitamin D supplementation and allergy development, where several vitamin D genes play a role [8].

The contribution of genetic factors involved in vitamin D metabolism to the development of allergic conditions comes from genetic studies investigating the association of vitamin D receptor polymorphisms with atopy and asthma [11, 12]. A study in 1064 individuals examined the association of 83 tagging polymorphisms (in 11 vitamin D–related genes) with atopy and asthma and found that a number of genes demonstrated modest levels of association with both the traits [13]. In a recent study on 649 children, vitamin D deficiency (VDD) increased the risk of food sensitization only among individuals with certain genotypes, providing evidence for gene–vitamin D interaction on food sensitization [14]. Using data from 9377 participants in the 1958 British birth cohort, we have previously reported evidence for a strong U-shaped association between 25(OH)D and total immunoglobulin E (IgE), with an association seen for polymorphism encoding the vitamin D activation enzyme, CYP27B1 [4].

Based on previous observations on gene–environment interactions in relation to food sensitization in children [14], we hypothesized that genetic variation may also affect vitamin D–related allergy responses in adults. Our study relies on an assumption that despite differences between the etiologies leading to food sensitization or elevations in systemic IgE levels, there may be shared influences on immune activation or modulation [5, 15]. In line with this hypothesis, we tested whether three single nucleotide polymorphisms (SNPs) [(rs2243250 (IL4), rs512555 (MS4A2), and rs2762934 (CYP24A1)] that showed a significant interaction with VDD in the previous study on children [14] also modify the association between VDD and total IgE/specific IgE in adults. We also systematically examined 27 potentially functional SNPs in/near 12 genes that affect total IgE or 25(OH)D for their associations with total IgE and specific IgE.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

Study population

Study participants are from the 1958 British birth cohort (1958BC) (Table 1), which initially included all births in England, Scotland, and Wales during first week in March 1958 (= 16 751) [16]. Most recently, participants were contacted between September 2002 and April 2004, when the majority was aged 45 years [range 44 (31.1%)–46 (0.4%)]. The target population for the survey consisted of 11 971 individuals currently living in Britain. Seventy-eight percent (= 9349) of the participants filled in a questionnaire, of whom 8302 (88.5%) also provided blood samples. The 1958BC is largely a white European population (98%), and despite some data attrition, it has been evaluated to be broadly representative of the surviving cohort [17].

Table 1. Characteristics of the participants in the 1958 British birth cohort (n = 4921)
Number (%)25(OH)D (nmol/L)Immunoglobulin E (kU/l)
Geometric mean (95% CI)25(OH)D ≤50 (%)Geometric mean (95% CI)Specific production (%)a
  1. P-values from likelihood ratio test, adjusted for sex, month of blood sample, and region, from linear regression for continuous outcomes and logistic regression for dichotomous outcomes.

  2. a

    Number of individuals with at least one specific IgE (cat, grass, or dust) >0.35 kU/l.

Sex
Male2464 (50.1)52.9 (51.9, 53.8)1013 (41.1)35.6 (33.7, 37.7)781 (31.8)
Female2457 (49.9)51.0 (50.0, 51.9)1052 (42.8)23.9 (22.6, 25.2)532 (21.7)
P-value 0.0010.161<0.001<0.001
Month in which blood sample was taken
January380 (7.7)39.5 (37.7, 41.4)258 (67.9)31.5 (27.5, 36.2)97 (25.5)
February280 (5.7)36.5 (34.6, 38.5)203 (72.5)30.1 (25.5, 35.6)67 (24.0)
March381 (7.7)39.5 (37.7, 41.3)252 (66.1)24.9 (21.5, 28.9)92 (24.2)
April313 (6.4)40.2 (38.4, 42.1)207 (66.1)27.0 (23.1, 31.5)87 (27.9)
May384 (7.8)44.2 (42.2, 46.3)225 (58.6)27.3 (23.7, 31.6)99 (25.8)
June471 (9.6)49.5 (47.7, 51.3)213 (45.2)28.0 (24.6, 31.7)118 (25.1)
July392 (7.9)61.2 (59.0, 63.5)89 (22.7)30.0 (26.3, 34.1)105 (26.8)
August285 (5.8)65.3 (62.7, 68.1)58 (20.4)35.0 (29.6, 41.3)81 (28.4)
September605 (12.3)71.9 (69.8, 74.1)85 (14.0)32.6 (29.0, 36.7)183 (30.4)
October623 (12.7)62.4 (60.4, 64.4)158 (25.4)26.5 (23.5, 29.8)167 (26.8)
November629 (12.8)54.2 (52.5, 5.9)241 (38.3)30.4 (27.1, 34.0)170 (27.0)
December178 (3.6)52.4 (49.1, 55.9)76 (42.7)28.9 (23.6, 35.5)47 (26.4)
P-value <0.001<0.0010.0210.745
Region
South England (including London)1850 (37.6)52.7 (51.6, 53.9)748 (40.4)28.6 (26.8, 30.5)502 (27.2)
Middle England and Wales1290 (26.2)52.3 (51.1, 53.6)516 (40.0)29.4 (27.2, 31.8)338 (26.2)
North England1293 (26.3)53.5 (52.1, 54.9)525 (40.6)29.8 (27.6, 32.2)352 (27.2)
Scotland488 (9.9)44.1 (42.3, 46.0)276 (56.6)29.3 (25.7, 33.4)121 (24.9)
P-value <0.001<0.0010.8340.441

Target sample for the current analyses arises from two genome-wide substudies, where a random sample of cohort members (white European ancestry only) was selected for use as a control population for genome-wide studies. Firstly, 3000 participants were randomly selected as part of the Wellcome Trust Case Control Consortium (WTCCC2) and genotyped on the Affymetrix 6.0 platform [18]. Secondly, 2592 participants from the 1958BC were randomly selected as controls for a type 1 diabetes case–control (T1DGC) study and genotyped using the Illumina Infinium 550 K chip [19]. From the 5233 participants with genome-wide data that passed quality control (QC) measures (2530 samples passing QC in T1DGC and 2703 after QC in WTCCC2), 94% (n = 4921) also had information on serum 25(OH)D concentrations and total IgE, qualifying for the current analyses. Written consent for the use of information in medical studies was obtained from the cohort members. The 45-year biomedical survey and genetic studies were approved by the South-East Multi-Centre Research Ethics Committee (ref: 01/1/44) and the joint UCL/UCLH Committees on the Ethics of Human Research (Ref: 08/H0714/40).

Measures

The 25(OH)D was measured using automated application of an enzyme-linked immunosorbent assay (IDS OCTEIA ELISA; IDS, Bolton, UK) and an analyser (BEP2000; Dade-Behring, Milton Keynes, UK) with sensitivity of 5.0 nmol/L, linearity ≤155 nmol/L, and intra-assay CV 5.5–7.2%. The 25(OH)D concentrations were standardized according to the mean of the values found by the Vitamin D External Quality Assurance Survey (DEQAS) [20]. Total IgE was assayed using the HYTEC-automated enzyme immunoassay [21]. Specific IgE to cat, grass, and dust were measured if total IgE was ≥30 kU/l. Specific IgE production was defined as at least one specific IgE > 0.35 kU/l. Information on specific IgE was missing in five individuals who had total IgE > 30 kU/l.

SNP selection and genotyping

Initially, 35 SNPs were chosen based on the study by Liu et al. [14]. Seven SNPs were excluded from the analysis as they had a minor allele frequency (MAF) of <5% in the population (Table 2). However, MS4A2 (Fc epsilon receptor 1 beta-chain) SNP, rs512555, was included in our analysis (MAF = 0.02) because of previous evidence for effect modification [14]. Genotype data were not available for FCER1G SNP rs2070901, which had also shown evidence for effect modification. In total, 27 SNPs were considered for the analysis.

Table 2. Association of the potentially functional single nucleotide polymorphisms (SNPs) with total immunoglobulin E (IgE) levels and specific IgE production in the 1958 British birth cohort
Genes/SNPsChromosomeMAFSNP–total IgE association (n = 4921)a SNP–specific IgE association (n = 4916)a
Coefficient (95% CI)b P-valuec Odds ratio (95% CI) P-valuec
  1. MAF, minor allele frequency.

  2. All models were adjusted for sex, month of measurement, region, and genotyping platform.

  3. a

    N for analysis varies according to the call rate of each SNP.

  4. b

    Coefficients presented as percentage change in the outcome.

  5. c

    False discovery rate-adjusted P-value in brackets, where appropriate.

  6. d

    Excluded from analysis because of MAF < 0.05.

IgE synthesis and IgE receptor complex
FCER1A (IgE fc receptor subunit alpha)
rs242783710.30−26.2 (−34.2, −18.2)<0.001 (<0.001)0.83 (0.73, 0.95)0.006 (0.08)
FCER1G (IgE fc receptor subunit gamma)
rs1158721310.173.9 (−4.7, 12.5)0.391.06 (0.92, 1.22)0.40
rs1142110.151.9 (−7.1, 10.8)0.681.00 (0.87, 1.15)0.99
IL-4 (interleukin 4)
rs224324850.06−6.8 (−18.9, 5.3)0.270.79 (0.65, 0.97)0.026 (0.14)
rs224325050.138.9 (−0.4, 18.2)0.06 (0.18)1.22 (1.06, 1.42)0.007 (0.05)
rs207087450.139.0 (−0.2, 18.3)0.06 (0.19)1.23 (1.06, 1.42)0.006 (0.05)
IL-13 (interleukin 13)
rs129568750.0612.6 (−4.7, 29.9)0.160.88 (0.73, 1.07)0.19
rs2069744d 50.00
rs2054150.1817.2 (8.7, 25.6)<0.001 (0.001)1.16 (1.02, 1.33)0.03 (0.13)
rs84850.1717.5 (9.0, 26.0)<0.001 (0.001)1.16 (1.01, 1.33)0.04 (0.14)
MS4A2 (Fc epsilon receptor 1 beta-chain)
rs512555110.022.3 (−16.9, 21.5)0.831.13 (0.84, 1.53)0.418
IL4R (interleukin 4 receptor alpha-chain)
rs2057768160.309.6 (1.6, 17.5)0.02 (0.09)0.97 (0.85, 1.10)0.59
rs3024633160.084.1 (−7.0, 15.2)0.461.08 (0.91, 1.29)0.39
rs1805010160.44−0.4 (−13.1, 12.3)0.960.83 (0.68, 1.01)0.06 (0.18)
rs1805016d 160.04
rs2074570d 160.03
rs3024682d 160.00
Vitamin D metabolic pathway
Near DHCR7 (7-dehydrocholesterol reductase)
rs12785878110.21−4.9 (−13.1, 3.3)0.230.99 (0.86, 1.12)0.83
Near CYP2R1 (cytochrome P450, family 2, subfamily R, polypeptide 1)
rs10741657110.39−10.6 (−19.0, −2.3)0.01 (0.07)0.9 (0.79, 1.03)0.12
VDR (vitamin D (1,25-dihydroxyvitamin D3) receptor)
rs2853563d 120.02
rs7954412d 120.00
rs3858733d 120.02
rs11168293120.35−3.5 (−11.6, 4.5)0.380.92 (0.81, 1.04)0.19
CYP27B1 (cytochrome P450, family 27, subfamily B, polypeptide 1)
rs8176351d 120.00
rs10877012120.322.5 (−5.7, 10.7)0.551.05 (0.92, 1.19)0.49
CYP24A1 (cytochrome P450, family 24, subfamily A, polypeptide 1)
rs6013897200.194.2 (−4.1, 12.5)0.321.08 (0.95, 1.24)0.24
rs4809957200.21−8.6 (−20.0, 2.7)0.141.16 (0.88, 1.53)0.29
rs2762934200.186.8 (−1.7, 15.2)0.111.10 (0.96, 1.26)0.17
rs2296239200.21−8.2 (−16.4, −0.0)0.05 (0.19)0.91 (0.79, 1.04)0.15
rs2296241200.453.4 (−5.6, 12.5)0.470.97 (0.84, 1.13)0.72
rs2248359200.36−11.7 (−20.2, −3.2)0.007 (0.05)0.80 (0.70, 0.92)0.002 (0.04)
GC (group-specific component – vitamin D binding protein)
rs228267940.290.5 (−8.4, 7.4)0.910.89 (0.78, 1.01)0.06
rs458840.26−0.9 (−9.3, 7.5)0.840.87 (0.76, 0.99)0.04 (0.15)
rs115556340.261.3 (−7.1, 9.8)0.760.93 (0.81, 1.07)0.31
rs373335940.064.3 (−8.4, 16.7)0.501.12 (0.92, 1.22)0.40

Of the 27 SNPs, two SNPs, rs10877012 and rs4588, were genotyped using the Taqman platform (Applied Biosystems, Warrington, UK), while others were obtained from the genome-wide platforms [18, 19]. The overall call rate for the genotyping was >97%, and the genotype frequencies of all SNPs were in Hardy–Weinberg equilibrium (P > 0.01).

Statistical analysis

Variations in total IgE were evaluated by linear regression for continuous total IgE and by logistic regression for the dichotomous outcome of specific IgE production. Natural log transformation was applied to 25(OH)D and total IgE to calculate geometric means and to improve the approximation of normal distribution. All models were adjusted for sex, month of measurement, and region (Scotland or South, Middle or North of England). For analyses including SNPs, a further adjustment was made for the genotyping platform. All carriers of minor alleles were grouped together to improve the stability of the estimates. A cut-off value of 50 nmol/L was used to define VDD [22], and individuals with 25(OH)D > 135 nmol/L (n = 33) were excluded to reflect the nonlinear relationship between 25(OH)D and total IgE [4]. Interactions between each SNP and VDD were assessed using the likelihood ratio test. A stratified analysis was performed when significant interactions were detected. A genetic risk score was constructed by combining the risk genotypes of SNPs that were found to interact with VDD (rs2443250 and rs512555), with risk genotypes defined as the groups in which VDD was associated with increasing total IgE levels. A chi-square goodness of fit test was used to assess deviation from Hardy–Weinberg equilibrium (P > 0.01 for all SNPs). All P-values were derived from two-sided tests. Multiple testing was corrected for all single SNP and exploratory interaction analyses using the false discovery rate method [(≤0.05 × 27 SNPs)/rank]. All analyses were carried out using STATA, version 12 [23].

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

The characteristics of the study participants are presented in the Table 1. Total IgE concentrations appeared higher in participants with VDD (<50 nmol/L) compared to participants with 25(OH)D levels 50–135 nmol/L [coefficient (95% CI): 8.1% (−0.7%, 16.9%), P = 0.07, while no association was seen for specific IgE [odds ratio (95% CI): 1.07 (0.93, 1.24), P = 0.33] after adjusting for sex, region and month of blood sample.

We investigated the association of the 27 SNPs with total IgE and specific IgE and found that only four SNPs [rs2427837 (FCER1A), P < 0.001; rs20541 (IL13), P = 0.001; rs848 (IL13), P = 0.001; and rs2248359 (CYP24A1), P = 0.05] remained associated with total IgE and 3 SNPs [rs2243250 (IL4), P = 0.05, rs2070874 (IL4), P = 0.05; and rs2248359 (CYP24A1), P = 0.04] with specific IgE, after correction for multiple testing (Table 2).

Of the three SNPs [(rs2243250 (IL4), rs512555 (MS4A2), and rs2762934 (CYP24A1)] that showed an interaction with VDD in an earlier study on children [14], two also modified the association between VDD and total IgE concentrations in our study (IL4, rs2243250, Pinteraction = 0.02; MS4A2, rs512555, Pinteraction = 0.04), while no significant associations were seen for specific IgE (Table 3). Vitamin D deficiency was associated with higher total IgE only in the carriers of the ‘T’ allele in MS4A2 [coefficient (95% CI): 52.6% (0.1%, 105.1%), P = 0.04] and ‘C’ allele in IL4 [coefficient (95% CI): 12.8% (2.7%, 22.8%), P = 0.01] (Table 3). In the HapMap [24], 86% of Caucasian populations carry the ‘C’ allele in IL4 (87% current study), while only 20% of the Yoruban and 26%/27% Chinese/Japanese are ‘C’ allele carriers (Table 4).

Table 3. Interaction between SNPs and vitamin D deficiency (VDD) (<50 vs 50–135 nmol/L) on total IgE and specific IgE production
Interaction between SNPs and VDD on total IgE levels using linear regression analysis (n = 4888)
GenesGenotypes

25(OH)D

nmol/L

N a Coefficient (95% CI)b P-value from LRTInteraction P-value
  1. LRT, likelihood ratio test.

  2. All models were adjusted for sex, month of measurement, region, and genotyping platform.

  3. a

    N for analysis varies according to the call rate of each SNP.

  4. b

    Coefficients presented as percentage change in the outcome.

  5. c

    Cases defined as those with at least one specific IgE (cat, grass, or dust) >0.35 kU/l.

IL4 rs2243250 = CC50–1352127Reference0.010.02
<50155012.8 (2.7, 22.8)
rs2243250 = CT/TT50–135671Reference0.68
<50503−3.9 (−22.6, 14.8)
MS4A2 rs512555 = CC50–1352697Reference0.160.04
<5019726.4 (−2.5, 15.4)
rs512555 = CT/TT50–135126Reference0.04
<509152.6 (0.1, 105.1)
CYP24A1 rs2762934 = GG50–1351875Reference0.560.24
<5013723.2 (−7.7, 14.1)
rs2762934 = GA/AA50–135921Reference0.01
<5067719.4 (4.1, 34.6)
Interaction between SNPs and VDD on specific IgE using logistic regression analysis (n = 4883)
GenesGenotypes25(OH)Dnmol/LSpecific IgE casesc/controlsa OR (95% CI) P-value from LRTInteraction P-value
IL4 rs2243250 = CC50–135529/1595Reference0.060.15
<50404/11441.17 (0.99, 1.38)
rs2243250 = CT/TT50–135209/462Reference0.51
<50141/3620.91 (0.68, 1.21)
MS4A2 rs512555 = CC50–135717/1977Reference0.450.19
<50517/14531.06 (0.92, 1.22)
rs512555 = CT/TT50–13534/92Reference0.13
<5030/611.76 (0.84, 3.66)
CYP24A1 rs2762934 = GG50–135497/1376Reference0.990.19
<50347/10241.00 (0.83, 1.19)
rs2762934 = GA/AA50–135252/668Reference0.08
<50198/4781.25 (0.98, 1.66)
Table 4. Allele frequencies of the interleukin-4 ( IL4 ) SNP rs2243250 in HapMap populations (HapMap Data Rel 28/phase II + III August 10, on NCBI B36 assembly, dbSNP b126)
HapMap population‘C’ allele frequency of the IL4 SNP rs2243250 (%)
Gujarathi Indians, Houston, Texas89
Tuscans, Italy87
European ancestry, Utah86
Mexican ancestry, Los Angeles, California60
Maasai, Kinyawa, Kenya46
African ancestry, Southwest USA44
Japanese, Tokyo, Japan27
Han Chinese, Beijing, China26
Chinese, Metropolitan Denver, Colorado25
Yoruban in Ibadan, Nigeria20
Luhya in Webuye, Kenya18

There was a strong interaction between the genetic risk score (comprising IL4 and MS4A2) and VDD on total IgE levels (P interaction = 0.009), such that the elevated total IgE concentrations associated with VDD were most pronounced in the carriers of the two risk alleles (‘T’ allele in MS4A2 and ‘C’ allele in IL4) (Fig. 1). Further exploratory analyses provided no evidence for interaction between the remaining 24 SNPs and VDD on total IgE (P > 0.16 for all comparisons).

image

Figure 1. Interactive effect between the combined risk genotypes of the genes IL4 (rs2243250) and MS4A2 (rs512555) and vitamin D deficiency in the 1958 British birth cohort. Risk genotypes refer to the groups in which vitamin D status was associated with increasing total immunoglobulin E levels.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

In this study, we have examined the association of 27 potentially functional SNPs [14] with total IgE and specific IgE in the 1958BC and found that after accounting for multiple testing, four SNPs were significantly associated with total IgE levels and three SNPs with specific IgE. Interestingly, one of the genes implicated with both outcomes (CYP24A1) has a primary effect on vitamin D metabolism [25] rather than on immunological factors more closely related to allergy risk. Our study also confirmed that modification by other immune-related genetic factors is likely to affect the influence of vitamin D status on allergy risk. A previous study had shown that VDD increases the risk of food sensitization among children, depending on their IL4 and MS4A2 genotypes [14]. Our findings for total IgE suggest that such effect modification is implicative of a gene–environment interaction that is also relevant to adults.

It is important to note that while our study was conducted in an adult British white population, the study by Liu et al. [14] was performed in children comprising approximately 57% Blacks, 22% Hispanics, 6% Whites, and 15% others. Differences in the ethnic mix between the studies are likely to explain the difference in the proportion carrying the ‘C’ allele of the IL4 SNP rs2243250 between the two studies [87% in the 1958BC vs 37% in the Liu et al. [14]]. This difference in allele distribution is in accordance with large differences in allele frequencies between populations, as reported in the recent HapMap [24] (Table 4), which has been shown to be a reliable tool in making inferences based on the comparison of allele frequencies and genetic association data across populations [26]. Despite these differences in the frequency of risk allele carriers, the association between VDD and increased allergy risk only for ‘C’ allele carriers appears to be consistent across diverse populations [14]. However, as most individuals in white European groups, or indeed individuals of Indian origin, carry the risk allele, VDD is likely to be a more important determinant for allergy risk for them than for individuals from some other ethnic groups.

Similar large variation in allele frequencies was not seen for the minor ‘T’ allele of the MS4A2 SNP, rs512555, which was relatively rare across populations (<20% across all) [24]. However, in our study, VDD was associated with higher total IgE among the carriers of MS4A2 ‘T’ allele, while in the earlier study, VDD had a protective effect [14]. This could suggest possible variation also in the alleles conferring the risk between populations or perhaps variation arising from differences in the etiology of allergy-related outcomes. A recent study in the Boston Birth Cohort (n = 1104) has also highlighted the importance of ethnic differences in relation to the genetic risk of developing food sensitization, which could possibly be attributed to the vitamin D status [27].

Our findings also confirm the results from the recent genome-wide association study for plasma total IgE levels [28], with FCER1A (IgE fc receptor subunit alpha) and IL13 (interleukin 13) SNPs having significant associations after accounting for multiple testing. Our findings on CYP24A1, a known vitamin D–related candidate gene affecting the clearance of 25(OH)D and other vitamin D metabolites [29], are extremely intriguing and could provide further evidence for a causal role of vitamin D status in determining total IgE concentrations [25]. IL4 is a well-known candidate gene for total IgE [30, 31], and the locus for asthma and atopy has been linked to chromosome 5q31, the region where IL4 gene has been mapped [32]. Previous linkage studies have shown that gene predisposing to atopy is localized on chromosome 11q13 [33, 34], the region where MS4A2 is located. Hence, our finding in the 1958BC further confirms the role of IL4 and MS4A2 in contributing to the risk of allergy, in part by altering the risk associated with VDD. Unfortunately, genotype data were not available for the FCER1G SNP rs2070901 from our study; hence, we were unable to investigate the related interaction also identified by Liu et al. [14].

The lack and difficulty of replication is a major drawback to obtaining evidence on gene–environment interaction affecting allergy risk but also more generally [35]. One explanation is in the large sample size requirements, and even with our study which is the largest to date, we did not have sufficient power to detect interactions in relation to specific IgE as an outcome. Furthermore, previous evidence for genetic interaction by VDD arose from a study investigating determinants for food sensitization, and although it is possible that VDD could affect both food sensitization and systemic IgE levels through a shared pathway related to immune activation or modulation [5, 15], also differences in the etiology could explain the lack of association with specific IgE to inhalant allergens as well as the observed opposing effects of VDD in the carriers of ‘T’ allele of the MS4A2 SNP rs512555 [14].

In summary, our findings confirm the association of FCER1A and IL13 SNPs with total IgE levels and the previously reported interactions of the IL4 and MS4A2 SNPs with VDD on allergy risk. Together with a previous study [14] and known variation in allele frequencies across populations [24], these data highlight the need to consider possible ethnic differences in the allergy-related responsiveness to VDD.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

We are deeply grateful to all the participants in the 1958BC. We thank Professor Christine Power (UCL Institute of Child Health, London, UK), co-PI of the biomedical survey of the 1958 British Cohort, for her support. We acknowledge Professor Ian Gibb, Dr Steve Turner, and Marie-Claude Fawcett (Royal Victoria Infirmary, Newcastle-upon-Tyne, U.K.) for carrying out the laboratory assays and the Centre for Longitudinal studies, Institute of Education (original data producers), for providing the data. The Medical Research Council funded the 2002–2004 clinical follow-up of the 1958 birth cohort (grant G0000934). This work was undertaken at the Centre for Paediatric Epidemiology and Biostatistics, which benefits from funding support from the MRC in its capacity as the MRC Centre of Epidemiology for Child Health. Research at the University College London Institute of Child Health and Great Ormond Street Hospital for Children NHS Trust benefits from R&D funding received from the NHS Executive.

Author contributions

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

KSV wrote the first draft and contributed to the study hypothesis, literature search, and data interpretation. AC carried out all analyses and contributed to drafting and interpretation. EH had the study idea and formed the hypothesis, obtained funding, supervised the study and contributed to the interpretation, drafting and critical revision.

Funding

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References

Funding for the project was provided by the British Heart Foundation (grant PG/09/023) and the UK Medical Research Council (grant G0601653 and SALVE/PREVMEDSYN with Academy of Finland).

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgments
  7. Author contributions
  8. Funding
  9. Conflict of interest
  10. References