Cytokine polymorphisms in the Th1/Th2 pathway and susceptibility to non-Hodgkin lymphoma

Cytokines play a crucial role in the regulation of key pathways of immunity, the balance between cell-mediated (Th1) and humoral (Th2) responsiveness. Th1 cells drive cellular immunity to fight intracellular pathogens including viruses and to remove cancerous cells, whereas Th-2 cells control humoral immunity by up-regulating antibody production to protect against extracellular pathogens.^1,2  Select cytokines regulate the subpopulation of T-cell lymphocytes responsible for this balance. T helper 1 (Th1) lymphocyte cells produce IL-2 and interferon-gamma and promote cell-mediated immune responses. T helper 2 (Th2) lymphocyte cells produce IL-4, IL-5, IL-6, IL-10, and IL-13, which favor B-cell activation and immunoglobulin production. These cytokines can modulate lymphoid development and immune function.^3-10  Since immune dysfunction is thought to be at the underlying basis of lymphomagenesis, an imbalance in the regulation and expression of Th1 and Th2 cytokines, which are the fundamental messengers of adaptive immunity, could play an important role in the etiology of non-Hodgkin lymphoma (NHL) and its major subtypes.^11,12 

Genetic variation is common across the human genome. It is estimated that there are more than 7 million single nucleotide polymorphisms (SNPs) with a minor allele frequency of 5% to 10%. Although most SNPs are not functionally important, there is a subset of variants that alter the expression or function of a gene product.¹³  These functional variants may alter disease risk, affect the observed phenotype, contribute to the pathogenesis of the disease, or alter the response to treatment.^3,6,10,14-19  Common genetic variants in candidate genes, or for that matter, pathways of genes, can be applied to genetic association studies in search of genetic risk factors for disease susceptibility. Since the expression of Th1 and Th2 cytokines can be altered by germ-line genetic variants,⁵  we studied the association between common polymorphisms in Th1 and Th2 cytokine genes and the risk of NHL.

Here, we report on the results of 39 SNPs drawn from 20 distinct Th1 and Th2 genes in a population-based case-control study among women in Connecticut. The SNPs analyzed in this study were chosen on the basis of prior functional data and previous association studies^15-18,20-28  or to help characterize the haplotype structure of the gene of interest.

The study population has previously been described.^29-31  Briefly, between 1996 and 2000, all histologically confirmed, incident cases of NHL (ICD-O, M-9590-9642, 9690-9701, 9740-9750) from New Haven, CT, were identified through the Yale Cancer Center's Rapid Case Ascertainment Shared Resource (RCA). Enrollment criteria included age 21 to 84 years, residence in Connecticut, female, alive at the time of interview, and without a previous diagnosis of cancer except for nonmelanoma skin cancer. Of 832 eligible cases, 601 (72%) completed in-person interviews. Pathology slides (or tissue blocks) from all patients were obtained from the original pathology departments and specimens were classified using the Revised European-American Lymphoma (REAL) system by central review.

Female population-based controls from Connecticut were recruited by: (1) random-digit dialing methods for those younger than 65 years of age; or (2) random selection from Health Care Financing Administration files for those aged 65 years or older. The participation rate was 69% for those contacted by random-digit dialing and 47% for those contacted through health care records. Cases and controls were frequency matched on age (± 5 years) by adjusting the number of controls randomly selected in each age stratum once every several months during the period of recruitment.

The study was approved by the institutional review boards at Yale University, the Connecticut Department of Public Health, and the National Cancer Institute. Participation was voluntary, and written informed consent was obtained from all participants. Those who signed consent were interviewed by trained study nurses either at the subject's home or at a convenient location. Subjects were administered a questionnaire requesting information on demographic characteristics, family history of cancer, past medical condition and medication use, diet, occupation, smoking, and drinking.

Following the interview, subjects provided a 10-mL peripheral-blood sample. Subjects for whom the blood draw was contraindicated or who refused to participate in the blood-draw were offered the option to provide buccal cell cotton swab samples instead. In total, 76.7% (461/601) of interviewed cases and 74.6% (535/717) of interviewed controls provided a blood sample, and 11.0% (57) of cases and 10.4% (62) of controls provided buccal cell samples.

Genotyping was performed at the National Cancer Institute Core Genotyping Facility (http://cgf.nci.nih.gov). All TaqMan assays (Applied Biosystems, Foster City, CA) for this study were optimized on the ABI 7900HT detection system with 100% concordance with sequence analysis of 102 individuals as listed on the SNP500Cancer website (http://snp500cancer.nci.nih.gov).³²  We selected 39 SNPs in 20 Th1/Th2 immune genes based on the following criteria: minor allele frequencies more than 5%, laboratory evidence of function, or prior association with human disease studies (Table 1). Due to a limited amount of DNA available for subjects who provided only buccal cells, we first genotyped subjects who provided a blood sample. If there was suggestive evidence, or if we had a relatively high prior that a given SNP was associated with risk of NHL, genotype analysis proceeded to include subjects who provided buccal cell samples. Among the 39 SNPs, 15 were genotyped only in subjects who provided a blood sample (Table 1). We observed a random drop-out for amplification and for samples that amplified yet could not be determined due to ambiguous results. In addition, the total number of cases and controls for which genotyping data were available varied because there was a limited amount of DNA available. Data on 8 of the 39 SNPs for white study subjects (noted in Table 1) were contributed to a pooling effort by the InterLymph case-control consortium of lymphoma studies.³³ 

Duplicate samples from 100 study subjects and 40 replicate samples from each of 2 blood donors were interspersed throughout the plates used for genotype analysis. The concordance rates for quality control (QC) samples were between 99% and 100% for all assays. The genotype frequencies for 4 SNPs (IFNGR2 [Ex7-128C>T], CTLA4 [Ex1-61A>G], IL4 [-588C>T and Ex1-168C>T]) were not consistent with Hardy-Weinberg equilibrium (HWE) among non-Hispanic white controls using a chi-square test (P < .05; Table 1). It is notable that the statistical analysis of deviation for the P values for these SNPs fell between .05 and .01. Quality control data were rechecked and the accuracy of each assay not in HWE among controls was confirmed. Evaluation of all SNPs analyzed to date in the study showed that approximately 5% were not consistent with HWE, as expected.

Unconditional logistic regression was used to estimate the odds ratios (ORs) and 95% confidence intervals (CIs), adjusting for age (< 50 years, 50-70 years, > 70 years), race (white, African-American, other). Analyses adjusted for family history resulted in similar estimates. Analyses limited to non-Hispanic whites (representing 93.2% and 91.6% of all cases and controls, respectively) are shown in Supplemental Tables S1 and S2 (available at the Blood website; click on the Supplemental Tables link at the top of the online article) and were comparable to analyses that included all study subjects (Tables 3,4). In this report, we provide results of individual SNP analyses from models that include all study subjects adjusting for age and sex.

The most prevalent homozygous genotype was used as the reference group. Tests for trend were conducted by assigning the ordinal values 1, 2, and 3 to the most prevalent genotypes in rank order of wild type, heterozygous, and variant homozygous genotypes, respectively. Risks for NHL subtypes were carried out using all controls as the comparison group, to maximize statistical power. NHL subtype comparisons were conducted using unconditional logistic regression by treating one subtype as a “case” and the other one as a “control” in the model.

Since we conducted multiple tests within this data set and there was a chance that some of our results could be false-positive findings, we used the Benjamini-Hochberg method to control for the false discovery rate (FDR).³⁴  The FDR is defined as the expected ratio of erroneous rejections of the null hypothesis to the total number of rejected hypotheses. We applied the FDR method to the P values of the risk for homozygous carriers of the rare versus common allele, as this provides the greatest potential contrast in effects across genotypes. All P values presented are 2-sided and all analyses were carried by the Statistical Analysis Software, version 8.02 (SAS Institute, Cary, NC).

Haplotype analyses were conducted within non-Hispanic whites for all genes in which more than one SNP was genotyped. Haplotype block structure was evaluated with the program HaploView (Whitehead Institute, Cambridge, MA) using the 4-gamete rule with a minimum frequency of 0.005 for the fourth gamete. Haplotypes were estimated using the estimation-maximization algorithm,³⁵  and overall differences in haplotype frequencies between non-Hispanic white cases and controls were assessed using the global score test implemented in HaploStats³⁶  (R Version 1.2.0), adjusting for age. A logistic regression model was used to estimate the effect of individual haplotypes assuming an additive model by using posterior probabilities of the haplotypes as weights to update the regression coefficients in an iterative manner.³⁶ 

NHL cases were comparable to controls with regard to age and ethnicity. However, cases were more likely to have a positive family history of cancer (Table 2). Analyses adjusted for family history resulted in similar estimates (data not shown). SNPs that were significantly associated with risk of all NHL, B-cell, or T-cell lymphoma are shown in Table 3. In addition, we present results for all IL10 promoter SNPs, as these are used for subsequent haplotype analyses.

SNPs in each of 3 Th2-related cytokines, IL4, IL5, and IL10, were significantly associated with an increased risk for NHL overall (Table 3). Significant trends were observed for SNPs in IL10 (-1082A>G; -3575T>A), IL4 (-1098T>G), and borderline effects for IL5 (-745C>T). There was no evidence of gene-gene interactions between IL10, IL4, and IL5, although the power was too low to detect such effects. Subsequent analyses by NHL subtype showed that variants in IL10 and IL5 were significantly associated with an increased risk of B-cell lymphoma, and variants in IL4, IL4R, and IL6 were significantly associated with an altered risk of T-cell lymphoma (Table 3). The risk estimates for IL4 (-1098T>G) and IL6 (-598G>A) SNPs for T-cell versus B-cell lymphoma differed significantly (P = .003 and P = .016 for test of heterozygote/homozygote carriers of the rare allele versus subjects homozygous for the common allele, respectively).

The FDR method³⁴  was used to adjust the P values from the 39 tests (ie, total number of SNPs studied) evaluating the association between each SNP and the risk of NHL and from the 78 tests of association for risk of B- and T-cell lymphoma. After accounting for multiple comparisons, the IL10 (-3575T>A) SNP remained significantly associated with all NHL (ie, original P value of .00098 was adjusted to .034) and B-cell lymphoma (ie, original P value of .00067 was adjusted to .032). In addition, the IL4 (-1098T>G) SNP remained significantly associated with risk of T-cell lymphoma (ie, original P value of .00054 was adjusted to .023). FDR-adjusted P values for all other associations shown in Table 3 were above .10.

In an exploratory analysis, we further examined the risk of the major B-cell histologic subtypes for SNPs that showed significant effects for all B-cell lymphoma. The IL10 (-3575T>A) SNP was significantly associated with increased risk for diffuse large B-cell lymphoma (DLBCL) and follicular lymphoma (Table 4). In contrast, the IL5 variant was associated with increased risk for DLBCL only.

When possible, we estimated haplotypes and analyzed for risk of NHL. For instance, a haplotype analysis of the 4 SNPs located in the IL10 proximal and distal promoter regions was conducted. The IL10 haplotype (-3575T>A, -1082A>G, -819C>T, and -592C>A) A-G-C-C, containing the minor alleles -3575A and -1082G (both previously associated with disease risk) and the common alleles -819C and -592C, was associated with an increased risk for NHL, particularly B-cell lymphoma (OR = 1.54, 95% CI = 1.21-1.96; Table 5). Since an alternative haplotype T-G-C-C bearing only the -1082G variant was not associated with increased risk for B-cell lymphoma, we interpret the results to suggest that the -3575A variant was important in driving the increased risk observed for B-cell lymphoma. Interestingly, the T-A-T-A haplotype, containing the common alleles -3575T and -1082A and the rare variants -819T and -592A in the proximal promoter, was also associated with increased risk for B-cell lymphoma (OR = 1.37, 95% CI = 1.05-1.79). Single analysis of the SNPs IL10 -819T and -592A did not demonstrate an association (Table S1), which suggests that another unobserved variant lying on the same haplotype structure or in linkage disequilibrium with the T-A-T-A haplotype could be driving the observed association between the T-A-T-A haplotype and NHL risk. The inclusion of SNPs outside the promoter region did not yield additional insight into the association between IL10 and NHL risk (data not shown). Similarly, in a haplotype analysis of the 5 SNPs studied in IL4, we observed a comparable risk to that observed for IL4 (-1098T>G) SNP (Table 3).

Although additional SNPs were significantly associated with one or more NHL subtypes (ie, IL7R Ex4 + 33G>A and JAK3 Ex23 + 291A>G and T-cell lymphoma; IFNGR1 IVS6-4A>G and follicular lymphoma; Tables S3 and S4), the number of subjects with the homozygous genotype was relatively small. Further, none of these SNPs was associated with risk for these subtypes at P values less than .05 after correction for multiple comparisons.

We report one of the first analyses of the association between SNPs chosen from key immunologic cytokine genes and NHL. This study represents a pathway-based approach toward investigating common genetic variants in the Th1/Th2 cytokine network in a population-based case-control study of non-Hodgkin lymphoma in women in Connecticut. In total, we analyzed 17 SNPs drawn from 11 Th1 genes, 21 SNPs from 8 Th2 genes, and 1 SNP from 1 Th1/Th2 gene. Overall, we observed that common genetic variants in Th2 cytokine genes were associated with risk for NHL. SNPs in the Th2 genes IL4, IL5, and IL10 were significantly associated with increased risk of NHL overall. However, there was evidence that the effects were specific to lymphoma subtypes. In particular, SNPs in IL10 and IL5 were associated with B-cell lymphoma, whereas SNPs in IL4, IL4R, and IL6 were associated with T-cell lymphoma. A specific haplotype spanning the promoter region of IL10 was associated with an increased risk for B-cell lymphoma, an effect not apparent in single SNP analyses. This observation suggests that the `causal' SNP(s) probably are in linkage disequilibrium with SNPs chosen for this study. Follow-up analysis will need to analyze additional SNPs across the ancestral haplotypes of the IL10 and IL4 genes.

A pooled analysis of the international InterLymph Consortium of case-control studies of NHL, which included a subset of data on whites from the current study (Table 1), presented results for DLBCL and follicular lymphoma and determined that 2 promoter SNPs in IL10, namely, the IL10 -3575T>A and -1082A>G SNPs, were associated with risk of NHL, particularly DLBCL.³³  Even when the robust results from our study were excluded from the pooled analysis, the effect remained significant for DLBCL. Here, we have extended the InterLymph analysis of IL10 variants by evaluating 2 additional promoter SNPs, -819C>T and -592C>A, and report for the first time that a haplotype containing both variant alleles was associated with B-cell lymphoma, and both for DLBCL and follicular lymphoma (Table 5). The results from our study and others^17,33  provide compelling evidence that genetic variation in the IL10 promoter region plays an important role in the etiology of NHL and deserves further investigation. There is substantial evidence in support of a role for IL-10 in lymphomagenesis. It is a critical mediator of the Th1/Th2 balance, apoptosis potential, and regulation of inflammation.⁶  A IL10 knock-out mouse model showed that IL-10 is critical for B-cell lymphomagenesis.¹⁶  IL-10 serum levels have been shown to be prognostic factors for NHL, particularly the DLBCL subtype.^17,37  Elevated levels of IL-10 in the vitreous have been correlated with primary intraocular lymphoma.³⁸  In vitro laboratory work has explored the functional consequences of IL10 variants, such as the IL10-3575A allele, which appears to be associated with decreased levels of IL-10.¹⁸  Lower IL-10 levels could result in higher TNFα expression, shifting the balance toward Th1 cellular immunity.³⁹  Indeed, we also found an increased risk for the TNF-308A allele for DLBCL (Table S4) that was consistent with the pooled risk estimate in the InterLymph study.³³  However, the association was not significant, probably because of limited power.

Previous work has shown that proximal promoter IL10 haplotypes alter IL-10 secretion as well as expression of the gene.^26,27  In our haplotype analysis, we determined that SNPs either in the promoter or in linkage disequilibrium (LD) with those tested in our study, were associated with risk for NHL. It is notable that the same IL10 promoter haplotypes have been associated with a number of immune-mediated diseases, including progression of HIV infection and autoimmune diseases such as lupus erythematosus, graft-versus-host disease following marrow transplantation, and asthma, providing supportive evidence that common genetic variation in the IL10 genes could influence disease susceptibility further.^40-44 

The other noteworthy finding in our study was that variation in the IL4 gene is associated with risk of all NHL, particularly T-cell lymphoma. Although the number of T-cell lymphoma cases in our study was small, the effect was striking and merits follow-up in one of the large international consortia. These results are intriguing because IL4 plays a key role in the proliferation of T cells.^2,7,45-47  The IL4 promoter SNPs and haplotypes have been associated with juvenile idiopathic arthritis, severity of infection with respiratory syncytial virus in young children, asthma, fungal infection with Candida albicans in patients with leukemia, atopy, and inflammatory bowel disease.^48-53  Further, this same IL4 promoter haplotype has been shown to alter gene expression in vivo and in vitro.^25,54  The observation that an SNP in the IL4R gene was associated with increased risk of T-cell lymphoma also underscores the importance of regulating the Th1/Th2 balance. Previous work reported that genetic variants in IL4R can affect signal transduction and the level of gene expression of IL4R.^21,55 

In summary, we report that common genetic variants in 2 key genes of the Th2 pathway, namely, IL10 and IL4, could be associated with the risk of NHL and possibly one or more of the major subtypes. It is notable that Th2 cytokines have pleiotropic functions, including modulating the inflammatory process and response to viral infection and other agents, as well as function as autocrine growth factors.^56-58  Our results raise the possibility that a shift in the balance of the Th1/Th2 response caused by genetic variants in key cytokine genes could have important consequences for the pathogenesis of NHL. More extensive genomic analysis of the genes evaluated here, as well as additional genes in the Th2 pathway, is warranted. These findings, although intriguing, require replication in larger studies and ultimately in pooled analyses.

We gratefully acknowledge the assistance of Peter Hui (Information Management Services, Silver Spring, MD) for programming support.