Interactions of IL-12A and IL-12B Polymorphisms on the Risk of Cervical Cancer in Chinese Women

Purpose: Accumulative evidence suggests that interleukin-12 (IL-12) plays a central role in the Th1 responses and thus participates in the carcinogenesis of human papillomavirus–related cervical cancer. We hypothesized that potentially functional polymorphisms in IL12A and IL12B may individually and jointly contribute to cervical cancer risk.

Experimental Design: We genotyped IL12A rs568408 [3′ untranslated region (UTR) G>A] and rs2243115 (5′UTR T>G) and IL12B rs3212227 (3′UTR A>C) in a hospital-based study of 404 cervical cancer cases and 404 cancer-free controls.

Results: The IL12A rs568408 GA/AA and IL12B rs3212227 AC/CC variant genotypes were associated with a significantly increased risk of cervical cancer [adjusted odds ratio, 1.43; 95% confidence interval (CI), 1.06-1.93; and adjusted odds ratio, 1.30; 95% CI, 0.97-1.75, respectively], compared with their corresponding wild-type homozygotes. Moreover, a significant gene-gene interaction of these 2 loci were evident in the risk of cervical cancer, and subjects carrying variant genotypes of both loci had a 1.82-fold (95% CI, 1.28-2.57) increased risk of cervical cancer. In the stratified analyses, the combined genetic effect was more pronounced in patients who had early-stage tumors or more parities. Subjects carrying rs568408 AG/AA and rs3212227 AC/CC genotypes and having >2 parities showed a 6.00-fold (95% CI, 2.86-12.56) elevated cervical cancer risk (P for multiplicative interaction = 0.046).

Conclusion: These findings suggest that IL12A rs568408 and IL12B rs3212227 may individually and jointly contribute to the risk of cervical cancer and may modify cervical cancer risk associated with parity, but these data need further validation.

Cervical cancer is the second most common cancer among women worldwide, with an estimated 493,000 new cases and 274,000 deaths in the year 2002, and 83% of the cases had occurred in developing countries (1). Although there have been substantial declines in both incidence and mortality rates in the past 30 years, >100,000 cervical cancer cases are diagnosed in China every year (2). Epidemiologic and laboratory-based studies have identified the infection with one of 15 high-risk, or oncogenic, HPV types as a necessary but not sufficient cause of cervical cancer (3–5). The prevalence of genital HPV infection in developing countries is very high in young women, but most of the infected regress without intervention (6). Given the mounting evidence that long-term HPV infection is a prerequisite for cervical carcinogenesis (7, 8), genetic variations that influence the host primary immune response may determine the outcome of high-risk HPV infection.

Recently, cell-mediated immunity has been thought as one of the important control mechanisms in the HPV-associated carcinogenesis. There is evidence that the cell-mediated immunity response of the host, both systemically and locally, are important determinants for the course of the infection (9). The cell-mediated immunity response to the NH₂-terminal E6 and E7 peptides has been shown to be associated with regression of cervical intraepithelial neoplasia and loss of associated HPV infection (10). Persistently infected individuals, usually with focally high levels of HPV DNA replication, have an increased probability of progressing to a high-grade cervical intraepithelial neoplasia and an invasive carcinoma (6, 7, 11). Th type 1 (Th1) cells, which promote cell-mediated immunity responses to intracellular pathogens by cytokine production, are necessary for the clearance of HPV (12).

Interleukin-12 (IL-12) is a heterodimeric proinflammatory cytokine formed by a 35,000 dalton light chain (known as p35 encoded by IL12A) and a 40,000 dalton heavy chain (known as p40 encoded by IL12B), which induces the production of IFN-γ, favors the differentiation of Th1 cells and forms a link between innate resistance and adaptive immunity (13). Studies showed that IL-12 had a central role in Th1 responses (14–16), and it is required for optimal Th1 cell development during the immune response to pathogens (17, 18). Besides the activity of antivirus, IL-12 is also important for the host resistance to tumors. The antitumor activity of IL-12 has been extensively reported in mouse models of cancer, where it has been shown to inhibit tumorigenesis and induce regression of established tumors (19–21). The major antitumor activities of IL12 rely on its ability to promote Th1 adaptive immunity and CTL responses (13).

The two subunits of IL12, p35 and p40 encoded by IL12A and IL12B, respectively, are located on separate chromosomes (3p12-q13.2 and 5q31-33). Because of the functional relevance of IL12, several molecular epidemiologic studies have been done to investigate the association between the IL12A and IL12B polymorphisms and risk of cancers, including non–Hodgkin lymphoma (22–24), hepatocellular carcinoma (25), lung cancer (26), gastric cancer (27, 28), and others (29–32). In the present study, we hypothesized that IL12A and IL12B polymorphisms were associated with cervical cancer risk. To test this hypothesis, we did a genotyping analysis for rs2243115 (5′UTR T>G) and rs568408 (3′UTR G>A) in IL12A and rs3212227 (3′UTR A>C) in IL12B in a hospital-based study of 404 cervical cancer cases and 404 age frequency–matched controls in Chinese women.

Study population. This hospital-based case-control study was approved by the institutional review board of Nanjing Medical University. Four hundred and four newly diagnosed, histologically confirmed cervical cancer patients were consecutively recruited between March 2006 and April 2007 from the First Affiliated Hospital of Nanjing Medical University and the Nantong Tumor Hospital, Jiangsu, China with a response rate of 93.1% (404 of 434). A total of 404 controls were randomly selected from a pool of >30,000 individuals (overall response rate, 86.3%) who participated in a community-based screening program for noninfectious diseases conducted in Jiangsu Province during the same time period as the cases were recruited. These control subjects had no self-reported cancer history and were frequency-matched to the cases on age (±5 y) and residential areas (urban and rural). The cases and control subjects were all genetically unrelated Han Chinese women. After informed consent was obtained, each subject was personally interviewed to provide information on demographic data, tobacco smoking, menstrual and reproductive history, and family history of cancer (any reported cancer in first-degree relatives). After interview, a 5-mL venous blood sample was collected from each subject and used for genotyping assays.

Single nucleotide polymorphism selection and genotyping. Because no common (minor allele frequency, >0.05) nonsynonymous single nucleotide polymorphism (SNP) was found in the dbSNP database (build 129) for both IL12A and IL12B, we searched all the common potentially functional SNPs reported for Asians, which were located at 5′-flaking regions, 5′-, and 3′-untranslated regions (UTR). As a result, we selected three common SNPs: rs2243115 (5′UTR T>G) and rs568408 (3′UTR G>A) in IL12A and rs3212227 (3′UTR A>C) in IL12B.

Genomic DNA was extracted from a leukocyte pellet by traditional proteinase K digestion and was followed by phenol-chloroform extraction and ethanol precipitation. The PCR-RFLP assay was used to detect all the three SNPs. The primers of rs3212227 were 5′-GATATCTTTGCTGTATTTGTATAGTT-3′ (forward) and 5′-AATATTTAAATAGCATGAAGGC-3′ (reverse), which generated a 118-bp fragment. The fragment was then digested by TaqI (New England BioLabs) and separated on a 3% agarose gel. The variant allele rs3212227C produced 2 fragments of 92 and 26 bp, and the wild-type allele rs3212227A resulted in a single 118-bp fragment. For rs2243115 and rs568408, the forward primers were introduced a mismatched A to replace C and a mismatched T to replace C, respectively, at −3 bp from the polymorphic sites to create BsenI and NdeI (New England BioLabs) restriction sites. The primers were 5′-AGAAAAGACCTGTGAACAAAACGACT-3′ (forward) and 5′-AGATGGCTCACTAGATGCCAGG-3′ (reverse) for rs2243115, and 5′-GAAGGATGGGACYATTACATCCATAT-3′ (forward) and 5′-CAGGATGGATATTTTCCCTTCT-3′ (reverse) for rs568408. The variant allele rs2243115G produced two fragments of 93 and 29 bp and the wild-type allele rs2243115T generated 1 fragment of 122 bp. Similarly, the wild-type allele rs568408G produced 2 fragments of 98 and 23 bp and the variant allele rs568408A resulted in one fragment of 121 bp.

All the genotyping assays was done without knowing the subjects' case and control status, and approximately equal numbers of cases and controls were assayed in each 96-well PCR plate with a positive control of a DNA sample with the known heterozygous genotype. Ten percent of the samples (48 cases and 48 controls) were randomly selected to perform the repeated assays, and the results were 100% concordant.

Statistical analyses. Differences in the distributions of demographic characteristics, selected variables, and genotypes of the IL12A and IL12B variants between the cases and controls were evaluated using the χ² test or the Student's t test. The associations between IL12A and IL12B genotypes and risk of cervical cancer were estimated by computing the odds ratios (OR) and their 95% confidence intervals (95% CI) using logistic regression analyses. The Hardy-Weinberg equilibrium was tested by a goodness-of-fit χ² test to compare the observed genotype frequencies to the expected ones among the control subjects. A more-than-multiplicative gene-environment or gene-gene interaction was evaluated by logistic regression analysis. When the test for multiplicative interaction was not rejected, further test for an additive interaction was done by a bootstrapping test of goodness of fit of the null hypothesis for no departure from an additive model versus an alternative hypothesis for a departure from an additive model by using Stata software (version 8.2; StataCorp LP). All the other statistical analyses were done with SAS 9.1.3 (SAS Institute).

Selected characteristics of the 404 cervical cancer cases and 404 cancer-free controls are summarized in Table 1. Of the 404 cervical cancer cases, 363 (89.8%) were squamous cell carcinoma, 32 (8.0%) adenocarcinoma, 4 (1.0%) adenosquamous carcinoma, and 5 (1.2%) were undifferentiated carcinomas or others. Only 2 (0.5%) patients had cervical intraepithelial neoplasia 3, 114 (28.2%) had stage I carcinoma, 208 (51.5%) had stage II, 52 (12.9%) had stage III, 2 (0.5%) had stage IV, and 26 (6.4%) had unknown stage information. Compared with the control subjects, the cervical cancer cases had a significantly lower age at first live birth (P < 0.001), higher parity (P = 0.010), and higher family history of any cancers (P = 0.003). Only 7 cases (1.7%) reported to have family history of cervical cancer.

The genotype distributions of IL12A rs568408, rs2243115, and IL12B rs3212227 in the cases and controls are shown in Table 2. The observed genotype frequencies for these three polymorphisms in the controls were all in Hardy-Weinberg equilibrium (P = 0.286, 0.984, and 0.357 for rs568408, rs2243115, and rs3212227, respectively). The logistic regression analysis revealed that the variant rs568408 GA/AA genotypes were associated with a significantly increased risk of cervical cancer in a dominant genetic model (adjusted OR, 1.43; 95% CI, 1.06-1.93), compared with the wild-type rs568408 GG. Similarly, variant rs3212227AC/CC genotypes were associated with a 30% elevated risk of cervical cancer with borderline significance (adjusted OR, 1.30; 95% CI, 0.97-1.75), compared with the wild-type rs3212227 AA (Table 2). However, no evidence of the association was observed between the IL12A rs2243115 polymorphism and cervical cancer risk (Table 2).

We then examined the combined effects of IL12A rs568408 (3′UTR G>A) and IL12B rs3212227 (3′UTR A>C) variants on cervical cancer risk. As shown in Table 2, 27.0% of the cases and 17.1% of the controls had variant genotypes at both loci (rs568408 GA/AA and rs3212227 AC/CC), and the carriers of these loci had a 1.85-fold increased risk of cervical cancer (95% CI, 1.21-2.84), compared with those having both wild-type genotypes (rs568408 GG and rs3212227 AA). Furthermore, significant more-than-multiplicative (0.048) and more-than-additive (0.014) gene-gene interactions of these two loci were found in relation to risk of cervical cancer. When we dichotomized the combined genotypes of rs568408 and rs3212227, we found that subjects carrying variant genotypes at both loci had a 1.82-fold (95% CI, 1.28-2.57; P = 0.0008) increased cervical cancer risk, compared with those carrying other combined genotypes (Table 2). Although none of the results for 3 SNPs attained significance when considering multiple testing, the combination effects of dichotomized genotype groups achieved the significant level after 13 times Bonferroni corrections (0.05/13 = 0.0038).

Further stratified analyses showed that the elevated joint effects of rs568408 and rs3212227 were more evident in younger subjects (<55 years; adjusted OR, 1.96; 95% CI, 1.22-3.16), premenopausal women (adjusted OR, 2.22; 95% CI, 1.24-3.97), women with an early age at first live birth (<25 years; adjusted OR, 2.33; 95% CI, 1.42-3.81), women with more parities (adjusted OR, 2.05; 95% CI, 1.01-4.15 for 2 parities; and adjusted OR, 3.01; 95% CI, 1.50-6.07 for >2 parities), those without family cancer history (adjusted OR, 1.92; 95% CI, 1.29-2.85), patients with histology types other than squamous cell carcinoma (adjusted OR, 2.13; 95% CI, 1.02-4.46) and patients with early-stage tumors (Table 3). However, because of the limited study sample size, all the results from the stratified analyses were preliminary.

In cases, rs568408 AG/AA and rs3212227 AC/CC carriers were more likely to be early-stage patients compared with those carrying other combined genotypes, indicating that these two SNPs may be implicated in risk but not in progression of cervical cancer. Furthermore, subjects carrying both rs568408 AG/AA and rs3212227 AC/CC genotypes and simultaneously having >2 parities had a significantly 6-fold (95% CI, 2.86-12.56) elevated cervical cancer risk. Test for interactions between the combined genotypes of rs568408 and rs3212227 and selected variables showed some statistical evidence that parity interacted with the genetic effects (P for multiplicative interaction = 0.046) in relation to cervical cancer risk.

In this hospital-based case-control study, we investigated the association of IL12A rs568408 (3′UTR G>A) and rs2243115 (5′UTR T>G) and IL12B rs3212227 (3′UTR A>C) polymorphisms and risk of cervical cancer in Chinese women. We found, for the first time, that IL12A rs568408 and IL12B rs3212227 interacted to contribute to the risk of cervical cancer, especially among early stage patients. Stratified analyses indicated that the joint effects varied by host menstrual and reproductive histories and interacted with parities. These findings support our hypothesis that potentially functional polymorphisms in IL12 may play a role in the etiology of cervical cancer.

There is an increasing volume of literature published to date on the function of IL12B rs3212227 (3′UTR A>C). Morahan et al. (33) reported that rs3212227AA genotype was associated with significantly elevated expression of IL12 in EBV-transformed human cell lines. Similar results were observed using peripheral lymphocytes (34). However, controversial results were reported in other studies (35–37). We searched the PupaSuite online Web site⁷

to identify functional relevance of the three loci included in the present study. We found that all the three SNPs are located in mouse conserved regions, which rs3212227 may disrupt exonic splicing silencers, and that rs568408 may disrupt exonic splicing enhancers and miRNAs binding. Therefore, our results on these two SNPs seems to be biologically plausible, but further functional characterization of these SNPs is warranted.

There was only one small case-control study of 154 cases and 191 controls published recently, in which the authors investigated the role of IL12B variants and cervical cancer susceptibility. They found that rs3212227 AC/CC genotypes were associated with a nonsignificantly increased risk of cervical cancer in Korean women (32), which was consistent with our findings in Chinese women. Similar associations were also reported in a case-control study of HBV-related hepatocellular carcinoma in a Chinese population (25). However, no published studies have investigated to date the association between IL12A polymorphisms and cervical cancer risk. Several recent studies have evaluated the association between IL12A polymorphisms and risk of other kind of cancers, but the results were inconsistent (26–29). The discrepancies between our current study and published studies might reflect differences in disease etiology, the underlying molecular mechanisms and/or environmental exposures in different populations.

Interestingly, in the present study, we found that the two 3′UTR SNPs in IL12A and IL12B interactively contributed to the risk of cervical cancer. This finding is also biologically plausible because the production of IL12 heterodimer is reportedly to require the coordinated expression of both p40 and p35 chains (13). We further found that the combined genotypes of these two loci significantly interacted with parities, a known risk factor of cervical cancer, although how parity is related to these genotypes, HPV infection, inflammation, immunity, and cervical cancer remains unknown.

Several limitations of our study need to be addressed. First, our study lacked the measurement of local IL12 expression levels, and thus, we could not evaluate the associations between genotypes and phenotypes of IL12. Second, the number of subjects in our study was moderate, and the statistical power of the study was limited. Third, we did not have the tissue sample for HPV typing. Although HPV infection is a necessary pathogenic factor for cervical cancer, it is possible that associations may be stronger with one particular subtype than another. Finally, our study was a hospital-based, using the cases from hospitals and the controls from a screening program of the surrounding community. Therefore, the study subjects, particularly cancer cases, may not be representative for the target population, and our findings might not be generalizable to the general population. However, potential confounding might be minimized by matching the controls to the cases on age and residential areas and by adjusting for some potential confounders in final data analyses. Nevertheless, our findings need to be interpreted cautiously, and the associations need to be replicated in larger, preferably population-based studies.

No potential conflicts of interest were disclosed.