Genes & Genetic Systems
Online ISSN : 1880-5779
Print ISSN : 1341-7568
ISSN-L : 1341-7568
Full papers
Single-nucleotide polymorphisms of the IL-12 gene lead to a higher cancer risk: a meta-analysis based on 22,670 subjects
Xiaohan ShiYingxian JiaXiaochuan XieShangwei Li
ジャーナル オープンアクセス HTML

2017 年 92 巻 4 号 p. 173-187


The associations between interleukin-12 (IL-12) gene polymorphisms and cancer risk have been discussed extensively, with controversial results. Therefore, we conducted the present meta-analysis to better assess the potential roles of IL-12 gene variation in cancer occurrence. Eligible articles were found via PubMed, Medline, EMBASE, Google Scholar and CNKI. Odds ratios and 95% confidence intervals were used to evaluate the associations between IL-12 gene polymorphisms and cancer risk. Thirty-one studies with 10,749 cancer patients and 11,921 healthy subjects were included in the analyses. The overall results showed that cancer risk was increased by IL-12A rs568408 (GG versus GA + AA: P = 0.004; G versus A: P = 0.005) and IL-12B rs3212227 (AA versus AC + CC: P = 0.004; CC versus AA + AC: P = 0.03; A versus C: P = 0.007) polymorphisms. Further subgroup analyses for IL-12A rs568408 and IL-12B rs3212227 revealed that the positive results could be impacted by the ethnicity of the population, cancer type and/or genotyping methods. However, we failed to detect any significant associations between the IL-12A rs2243115 polymorphism and cancer risk in either the overall or the subgroup analyses. The current study suggests that certain IL-12 gene polymorphisms serve as biological markers of cancer susceptibility.


Cancer is a major threat to public health. According to a recent investigation, over 14.1 million new cases and 8.2 million deaths are caused annually by cancer (Siegel et al., 2016). Furthermore, by 2020, the burden of cancer is expected to rise by 50%, due to the rapidly aging population (Popat et al., 2013). To date, the exact pathogenic mechanisms of cancer remain ambiguous. Although tobacco use, heavy alcohol intake, high caloric diet and chemical dyes have been identified as risk factors for cancer (Jemal et al., 2010), the fact that only a small proportion of individuals exposed to these carcinogenic agents ultimately develops cancer suggests that genetic factors play a crucial part in cancer pathogenesis. In addition, the familial aggregation tendency of cancer has long been acknowledged (Risch, 2001), and a number of genetic variants have been found to be associated with cancer susceptibility in different populations (Mtoert et al., 2014; Jamieson et al., 2015). Nevertheless, the mechanisms of cancer pathogenesis are highly complex, and the genetic determinants involved in cancer development are not fully clarified.

Interleukin-12 (IL-12) is a pro-inflammatory cytokine that is mainly secreted by antigen-presenting cells. IL-12 targets T-helper (Th) cells and natural killer cells, and stimulates the synthesis and secretion of interferon gamma (IFN-γ), which is a well established anti-tumor factor (Croxford et al., 2014). Additionally, lower serum IL-12 levels have been observed in patients with various types of cancer (Green et al., 2012; Stanilov et al., 2012; Tao et al., 2012; Wang et al., 2013; Fang et al., 2015), suggesting that IL-12 functions as a potent tumor-suppressive factor. Biologically active IL-12 consists of two functional subunits, p35 and p40, which are encoded by the IL-12A and IL-12B genes, respectively (Croxford et al., 2014). Since the tumor-suppressive effect of IL-12 is well documented, functional polymorphisms of the IL-12A and IL-12B genes are thought to be good genetic candidates for cancer susceptibility.

Recently, extensive studies have explored the potential associations of IL-12A and IL-12B genetic variants with cancer risk. Among these, 3’UTR A > C (1188A > C; rs3212227) in the IL-12B gene and 3’UTR G > A (277 G > A; rs568408) and 5’UTR T > G (564 T > G; rs2243115) in the IL-12A gene are the most intensively studied sites. It is well established that the 3’UTR and 5’UTR are important to mRNA stability, which may contribute to gene expression (Wu and Brewer, 2012). Thus, although these variants are located in untranslated regions, some studies note that they are able to impact IL-12 levels. Morahan et al. (2001) reported that IL-12B rs3212227 was responsible for altered levels of IL-12B mRNA expression in cell lines, and, compared to the AA genotype cell line, decreased IL-12B levels were observed in the CC line. Other scientists have subsequently found similar phenomena in different diseases, including cancer, suggesting that serum IL-12 levels could be modulated by this SNP (Seegers et al., 2002; Windsor et al., 2004; Yilmaz et al., 2005; Stanilova et al., 2007; Wang et al., 2013). For IL12A rs568408 and rs2243115, studies on their potential association with IL-12 expression are still at early stage, and no evidence was detected that they affect IL-12 production (Tao et al., 2012; Wang et al., 2013). It is notable that most studies have simply discussed whether the distributions of IL-12 SNPs in cancer patients are statistically different from that in healthy individuals, and the exact underlying mechanisms between these three SNPs and cancer remain unclear. However, the results of relevant studies were inconsistent and inconclusive. Thus, we conducted the present meta-analysis to better assess the relationship between the IL-12 gene polymorphisms and the risk of cancer.


Literature search

To retrieve all relevant articles, we searched the electronic databases PubMed, Medline, EMBASE, Google Scholar and China National Knowledge Infrastructure (CNKI) without a time limit using the following terms: Interleukin-12, IL-12, Interleukin 12, IL 12, polymorphism, variant, genotype, allele, cancer, tumor, carcinoma, neoplasm and malignancy. Moreover, the reference lists of all retrieved articles were reviewed manually for other potentially eligible articles.

Inclusion criteria and exclusion criteria

Literature matching all the following criteria was included in the present meta-analysis: (1) published articles; (2) a case-control study of associations between IL-12 gene variants and cancer susceptibility in unrelated cancer patients and control subjects; (3) identification of both genotypic and allelic distributions of the IL-12 gene polymorphisms; and (4) availability of full text in English or Chinese. Abstracts, case reports, case series, pedigree studies, reviews and experts’ opinions were intentionally excluded. Additionally, if the same patients were enrolled in multiple studies, only the most recent and complete study was included. It should be particularly noted that, since there is no consensus on how to handle studies with a control group that is not in Hardy-Weinberg equilibrium (HWE), in this meta-analysis we did not exclude studies deviating from HWE as long as they were eligible according to the inclusion criteria and were not of poor quality (Zintzaras and Lau, 2008).

Data extraction

Two of the authors (Shi and Jia) extracted the following information from every included study: name of first author, year of publication, country of origin, ethnicity of the study population, type of cancer, number of cases and controls, genotypic and allelic frequencies in the cases and controls, and P values of HWE in the control group.

Quality assessment of included studies

The Newcastle-Ottawa quality assessment scale (NOS) was used to evaluate the quality of each included study (Stang, 2010). As a classical rating tool of observational studies, NOS assesses studies from the following three aspects: selection, comparability and exposure. The score range of NOS is from 0 to 9, and studies with a score greater than 7 are assumed to be high-quality. The quality assessment was conducted independently by two of the authors (Shi and Jia), and any discordance between the investigators was resolved by discussion with a third author (Xie) until an agreement was achieved.

Statistical analysis

Odds ratios (ORs) and 95% confidence intervals (CIs) were used to evaluate the associations between the IL-12 gene variants and cancer susceptibility in dominant, recessive and allelic models. In addition, a raw probability value (P value) of 0.05 or less was considered statistically significant, and was further subjected to Bonferroni correction to account for multiple statistical tests. The significance threshold was set at 0.017 (0.05/3) for a single SNP because three statistical models were performed for each SNP. The chi-square test was used to explore HWE. Q test and I2 statistic were applied to assess the heterogeneity between studies. If the P value of the Q test was less than 0.1 or I2 was greater than 50%, the between-study heterogeneity was considered statistically significant and the random-effect model (REM) was employed for the analyses. Otherwise, the analyses were carried out with the fixed-effect model (FEM). Furthermore, subgroup analyses by cancer type, genotype method and ethnicity were conducted to trace the source of the heterogeneity. Sensitivity analyses were performed to examine the stability of the results. Publication bias was tested by funnel plots. All data were analyzed with ReviewManager Version 5.3.3 (The Cochrane Collaboration, Software Update, Oxford, U.K.).


Included studies

The literature search generated 1,212 results, and 1,143 citations were excluded after reading the titles and abstracts. Among the remaining 69 articles, 31 studies, containing 10,749 cancer patients and 11,921 control subjects, were finally included in the present meta-analysis (Howell et al., 2003; Amirzargar et al., 2005; García-González et al., 2007; Hou et al., 2007; Lee et al., 2007; Crusius et al., 2008; Han et al., 2008; Chen et al., 2009; Miteva et al., 2009; Tamandani et al., 2009; Wei et al., 2009; Wu et al., 2009; Zhao et al., 2009; Ben Chaaben et al., 2011; Liu et al., 2011; Welsh et al., 2011; Yang, et al., 2011; do Carmo Vasconcelos de Carvalho et al., 2012; Huang et al., 2012; Kaarvatn et al., 2012; Roszak et al., 2012; Sima et al., 2012; Tao et al., 2012; Jaiswal et al., 2013; Sun et al., 2013; Wang et al., 2013; Saxena et al., 2014; Jafarzadeh et al., 2015; Sun et al., 2015; Winchester et al., 2015; Yin et al., 2015) (Fig. 1). Of these studies, 29/31 concerned the IL-12B rs3212227 polymorphism, 7/31 concerned the IL-12A rs2243115 polymorphism, and 8/31 concerned the IL-12A rs568408 polymorphism. All articles were published between 2003 and 2015. Fourteen studies were performed in Caucasians, 15 in Asians, and two in populations with mixed ethnicities. Twenty-nine articles were published in English, and the other two were in Chinese. A violation of HWE was found in three studies, which were nevertheless included in further analyses because of their high quality. The characteristics of the included studies are summarized in Table 1 (IL-12A rs568408 variant), Table 2 (IL-12A rs2243115 variant) and Table 3 (IL-12B rs3212227 variant).

Fig. 1.

Flowchart of study selection for the present meta-analysis.

Table 1. Characteristics of the included studies for IL-12A rs568408 polymorphism and cancer risk
First author, yearCountryEthnicityCancer typeGenotype methodCaseControlPvalue HWENOS score
nGenotypes GG/GA/AAAlleles G/A (%)nGenotypes GG/GA/AAAlleles G/A (%)
Chen 2009ChinaAsianCervical cancerPCR-RFLP404253/145/680.6/19.4404285/112/784.4/15.60.2868
Hou 2007PolandCaucasianGastric cancerTaqMan302211/83/883.6/16.4428306/112/1084.6/15.40.9488
Lee 2007ChinaAsianLung cancerTaqMan118101/17/092.8/7.211381/31/185.4/14.60.2887
Liu 2011ChinaAsianHepatocellular carcinomaPCR-RFLP802504/277/2180.1/19.9861631/220/1086.1/13.90.0567
Roszak 2012PolandCaucasianCervical cancerPCR-RFLP405253/128/2478.3/21.7450306/125/1981.9/18.10.1787
Sun 2015ChinaAsianColorectal cancerPCR-RFLP257172/83/283.1/16.9236181/55/088.3/11.70.0437
Tao 2012ChinaAsianEsophageal cancerPCR-RFLP426267/147/1279.9/20.1432304/112/1683.3/16.70.1668
Wang 2013ChinaAsianOsteosarcomaPCR-RFLP10666/37/379.7/20.3210156/47/785.5/14.50.1537

HWE: Hardy-Weinberg equilibrium; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism; NOS: Newcastle-Ottawa quality assessment scale.

The figure in bold type indicates that the distributions of IL-12A rs568408 in controls deviate from HWE.

Table 2. Characteristics of the included studies for IL-12A rs2243115 polymorphism and cancer risk
First author, yearCountryEthnicityCancer typeGenotype methodCaseControlPvalue HWENOS score
nGenotypes TT/TG/GGAlleles T/G (%)nGenotypes TT/TG/GGAlleles T/G (%)
Chen 2009ChinaAsianCervical cancerPCR-RFLP404342/60/292.1/7.9404337/64/391.3/8.70.9738
Liu 2011ChinaAsianHepatocellularcarcinomaPCR-RFLP861735/125/192.6/7.4874732/137/591.6/8.40.6047
Sima 2012ChinaAsianBrain tumorPCR-RFLP170140/25/589.7/10.3222198/24/094.6/5.40.3947
Sun 2013ChinaAsianEsophageal cancerTaqMan368311/56/192.1/7.9370308/58/491.1/8.90.4997
Sun 2015ChinaAsianColorectal cancerPCR-RFLP257225/31/193.6/6.4236212/24/094.9/5.10.4117
Wang 2013ChinaAsianOsteosarcomaPCR-RFLP10689/16/191.5/8.5210175/31/490.7/9.30.0737
Yin 2015ChinaAsianGastric cancerTaqMan235197/36/291.5/8.5466387/75/491.1/8.90.8628

HWE: Hardy-Weinberg equilibrium; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism; NOS: Newcastle-Ottawa quality assessment scale.

Table 3. Characteristics of the included studies for IL-12B rs3212227 polymorphism and cancer risk
First author, yearCountryEthnicityCancer typeGenotype methodCaseControlPvalue HWENOS score
nGenotypes AA/AC/CCAlleles A/C (%)nGenotypes AA/AC/CCAlleles A/C (%)
Amirzargar 2005IranCaucasianCMLPCR-SSP3017/11/275.0/25.04024/13/376.3/23.70.5167
Ben Chaaben 2011TunisiaCaucasianNPCTaqMan24774/124/4955.1/44.9284135/118/3168.3/31.70.4977
Chen 2009ChinaAsianCervical cancerPCR-RFLP404127/199/7856.1/43.9404150/185/6960.0/40.00.3578
Crusius 2008NetherlandsCaucasianGastric cancerTaqMan230139/74/1776.5/23.51060677/343/4080.0/20.00.6728
do Carmo 2012BrazilMixedCervical cancerPCR–RFLP97/1/183.3/16.77631/37/865.1/34.90.5317
García-González 2007SpainCaucasianGastric cancerTaqMan404247/133/2477.6/22.4404256/130/1879.5/20.50.7738
Han 2008KoreaAsianCervical cancerDNA sequencing15032/87/3150.3/49.717952/88/3953.6/46.40.8777
Howell 2003UKCaucasianSkin cancerARMS-PCR14595/42/880.0/20.0229139/77/1377.5/22.50.5917
Huang 2012ChinaAsianColorectal cancerPCR-RFLP41093/219/9849.4/50.6450124/230/9653.1/46.90.5788
Jafarzadeh 2015IranCaucasianBreast cancerPCR-RFLP10058/32/1074.0/26.010059/33/876.0/24.00.2808
Jaiswal 2013IndiaCaucasianBladder cancerPCR-RFLP200111/74/1574.0/26.020087/94/1967.0/33.00.3748
Kaarvatn 2012CroatiaCaucasianBreast cancerTaqMan191126/59/681.4/18.6194104/73/1772.4/27.60.4197
Liu 2011ChinaAsianHCCPCR-RFLP831249/422/16055.4/44.6844272/414/15856.8/43.20.9837
Miteva 2009BulgariaCaucasianColorectal cancerPCR-RFLP8550/31/477.1/22.913480/45/976.5/23.50.4438
Roszak 2012PolandCaucasianCervical cancerPCR-RFLP405212/174/1973.8/26.2450289/151/1081.0/19.00.0567
Saxena 2014IndiaCaucasianHCCPCR-RFLP5919/31/958.5/41.514863/71/1466.6/33.40.3457
Sima 2012ChinaAsianBrain tumorPCR-RFLP17066/86/1864.1/35.922271/116/3558.1/41.90.2757
Sun 2013ChinaAsianEsophageal cancerTaqMan368116/176/7655.4/44.6370112/179/7954.5/45.50.6357
Sun 2015ChinaAsianColorectal cancerPCR-RFLP25769/140/4854.1/45.923671/124/4156.4/43.60.2957
Tamandani 2009IndiaCaucasianCervical cancerPCR-RFLP20063/134/365.0/35.020088/104/870.0/30.00.0018
Tao 2012ChinaAsianEsophageal cancerPCR-RFLP426109/216/10150.9/49.1432148/213/7158.9/41.10.7018
Wang 2013ChinaAsianOsteosarcomaPCR-RFLP10627/50/2949.1/50.921078/101/3161.2/38.80.8557
Wei 2009ChinaAsianNPCPCR-RFLP30258/165/7946.5/53.531098/152/6056.1/43.90.9388
Welsh 2011UKCaucasianSkin cancerTaqMan1483971/446/6680.5/19.5763500/230/3380.6/19.40.3208
Winchester 2015USAMixedProstate cancerMassArray866535/274/5777.6/22.4830568/227/3582.1/17.90.0468
Wu 2009ChinaAsianGastric cancerPCR-RFLP1035347/508/18058.1/41.91073333/554/18656.8/43.20.0867
Yang 2011ChinaAsianHCCTaqMan772220/397/15554.2/45.8852236/417/19952.2/47.80.5748
Yin 2015ChinaAsianGastric cancerTaqMan23460/132/4253.8/46.2466152/221/9356.3/43.70.4368
Zhao 2009ChinaAsianBrain tumorPCR-RFLP21038/115/5745.5/54.522070/106/4455.9/44.10.7368

HWE: Hardy-Weinberg equilibrium; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism; CML: chronic myelogenous leukemia; NPL: nasopharyngealcarcinoma; HCC: hepatocellularcarcinoma; ARMS-PCR: amplification refractory mutation system-polymerase chain reaction.

Quality assessment of included studies

The average NOS score of the included studies was 7.53 (range, 7 to 8), indicating that all enrolled articles were of high quality. Potential biases in the eligible studies mainly originated from unmatched baseline characteristics of the cases and controls, such as age and gender.

IL-12A rs568408 polymorphism and cancer susceptibilit

Eight studies containing 2,820 cancer patients and 3,134 control subjects were included to assess the association between the IL-12A rs568408 variant and cancer risk. For GG versus GA + AA (dominant model) and G versus A (allelic model), analyses were conducted with REMs because of obvious between-study heterogeneity. For AA versus GG + GA (recessive model), the heterogeneity between studies was mild, and the FEM was therefore used for the analysis. A significant association between the IL-12A rs568408 polymorphism and cancer susceptibility was detected in GG versus GA + AA (P = 0.004, OR = 0.75, 95%CI 0.62–0.91) and G versus A (P = 0.005, OR = 0.79, 95%CI 0.67–0.93) (Fig. 2). Further subgroup analyses were conducted by ethnicity, cancer type and genotype method. By subdividing the studies according to different ethnicities, we found that six studies were conducted in Asians and two in Caucasians. The between-study heterogeneity remained significant in the Asian subgroup, but was no longer notable in the Caucasian subgroup. Similar positive results were observed in the Asian population (GG versus GA + AA: P = 0.01, OR = 0.73, 95%CI 0.57–0.93; G versus A: P = 0.02, OR = 0.78, 95%CI 0.63–0.97) while no significant associations were found in the Caucasian population. Subgroup analysis by cancer type could only be performed for cervical cancer with two studies that were homogeneous, and the results were consistent with the overall results (GG versus GA + AA: P = 0.004, OR = 0.74, 95%CI 0.61–0.91; G versus A: P = 0.006, OR = 0.78, 95%CI 0.66–0.93). By subdividing the studies according to genotype method, we observed that polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) was used in six studies, and TaqMan was used in two. The heterogeneity between the studies was trivial in the PCR-RFLP group, and FEM-effects meta-analysis pooled significant results in both GG versus GA + AA (P < 0.00001, OR = 0.67, 95%CI 0.59–0.76) and G versus A (P < 0.00001, OR = 0.72, 95%CI 0.65–0.80), which were consistent with the overall results. In the TaqMan group, REMs were carried out in the analyses because the between-study heterogeneity remained obvious. However, no significant association was detected in this subgroup (Table 4).

Fig. 2.

Forest plots on the association between IL-12A rs568408 polymorphism and cancer risk. (A). Forest plot of GG versus GA + AA for IL12A rs568408 polymorphism and cancer risk. (B). Forest plot of AA versus GG + GA for IL12A rs568408 polymorphism and cancer risk. (C). Forest plot of G versus A for IL12A rs568408 polymorphism and cancer risk.

Table 4. Subgroup analyses for IL-12B rs3212227polymorphism and cancer risk
PolymorphismVariablesP valueOR (95%Cl)I-square (%)P for the heterogeneity
IL-12A rs568408Ethnicity
Asian (No.a: 6)
 GG versus GA + AA0.010.73 (0.57–0.93)670.009
 AA versus GG + GA0.421.19 (0.78–1.83)200.28
 G versus A0.020.78 (0.63–0.97)670.01
Caucasian (No.a: 2)
 GG versus GA + AA0.110.84 (0.68–1.04)00.45
 AA versus GG + GA0.271.33 (0.80–2.23)00.69
 G versus A0.080.85 (0.71–1.02)00.41
Genotype method
PCR-RFLP (No.a: 6)
 GG versus GA + AA0.000010.67 (0.59–0.76)00.76
 AA versus GG + GA0.161.29 (0.91–1.84)140.33
 G versus A0.000010.72 (0.65–0.80)00.68
TaqMan(No.a: 2)
 GG versus GA + AA0.461.41 (0.57–3.49)840.01
 AA versus GG + GA0.991.01 (0.41–2.46)00.45
 G versus A0.461.37 (0.59–3.17)840.01
Cancer type
Cervical cancer (No.a: 2)
 GG versus GA + AA0.0040.74 (0.61–0.91)00.59
 AA versus GG + GA0.391.26 (0.74–2.16)00.42
 G versus A0.0060.78 (0.66–0.93)00.83
IL-12A rs2243115Genotype method
PCR-RFLP (No.a: 5)
 TT versus TG + GG0.961.00 (0.84–1.20)270.24
 GG versus TT + TG0.931.03 (0.46–2.32)380.17
 T versus G0.620.93 (0.70–1.24)550.06
TaqMan(No.a: 2)
 TT versus TG + GG0.601.08 (0.81–1.44)00.90
 GG versus TT + TG0.370.55 (0.15–2.03)00.33
 T versus G0.491.10 (0.84–1.44)00.76
IL-12B rs3212227Ethnicity
Asian (No.a: 14)
 AA versus AC + CC0.0050.81 (0.70–0.94)670.0002
 CC versus AA + AC0.161.10 (0.96–1.26)490.02
 A versus C0.020.89 (0.80–0.98)700.0001
Caucasian (No.a: 13)
 AA versus AC + CC0.270.89 (0.73–1.09)740.00001
 CC versus AA + AC0.311.17 (0.86–1.58)490.03
 A versus C0.300.92 (0.78–1.08)740.00001
Genotype method
PCR-RFLP (No.a: 17)
 AA versus AC + CC0.010.81 (0.68–0.96)720.00001
 CC versus AA + AC0.021.19 (1.03–1.38)360.07
 A versus C0.0090.86 (0.77–0.96)700.00001
TaqMan(No.a: 8)
 AA versus AC + CC0.380.92 (0.75–1.11)730.0004
 CC versus AA + AC0.611.08 (0.81–1.43)690.002
 A versus C0.520.95 (0.81–1.12)790.0001
Cancer type
Nasopharyngeal carcinoma (No.a: 2)
 AA versus AC + CC0.000010.49 (0.38–0.64)00.75
 CC versus AA + AC0.00091.66 (1.23–2.24)00.32
 A versus C0.000010.63 (0.53–0.75)80.30
Hepatocellular carcinoma (No.a: 3)
 AA versus AC + CC0.460.94 (0.79–1.11)170.30
 CC versus AA + AC0.780.97 (0.75–1.24)430.17
 A versus C0.740.97 (0.83–1.14)540.11
Cervical cancer (No.a: 5)
 AA versus AC + CC0.0060.69 (0.53–0.90)480.10
 CC versus AA + AC0.581.12 (0.76–1.65)320.21
 A versus C0.010.80 (0.68–0.95)410.15
Colorectal cancer (No.a: 3)
 AA versus AC + CC0.090.83 (0.66–1.03)00.77
 CC versus AA + AC0.421.11 (0.86–1.44)00.71
 A versus C0.120.89 (0.77–1.03)00.76
Brain tumor (No.a: 2)
 AA versus AC + CC0.670.80 (0.29–2.24)910.0008
 CC versus AA + AC0.991.00 (0.43–2.30)800.03
 A versus C0.800.92 (0.48–1.77)910.0009
Skin cancer (No.a: 2)
 AA versus AC + CC0.731.03 (0.87–1.22)00.38
 CC versus AA + AC0.921.02 (0.69–1.50)00.91
 A versus C0.801.02 (0.88–1.18)00.44
Breast cancer (No.a: 2)
 AA versus AC + CC0.331.31 (0.76–2.26)590.12
 CC versus AA + AC0.520.65 (0.18–2.42)730.06
 A versus C0.421.27 (0.71–2.26)760.04
Esophageal cancer (No.a: 2)
 AA versus AC + CC0.440.83 (0.52–1.33)790.03
 CC versus AA + AC0.401.23 (0.76–2.01)750.05
 A versus C0.430.87 (0.61–1.24)840.01
Gastric cancer (No.a: 4)
 AA versus AC + CC0.410.92 (0.76–1.12)510.10
 CC versus AA + AC0.371.15 (0.85–1.58)530.09
 A versus C0.330.94 (0.83–1.06)340.21

OR: odds ratio; CI: confidence interval; NA: not applicable; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism.

a  Number of articles.

Bold-type numbers indicate astatistically significant difference between cases and controls.

IL-12A rs2243115 polymorphism and cancer susceptibility

Estimation of the association between the IL-12A rs2243115 polymorphism and cancer risk was based on seven studies, including 2,401 cases and 2,782 controls. FEMs were applied for all analyses, because no obvious between-study heterogeneity was detected. No significant associations with cancer risk were observed for the IL-12A rs2243115 polymorphism (Fig. 3). Subgroup analyses were performed by the genotype method and were mostly under FEMs, because there was negligible heterogeneity between the studies. Nevertheless, neither the PCR-RFLP group nor the TaqMan group showed positive results in any of the comparison models (Table 4).

Fig. 3.

Forest plots on association between IL12A rs2243115 polymorphism and cancer risk. (A). Forest plot of TT versus TG + GG for IL12A rs2243115 polymorphism and cancer risk is shown. (B). Forest plot of GG versus TT + TG for IL12A rs2243115 polymorphism and cancer risk is shown. (C). Forest plot of T versus G for IL12A rs2243115 polymorphism and cancer risk is shown.

IL-12B rs3212227 polymorphism and cancer susceptibility

For the IL-12B rs3212227 polymorphism, a total of 29 studies, including 10,329 cancer patients and 11,380 control subjects, were analyzed. Due to the significant heterogeneity, comparisons were performed with REMs in AA versus AC + CC (dominant model), CC versus AA + AC (recessive model), and A versus C (allelic model). We found a significant association between the IL-12B rs3212227 variant and cancer risk in all three models (AA versus AC + CC: P = 0.004, OR = 0.85, 95%CI 0.75–0.95; CC versus AA + AC: P = 0.03, OR = 1.15, 95%CI 1.02–1.31; A versus C: P = 0.007, OR = 0.89, 95%CI 0.82–0.97) (Fig. 4). Further stratified analyses of the IL-12B rs3212227 variant were conducted by ethnicity, cancer type and genotype method. Regarding the ethnic background, there were 14 studies of Asian ethnicity and 13 studies of Caucasian ethnicity. The heterogeneity between the studies remained obvious in both subgroups. Consistent with our overall results, we found a significant association between the IL-12B rs3212227 polymorphism and cancer risk in the Asian subgroup for AA versus AC + CC (P = 0.005, OR = 0.81, 95%CI 0.70–0.94) and A versus C (P = 0.02, OR = 0.89, 95%CI 0.80–0.98). However, no associations were observed in the Caucasian subgroup. When considering the different genotype methods, the numbers of studies using PCR-RFLP or the TaqMan method were 17 and eight, respectively. REMs were applied in both subgroups due to the significant heterogeneity. These models demonstrated that the PCR-RFLP group was responsible for the positive results (AA versus AC + CC: P = 0.01, OR = 0.81, 95%CI 0.69–0.96, CC versus AA + AC: P = 0.02, OR = 1.19, 95%CI 1.03–1.38, A versus C: P = 0.009, OR = 0.86, 95%CI 0.77–0.96), whereas the TaqMan group showed no cancer susceptibility. Moreover, we performed a subgroup analysis by cancer type for the IL-12B rs3212227. As shown in Table 4, the between-study heterogeneity was reduced in some genetic models after stratifying the available data according to cancer types. Compared to our overall findings, similar results were observed for nasopharyngeal carcinoma (AA versus AC + CC, P = 0.00001, OR = 0.49, 95%CI 0.38–0.64; CC versus AA + AC: P = 0.0009, OR = 1.66, 95%CI 1.23–2.24; A versus C: P = 0.00001, OR = 0.63, 95%CI 0.53–0.75) and cervical cancer (AA versus AC + CC: P = 0.006, OR = 0.69, 95%CI 0.53–0.90; A versus C: P = 0.01, OR = 0.80, 95%CI 0.68–0.95) (Table 4).

Fig. 4.

Forest plots on association between IL12B rs3212227 polymorphism and cancer risk. (A). Forest plot of AA versus AC + CC for IL12B rs3212227 polymorphism and cancer risk is shown. (B). Forest plot of CC versus AA + AC for IL12B rs3212227 polymorphism and cancer risk is shown. (C). Forest plot of A versus C for IL12B rs3212227 polymorphism and cancer risk is shown.

Bonferroni correction of overall results

Since multiple statistical tests were conducted for each SNP, Bonferroni corrections of overall results were further performed, and only the CC versus AA + AC comparison (recessive model) of IL-12B rs3212227 yielded the opposite outcome (raw P value = 0.03 > 0.017), suggesting that most of our results were statistically robust.

Sensitivity analyses and publication bias

Sensitivity analyses were performed by removing one individual study at a time. The overall results were not impacted by any individual study, including those that were not in HWE for all of the three analyzed SNPs. No other changes in the results were found in the subgroup analyses. Funnel plots were used to evaluate the potential publication bias. A visual inspection of the funnel plots revealed no apparent asymmetry for any of the IL-12 polymorphisms, and these results suggest that there was no significant publication bias in the present meta-analysis (Supplementary Fig. S1).


IL-12 is an anti-tumor cytokine with multiple functions. First, IL-12 can stimulate the synthesis and secretion of IFN-γ, which serves as a pro-inflammatory and tumor-suppressive factor (Croxford et al., 2014). In addition, it was found that p53, which triggers apoptosis in cancer cells, was up-regulated by the secretion of IFN-γ, thus, subsequently, inhibiting tumor growth (Takaoka et al., 2003). Second, IL-12 can promote Th1 cell differentiation by stimulating the synthesis of interferon regulatory factors 1 (IRF1) and 4 (IRF4) (Lehtonen et al., 2003). Additionally, when IL-12 binds to its receptor (IL-12R), binding sites form for Tyk2 and Jak2 kinases. These kinases are able to recruit transcription factor STAT4, which accumulates at the promoter region of IL-2R. Consequently, the expression levels of IL-2R increase, leading to enhanced Th1 cell proliferation and Th1-cell related immune responses, especially cytotoxic reactions (Lee et al., 2012; Stark et al., 2013). Third, IL-12 can inhibit the formation of tumor microvessels by reducing the production of vascular endothelial growth factor (Cavallo et al., 2001). As mentioned above, the IL-12 p35 and IL-12 p40 subunits are encoded by the IL-12A and IL-12B genes, respectively (Croxford et al., 2014). Consequently, IL-12A and IL-12B genetic polymorphisms, which may affect the expression levels and normal functioning of IL-12, are thought to be implicated in the occurrence and development of various types of cancer (Yuzhalin and Kutikhin, 2012). Among these, IL-12A rs568408, IL-12B rs2243115 and IL-12B rs3212227 polymorphisms were the three most investigated sites. IL-12A rs568408 and IL-12B rs3212227 are located in the 3’-untranslated region (3’UTR). These two polymorphisms may disrupt exonic splicing and influence the production level of IL-12 (Chen et al., 2009; Liu et al., 2011). IL-12A rs2243115, however, is situated in the 5’UTR, and its functional significance has not been reported. Despite these potential molecular mechanisms, the results of relevant studies remain conflicting. Therefore, we conducted the present study in an attempt to obtain a more conclusive result.

The results of the current meta-analysis support the idea that the IL-12A rs568408 and IL-12B rs3212227 polymorphisms significantly correlate with cancer risk in the Asian ethnicity, especially when the PCR-RFLP genotype method is used. In addition, the A allele of the IL-12A rs568408 variant and the C allele of theIL-12B rs3212227 variant contribute to the increased risk of cancer. However, no associations were detected between the IL-12A 2243115 variant and cancer risk. It is worth noting that Zhou et al. (2012) performed a meta-analysis for these three IL-12 polymorphisms in 2012. Compared with that study, similar results were found for the IL-12A 2243115 and IL-12B rs3212227 polymorphisms in the present meta-analysis, whereas the results for IL-12A rs568408 were opposite. Since the sample size of our analysis was much larger than the previous one (22,670 subjects versus 13,875 subjects), our findings should be more reliable. Further subgroup analyses yielded similar positive results for the IL-12A rs568408 and IL-12B rs3212227 polymorphisms in Asians. However, no significant associations were found in Caucasians. In addition, the IL-12B rs3212227 polymorphism was found to be implicated in nasopharyngeal carcinoma and cervical cancer but not in hepatocellular carcinoma, colorectal cancer, brain tumor, skin cancer, breast cancer, esophageal cancer or gastric cancer. The between-study heterogeneity remained significant for the IL-12A rs568408 and IL-12B rs3212227 polymorphisms in different ethnicities, TaqMan subgroups and cancer subgroups, suggesting that the correlations between the IL-12 genetic polymorphisms and cancer risk may also be modulated by other unevaluated factors.

Several limitations of our study should be noted. First, the number of studies investigating the associations between the IL-12A rs568408 and rs2243115 polymorphisms and cancer risk is still limited, and the sample sizes of several of the included studies were not sufficient, which precluded us from drawing definite conclusions. Second, our results were based on unadjusted estimates, because the majority of the included studies failed to report baseline characteristics of the individuals, such as age, sex, smoking status and eating habits, and the lack of analyses adjusted for these potential confounding factors might affect the reliability of our results. Third, although funnel plots revealed no apparent publication bias, we still could not eliminate the possibility of publication bias, because only published studies were included. Fourth, all included studies were published in English or Chinese; therefore, some qualified articles in other languages may have been missed. Fifth, the relationship between a certain gene polymorphism and cancer risk could be affected by gene-gene or gene-environment interactions. It is possible that a particular polymorphism may be associated with cancer susceptibility, but, due to interactions with multiple genetic and environmental factors, the association would no longer be observed. Given these limitations, we should interpret the current results with caution.

In conclusion, the present meta-analysis of 31 studies demonstrated that the IL-12A rs568408 and IL-12B rs3212227 polymorphisms serve as promising candidate biomarkers of cancer susceptibility in the Asian population. Further studies with larger sample sizes from different populations are warranted to confirm our results. Moreover, since interleukins play a crucial role in regulating anti-tumor immune responses, future investigations are needed to explore the potential roles of other interleukin gene polymorphisms in cancer pathogenesis.

© 2017 by The Genetics Society of Japan