Genetic variants and risk of cervical cancer: epidemiological evidence, meta-analysis and research review


  • X Zhang,

    1. Department of Epidemiology and Health Statistics, School of Public Health, Shandong University, Jinan, Shandong, China
    2. Hangzhou Center for Disease Control and Prevention, Hangzhou, Zhejiang, China
    Search for more papers by this author
  • L Zhang,

    1. Department of Epidemiology and Health Statistics, School of Public Health, Shandong University, Jinan, Shandong, China
    Search for more papers by this author
  • C Tian,

    1. Kunshan Municipal Center for Disease Control and Prevention, Suzhou, Jiangsu, China
    Search for more papers by this author
  • L Yang,

    1. Hangzhou Center for Disease Control and Prevention, Hangzhou, Zhejiang, China
    Search for more papers by this author
  • Z Wang

    Corresponding author
    1. Department of Epidemiology and Health Statistics, School of Public Health, Shandong University, Jinan, Shandong, China
    • Correspondence: Z Wang, Department of Epidemiology and Health Statistics, School of Public Health, Shandong University, 44 Wenhua Xi Road, Jinan, Shandong, 250012, China. Email

    Search for more papers by this author



More than 200 articles have been published in the past 20 years on associations between genetic variants and risk of cervical cancer but the results have generally been inconsistent.


To provide a synopsis of the current understanding of the genetic architecture of the risk of cervical cancer by conducting a systematic review and meta-analysis.

Search strategy

We conducted a systematic literature search by a two-stage strategy using PubMed and other databases on or before 31 March 2012.

Selection criteria

Cross-sectional, case–control or cohort studies about the relationship between genetic variants and cervical cancer were included.

Data collection and analysis

Study outcomes were presented as odds ratios (ORs) with a 95% confidence interval.We did the meta-analysis for genetic variants which had at least three data sources and for which the significant associations were assessed using the Venice criteria.

Main results

A total of 5605 publications were screened, of which 286 were eligible. Meta-analysis was conducted for 58 variants in 25 genes or loci. Fourteen variants in 11 genes or loci could increase the risk of cervical cancer and five variants in three genes or loci could decrease the risk. The epidemiological evidence of the association was graded as strong for four variants in CTLA4 and HLA DQB1, moderate for five variants in IL-1B, IL-10, XRCC3 and HLA DQA1, and weak for 10 variants.


Many genetic variants were associated with the risk of cervical cancer as supported by the epidemiological evidence in this meta-analysis.


The incidence of and mortality from cervical cancer has declined substantially in developed countries since the 1950s but cancer of the cervix remains the third most common cancer among women worldwide.[1] Overwhelming evidence now supports the role of human papillomavirus (HPV) in cervical carcinogenesis,[2] and persistent infection by certain HPV genotypes has been recognised as a necessary step for the progression of cervical cancer.[3] However, as shown by prospective studies, HPV infection alone is not sufficient to cause cervical cancer – only a small fraction of women infected with HPV will eventually develop cervical cancer.[4] HPV is believed to interact with other factors, including reproductive[5] and nutritional factors.[6] Genome-wide association studies (GWAS) have widely been done to identify the association between genetic variants and diseases such as breast cancer[7] and Kawasaki disease.[8] But so far no GWAS have been conducted to identify the association between genetic variants and risk of cervical cancer. Over 200 candidate genes, involving more than 500 genetic variants, have been studied over the past 20 years for an association with cervical cancer. Some of these genes might have true associations with cervical cancer, but many false-positive associations which are not replicated in other populations have also been identified. A meta-analysis is generally done to determine whether an association is real. About 6 years ago, the meta-analyses usually included one variant, or several variants within one gene, but recently the size and scope of meta-analyses has increased.[9, 10] Such investigations have already been carried out for breast cancer[11] and we decided to investigate cervical cancer in the same way. We performed a meta-analyis for the genetic variants for which there are sufficient data. We also summarise the current understanding of the genetic variants associated with an increased risk of cervical cancer.


We followed the PRISMA guidelines and also the proposals of the Human Genome Epidemiology Network for systematic review of studies of genetic association.[12-15]

Literature searches

We used a two-stage search strategy to identify the relevant publications. First, studies in English were identified through PubMed from their earliest available date to 1 October 2011. Reports in Chinese that included associations with cervical cancer were found through three commonly used databases, China National Knowledge Infrastructure (CNKI), the Database of Chinese Scientific and Technical Periodicals (VIP) and the China Biology Medical Literature database (CBM); these were searched from 1979, 1989 and 1970, respectively, to 1 October 2011. The keywords (‘uterine cervical neoplasm’) and (‘gene’ or ‘single nucleotide polymorphism’ or ‘DNA polymorphism’ or ‘genetic polymorphism’ or ‘genetic variation’) were used in combination to retrieve the relevant literature in all Chinese databases. Studies in English were selected using the strategy outlined below:

  1. Restriction fragment length polymorphism OR genetic polymorphism OR single nucleotide polymorphism OR DNA polymorphism OR genetic variation.
  2. Genetic association.
  3. Single nucleotide polymorphism OR SNP.
  4. Cervical cancer OR uterine cervical neoplasm.
  5. 1 OR 2 OR 3.
  6. 4 AND 5.
  7. Limit 6 to (Human and English language).

Titles and abstracts of the articles found were screened, and the full texts of articles of interest were evaluated. We included only original articles that reported associations between cervical cancer and particular human gene polymorphisms or variability in a case–control or population-based study and were available in English or Chinese. Case reports and series, reviews and other publications, such as editorials and animal studies, were excluded from the analysis.

Secondly, targeted monthly searches of the databases were done using the above genes, e.g. ‘MTHFR’ in conjunction with ‘cervical cancer’ as query terms between 1 October 2011 and 31 March 2012. We also carried out general monthly searches between 1 October 2011 and 31 March 2012 using the MeSH term ‘cervical cancer’.

As a final step we screened the references of all the included studies, reviews and meta-analyses, and approached the authors for help if necessary. These last components of the search strategy identified any remaining publications.

Eligibility criteria

Studies were included in the meta-analysis if they met the following criteria:

  1. They appeared online or in a peer-reviewed journal published in English or Chinese on or before 31 March 2012.
  2. They were cross-sectional, case–control or cohort studies in human beings.
  3. The cases were diagnosed by pathological or histological examination.
  4. The methods and procedures used to determine the genotype in each study are universally acknowledged.
  5. The study provided sufficient information such that the genotype frequencies for cases and controls could be obtained. If data were duplicated in more than one study, we included the study with the most cases or the most recent analysis.

Data extraction

Data were extracted by two authors independently (X.Z. and L.Z.), with any disagreements being resolved by consensus. Data were extracted separately for reports that included several sources or study populations. The following data were extracted from each study: first author, year of publication, sample size, study design, mean ages of cases and controls, method of case selection, genes, variants and genotype counts for cases and controls, genotyping methods, the Hardy–Weinberg equilibrium (HWE) among controls, matching factors, the country where the study was performed and the ethnicity of participants. We classified case selection as population-based, hospital-based or mixed. We used four classes of ethnicity, based on the ethnicity of at least 80% of the study population:[16] African, Asian, White British or Other (including mixed). However, if the ethnicity of the cases was not made clear, we classified ethnicity based on the country in which the study was performed.[16] We used the current gene names and variant accession numbers according to the database of the National Center for Biotechnology Information (NCBI).[17] The most common name was used for variants without accession numbers, such as 7271T>G or Val2424Gly. The dbSNP database was used to determine minor and major alleles of variants among populations of specific ethnicities. The study authors were contacted if reported alleles were different from those in the dbSNP. Authors were also contacted concerning publications which were initially ineligible because of insufficient information for data extraction; if we got responses, these publications were included. The quality of the studies was assessed using the nine-star Newcastle–Ottawa Scale.[18]

Statistical analysis

Unless otherwise stated, two-tailed tests were used throughout and P ≤ 0.05 was considered significant. Statistical analysis was done with stata, version 9.2 (Stata Corporation, College Station, TX, USA). A minimum of three data sources were required for meta-analysis of any variant. The HWE among control groups in each study was assessed to compare observed and expected genotype frequencies.[19] Estimates of association were evaluated by odds ratios (ORs) and corresponding 95% confidence intervals (CIs). For the genetic variants, allelic, dominant and recessive models were computed. Meta-analysis stratified by ethnicity was done if permitted.

A heterogeneity test, sensitivity analysis and examination for bias were used to evaluate the results of the meta-analysis. Heterogeneity between studies was assessed by the Cochran Q statistic, with < 0.10 indicating significant heterogeneity.[20] The I2 statistic was also used the to quantify heterogeneity.[21] Generally, I2 < 25% corresponds to mild heterogeneity; I2 between 25 and 50% corresponds to moderate heterogeneity, and I2 > 50% corresponds to large heterogeneity. If the data were heterogeneous, a random effects model was adopted[22]; if the data were homogeneous, a fixed effects model was applied. Meta-regression with restricted maximum likelihood estimation was performed to assess the potentially important covariates that could have a substantial impact on between-study heterogeneity. Sensitivity analysis was performed to strengthen the result of the meta-analysis. Potential publication bias was assessed with funnel plots and a modified Egger linear regression test, as proposed by Harbord et al.,[23] was used to identify significant asymmetry. An analysis of influence was conducted; this describes how robust the pooled estimator is to the removal of individual studies. An individual study is suspected of excessive influence if the point estimate of its omitted analysis lies outside the 95%CI of the combined analysis.


The detailed steps of our literature search are shown in Figure 1. The first approach yielded 4909 publications; these were screened by title and abstract, or a full-text browse, to identify 249 articles (207 English and 42 Chinese articles) that met our eligibility criteria. The second-stage search identified 696 articles, of which 37 (24 English and 13 Chinese studies) met our eligibility criteria. Thus, 5605 publications were screened by title, abstract or full text, identifying a total of 286 eligible articles (Table S1) that included 114 genes and 349 variants. The quality score of studies ranged from six to eight stars according to the nine-star Newcastle–Ottawa Scale (58% of the studies were given eight stars, 36% seven stars, and 6% six stars). The quality of the included studies was therefore good.

Figure 1.

Selection of studies for inclusion in meta-analysis.

Meta-analysis was conducted for 58 variants in 25 genes or loci. Nineteen variants within 13 genes (CTLA4, CYP1A1, CYP2E1, GSTM1, GSTT1, IFN-G, IL-1B, IL-10, TP53, TNF, XRCC3, HLA-DQA1 and HLA-DQB1) had significant associations with a risk of cervical cancer (Table 1). The meta-analysis included an average of 11.5 studies (range 3–60) and 3699 subjects (951–14,541). Strong associations (OR > 2) were seen for three variants (CYP1A1 MspI, CYP2E1 Ile/Val and XRCC3 rs861539), and moderate associations (OR > 1.5) for two common variants, including polymorphisms in IL-1B (C-511T) and TNF (rs1800629). Significant associations were also found for four single-nucleotide polymorphisms (CTLA4 rs231775, IFN-G rs2430561, IL-10 592C/A, p53 codon72; Figure 2), one deletion (GSTT1), two variants in HLA-DQA1 (0101 and 0201) and five variants in HLA-DQB1 (0301, 0303, 0501, 06 and 0602). In several of the meta-analyses the available data were limited to a single ethnicity. Three variants with significant associations (CYP2E1 Ile/Val, IL-1B C-511T and XRCC3 rs861539) were studied only among Asian women.

Table 1. Genetic variants with a significant association with cervical-cancer risk in meta-analysis
VariantComparisonEthnicityNumber assessedCervical cancer riskHeterogeneityVenice criteria gradeaCumulative evidence of associationb
StudiesCasesControlsOR (95%CI)P-value I 2 P-value
  1. A, adenine; C, cytosine; G, guanine; OR, odds ratio; T, thymine.

  2. a

    Venice criteria grades are for amount of evidence, replication of the association, and protection from bias.

  3. b

    Cumulative epidemiological evidence as graded by Venice criteria as strong (+++), moderate (++), or weak (+) for association with breast-cancer risk.

  4. c

    Only data on White or Asian women were available for meta-analysis.

CTLA4rs231775AA vs (GG+GA)All ancestries6193622761.24 (1.04–1.48)0.0155.4%0.382AAA+++
A vs GAll ancestries6193622761.13 (1.03–1.25)0.010.0%0.446AAC+
CYP1A1MspIm2 m2 vs (m1 m1 + m1 m2)All ancestries56507122.02 (1.00–4.06)0.04961.0%0.036BCA+
(m2 m2 + m1 m2) vs m1 m1All ancestries56507122.29 (1.24–4.23)0.00883.3%0.000BCA+
m2 vs m1All ancestries56507121.98 (1.18–3.32)0.0186.6%0.000BCA+
CYP2E1Ile/Val(Val/Val+Ile/Val) vs Ile/IleAsianc43985882.60 (1.41–4.81)0.00276.2%0.006BCA+
Val vs IleAsianc43985882.08 (1.31–3.30)0.00274.4%0.008BCA+
GSTM1DeletionNull vs presentAll ancestries19200324541.44 (1.18–1.75)0.00059.5%0.000ACA+
GSTT1DeletionNull vs presentAll ancestries15159120381.39 (1.06–1.82)0.01867.0%0.000ACA+
IFN-γrs2430561AA vs (TT+TA)All ancestries3153210890.57 (0.33–0.98)0.04380.9%0.005ACA+
A vs TAll ancestries3153210890.77 (0.59–0.99)0.03958.8%0.088ACA+
AA vs TTAll ancestries3153210890.70 (0.56–0.89)0.0030.0%0.381AAA+++
IL-1βC-511T(TT+TC) vs CCAsianc37369301.69 (1.30–2.18)0.00014.9%0.309BAA++
TT vs CCAsianc37369301.64 (1.19–2.25)0.00218.6%0.293BAA++
TC vs CCAsianc37369301.69 (1.29–2.22)0.00042.7%0.174BBA++
IL-10592C/AA vs CAll ancestries4218311881.17 (1.04–1.33)0.01242.1%0.159ABA++
p53codon72Arg/Arg vs (Arg/Pro+Pro/Pro)All ancestries60604584961.25 (1.08–1.44)0.00167.6%0.000ACC+
Arg vs ProAll ancestries59599282101.12 (1.02–1.22)0.01559.1%0.000ABC+
Arg/Arg vs Pro/ProAll ancestries59599282101.20 (1.07–1.35)0.00232.5%0.01ABA++
TNFrs1800629(AA +GA) vs GGAll ancestries9232119831.63 (1.22–2.18)0.00162.7%0.006ACA+
A vs GAll ancestries8227918891.57 (1.21–2.04)0.00163.9%0.007ACC+
XRCC3rs861539(TT+TC) vs CCAsianc34694826.27 (2.59–15.19)0.00087.7%0.000BCA+
T vs CAsianc34694821.67 (1.24–2.24)0.00136.4%0.208BBA++
TC vs CCAsianc34694821.50 (1.04–2.14)0.02941.7%0.180BBA++
HLA-DQA10101Carriers vs non-carriersAll ancestries54116120.76 (0.60–0.97)0.03029.0%0.228BBA++
0201Carriers vs non-carriersAll ancestries54116120.76 (0.60–0.95)0.0190.0%0.606BAA++
HLA-DQB10301Carriers vs non-carriersAll ancestries13194424591.20 (1.08–1.34)0.00134.9%0.101ABC+
0303Carriers vs non-carriersAll ancestries10130321021.33 (1.10–1.60)0.00423.9%0.223AAA+++
0501Carriers vs non-carriersAll ancestries12194422110.81 (0.70–0.93)0.00318.8%0.259AAA+++
06Carriers vs non-carriersAll ancestries19276631651.29 (1.06–1.58)0.01274.7%0.000ACA+
0602Carriers vs non-carriersAll ancestries13198023331.39 (1.05–1.82)0.02069.7%0.000ACA+
0603Carriers vs non-carriersAll ancestries10179521370.59 (0.47–0.73)0.0000.0%0.535AAA+++
Figure 2.

Forest plot of ORs for cervical cancer in recessive model (Arg/Arg versus Arg/Pro + Pro/Pro) of p53 codon72 polymorphism.

Fifteen variants in eight genes were significantly associated with the risk of cervical cancer in the meta-analysis stratified by ethnicity (Table 2). Six common polymorphisms (CTLA-4 rs231775, CYP1A1 MspI, Fasl-844 T→C, GSTM1 deletion, TNF rs1800629 and XRCC1 rs25487) were associated with cervical cancer among Asian women but not among White British women. The HLA DQB1 0602 variant had an association with the risk of cervical cancer among White British women but not among Asian women. HLA DQB1 06 and 0601 had associations with the risk of cervical cancer among Asian and White British women, and p53 codon 72 polymorphism was associated with the risk of cervical cancer among White British and African women. HLA DQB1 0501 in White British women had a protective effect, and Other women carrying HLA DQB1 0302 had a decreased risk of cervical cancer. In addition, HLA DQB1 0603 had a protective effect in White British and Other women.

Table 2. Genetic variants with a significant association with cervical-cancer risk by ethnic group in meta-analysis
VariantComparisonEthnicityNumber assessedCervical cancer riskHeterogeneityVenice criteria gradeaCumulative evidence of associationb
StudiesCasesControlsOR (95%CI)P-value I 2 P-value
  1. A, adenine; C, cytosine; G, guanine; OR, odds ratio; T, thymine.

  2. a

    Venice criteria grades are for amount of evidence, replication of the association, and protection from bias.

  3. b

    Cumulative epidemiological evidence as graded by Venice criteria as strong (+++), moderate (++), or weak (+) for association with breast-cancer risk.

CTLA4rs231775AA vs GG/GAAsian4120415141.37 (1.07–1.76)0.0140.0%0.414AAA+++
AA/GA vs GGAsian4120415141.23 (1.05–1.44)0.0115.0%0.317AAA+++
A vs GAsian4120415141.21 (1.07–1.36)0.0020.0%0.642AAA+++
CYP1A1MspI(m2 m2 + m1 m2) vs m1 m1Asian44955571.84 (1.04–3.24)0.03572.6%0.012BCA+
m2 vs m1Asian44955571.66 (1.03–2.70)0.0477.6%0.004BCA+
Fasl844 T→CC vs TAsian377110911.48 (1.13–1.93)0.00464.3%0.061BCA+
CC vs TTAsian377110911.67 (1.10–2.54)0.01643.30.171BBA++
GSTM1DeletionNull vs presentAsian15159918871.54 (1.23–1.94)0.00062.0%0.001ACA+
p53codon72Arg/Arg vs (Arg/Pro+Pro/Pro)Caucasian15149222081.49 (1.13–1.97)0.00566.7%0.000ACA+
Arg/Arg vs (Arg/Pro+Pro/Pro)African56668541.38 (1.03–1.85)0.0320.0%0.495BAA++
Arg vs ProCaucasian15149222081.31 (1.05–1.64)0.01666.3%0.000ACA+
Arg/Arg vs Pro/ProCaucasian15149222081.45 (1.07–1.95)0.01537.5%0.071ABA++
Arg/Arg vs Pro/ProAfrican56668541.40 (1.02–1.92)0.03837.9%0.169BBA++
TNFrs1800629(AA +GA) vs GGAsian44745122.56 (1.81–3.61)0.0000.0%0.612BAA++
A vs GAsian44745122.15 (1.60–2.89)0.0000.0%0.861BAA++
XRCC1rs25487AA vs (GG+GA)Asian494314211.40 (1.02–1.92)0.03945.6%0.138BBA++
HLA-DQB10302Carriers vs non-carriersOther45367370.80 (0.64–1.00)0.05125.8%0.257BBA++
0501Carriers vs non-carriersCaucasian49167360.76 (0.61–0.95)0.01522.6%0.275BAA++
06Carriers vs non-carriersAsian890310261.57 (1.16–2.12)0.00367.6%0.003BCA+
06Carriers vs non-carriersCaucasian512069481.41 (1.01–1.96)0.04174.4%0.004BCA+
0601Carriers vs non-carriersAsian34074081.99 (1.51–2.63)0.00030.4%0.238BBA++
0602Carriers vs non-carriersCaucasian49167361.91 (1.13–3.23)0.01583.9%0.000BCA+
0603Carriers vs non-carriersCaucasian49167360.49 (0.36–0.66)0.0000.0%0.407BAA++
0603Carriers vs non-carriersOther35007130.69 (0.49–0.97)0.0330.0%0.445BAA++

Meta-analysis also detected no changes in the risk of cervical cancer for 39 genetic variants in 12 genes (data not shown). To provide solid evidence for a null association, we conducted a cumulative meta-analysis for the variants which had sufficient study sources and large sample sizes (the number of cases and controls was more than 1000 and the number of studies was more than five). For these, the results of the null association between HLA DQB1 02, 03, 04 and 05 variants and the risk of cervical cancer were stable. However, we could not exclude an association between other variants, such as CCND1 rs9344, FAS 670, IL-10 1082A/G, MTHFR rs1801133, p21 codon31, TNF 238, HLA DQB1 0402, 0503 and 0604, and the risk of cervical cancer because of the unstable results of cumulative meta-analysis.

Publication bias was assessed by funnel plots and modified Egger tests, and three variants showed possible publication bias (HLA DQB1 0301, dominant association for p53 codon 72, and allelic association for TNF rs1800629).

In the assessment of the epidemiological evidence for the 23 significant associations with the risk of cervical cancer ( 0.05) (Tables 1 and 2), an A grade was given to 13, 7 and 21 variants, respectively, for the amount of evidence, replication of the association, and protection from bias, a B grade to 9, 8 and 0 variants, respectively, and a C grade to five variants for protection from bias [mainly because of a small effect with cervical cancer risk (OR < 1.15, = 2) and possible publication bias (= 3). No variant received a C grade for amount of evidence and eight variants received a C for replication of the association. Finally, five variants received A grades for all three criteria and can therefore be regarded as having strong evidence, including CTLA4 rs231775, IFN-G rs2430561, HLA-DQB1 0301, 0501 and 0603. Twenty-one associations received a grade of either A or B for the three criteria and were therefore scored as having moderate evidence, including IL-1B C-511T, XRCC3 rs861539, HLA-DQA1 0101, 0201, a recessive effect among Asian women for TNF rs1800629 and so on. The remaining variants were scored as having weak evidence based on the Venice criteria.

However, we did not perform meta-analysis for some of the other variants because of fewer than three data sources for AKNA rs3748178,[24] APE1,[25], CASP8,[26], CDC6 G1321A,[27] EGF +61 A/G,[28] HER2,[29] IL-4,[30] MDM2,[31] NAT2,[32] REV1,[33] TGF-B 509,[34] VEGF[35] and other variants.


Main findings

In our comprehensive research synopsis and meta-analysis, 23 variants were significantly associated with the risk of cervical cancer. Additionally, the null association with the risk of cervical cancer for four variants was confirmed by cumulative meta-analysis. In addition, some variants were not included in the meta-analysis because of fewer than three data sources. Moreover, we evaluated the epidemiological credibility of all significant associations identified by the meta-analysis. Finally, five variants were regarded as having strong evidence and 21 associations were scored as having moderate evidence.

Strengths and limitations

To the best of our knowledge, this is the first time that an association with the risk of cervical cancer has been identified by meta-analysis for most of the variants mentioned in our study, including CCR2, CD28, CTLA-4, CYP1A1, CYP2D6, CYP2E1, FAS, GSTP1, HIF, IL-1, IL-6, IL-10, MS, MMP, MTHFR, XRCC1 and XRCC3. More importantly, our work has provided robust evidence for personalised clinical assessment of the risk of cervical cancer and cervical cancer genetic risk assessment.

Our study has a few limitations. Although a systemic literature search was performed, it is likely that some publications were overlooked. However, these reports would probably not have altered the overall results of our large-scale meta-analysis. We used genotype counts and crude estimates of effect rather than adjusted estimates of association, but we did not assess gene–gene or gene–environment interactions. In addition, heterogeneity was common in the 58 meta-analyses in this study, and although we could remove some of this with subgroup analysis, other sources of heterogeneity, such as the type of cancer or clinical stage of tumour, were not examined. Moreover, any assessment of cumulative evidence is only temporary and needs to be continuously refined as new data are gathered.[36] Finally, publication bias could be of concern, because small studies with null results tend not to be published.


Our study found that 14 variants in 11 genes or loci could increase the risk of cervical cancer. These genes are related to immunosurveillance, immunoregulation, DNA repair, cell metabolism and other mechanisms. CTLA-4 is a vital negative regulator of T-cell activation and proliferation. Activated T cells have the function of immune surveillance in carcinogenesis, and the CD28–CTLA-4–B7 pathway is the most important co-stimulatory means of T-cell activation and proliferation. The 49G>A polymorphism leading sequence caused an Ala17 to Thr17 amino acid substitution.[37] Biochemical analyses showed that CTLA-4 Thr17 had a higher ability to bind B7 and a stronger inhibitory effect on T-cell activation than with Ala17. T cells carrying the 49AA genotype had significantly lower activation and proliferation rates than T cells carrying the 49GG genotype upon stimulation. Several functional studies found that the A allele in TNF 308 induced higher levels of TNFα secretion.[38] In vivo studies showed that tumour necrosis factor (TNF)α promotes immortalisation of HPV and malignant changes in cervical epithelial cells, mediating the escape of HPV-infected cells from the host's immune surveillance mechanism.

Since chemical carcinogens generally require activation to electrophilic reactive forms to produce DNA adducts, individuals with increased metabolic activity are at a higher risk of developing cancer.[39] Many carcinogens require metabolic activation by phase I enzymes such as CYP1A1 that convert environmental procarcinogens to reactive intermediates with carcinogenic effects.[40] These active metabolites are then detoxified by phase II enzymes. The glutathione S-transferases (GSTs) are the most important members of the phase II superfamily of metabolic enzymes and have a crucial function in the detoxification of a variety of both endogenous products of oxidative stress and exogenous carcinogens.[41, 42] Both GSTM1 and GSTT1 exhibit an inherited homozygous deletion polymorphism (null genotype) that is associated with an absence of enzyme activity. Hence, individuals with high phase I enzyme activity and low phase II enzyme activity would theoretically produce higher levels of active metabolites and consequently more DNA damage.

Cytokines play an important role in defence against viral infection. Interleukin (IL)-1 is an important part of the innate immune system,[43] and IL-10 has a suppressive effect on cell-mediated immunity.[44, 45] It is well known that DNA repair is a very important mechanism for maintaining genetic stability and protection against cancer initiation, and defects in the DNA repair mechanism are associated with many diseases, including cancer. The variant Gln allele in XRCC1 was thought to reduce DNA repair activity and hence lead to increased DNA damage.[46] XRCC3 encodes a protein involved in the homologous recombinational repair pathway of double-stranded DNA repair, and plays a key role in preventing mutations, chromosomal instability and cancer.[47]

HLAs play a pivotal role in presenting foreign antigens to immune cells that are responsible for the clearance of virus-infected cells and tumour cells. In our study, we found that the HLA DQA1 0101, 0201 alleles and HLA DQB1 0501 and 0603 could decrease the risk of cervical cancer, whereas 0301, 0303, 06 and 0602 alleles in HLA-DQB1 could increase the risk. Further studies are needed to focus on HLA alleles, and these could have a positive influence on the development of protective and therapeutic vaccines.


We have carried out a detailed research review and meta-analysis of genetic variants that have a possible association with the risk of cervical cancer: significant associations were found for 23 variants. To the best of our knowledge, our study is the largest and most comprehensive assessment so far of the literature on the genetic basis of susceptibility to cervical cancer. Meta-analyses are important for evaluating the consistency of study findings because they assess cumulative evidence from previous studies and provide pointers for future studies; however, further research is needed to assess gene–gene and gene–environment interactions.

Disclosure of interest


Contribution to authorship

XZ and ZW formulated the idea for conducting a systematic review, did the meta-analysis, and wrote the initial draft; XZ, LZ and LY performed literature searches, study selection, data extraction and risk of bias assessment; CT solved disagreements regarding study selection and risk of bias assessment, and critically revised the article.

Details of ethics approval

Ethics approval was not required for this research.