Introduction

Cervical cancer is one of the leading cancers in women worldwide, with529,800 new-diagnosed cancer cases and 275,100 cancer deaths in 20081. More than 85% of these cases and deaths occur in developing countries,including China1. Accumulated molecular epidemiologic datasupport the hypothesis that persistent infection with oncogenic high-risktypes of human papillomavirus (HPV) is the primary, even necessary cause ofcervical cancer1,2. However, only a small fraction of womenwith HPV infection eventually develop cervical cancer, suggesting a wide rangeof inter-individual genetic variability in cervical cancer susceptibility3. Recently, two genome-wide association studies showed that somesingle nucleotide polymorphisms (SNPs) in the major histocompatibility complexregion were associated with cervical cancer risk in both Caucasian and ChineseHan populations4,5. Despite these successes in identifyinggenetic variants for cervical cancer risk, the causal variants and/or mechanismsunderlying the etiology have been determined for only a small fraction ofthese associations6. Recently, investigations of potentiallyfunctional SNPs have now been increasingly advocated across diseases. Forexample, SNPs at microRNA (miRNA)-binding sites in the 3′-untranslatedregion (UTR) can remarkably alter the biogenesis and/or function of the correspondingmiRNAs and thus contribute to cervical carcinogenesis7.

Caspases, at the heart of the apoptotic machinery, encode an evolutionaryconserved family of cysteine-aspartic acid proteases and coordinate in cellularregulation and execution of apoptosis8. Together with caspase-3and -6, caspase-7 belongs to the subgroup of executioner caspases9,and it executes a coordinated program of proteolysis that leads to the finalprogrammed cell death10. Besides its activation during apoptosis,proteolytic maturation of caspase-7 has also been observed in inflammatoryconditions11, which indicates a potential mechanism of caspase-7involving the process of HPV infection and host immune response in cervicalcancer. Previous genetic association studies had revealed that polymorphismsin the Caspase-7 (CASP7) gene may modulate the default programmedcell death, thus leading to genomic instability and contributing to inter-individualvariation in cancer susceptibility12,13.

To date, no published studies have investigated associations between functional CASP7SNPs and cervical cancer risk, besides genome-wide association studies. Herein,we performed a relatively large case-control study to test the hypothesisthat potentially functional SNPs in the CASP7 3′-UTR are independentlyand/or jointly associated with cervical cancer risk.

Results

Population characteristics

The selected characteristics of the study subjects are listed in SupplementaryTable S1. There was no significant difference in distributions of agebetween the 1,486 cases and 1,301 controls (P = 0.126) as result ofmatching. However, the differences in age at primiparity, menopausal statusand body mass index (BMI) were significant between cases and controls. Therefore,we subsequently adjusted these variables for any residual confounding effectin multivariate logistic regression analyses.

Association of CASP7 SNPs with cervical cancer risk

As shown in Table 1, compared with CC/CT genotypes,the rs4353229TT genotype was associated with a significantly increased riskof cervical cancer [crude odds ratio (OR) = 1.20, 95% confidence interval(CI) = 1.02–1.40], but after adjustment for age, age at primiparity,menopausal status and BMI, this association was no longer statistically significant.In addition, the rs10787498GT genotype was associated with an increased riskof cervical cancer, compared with the TT genotype (adjusted OR = 1.19, 95%CI = 1.00–1.41). No risk association was observed for the other twoSNPs (i.e., rs12247479 and rs1127687), nor for the haplotypes of these four CASP7SNPs (Supplementary Table S2). However, when combiningthese four SNPs and assuming a dominant genetic model, we found that thosewomen who carried four putative risk genotypes had a 1.54-fold increased risk(95% CI = 1.07–2.22) of cervical cancer, compared with those who carriedthree or less putative risk genotypes (Table 1). Furtherstratified analyses showed that the significantly increased risk of cervicalcancer associated with the rs10787498 GT/GG genotype was more prominent inwomen younger at primiparity (adjusted OR = 1.40, 95% CI = 1.08–1.82, Pfor homogeneity test = 0.014; Supplementary Table S3).

Table 1 Associations of CASP7genotypes with the risk of cervical cancer
Full size table

In further logistic regression analyses, we observed a significant locus-locusmultiplicative interaction between rs1127687 and rs12247479 as well as betweenrs1127687 and rs10787498 (P = 0.016 and 0.007, respectively; data notshown). We then explored their two-locus joint effects. As shown in Table 2, women who carried rs1127687AG/AA-rs12247479AG/AAgenotypes and those who carried rs1127687AG/AA-rs10787498GT/GG genotypes hada significantly increased risk of cervical cancer, compared with carriersof rs1127687GG-rs12247479GG and rs1127687GG-rs10787498TT, respectively (adjustedOR = 1.48 and 1.52, 95% CI = 1.02–2.15 and 1.13–2.06, Pfor homogeneity = 0.016 and 0.007; respectively). Although no difference inrisk estimates by the homogeneity tests, we observed a highly associated riskfor the presence of rs10787498GT/GG and rs4353229TT genotypes (adjusted OR= 1.27, 95% CI = 1.04–1.55) as well as a board-line significance forthe joint effect of rs12247479AG/AA with rs4353229TT genotypes and for thatof rs10787498G with/without rs12247479A allele (adjusted OR = 1.23 and 1.15,95% CI = 0.97–1.55 and 0.98–1.34; respectively).

Table 2 Locus-locus jointeffects in associations between CASP7 genotypes and cervical cancerrisk
Full size table

We then calculated false-positive report probability (FPRP) values forall observed significant associations (Table 3). Thers4353229TT genotype was associated with an increased risk of cervical cancerwith a statistical power of 99.8%, compared with CT/CC genotypes. When theassumption of prior probability was 0.1, the association with rs4353229 wasnoteworthy in all patients and in the subgroup of postmenopausal women (FPRP= 0.190 and 0.191, respectively), similar for the association of rs10787498GT/GGgenotypes in the subgroup of younger at primiparity (FPRP = 0.043) as wellas for that of the four putative risk genotype combination effect (FPRP =0.162). Meanwhile, the two-locus joint effect was still noteworthy for rs10787498-rs4353229and rs1127687-rs10787498 (FPRP = 0.020 and 0.024, respectively).

Table 3 False-positive reportprobability values for associations between CASP7 genotypes and cervicalcancer risk
Full size table

Association of high-order interactions with cervical cancer risk

We further performed the multifactor dimensionality reduction (MDR) analysisand found that rs4353229 was the best one-factor model with the highest cross-validationconsistency (CVC) (89%) and the lowest prediction error (48.0%) among allfour SNPs. Additionally, the four-locus model had a maximal CVC (100%) anda minimal prediction error (46.8%), suggesting a better prediction than othermodels (Figure 1A). Subsequent hierarchical clusteranalysis placed rs4353229 and rs10787498, rs12247479 and rs1127687 on thesame branch (Figure 1B), suggesting that this four-locusmodel might have an interaction effect by modulating cervical cancer risk,which is also supported by the interaction graph (Figure 1C).Moreover, consistent with the findings in the single locus analysis, rs4353229(0.13%) and rs10787498 (0.12%) showed a strong effect on cervical cancer risk(Figure 1C).

Figure 1
figure 1

High-order interaction analyses for the four CASP7 SNPs.

(A) The best multifactor dimensionality reduction interactionmodels. The multi-locus model with maximum cross-validation consistency andminimum prediction error rate is indicated in bold. (B) Interaction dendrogram.The color indicates the strength of the dependence: green is weak and redis strong. (C) Interaction entropy graph. Each SNP is shown in a box withthe percent of entropy (main effect). Two-way interactions between SNPs aredepicted as an arrow accompanied by a percent of entropy (interaction effect).In the interaction graph, rs4353229 alone eliminates 0.13% of class entropyand has the largest univariate effect. Only small percentages of entropy wereexplained by rs12247479 (0.04%) or rs1127687 (0.03%) when considered independently,while a large percentage of entropy was explained by their pairwise interactions(0.11%), indicating a synergistic interaction. (D) Classification and regressiontree. Terminal nodes are thick bordered. W, wild type genotype; V, variantgenotype; TN, terminal node; #, P value <0.05.

Full size image

By the classification and regression tree (CART) analysis, we found rs4353229to be the initial split of root nodes, indicating that rs4353229 was the strongestrisk factor for cervical cancer among these four SNPs. Further inspectionof the tree structure revealed distinct interaction patterns. Women carryingrs4353229TT, rs12247479GG, rs10787498GT/GG and rs1127687AG/AA genotypes [terminalnode (TN) 6] had a 1.67-fold increased risk of cervical cancer, comparedwith the TN3 group at the lowest risk (P = 0.041; Figure1D).

Correlation between CASP7 genotypes and mRNA expression levels

In 270 HapMap individuals whose mRNA expression data were available, althoughthere was no correlation of CASP7 mRNA expression levels with the riskloci, we did observe a board-line significant correlation of CASP7mRNA expression levels with the joint effect of rs10787498 and rs12247479 [generalizedlinear model (GLM), P = 0.056; Figure 2G].Moreover, CASP7 mRNA expression levels showed an increased trend forrs10787498TT-rs12247479GG carriers and a decreased trend for rs1127687AG/AA-rs10787498GT/GGcarriers (Student's t test, P = 0.018 and 0.064, respectively; Figure 2G, 2J).

Figure 2
figure 2

The relative expression levels of CASP7 mRNA by different genotypesin 270 HapMap subjects.

(A) rs4353229, (B) rs10787498, (C) rs12247479, (D) rs1127687,as well as the joint effects of (E) rs12247479 with rs4353229, (F) rs10787498with rs4353229, (G) rs10787498 with rs12247479, (H) rs1127687 with rs4353229,(I) rs1127687 with rs12247479 and (J) rs1127687 with rs10787498 are evaluatedby generalized linear models and Student's t tests.

Full size image

Discussion

In this case-control study of 1,486 cervical cancer cases and 1,301 femalecontrols, we found that the rs4353229TT genotype was associated with an increasedrisk of cervical cancer with a statistical power of 99.8%. Moreover, we alsoobserved significant joint effects and locus-locus interactions of the CASP7SNPs on cervical cancer risk. This is, to the best of our knowledge, the firstreport that describes the associations between potentially functional SNPsin CASP7 and cervical cancer risk. Our study is also among the fewthat have examined the locus-locus interaction in the etiology of cervicalcancer.

CASP7, located at chromosome 10q25, encodes a member of cysteinepeptidase and has been identified as one of the three downstream effectorsin the apoptosis pathway in mammalian cells9, involved in theexecution-phase process of cellular apoptosis. Previous data demonstratedthat genetic variations in apoptosis genes might modulate the programmed celldeath in various biological systems and alter tissue response to irradiationand cytotoxic chemotherapy14, thus eventually leading to genomicinstability and tumorigenesis in humans15. It is of note thatthe resistance to apoptosis is an important indicator related to cervicalcarcinogenesis16. In cervical cancer cells, the lack of caspase-mediatedapoptosis due to unresponsiveness to pro-apoptotic stimuli causes uncontrolledcell proliferation17.

Recently, Wang et al. reported that the rs4353229TT genotype was associatedwith 0.83-fold decreased risk of gastric cancer13. Inversely,in the current study, we found a possibly increased risk of cervical cancerfor the rs4353229TT genotype. This discrepancy might be partly due to tumorspecificity and population stratification. On the other hand, we also observedthat this risk association might be modified by environmental variables andthat the effect of one single CASP7 locus on cervical cancer risk mightbe weak. Indeed, for cancer biology, the functional characterization of riskloci as well as the complex interplay among multiple loci in many cancersposes a particular exciting challenge for the era of post genome-wide associationstudy.

In the present study, we did find that potentially functional SNPs at CASP73′-UTR might be jointly associated with cervical cancer risk. Furthergenotype-phenotype analyses suggested an association of CASP7 mRNAexpression levels with the joint effect between rs10787498 and rs12247479as well as between rs1127687 and rs10787498. Consistently, the locus-locusjoint effect association analyses demonstrated that there was a super-multiplicativejoint effect between rs1127687 and rs10787498 as well as possibly betweenrs10787498 and rs12247479 on cervical cancer risk. These findings indicatedthat CASP7 SNPs might interact to modify cervical cancer risk by affecting CASP7mRNA expression. Subsequent high-order interaction analyses also helped toexplain this paradigm. The best interaction model revealed that the four CASP7SNPs interacted with a maximal CVC and a minimal prediction error, which wasmore evident in the interaction entropy analysis. Additionally, the CART analysisidentified subsets of individuals with cervical cancer risk based on variouscombinations of genotypes and the OR for individuals in each TN ranged from1.11 to 1.67, which also suggests a synergistic interaction between thesefour SNPs.

Despite the strengths and biologic plausibility of the associations observedin the current study, several limitations need to be addressed. Firstly, theremay be selection and information bias originated from a retrospective studydesign, which may have been minimized by frequency-matching for cases andcontrols as well as the adjustment for potential confounding factors in multivariateanalyses. Secondly, the P value of Hardy-Weinberg equilibrium (HWE)was 0.034 for rs10787498, but given that the deviation from HWE among controlswas defined as a significance level of α <10−3or 10−4, all the SNPs in our analyses were in agreementwith HWE. Finally, because the lack of routine HPV screening for all casesand controls in our hospital, we could not evaluate HPV infection as the potentialconfounder in risk estimates of cervical cancer.

In summary, in the current case-control study of 1,486 cases and 1,301controls, we found that CASP7 SNPs might be associated with cervicalcancer risk in Eastern Chinese women. There were substantial joint effectsand locus-locus interactions among these SNPs and such effects may contributeto cervical cancer risk by affecting CASP7 mRNA expression. However,well-designed, larger and prospective studies with detailed information aboutHPV infection are warranted to validate our findings.

Methods

Study subjects

The recruitment of the cases and controls was partly described previously7. Briefly, all subjects were unrelated ethnic Han Chinese and residentsin the Eastern China. The 1,486 newly diagnosed and histopathologically confirmedprimary cervical cancer patients were consecutively recruited and collectedby the tissue bank of Fudan University Shanghai Cancer Center (FUSCC). The1,301 frequency-matched healthy controls without history of cancers were recruitedfrom women who had come to FUSCC for breast cancer screening. After a writteninformed consent was obtained, all subjects were interviewed to collect theirdemographic and risk factor information. Because most Chinese women did notsmoke cigarettes or drink alcohol, all participants included in the analysiswere non-smokers and non-drinkers and provided a one-time 10 mL ofvenous blood sample (after diagnosis and before the initiation of treatmentfor cases). The experimental and research protocols were approved by the InstitutionalReview Board of FUSCC and all experiment on humans was performed in accordancewith relevant guidelines and regulations.

SNP selection and genotyping

By searching the NCBI dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP) and the International HapMap Project database (http://hapmap.ncbi.nlm.nih.gov/),we found that there were 22 SNPs in CASP7 3′-UTR, of which fourwere finally selected for genotyping, based on the following criteria: 1)minor allele frequency of at least 5% in Chinese populations, 2) with lowlinkage disequilibrium by using an r2 threshold of <0.8for each other, 3) predicted to be potentially functional by the SNP functionprediction platform (http://snpinfo.niehs.nih.gov/snpinfo/snpfunc.htm),and 4) not included and published in genome-wide association studies. Thus,the selected SNPs were rs4353229 T> C, rs12247479 G> A, rs10787498 T>G and rs1127687 G> A. Genomic DNA extraction and genotyping were conductedas described previously18. As a result, the discrepancy ratein all positive controls (i.e., duplicated samples, overlapping samples fromprevious studies and samples randomly selected to be sequenced) was less than0.1%.

Genotype-phenotype correlation analysis

To evaluate biological plausibility of our findings, we used the data on CASP7genotypes and CASP7 mRNA expression levels both available for 270 HapMapsubjects by SNPexp online tool (http://app3.titan.uio.no/biotools/help.php?app=snpexp) and conducted genotype-phenotype correlation analysis as described previously18,19.

Statistical analysis

HWE was tested by χ2-test for each SNP. We performedthe Pearson's χ2-test for the differences in selectedvariables between cases and controls. The association of CASP7 genotypeswith cervical cancer risk was estimated by computing ORs and their 95% CIsfrom both univariate and multivariate logistic regression models. We alsoevaluated the associations in subgroup and joint effect analyses. The PROCHAPLOTYPE procedure in SAS software was applied to infer haplotype frequenciesamong the four SNPs. To avoid false positive associations in this study, wecalculated the FPRP with the assumption of different prior probabilities (0.0001,0.001, 0.01, 0.1 and 0.25). FPRP values <0.2 were considered to be noteworthy20. We used GLM for the genotype-phenotype correlation and usedstudent's t test and analysis of variance test to evaluate thedifferences in the relative mRNA expression levels among different genotypegroups.

The MDR and CART analyses were conducted by the MDR V2.0 beta 8.2 program(http://www.multifactordimensionalityreduction.org/)and SAS software (version 9.1; SAS Institute, Cary, NC), respectively, asdescribed previously21. Briefly, we enrolled the four riskloci in the MDR analysis to identify the best n-factor interactionmodel. Then, we performed the interaction dendrograms and graphs22.The color of branches and lines is referred to the type of interaction, green-to-yellow-to-redindicates a weak-to-strong interaction. CART creates a decision tree thatdepicts how well each genotype predicts disease and ends up with TNs.

All statistical analyses were performed with SAS 9.1 software (SAS Institute,Cary, NC), unless stated otherwise. All P values were two-sided witha significance level of P <0.05.