Genetic variants in telomere‐maintenance genes are associated with ovarian cancer risk and outcome

Abstract Most ovarian cancer patients present at an advanced stage with poor prognosis. Telomeres play a critical role in protecting chromosomes stability. The associations of genetic variants in telomere maintenance genes and ovarian cancer risk and outcome are unclear. We genotyped 137 single nucleotide polymorphisms (SNPs) in telomere‐maintenance genes in 417 ovarian cancer cases and 417 matched healthy controls to evaluate their associations with cancer risk, survival and therapeutic response. False discovery rate Q‐value was calculated to account for multiple testing. Eleven SNPs from two genes showed nominally significant associations with the risks of ovarian cancer. The most significant SNP was TEP1: rs2228026 with participants carrying at least one variant allele exhibiting a 3.28‐fold (95% CI: 1.72‐6.29; P < 0.001, Q = 0.028) increased ovarian cancer risk, which remained significant after multiple testing adjusting. There was also suggested evidence for the associations of SNPs with outcome, although none of the associations had a Q < 0.05. Seven SNPs from two genes showed associations with ovarian cancer survival (P < 0.05). The strongest association was found in TNKS gene (rs10093972, hazard ratio = 1.88; 95% CI: 1.20‐2.92; P = 0.006, Q = 0.076). Five SNPs from four genes showed suggestive associations with therapeutic response (P < 0.05). In a survival tree analysis, TEP1:rs10143407 was the primary factor contributing to overall survival. Unfavourable genotype analysis showed a cumulative effect of significant SNPs on ovarian cancer risk, survival and therapeutic response. Genetic variations in telomere‐maintenance genes may be associated with ovarian cancer risk and outcome.


Introduction
Ovarian cancer is the most frequent cause of cancer-related death among gynaecological malignancies. In 2015, the estimated new cases were 21,290 in the United States, and the estimated deaths were 14,180 [1]. Non-Hispanic Whites have the highest incidence rate of ovarian cancer in the U.S. Current surveillance strategy, by transvaginal ultrasound and serum tumour marker cancer antigen 125 (CA 125 ), is ineffective in detecting ovarian cancer at an early stage [2,3]. As a result of the absence of obvious clinical symptoms and sensitive screening tests, most ovarian cancer patients (61%) are diagnosed at advanced stages. The 5-year relative survival rates of ovarian cancer patients with local, regional and distant stages are 92.3%, 71.7% and 27.4% respectively [4].
The aetiology of ovarian cancer remains poorly understood. Many factors are thought to be associated with ovarian cancer, including smoking, infertility, endometriosis, oestrogen use for menopause hormone therapy, family ovarian cancer history, Lynch syndrome and mutations in BRCA1 or BRCA2 genes [5][6][7]. Genome-wide association studies (GWAS) have identified a number of susceptibility loci for ovarian risk and clinical outcome [8][9][10][11][12][13]. Previous candidate gene studies also reported nucleotide excision repair pathway, microRNA biosynthesis pathway, transforming growth factor-b pathway and fibroblast growth factor pathway genetic variants may be associated with ovarian cancer risk or clinical outcome [14][15][16][17].
Telomeres are nucleoprotein complexes at the ends of chromosomes and consist of short repetitive sequences (TTAGGG in humans) and a set of specialized proteins [18]. Telomeres play a critical role in protecting chromosomes from degradation, end-to-end fusion, abnormal recombination and other detrimental chromosomal events. In normal somatic cells, telomeres are progressively eroded by  bp during each mitotic cell division [19]. Many proteins are involved in regulating telomere functions. When telomere lengths become critically short, the process of cell senescence is initiated, resulting in cell-cycle arrest or apoptosis in normal cells [20]. Telomerase is activated in the majority of cancer cells that compensates telomere erosion and gives cancer cells growth advantage [21]. Telomere shortening and telomerase activation are hallmarks of human cancers. Higher telomerase activity has been observed in poorly differentiated ovarian tumours tissue [22,23]. Several studies have suggested that leucocyte telomere length is associated with ovarian cancer risks [24][25][26]. Genome-wide association studies and candidate gene study have shown that single nucleotide polymorphisms (SNPs) in human telomerase reverse transcriptase (hTERT) gene were associated with the risk of ovarian cancer [27][28][29][30]. There were scarce studies evaluating the associations of hTERT and other telomere-maintenance genes with ovarian cancer outcome. One study of 40 tagging SNPs from five telomere-maintenance genes showed no associations between these SNPs and ovarian cancer survival, but there were some suggestive associations in subgroup analyses [31].
We hypothesize that common SNPs in telomere-maintenance genes are significantly associated with ovarian cancer risk, survival and therapeutic response. We used a case-control study to test our hypothesis.

Study population
Patients (n = 417) with pathologically confirmed ovarian cancer were recruited from the University of Texas MD Anderson Cancer Center from 1998 to 2011. All case participants were newly diagnosed, histologically confirmed ovarian cancer and previously untreated before enrolment. There were no age, ethnicity or cancer stage restrictions on recruitment. Healthy control participants (n = 417) were recruited from Kelsey-Seybold Clinic, a large multi-specialty physician group in Houston metropolitan area. Controls without cancer history other than non-melanoma skin cancer were recruited during the same time period as the cases, and were matched to cases on age (AE5 year) and ethnicity. All study participants signed written informed consent before participation. The study was approved by the institutional review boards of MD Anderson and Kelsey-Seybold Clinic. Informed consents were obtained from all participants. Epidemiologic data including demographics, tobacco use history, bw and height, history of cancer, and medical history were collected for all cases and controls. Information on vital status was obtained from the medical records and the Social Security Death Index. For each participant, a blood sample was drawn into coded heparinized tubes for lymphocyte isolation and DNA extraction.

SNP selection and genotyping
The details of SNP selection and array construction were described in our previous publication [32]. Briefly, selected tagging SNPs have an r 2 threshold of 0.8 and minor allele frequency (MAF) greater than 0.05 in Caucasians. For each gene with a high priority score, we identified the tagging SNPs ranging from 10 kb upstream of transcriptional start site to 10 kb downstream of translational end site [33]. We also identified potentially functional SNPs, which are located in the functional region of genes, including coding SNPs (synonymous SNPs and non-synonymous SNPs) and regulatory [promoter, splicing site, 5 0 untranslated region (5 0 UTR) and 3 0 UTR] regions. A complete set of SNPs was sent to Illumina technical support for custom iSelect, Infinium II BeadChip design using a proprietary program developed by Illumina. A total of 145 SNPs from 11 telomere-maintenance genes were identified. The number of SNPs for each gene regions was as follows: PINX1, 27; PTOP, 4; POT1, 7; TEP1, 46; TERF1, 5; TERF2, 4; TERF2IP, 4; TERT, 15; TNKS, 21; TNKS1BP1, 6; and TNKS2, 6. Genomic DNA was extracted from peripheral blood lymphocytes using QIAmp DNA extraction kit (Qiagen, Hilden, Germany) and genotyped according to the standard protocol for Illumina's Infinium iSelect HD custom Genotyping Beadchip provided by Illumina (San Diego, CA, USA). The genotypes were auto-called using the BeadStudio software. All the laboratory personnel performing the experiments described above were blinded to the case-control and outcome status of the DNA samples. All the laboratory personnel performing genotyping were blinded to the case-control and outcome statuses.

Statistical analysis
Statistical analysis was performed using STATA software (version 10; STATA Corporation, College Station, TX, USA). The difference between participant groups with regard to categorical variables was compared by either Pearson's chi-squared test or Fisher's exact test. Student's t-test was used to assess continuous variables. Among the control participants, goodness-of-fit chi-squared analysis was used to test Hardy-Weinberg equilibrium to each SNPs. Unconditional logistic regression was used to estimate the odds ratio (OR) and 95% confidence interval (CI), adjusting for age, smoking status and body mass index (BMI). Three genetic models (dominant, recessive and additive) were tested for each SNPs and the model with the highest significance was used to determine the statistical significance of each SNP [34]. Overall survival (OS) was calculated from the date of diagnosis to the date of death or the end of patient follow-up. The effects of SNPs on ovarian cancer survival were estimated as hazard ratios (HR) and 95% confidence intervals (95% CI) using multivariate Cox proportional hazards regression analysis. The co-variants included were age, histology, clinical stage, tumour grade and treatment information. Kaplan-Meier curves and logrank tests were used to compare the OS differences by different genotypes. Higher order gene-gene interactions were explored using the Classification and Regression Tree analysis, performed using HelixTree Software (Golden Helix, Bozeman, MT, USA). Survival tree analysis was performed using the STREE program (http://masal.med.yale.edu/stree/), which also uses recursive partitioning method. Platinum-based therapeutic response was defined by whether there was evidence of residual disease as determined by various clinical measures, such as positron emission tomography and computed tomography scans, second-look surgery and post-chemotherapy CA 125 level. For response to therapy, unconditional multivariate logistic regression analysis was used while adjusting for age, histology, clinical stage, tumour grade and therapeutic information. Cumulative effects of multiple unfavourable genotypes were evaluated by counting the number of unfavourable genotypes from SNPs identified from in the main analysis (P < 0.05). According to the tertile distribution, the unfavourable genotypes were collapsed into high,  medium and low groups. All P-values reported were two-sided. P < 0.05 was considered statistically significant. As an adjustment for multiple testing, false discovery rate (FDR) based Q-value was calculated for each SNP using the R-package [35]. As previously suggested, we reported all those SNPs with Q < 0.20 to account for multiple testing while balancing the discovery nature of the study [36].

Characteristics of the study population
Details regarding participant recruitment and participant characteristics have been described in previous publications (Table S1) [14]. Briefly, a total of 417 case participants and 417 control participants were included. The cases and controls were matched on age (mean AE S.D., 60.7 AE 10. SNPs in the telomere-maintenance genes associated with overall ovarian cancer risk, survival and therapeutic response Among the genotyped 145 SNPs, 11 SNPs from two genes (TEP1 and TERT) showed significant associations with overall risk of ovarian cancer (P < 0.05 and Q < 0.10; Table 1). The most significant SNP was TEP1: rs2228026 with participants carrying at least one variant C allele exhibiting a 3.28-fold (95% CI: 1.72-6.29; P < 0.001, Q = 0.028) increased ovarian cancer risk.
For response to platinum-based adjuvant chemotherapy, four SNPs from three genes showed significant association (P < 0.05, Q < 0.20; Table 3). PINX1: rs7826180 displayed the greatest risk for poor treatment response. The variant A allele of PINX1: rs7826180 was associated with a 6.77-fold increased risk of poor response to chemotherapy (95% CI: 1.68-27.27; P = 0.007).
We also performed Kaplan-Meier curves and log-rank tests to compare OS differences in patients with different unfavourable genotypes. The results showed a trend towards decreased survival with increasing number of unfavourable genotypes. The MST for patients with 2-4 unfavourable genotypes was 26.7 months compared to 62.8 months for those without unfavourable genotype (P = 7.94 9 10 À6 , log-rank test; Fig. 1).
We also performed a survival tree analysis for these seven variants (Fig. 2). The first split on the survival tree was TEP1: rs10143407, indicating that this SNP is the primary factor contributing to OS. When we used individuals of terminal node 2 as reference, the HRs for the other three terminal nodes ranged from 1.95 to 6.97 ( Fig. 2A). Classifying these terminal nodes into three groups (low, medium and high), the MST for patients in the low-risk, medium-risk and high-risk groups were 128.9, 56.6 and 25.1 months respectively (P = 1.66 9 10 À6 , log-rank test; Fig. 2B).

Discussion
In this study, we took a pathway-based approach to comprehensively investigate the associations of genetic variants in telomere-maintenance genes with ovarian cancer risk and outcome. Eleven SNPs from two genes showed significant associations with overall ovarian cancer risk, 10 of which were located on TEP1 gene. The most significant SNP was TEP1: rs2228026, which remained significant after adjusting for multiple testing. TEP1: rs2228026 is a synonymous SNP located in the exon of TEP1. TEP1: rs2228041 and rs1713456 are missense SNPs. TEP1: rs2228041 changes Arginine to Glutamine at codon    1155 and rs1713456 changes codon 1468 from Cysteine to Tyrosine. TEP1 (telomerase-associated protein 1, 14q11.2) gene product is a component of the telomerase enzyme complex [37]. TEP1 SNPs have been associated with the risks of bladder [38], stomach [39], prostate [40], and breast cancer [41], the prognosis of liver [42] and prostate cancer [40], and the risk of type 2 diabetes [43]. Although TEP1 is a telomerase-binding protein, previously it was shown that TEP1 is not essential for telomerase activity or telomere length maintenance in a mouse model [44]. Whether TEP1 is essential for telomere length maintenance in human cells is not clear. On the other hand, TEP1 is essential for vault RNA stability and its association with the vault particle [45]. Vaults are evolutionary highly conserved ribonucleoprotein particles that are associated with several cellular processes such as cell motility and differentiation [46]. The associations of TEP1 SNPs with different diseases suggest that TEP1 may have general cellular functions, which when impaired, can have a broad range of physiological and pathological consequences. The molecular mechanisms underlying these associations warrant further studies. The other SNP that was associated with ovarian cancer risk was TERT: rs2853676. In a previous study, seven SNPs (rs2736122, rs4246742, rs4975605, rs10069690, rs2736100, rs2853676, rs7726159) in the TERT gene were associated with ovarian cancer risks [28]. TERT: rs2853676 was more strongly correlated with serous ovarian cancer, consistent with our findings [28]. Taken together, these data suggest that genetic variations in telomere-maintenance genes may modulate the risks of developing ovarian cancer. Two thirds of ovarian cancer patients die as a result of progressive disease and chemotherapy resistance. The cytotoxic effect of platinum is mediated through its interaction with DNA and formation of a variety of DNA adducts, followed by the induction of apoptosis and/or other mechanisms of cell death [47,48]. Many genes on telomere maintenance pathways have been found to be associated with chemo-resistance to platinum in vitro [49,50]. In this study, we found seven SNPs significantly associated with ovarian cancer survival and four SNPs with response to platinumbased chemotherapy. Interestingly, TEP1: rs10143407 and TEP1: rs2151753 were significantly associated with both survival and therapy response. Rs10143407 is located on the 3 0 UTR and rs2151753 is an intronic SNP. They may regulate the expression of TEP1 or serve as tag SNPs that are linked to causative SNPs. Further studies are needed to determine the biological mechanisms underlying the associations of these SNPs with ovarian cancer outcome. In addition to TEP1, the minor variant of PINX1: rs7826180 exhibited a nearly sevenfold increased risk of poor response for chemotherapy. PINX1: rs7826180 locates in an intron of PINX1. PINX1 (PIN2/TRF1 interacting, 8p23.1) encodes a protein of 328 amino acids and is a TRF1-interacting protein. PINX1 binds to the telomerase catalytic subunit TERT and inhibits telomerase activity [51]. PINX1 is a putative tumour suppressor and overexpression of PINX1 inhibits telomerase activity, shortens telomeres and induces crisis [51]. A previous study showed that PINX1 had lower expression in epithelial ovarian cancer tissues and was associated with shorter survival time [52]. Our finding provided epidemiologic evidence that PINX1 genetic variants could affect ovarian cancer outcome through telomere maintenance pathway.
There are a few limitations in our study. Firstly, the sample size was relatively small and we did not have a validation population. We used a FDR (Q-value)-based method to adjust for multiple testing and some of our findings had a Q-value of less than 0.05. Nevertheless, independent validation is the ultimate means to confirm that our observations are true. Secondly, as a retrospective, hospital-based case-control study, selection and recall bias may confound our observed associations. However, this study is a genetics-based study and the effect of environment on genetic association is minimal as demonstrated by numerous GWAS that often had heterogeneous study designs. Thirdly, because of technical issues of iSelect SNP custom array and evolving literature, we missed some important telomere maintenance genes, such as hTERC, TIN2 and REL1, and their roles in ovarian cancer risk and outcome warrant further study. Fourthly, many of the SNPs are tagging SNPs and are not the true functional variants. The biology underlying the observed associations is unclear.
In summary, this study provides epidemiologic evidence for the associations of telomere-maintenance gene variations with ovarian cancer risk and clinical outcome. Future studies are warranted to validate our findings and explore the biological mechanisms underlying the association of telomere maintenance gene variants with ovarian cancer risks and outcome.