HNF1B polymorphism influences the prognosis of endometrial cancer patients: a cohort study

To our knowledge, this was the first study aimed to test the hypothesis that rs4430796 influences the OS of patients with EC. Current retrospective cohort study showed that the OS of EC patients varied significantly according to rs4430796 genotypes and at multivariate analysis rs4430796 SNP genotype resulted an independent risk factor for OS in EC patients. In particular, GG genotyped patients showed the worst OS.

In order to explore this influence, our population was studied according to adjuvant treatment. In EC patients treated with post-surgical radio-chemotherapy, subjects with GG genotype had a significant HR of 39.1 for death, even if a wide CI was observed. Of note, a significant interaction between rs4430796 polymorphism and treatment was detected showing that the impact on OS is different from that expected from the sum of the individual effects of the two variables. Those findings were not confirmed for EC patients who did not receive adjuvant therapy and for those who received radiotherapy alone, suggesting that rs4430796 polymorphism could play a crucial role in tumor sensitivity to chemotherapy. To this regard, no other study is available in literature and our study was the first to suggest that A allele of rs4430796 in EC patients undergoing chemo-radiotherapy might reduce the risk of death by increasing the response to therapy.

Well-known prognostic factors for EC include age, race, stage, grade, ploidy, depth of invasion, tumor size, receptor status, and cell type [24]. However, these multiple factors failed to give an accurate estimate of prognosis, whereas some genetic variants, such as rs4430796, may be in the next future a useful marker to predict, alone or in combination, the prognosis of EC patients.

Although many studies have investigated HNF1B in gynecological malignancies, such as CCC, showing its relationship with tumor cell survival [25] and tumor-associated thrombosis [6], few data are reported on its role in EC. A GWAS demonstrated that G allele of rs4430796 is associated with decreased EC risk in European population [12] and the PAGE study [13,14] subsequently confirmed those findings in two independent studies. The associations were observed irrespectively of several demographic, anthropometric, hormonal, voluptuary patients’ characteristics [17]. Finally, a more recent GWAS [18] confirmed previous finding [12] and no new SNPs were found. Recently, HNF1B was found to be overexpressed in EC [9-11], particularly in CC EC [26].

No data were found in literature explaining the role of rs4430796 genotype on HNF1B function but some contradictory studies supposed that rs4430796 polymorphism could influence HNF1B gene expression [12,27-30]. Spurdle et al. [12] analyzed several lymphocyte-derived gene expression datasets and identified significant associations between rs4430796 genotype and HNF1B expression in individuals of European ancestry. In the same population, a RNA sequencing experiment suggested that HNF1B expression in patients with different rs4430796 genotypes presented a trend of increase in read depth with increasing number of G alleles [12].

Recent and not always concordant studies analyzed the influence of HNF1B gene expression on chemosensitivity in ovarian cancer [27,28,30]. In particular, some studies, observing that ovarian CCC characterized by HNF1B overexpression were often chemo-resistant, suggested a possible effect of the level of HNF1B on chemosensitivity. It was demonstrated that shRNA mediated downregulation of HNF1B sensitizes ovarian cancer cells to cisplatin- or paclitaxel-mediated cytotoxicity, through inverse regulation of HSulf1 expression [27]. In 2014, an interesting mechanism of inhibition of cell death by HNF1B transcription factor was also proposed [28]. Specifically, it was demonstrated that chemoresistance that characterizes CCC might be due to aberrant retention of the G2 checkpoint of cell cycle induced by HNF1B overexpression [28]. In addition, HNF1B might induce this aberrant retention through the upregulation of CHK1 kinase, a protein that plays a pivotal role in the G2 checkpoint [28].

It was not possible for us to evaluate HNF1B gene expression in FFPE samples. The supposition of a potential biological model will require further experiments but, based on published studies, we suppose that the correlation that we observed between rs4430796 GG genotype and reduced OS in patients treated with chemotherapy could support the hypothesis that G allele reduces chemosensitivity through HNF1B overexpression. Identification of EC patients candidate for chemotherapy with rs4430796 GG genotype could be useful in the clinical practice in the next future to schedule a different chemotherapy and/or the administration of a sensitization agent. A recent study showed that the administration of CHK1 inhibitors to human cell line of CC EC treated with bleomicin was followed by an increase cell death rate [28]. Similarly, CHK1 inhibitors administration might decrease the chemoresistance to several drugs [28]. However, new agents that enhance the chemosensitivity should be investigated.

Unlike to HNF1B, ARID1A downexpression, mutation or loss of function is associated with chemoresistance and deep myometrial invasion in EC [31,32]. ARID1A is a tumor suppressor gene. Inactivating mutations of ARID1A and loss of its expression was found especially in endometrium-derived tumors, including ovarian CCC, ovarian endometrioid carcinomas and EC [33,34]. Both ARID1A and HNF1B play a role in the pathogenesis of ovarian CCC arose in endometriosis and both were proposed as new experimental markers in EC [8]. Usually, ARID1A mutation is an early event in neoplastic transformation and it is present both in carcinoma and preneoplastic tissue whilst HNF1B is usually found only in cancer tissue [8]. ARID1A mutations co-occur with mutations of PTEN and PIK3CA. Furthermore, PI3K pathway activity is regulated by ARID1A through phosphorylation of AKT [35]. Also AKT is involved in chemoresistance of EC cells, particularly Akt2 and Akt3 isoforms are involved in cell chemoresistance independently of the drug used for treatment [36,37]. Both AKT and HNF1B take part in the signaling pathway that regulates insulin secretion in pancreatic beta cells and their involvement in EC could suggest a possible role, not yet investigated, of these genes in the regulation of tumor cells glucose homeostasis [38,39] since abnormal glucose homeostasis is a well-known risk factor for EC type 1. However, despite a direct interaction between HNF1B and ARID1A/AKT was not yet clarified, aforementioned evidences suggest that a common pathway could be implicated in the pathogenesis and biological behavior of EC. Further studies need to elucidate these potential interactions from a biological and clinical point of view.

Current study presents issues of strength and limitation. The strength of our study regards the selection of a very homogeneous population of EC patients who received an upfront surgery with adequate treatment and follow-up length. The centralization of treatment, follow-up, and pathology review are further study strengths that ensured a uniformity of treatment, staging procedures, post-treatment monitoring and histological classification. Instead, in other available studies [13] the treatment and follow-up protocols widely varied [40]. Furthermore, our study has also important limitations. In particular, it might be underpowered because of the small cohort, and actual sample size might not be sufficient to detect a synergistic effect in a replicate study. However, despite the small sample size, the genotype distribution in our population comply with Hardy-Weinberg equilibrium indicating that identified significant association can be considered representative and not-biased by patients selection. Regarding to small sample size of this study, we did not manage to recruit a sufficient number of EC patients with advanced disease to request chemo-radiotherapy because most EC is diagnosed at early stage, and chemo-radiotherapy is commonly delivered in less than 10% of subjects [20-22]. Chemo-radiotherapy is generally administered to EC patients with positive lymph nodes (stage III) or distant metastases (stage IV) [20-22]. That population represents the 6-7% and 3% of EC patients, respectively of the overall EC population. In addition, the rate of 5-year survival for stage I disease is approximately 80-90%, for stage II, 70- 80%, and for stages III and IV, 20-60% [22]. Hence the mortality is also low.