In search for biomarkers and potential drug targets for uterine serous endometrial cancer

Endometrial cancer (EC) is the most common gynaecological cancer in the developed world and the fourth most common cancer in women (Gentry-Maharaj and Karpinskyj 2020). Based on differences in histology and clinical outcomes, endometrial cancers have long been divided into the estrogen-dependent type I (endometrioid adenocarcinoma) with a favourable prognosis, and the estrogen-independent type II (predominantly serous and clear cell carcinoma) (Bokhman 1983; Dedes et al. 2011). Approximately 80–90% of EC are endometrioid-type carcinomas, and 2–10% are serous endometrial carcinomas (USC) (Dedes et al. 2011). Although these latter account for only a small percentage of EC, as high as 40% of EC deaths are attributed to USC (Hamilton et al. 2008; Del Carmen et al. 2012). This highly aggressive behaviour is mainly related to its tendency to metastasize, even when the primary tumour is small (Hamilton et al. 2006). Currently, the standard of care for most patients with USC includes surgery and platinum-based adjuvant chemotherapy with or without radiotherapy. Indeed, given its distinct biological behaviour, USC is typically excluded from large clinical trials (Keys et al. 2004; Nout et al. 2010).

During the last decade, we have significantly increased our understanding of biological pathways implicated in disease development and progression. Specifically, USC were found to harbour a specific genomic signature, which is different from other EC histologic subtypes, and it is partially overlapping with high-grade serous ovarian carcinoma (HGSOC) and basal-like breast cancer (Kuhn et al. 2012; Le Gallo et al. 2012; Cancer Genome Atlas Research Network 2013). Mutations in the TP53 (Tumor protein p53) tumour suppressor gene and/or stabilization of the p53 protein were the most frequent molecular aberrations in serous carcinomas, occurring at frequencies in excess of 85%. Notably, however, the mutation frequency of five genes [PIK3CA (Phosphatidylinositol-4,5-Bisphosphate 3-Kinase Catalytic Subunit Alpha), PIK3R1 (phosphoinositide-3-kinase regulatory subunit 1), PTEN (phosphatase and tensin homolog), PPP2R1A (protein phosphatase 2 scaffold subunit Aalpha) and FBXW7 (F-box and WD repeat domain containing 7)] was dramatically higher in USC as compared to serous ovarian cancer (Cancer Genome Atlas Research Network 2013). These five genes are part of the PI3K (phosphatidylinositol 3-kinase)-AKT (protein kinase B)-FBXW7 pathway. The FBXW7 tumour suppressor is a component of the FBXW7-SKP1 (S-phase kinase-associated protein 1)-CUL1 (Cullin-1) ubiquitin ligase complex, which mediates the ubiquitination of the protein products of multiple oncogenes, including cyclin E, AURKA (Aurora A kinase), PLK1 (Polo-like kinase-1) and FOXM1 (Forkhead box protein M1) among the others (Galindo-Moreno et al. 2020). Glycogen synthase kinase-3 (GSK3)-mediated phosphorylation of the targeted proteins is a necessary step for subsequent FBXW7-mediated ubiquitylation and proteasomal degradation (Hoxhaj and Manning 2020). GSK3 is a ubiquitously expressed protein kinase that exists in two isoforms, α and β. The protein is active under basal conditions and is inhibited in response to growth factors and insulin, via AKT-mediated phosphorylation. GSK3 and FBXW7 thus act in concert to regulate ubiquitination of many important factors associated with cell division and growth.

Pathway analysis and clustering of the endometrial tumours in the TCGA data set also revealed that dysregulation of mitotic processes is a frequent occurrence in USC (Cancer Genome Atlas Research Network 2013). In this context, it is worth noting that numerous studies have established a crucial role for E2F-1 (E2F transcription factor 1) in the control of cellular proliferation, through transcriptional activation of genes important for cell-cycle progression, such as cyclin B1 and cyclin E (Hallstrom et al. 2008). Therefore, E2F-1 may have profound implications in USC, acting as a master regulator of cell fate. In line with this concept, E2F-1 has been identified as a prognostic biomarker in endometrial cancer, being mostly expressed in the serous histotype (Alkushi et al. 2007). Actually, E2F-1 has been shown to induce both proliferation and apoptosis, being the balance of these events regulated by the PI3K/AKT signalling, that would selectively block the expression of genes in the apoptotic program, but not in the proliferative program (Hallstrom et al. 2008). TP73 (Tumor protein p73) is a target of E2F-1 apoptotic program and altered TP73 expression (due to aberrant DNA methylation) is considered a predictor of USC histology (Seeber et al. 2010). Notably, after DNA damage, p73 induction is regulated by the checkpoint kinases CHK1 and CHK2, through E2F-1 stabilization, in a pathway central to p53-independent apoptosis (Urist et al. 2004). Therefore, this signalling may be a major determinant of anti-cancer drug efficacy in USC patients carrying TP53 mutations. Overall, these data suggest that molecular pathways converging on p73 may play important roles in driving disease outcome.

TCGA data also demonstrated that up to 40% of type II EC tumours are associated with heterozygous missense mutations in PPP2R1A, an established tumour suppressor gene encoding the Aα subunit of PP2A (Protein Phosphatase 2A), one of the major cellular serine–threonine phosphatases, involved in the regulation of PI3K/AKT pathway (Kuo et al. 2008; Remmerie and Janssens 2019). It has been recently shown that also PPP2R1B (the gene encoding for the β isoform of subunit A) is mutated in human neoplasm, with a functional inactivation of the protein that may promote tumorigenesis, through its role in cell-cycle regulation and cellular growth control (Calin et al. 2000). The inclusion of both isoforms of subunit A in the genes mutated in human cancer support the potential role of PP2A in human tumorigenesis via a mechanism of inactivation of the phosphatase activity (Calin et al. 2000). In line with these findings, overexpression of CIP2A (Cellular Inhibitor of PP2A, also called p90), an endogenous PP2A-inhibitory protein, has been associated to worse outcome in gynaecologic cancers (Remmerie and Janssens 2019).

Finally, recent data supported the view that homologous recombination deficiency (HRD) occurs in endometrial cancers and is largely restricted to non-endometrioid, TP53-mutant endometrial cancers, the spectrum of germline mutations detected including BRCA1/2 (breast cancer type 1/2 susceptibility protein) and RAD51 (DNA repair protein RAD51 homolog 1) among others (Ring et al. 2016; De Jonge et al. 2019).

Based on the above-mentioned, well-known or emerging evidence of molecular determinants of cancer development, we selected a panel of proteins for tumour profiling of a homogeneous cohort of USC patients, to identify clinically relevant prognostic biomarkers. Results obtained provide new perspective and potential strategies for drug target discovery in USC therapy.

During January 2009 through December 2012, a total of 93 consecutive patients diagnosed with serous or mixed serous EC underwent primary surgery at our institution and met the following inclusion criteria: research authorization, no synchronous cancer or previous neoadjuvant therapies. Of these patients, 34 had mixed serous histology and 59 had serous histology. Among these 59 women, the results herein are based on the 36 patients whose specimens were available and there was adequate tissue in the paraffin block to analyze multiple markers with immunohistochemistry.

Results from the present study offer some important insights into USC, which is a rare aggressive subtype of endometrial cancer (Brooks et al. 2019).

The most interesting finding emerging from our investigation is the favourable prognostic role of CHK1 for both progression-free and overall survival, a result consistently observed for both the nuclear and the cytoplasmic protein fraction. Interestingly, analysis of a TCGA data set of 109 USC patients corroborates our results showing a favourable prognostic role of CHK1 after adjusting for stage. CHK1 is a key signal transducer in the DNA-damage response pathways, being implicated in the induction of cell-cycle arrest, DNA repair and apoptosis (Bartek and Lukas 2003). The protein is mainly expressed in the nucleus, but, following activation, it shuttles between the nucleus and the cytoplasm, regulating both nuclear and cytoplasmic checkpoints; importantly, cytoplasmic CHK1 does not support cell viability (Wang et al. 2012). Uncertainty exists in the literature about the role of CHK1 in cancer development and progression. Indeed, given the regulatory role it plays in DNA damage, CHK1 has long been recognized as a tumour suppressor. However, recent evidences indicate that CHK1 may contribute to tumour growth, thus representing a potential target in anti-cancer therapy. The notion that tumour cells (often p53-deficient and defective in G1 arrest), unlike normal cells, rely mainly on S or G2 checkpoints mediated by CHK1 to repair their damaged DNA and preserve their genomic integrity for basic viability, supports this view (Chen et al. 2006). In line with this concept, no homozygous loss-of-function mutation of CHK1 has been detected in a wide range of human tumours (Remmerie and Janssens 2019), an observation suggesting that cells with defective CHK1 are eliminated during tumorigenesis. Finally, the gene has been found overexpressed in variety of tumours and expression correlated with tumour grade and disease recurrence (Zhang and Hunter 2014). On the other hand, CHK1 frameshift mutations have been reported in genetically unstable colorectal and endometrial cancers, and these mutations might be involved in tumorigenesis, through a defect in response to DNA damage (Bertoni et al. 1999; Vassileva et al. 2002). Although these mutations predict for truncated CHK1 protein, their pathogenic implication is not evident since they have been always found as heterozygous aberrations. In keeping with our results, previous studies have also demonstrated that constitutive activation of CHK1, in the absence of DNA-damage, leads to cell-cycle arrest and eventually cell death (Wang et al. 2012; Zhang and Hunter 2014). Indeed, by querying the public TCGA USC sequencing data, we found that the top coding-gene associated with CHK1 was EI24, a putative tumour suppressor gene, whose reduced expression has been associated with the induction of EMT and tumour progression (Choi et al. 2013). Notably, EI24 has been also characterized as an E2F-1 target gene (Sung et al. 2013). Relevant to our results, Urist and colleagues (2004) demonstrated that CHK1 is required for induction of p73 following DNA damage and that E2F-1 is critical in this regulation. This implies that the CHK1–E2F-1–p73 pathway is central to p53-independent apoptosis, after DNA damage. Interestingly, literature data also unveiled Aurora A tumorigenic properties actually mediated by repression of CHK1 kinase activity, with a consequent impairment of the error-free homologous recombination pathway (Sourisseau et al. 2010). Overall, these findings might support the idea that in USC elevated constitutive levels of CHK1 play a protective role in disease development, through induction of permanent cell-cycle arrest and cell death. In addition, the outcome of CHK1 signalling following DNA-damaging therapies may also lead to increased apoptosis linked to p73 induction.

Besides CHK1, also FBXW7 and PPP2R1B were found to have a favourable prognostic role for progression, in analysis adjusted for stage. FBXW7 is a candidate driver gene somatically mutated in about 15–29% USC (Le Gallo et al. 2012; Cancer Genome Atlas Research Network 2013). Besides its role as tumour suppressor, emerging evidence suggest that lack of protein or loss-of-function mutations in FBXW7 confer resistance to antitubulin agents (Wertz et al. 2011), while sensitizing to HDAC inhibitors (Garnett et al. 2012). In this context, results of our study, if confirmed in a larger population, could have a clinical significance in guiding personalized therapy. With regard to PPP2R1B, our findings are in line with available data testifying a regulator function in a variety of cellular processes, including cell-cycle progression, whose alteration may be involved in tumorigenesis (Calin et al. 2000). However, to the best of our knowledge, no other evidences are available on its role as prognostic biomarkers in USC. Again, these findings could open new perspective and potential strategies for drug target discovery in USC therapy.

Results from the present study also suggest that Cyclin B1 is an unfavourable prognostic biomarker for death in USC. Cyclin B1 is a positive regulator of cell-cycle progression; protein overexpression reduces cell-cycle length and enables cells to override the G2 DNA damage checkpoint (Jin et al. 1998; Pomerening et al. 2005), finally leading to uncontrolled proliferation. At present, there are few studies regarding the role of Cyclin B1 in USC development and progression, although recent findings by Kwan and colleagues (Kwan et al. 2020) showed high Cyclin B1 expression levels in USC.

Finally, we also found that, at univariate analysis, higher nuclear BRCA1 expression was associated with an increased risk of progression and death. This result is in line with data from HGSOC, showing that patients with low BRCA1 expression had a more favourable outcome (Weberpals et al. 2011).

Beside the above-mentioned biomarkers, here we also give evidence that, although not having a role as prognostic factors, some proteins are expressed at high levels in USC tumours and specific pathways activated. In this respect, our findings therefore also provide hints for further studies specifically investigating those proteins found to be highly expressed in the disease and their potential value as new therapeutic targets.

Among the peculiar strengths of this study are that patients were managed in a tertiary centre, with well-documented electronic charts and dedicated personnel drawing up the historical EC database. In addition, the follow-up was sufficiently long to capture enough events, thus providing adequate statistical power to identify meaningful associations. Finally, all samples were treated and analysed with standard protocols. Besides the above-mentioned strengths, few limitations in our study need to be acknowledged. First, sample size was not large enough for detecting modest associations between biomarkers expression and the risk of progression or death. Moreover, pathologist-dependent interpretation of immunohistochemical staining might represent a setback, therefore, requiring all necessary steps to avoid any bias. Not least important, the retrospective nature of such a study design can lead to missing data.

In conclusion, results from the present study, although preliminary, highlight that molecular tumour profiling may provide a prognostic tool in serous endometrial cancer, representing an essential first step in drug discovery and development of personalized therapy.