NON-CODING RNAS IN HIGH-GRADE SEROUS EPITHELIAL OVARIAN CANCER

Abstract Introduction: High-grade serous ovarian carcinoma (HGSOC) is the most lethal gynecologic malignancy, mainly because the disease is frequently diagnosed at an advanced stage and is characterized by the early onset of chemoresistant recurrences. The lack of reliable diagnostic and prognostic markers, together with the lack of effective therapies, are the major obstacles to the clinical management of patients with HGSOC. A new class of non-coding RNAs (ncRNAs), such as microRNA (miRNAs) and long non-coding RNAs (LncRNAs), with a function of gene expression regulation, have been discovered to play an important role in human cancers. Increasing evidences suggest that ncRNAs are involved in cancer progression and development of chemoresistance, and support their role as potential diagnostic, predictive and prognostic biomarkers. The hypoxic condition within the tumor microenvironment, improving the tumor neovascularization, represents an essential event contributing to the development of a more aggressive HGSOC phenotype. Recently, a group of miRNAs, termed hypoxia regulated-miRNAs (HRMs), have been identified as key elements in response to hypoxia, regulating important mechanisms involved in tumor progression. The complexity of hypoxia molecular mechanisms has not been fully elucidated yet in HGSOC, therefore there is an urgent need to discover novel biomarkers clinically useful to select patients with hypoxic tumor, that may benefit of tailored treatments. Aims of the study: My PhD project aims at elucidating transcriptional and post-transcriptional signatures characterizing HGSOC, both at the serum and tissue levels. In detail, the research effort includes: i) the investigation of circulating miRNAs as novel potential biomarkers for HGSOC detection; ii) the analysis of mRNA, miRNA and lncRNA expression profiles of HGSOC and normal tissues; iii) the evaluation of hypoxia-regulated miRNA expression in HGSOC and normal tissues. Methods: Sera from 168 HGSOC stage III-IV patients and 65 healthy donors were gathered together from two independent collections and stratified into a training set, for miRNA marker identification, and a validation set, for data validation. Nine synthetic viral/C.Elegans spike-in oligos were added to serum samples before RNA extraction, to allow accurate normalization. miRNA expression profiles were obtained using Agilent Microarray Technologies®. An innovative statistical approach for microarray data normalization, based on the contribute of spike-in oligos and the most invariant miRNAs, was developed to identify, in the training set, differentially expressed miRNAs. Signature validation in both the training and validation sets was performed by Real Time quantitative PCR (RT-qPCR) and confirmed by droplet digital PCR (ddPCR). A total of 99 tumor biopsies were collected from HGSOC stage III-IV patients, partially matched with the serum sample cohort (n=76). Thirty normal tissues were obtained from normal ovary (HOSE) and luminal fallopian tube surface epithelia, both representing the normal counterpart for HGSOC, whose histogenesis is still a matter of debate. Gene and miRNA expression profiles were obtained using Agilent Microarray Technologies®. miRNA expression levels were correlated with patient outcomes, as overall survival (OS) and progression-free survival (PFS). Additionally, a subgroup of 14 chemo-resistant and 14 chemo-sensitive HGSOC patients, together with 10 normal tissues were deep sequenced for the discovery of novel HGSOC specific coding and non-coding transcripts. Results: A panel of 97 miRNAs emerged significantly differentially expressed (92 up-regulated and five down-regulated) between sera of HGSOC patients and healthy donors by microarray analysis. Among them, the following miRNAs, i.e., miR-1246, miR-595, miR-574-5p, miR-483-3p, miR-4290, miR-2278, miR-32, miR-4281, and miR-3148, exhibiting both the highest average expression and log fold change measured in patients compared to healthy donors, were selected for further validation. miR-1246, miR-595 and miR-2278 were confirmed as significantly over-expressed in serum of HGSOC patients compared to controls by RT-qPCR (all p-values<0.03), in both the training and validation sets. Receiver Operating Characteristic (ROC) curve analysis revealed miR-1246 as the best diagnostic biomarker, with a sensitivity of 87%, a specificity of 77% and an accuracy of 84%. The absolute quantification of circulating miR-1246 by ddPCR confirmed its potential as diagnostic biomarker in HGSOC. Microarray analysis of tissue miRNA profiling revealed a total of 265 miRNAs significantly dysregulated (123 up-regulated and 142 down-regulated) in HGSOC compared to normal tissues. A group of nine miRNAs (i.e., miR-199b-5p, miR-423-5p miR-455-3p, miR-22-3p, miR-199a-3p, miR-15b-5p, miR-140-5p, miR-1246, and miR-320c) were associated with platinum response and prognosis. In particular, among tumor samples, miR-1246 up-regulation was consistently associated with platinum-resistance, poor OS and poor PFS (p-values<0.05). Kaplan-Meier survival curves, according to miR-1246 expression levels obtained by RT-qPCR, showed that OS and PFS decreased in patients with high miR-1246 expression compared to those with low miR-1246 expression (p-value<0.001, HR=2.57; p-value=0.024, HR=1.68; respectively). In addition, multivariate analysis revealed miR-1246 over-expression as an independent prognostic factor for poor OS and PFS (p-value=0.002, HR=2.31; p-value<0.05, HR=1.59; respectively). Interestingly, compared to normal tissues, both with microarray and RT-qPCR techniques, miR-1246 showed a down-regulation compared to HOSEs (p-value<0.0001), but we did not detect a significantly differential expression compared to fallopian tubes. This result mirrors the global miRNA expression trend revealed by principal component analysis (PCA) on microarray data. Subsequently, we focused our analysis on a group of 16 miRNAs belonging to the group of hypoxia-regulated miRNAs (HRMs) emerged from literature as relevant in other solid tumors. Among them, we confirmed miR-210 and miR-27a-3p/23a-3p/24-3p cluster as significantly up-regulated in HGSOC vs normal tissues by RT-qPCR (all p-values≤0.002). More interestingly, we validated the significant over-expression of miR-23a-3p in the group of patients resistant to platinum-based chemotherapy compared to platinum-sensitive patients (p-value=0.03). In addition, in univariate survival analysis miR-23a-3p over-expression showed a significant correlation with decreased progression-free survival (p-value=0.009, HR=1.8), but not with overall survival variable. Importantly, miR-23a-3p over-expression has emerged as an independent prognostic marker for shortened progression-free survival in multivariate Cox regression analysis (p-value=0.01, HR=1.78). Finally, the preliminary analysis of the transcriptome sequencing allowed us to identify 1371 transcripts differentially expressed between platinum-resistant and platinum-sensitive samples. Among them, 125 transcripts showed a complete match of intron chain with known transcripts, 686 were potentially novel isoforms or showed a generic overlap with known transcripts. The remaining 560 sequences, if validated, could be novel intergenic transcripts or transcripts with an exonic overlap with reference ones. Conclusions: This study demonstrates, for the first time, miR-1246 as a potential diagnostic serum biomarker in HGSOC, as assessed by three independent technologies (microarray, RT-qPCR and ddPCR) and validated in two independent cohorts of patients. Moreover, high-throughput analysis reveals most of the gene and miRNA dysregulated in HGSOC biopsies compared to the normal counterpart. In particular, our findings indicate, for the first time, that miR-1246 over-expression correlates with a platinum-resistant HGSOC phenotype and may constitute a novel prognostic factor for HGSOC patients. Furthermore, our results regarding HRMs suggest an important role of miRNAs in response to hypoxic conditions within HGSOC. Particularly, the miR-23a-3p over-expression in the group of platinum-resistance patients may contribute to explain the importance of hypoxia in HGSOC mechanism of drug resistance and could represent an independent prognostic marker for HGSOC patients. Lastly, preliminary data emerged from transcriptome analyses, suggesting a prominent non-coding role in HGSOC platinum resistance, will be integrated with gene and miRNA expression profiles previously obtained, with the aim to identify tumor circuits associated with response to treatment and prognosis, as well as to better elucidate the molecular mechanisms characterizing HGSOC progression and adaptation to hypoxic tumor microenvironment.