ONCOTAG BIOMARKER PANEL TO PROGNOSIS ANALYSIS IN EPITHELIAL OVARIAN CANCER

Ovarian cancer (OC) comprises heterogeneous and complex diseases group and represents the most lethal gynecologic cancer. In 2018, it was estimated to be responsible for 295,414 new cases and 184,799 deaths worldwide. Late-stage diagnosis associated with a lack of effective molecular markers to detect chemotherapy resistance contributes to high mortality and poor prognosis of epithelial ovarian cancer (EOC). For invasive or recurrent EOC, individualized assessments and optimized management are especially required. In this way, OncoTag was developed as biopsy-based gene expression assay to discriminate bad or favorable prognosis in EOC patient. We measured the expression apoptosis-related pathway genes and associate them with clinical-pathological patients features to examine its prognosis significance, with the ultimate goal of advancing patient care. Prospectively defined observational study was performed. The cohort were stratified according to stage, grade and histological subtype of tumor, overall and disease-free survival. Tissues were obtained from 28 women, shared into primary and metastatic EOC and 18 normal ovary patients (as control) who underwent surgical treatment at Hospital das Clínicas/Faculdade de Medicina/UFMG, between 2008 and 2011, after informed consent. RNA was extracted from each sample, treated with DNase I to remove genomic DNA and total RNA content was quantified. Reverse transcription polymerase chain reaction (RT-PCR) was used to convert RNA to complementary DNA prior to determine gene expression profile by quantitative PCR. Using statistics methods and supervised and unsupervised machine learning approaches, we characterized and validated the molecular targets. Our results appointed that there were statistically significant differences in gene expression profile between patients groups (p < 0.05), demonstrating the EOC genetic heterogeneity. Molecular classifier algorithm proved the diagnostic and prognostic value of genes evaluated. Decisions trees and biomarker clusters were constructed. In the training set, the classifier reported 90% and 89.4% of specificity and sensibility to predict occult metastasis and platinum resistance, respectively. Besides of that, the hierarchical clustering allowed us to classify each patient using the cutoff obtained by classifier algorithm. Our results can be considered in clinical practice for help in finding EOC treatment options, improving outcomes and support a more informed clinical decision.