Establishing a Novel Qualitative Model to Predict Chemotherapy Response and Prognosis in Ovarian Cancer

Abstract

Background: Ovarian cancer is frequently associated with chemoresistance, which is the major cause of treatment failure. In this study, we utilized relative expression ordering (REO) of gene pairs to develop a novel model to predict chemotherapy response and prognosis in ovarian cancer. Moreover, we attempted to explore the mechanisms underlying ovarian cancer chemoresistance. Methods: Datasets were downloaded from publicly available databases, and differentially expressed gene pairs were filtered using Wilcoxon signed-rank test, Cox proportional hazards regression and Fishers test to develop the model. Subsequently, the efficacy was validated by Kaplan–Meier analysis in training and validation sets. Comprehensive investigations were performed to investigate pathway variation, immune infiltration, and single-cell analysis. Next, gene expression was measured in chemoresistant ovarian cancer cells and their parent cells, and risk scores were calculated. Finally, a series of experiments were conducted to evaluate the regulatory impacts on chemosensitivity of lysyl oxidase-like 4 (LOXL4), one of the upregulated genes in chemoresistant cells. Results: The developed model, comprising 19 genes for predicting chemoresistance and prognosis, demonstrated robust performance in training and five validation sets. Chemoresistant samples identified by this model exhibited enrichment of genes in four pathways and downregulation of genes in one pathway. Besides, chemoresistant samples displayed a lower abundance of various immune cell types, indicating immune suppression within the tumor microenvironment. Single-cell analysis indicated heterogeneity within samples, revealing cell populations that may survive after chemotherapy. Chemoresistant ovarian cells exhibited higher risk scores compared to their parent cells, and LOXL4 was found to modulate cisplatin sensitivity in ovarian cancer cells. Conclusions: This study presents a novel prognostic model and provides possible therapeutic targets for further research in ovarian cancer.