Identification of a novel chemotherapy benefit index for patients with advanced ovarian cancer based on Bayesian network analysis.

Abstract

Background: This study aims to evaluate the efficacy of chemotherapy and optimize treatment strategies for patients with advanced ovarian cancer.

Methods: Based on The Cancer Genome Atlas (TCGA) transcriptome data, we conducted correlation and Bayesian network analyses to identify key genes strongly associated with chemotherapy prognosis. Reverse Transcription Quantitative Polymerase Chain Reaction (RT-qPCR) was used to verify the expression of these key genes. The Chemotherapy Benefit Index (CBI) was developed using these genes via multivariable Cox regression analysis, and validated using both internal and external validation sets (GSE32062 and GSE30161) with a random forest model. Subsequently, we analyzed distinct molecular characteristics and explored additional immunotherapy in CBI-high and CBI-low subgroups.

Results: Based on the network and machine learning analyses, CBI was developed from the following ten genes: COL6A3, SPI1, HSF1, CD3E, PIK3R4, MZB1, FERMT3, GZMA, PSMB9 and RSF1. Significant differences in overall survival were observed among the CBI-high, medium, and low subgroups (P < 0.001), which were consistent with the two external validation sets (P < 0.001 and P = 0.003). The AUC of internal validation and two external validation cohorts were 0.87, 0.71 and 0.70, respectively. Molecular function analysis indicated that the CBI-low subgroup is characterized by the activation of cancer-related signaling pathways, immune-related biological processes, higher TP53 mutation rate, particularly with a better response to immune checkpoint blockade (ICB) treatment, while the CBI-high subgroup is characterized by inhibition of cell cycle, less response to ICB treatment, and potential therapeutic targets.

Conclusions: This study provided a novel CBI for patients with advanced ovarian cancer through network analyses and machine learning. CBI could serve as a prognostic prediction tool for patients with advanced ovarian cancer, and also as a potential indicator for immunotherapy.