Abstract 5570: Development of prognostic index based on twenty-three genes for predicting superficial-to-invasive progression of non-muscle invasive bladder cancer patients

Abstract

Abstract Background: Although recent advances in high-throughput technology and data-driven approach have provided many insights into non-muscle invasive bladder cancer (NMIBC), previous studies are still limited in their ability to predict the clinical behavior of NMIBC. Here, using deep learning method with long term follow-up data, we identified a prognostic gene set consisting of a small gene group for precisely predicting NMIBC heterogeneity. Methods: We sought to identify progression-associated genes in patients with NMIBC using Cox regression analysis and verified their predictive values using a fully connected neural network (FNN) algorithm in five independent cohorts comprising more than 800 NMIBC patients. Based on these genes, a prognostic index (PI) in NMIBC progression was also developed. The association between the PI and prognosis of NMIBC patients was evaluated using Kaplan-Meier plots and log-rank tests. Results: Gene expression profiling in NMIBC patients identified a prognostic gene set consisting of 1,789 genes for predicting NMIBC progression in a patient cohort (training set, n = 103). Their prognostic significances were validated based on FNN algorithm in other four independent cohorts (validation sets, n = 722). Pathway enrichment analysis revealed a twenty-three gene signature including known prognostic transcription factors such as FOXM1 and E2F1 along with novel genes. We incorporated these genes into the PI system, which was a significant prognostic indicator of NMIBC progression. The PI system was shown to be an independent risk factor by a multivariate analysis and subset stratification according to stage and grade (each P < 0.001). The subset analysis also revealed that the PI system could identify patients who would benefit from BCG immunotherapy. Conclusions: The twenty-three gene-based PI represents a promising diagnostic tool for the identification of high-risk NMIBC patients who would display different clinical behaviours as well as response to BCG immunotherapy. Citation Format: Jae-Yoon Kim, Seong-Hwan Park, Seon-Young Kim, Seok Joong Yoon, Seon-Kyu Kim. Development of prognostic index based on twenty-three genes for predicting superficial-to-invasive progression of non-muscle invasive bladder cancer patients. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5570.