IBCN 2023: Gene Expression-Based Classification of Urothelial Carcinoma For Research And Clinical Applications

(UroToday.com) Transcriptomic subtype prediction of urothelial carcinomas can be affected by the choice of expression platform, cohort composition, data preprocessing, and tumor purity. To address this, a single-sample-predictor (SSP) based on binary gene-pair rules was developed. The rule-based RandomForest (RF) SSP was trained on minimally preprocessed microarray and RNA-sequencing data and validated in four external bladder cancer datasets. The predicted subtype assignments captured cohesive gene expression signature patterns and significant associations with mutations concordant with the Lund Taxonomy.

An initial 5-class (Uro, GU, Basal, Mes-like, and ScNE) Lund Taxonomy RF classifier was trained on 307 samples with microarray gene-expression data (preprocessed with both RMA and SCAN using fixed BrainArray version 25 annotations). A new model was trained after identification and removal of outlier samples, which was then applied to two additional uniformly preprocessed microarray datasets and 265 RNA-sequenced samples (summarized to TPM using both Kallisto and Salmon). Prediction results were in good agreement with both previous subtype classifications of these cohorts and with immunohistochemistry-based subtype assignments. A final predictor was built using all four datasets (each with two preprocessing versions) as training data. A separate classifier for substratification of Uro into UroA, UroB, and UroC was designed in a similar manner.

Application of the classifier to the TCGA, IMvigor210, UC-Genome, and Robertson et al. T1 datasets gave confident subtype predictions which recapitulated the expression patterns of the training cohort. A significant enrichment of FGFR3 and RB1 mutations were seen in UroA+UroB and GU+ScNE respectively.

The rule-based single sample classification approach performed well across multiple external datasets with minimal preprocessing requirements. Evaluation across additional datasets (RNA-seq and microarrays) is ongoing, with final validation planned in a cohort of prospectively RNA-sequenced tumors with paired IHC.

Presented by: Elena Aramendía, Division of Oncology, Department of Clinical Sciences, Lund University, Lund, Sweden

Written by: Stephen B. Williams, MD, MBA, MS @SWilliams_MD on Twitter during the International Bladder Cancer Network (IBCN) Annual Meeting, September 29-30, 2023, Montreal, Canada