Modeling of prognostication based on genomic expression in the tumor microenvironment of muscle invasive urothelial carcinoma via machine learning.

Abstract

64 Background: Stroma in the tumor microenvironment (TME) influences prognosis and response to therapy. Few mathematical models exist to prognosticate patients (pts), based on mRNA expressivity in the TME. Methods: Clinical outcomes data and mRNA-seq of 401 pts with muscle invasive urothelial carcinoma were obtained from TCGA. Expressivity of 191 genes enriched in cellular and structural elements of TME and clinical data were analyzed by Kaplan-Meier (KM) analysis, correlation analysis, and multivariate nonlinear regression assisted by machine learning to achieve confined optimization with model-data minimization among multiple distribution functions. Results: Prognostication was modeled with higher risk score (RS) representing worse prognosis in stage 2-4 (Table, stage 1 data not available from TCGA). P/G is the ratio of genes associated with poor (19 genes) to good (11) prognosis (refer to presentation). Based on RS, pts in each stage were clustered into 2 groups (high and low RS), showing 2 KM curves with p < 0.01 in each stage, confirming the validity of RS modeling. Analysis of immune profiles in these 2 groups shows that regardless of stage, expression of genes associated with Desmoplasia, Angiogenesis, and Epithelial-mesenchymal transition (DAE) is higher in high RS groups. Furthermore, expression of DEA genes in stage 4 correlated more strongly with poor prognosis than observed in stage 2-3 as evidenced by smaller p-value. Among stage 4 tumors, expression of genes related to IFN response, NK cells, and T1 helper cells is higher in low RS groups. In stage 2 and 3, genes related to immune activation and inhibition have no association with prognosis (p > 0.05). Conclusions: Machine learning-assisted mathematical modeling of RS and gene analysis show that genes related to immune activation are associated with better prognosis, while DAE genes correlate with poorer prognosis among advanced stages. RS enables prognostication of pts encountered in the clinic, given genomic profiles. [Table: see text]