We aim to develop a molecular pathology tool that enables comprehensive cell type and subtype identification in formalin-fixed, paraffin-embedded (FFPE) tissue sections of treatment-naïve high-risk non-muscle invasive bladder cancer (HR-NMIBC) patient samples. Working from the hypothesis that the progression of HR-NMIBC can be predicted by spatial investigation of epithelial, immune, and stromal features within NMIBC, we hope to provide a predictive tool to stratify HR-NMIBC and thereby guide clinical choice, specifically a choice between Bacillus Calmette-Guerin (BCG) intravesical immunotherapy or radical cystectomy. First, we employ droplet-based single-cell transcriptomics (SCT) to provide an unbiased view of cell type and cell state heterogeneity within NMIBC. Next, we utilize this information to develop an antibody panel for use in imaging mass cytometry (IMC). IMC provides the capability to measure and spatially resolve up to 35 epitopes simultaneously at 1-micron resolution. To date, we have profiled 31,673 single-cell transcriptomes across 12 treatment-naïve NMIBC patient samples. On average, we detect 9,198 unique transcripts and 2,162 unique genes per cell, providing a rich resource to characterize cellular heterogeneity in NMIBC. Batch analysis of all cells from all samples indicates greatest variability across the urothelial populations—the presumed cancer cells—as would be expected due to patient-specific differences in the accumulation of somatic mutations within these cells. While percentages of distinct noncancer cell types (i.e., cells of the microenvironment) differed between patients, cells from across patients intermingled within their respective cell type-specific clusters. Cell types identified within the microenvironment included T cells, B cells, conventional dendritic cells 1 (cDC1) and 2 (cDC2), mature DCs, plasmacytoid DCs, macrophage, endothelial, and mesenchymal cells. Besides providing cell type-specific transcriptome signatures to assist in deconvoluting bulk transcriptome data sets, these single-cell data provide amply information to design a comprehensive antibody panel for use in IMC capable of detecting all represented cell types. This panel is currently under development and now contains 25 validated antibodies. Once established, this will be applied to a set of treatment-naïve NMIBC FFPE tissue sections from patients with follow-up clinical outcome data to test our hypothesis of cell-type content and spatial localization being predictive of outcomes.

Citation Format: Dylan Baker, Elise T. Courtois, Santhosh Sivajothi, Benjamin T. Ristau, Joshua J. Meeks, Paul Robson. Developing a multiplex imaging tool for cellular phenotyping to stratify non-muscle invasive bladder cancer patients [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2019 May 18-21; Denver, CO. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(15_Suppl):Abstract nr B17.