MP06-01: Non-muscle Invasive Bladder Cancer Subtypes with Differential Response to Intravesical Bacillus Calmette-Guerin Treatment

Florus C de Jong*, Rotterdam, Netherlands, Teemu D Laajala, Rotterdam, Finland, Robert F Hoedemaeker, Rotterdam, Netherlands, Sébastien Rinaldetti, Luxembourg, Luxembourg, Kimberley R Jordan, Denver, CO, Angelique CJ van der Made, Rotterdam, Netherlands, Bart Nieuwkamer, Delft, Netherlands, Deric KE van der Schoot, Breda, Netherlands, Egbert R Boevé, Rotterdam, Netherlands, Emiel AM Janssen, Stavanger, Norway, Tokameh Mahmoudi, Joost L Boormans, Rotterdam, Netherlands, Dan Theodorescu, Los Angeles, CA, James C Costello, Denver, CO, Tahlita CM Zuiverloon, Rotterdam, Netherlands

Introduction: The recommended treatment for high-risk non-muscle invasive bladder cancer (NMIBC) patients is intravesical Bacillus Calmette-Guérin (BCG) instillations, yet only 50% of patients benefit from BCG therapy. Risk stratification is based on clinicopathological characteristics. Molecular features of BCG treatment failure is understudied. We aimed to improve risk stratification of high-risk NMIBC and identify molecular correlates of treatment failure by whole transcriptome analysis.

Methods: Tumors from primary high-risk NMIBC patients treated with BCG were centrally reviewed by a uropathologist. Cohort A consisted of n=132 BCG-naïve tumors (n=69 non-responders vs n=63 responders) and n=44 matching post-BCG recurrences. RNA-sequencing was performed on RNA isolated from paraffin embedded tumor tissue containing >80% cancer cells. The primary end point was progression-free survival (PFS), defined as time from diagnosis until muscle-invasive or (lymph node) metastatic disease. Consensus clustering was used to identify molecular subtypes. Next, to investigate the subtypes associated with PFS, differential expression, pathway, immune deconvolution and regulon analyses were performed . Finally, findings were validated in an independent Cohort (B), which consisted of n=151 BCG-naïve tumors.

Results: Analysis of RNA-sequencing data revealed three BCG response subtypes (BRS). In multivariable analysis, patients with BRS3 tumors had reduced PFS (Cohort A, p<0.01) as compared to patients with BRS1/2 tumors. BRS3 tumors expressed high EMT-basal markers and had an immunosuppressive profile, which was confirmed with spatial proteomics (T regs, Macrophages). The correlation of BRS subtypes with PFS was validated in Cohort B. The BRS stratification improved current recommended risk stratification, which is based on clinicopathological variables. A commercially approved assay (Philips OncoSignal®) identified BRS3 tumors with high accuracy (AUROC 0.87), paving the way for routine clinical application. Tumors which recurred post-BCG were highly enriched for the BRS3 subtype. Several BRS3 druggable regulators (TGFß, DDR2 etc.) implicated in resistance to checkpoint inhibitors were identified and serve as attractive candidate molecular targets for novel treatments in patients that do not respond to BCG.

Conclusions: Molecular subtyping identified tumors from high-risk NMIBC patients with an aggressive phenotype and a poor response to BCG. Our findings provide a clinical tool for improved identification of high-risk NMIBC patients at high risk of progression, which can be used to select patients for early radical cystectomy or to develop novel subtype-directed therapies.

Source of Funding: Erasmus MC MRace; Anschutz Medical Foundation; Finnish Cancer Institute; Stichting De Drie Lichten.