Impact of gene expression signatures of immune infiltration on response to sequential intravesical gemcitabine and docetaxel versus bacillus Calmette-Guerin in high-risk non-muscle-invasive bladder cancer.

832 Background: Intravesical Bacillus Calmette-Guerin (BCG) is the guideline-recommended therapy for high-risk non-muscle invasive bladder cancer (HR-NMIBC). However, alternative therapies for HR-NMIBC are currently under investigation due to persistent global BCG shortages. Sequential intravesical gemcitabine and docetaxel (Gem/Doce) has demonstrated similar efficacy to BCG in this setting, and a large, randomized trial versus BCG is underway. As such, biomarkers for treatment selection are an unmet clinical need. Methods: We performed a retrospective analysis of a matched cohort study (McElree et al. JAMA Netw Open. 2023) of 143 patients with treatment-naïve HR-NMIBC treated at the University of Iowa with either BCG (n=92) or Gem/Doce (n=51). Gene expression data from archived bladder tumor specimens were generated using Decipher Bladder Genomic Subtyping Classifier (GSC, Veracyte, San Diego, CA), a clinical-grade, transcriptome-wide assay. Molecular subtypes were classified using the GSC and Consensus subtyping models. ESTIMATE (Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data) scores were used to stratify patients into higher (>median) and lower (<median) ESTIMATE immune scores. The primary endpoint was high-grade recurrence-free survival (HG-RFS). Associations between molecular subgroups and HG-RFS were evaluated using Cox-regression and Kaplan-Meier analyses. Results: Of the cohort, 34% were Ta, 64% T1, and 2% T is alone; overall 30% of patients had concomitant CIS. Median patient age, clinical tumor stage, presence of CIS and molecular subtype distributions did not differ significantly between BCG and Gem/Doce treated patients (all p>0.5). Median follow-up was 49 months for patients who received BCG and 22 months for patients who received Gem/Doce. HG disease recurrence was observed in 36% of BCG treated patients and in 16% of Gem/Doce treated patients. In total, 85% were classified as GSC luminal and 71% as Consensus luminal papillary subtypes and median ESTIMATE immune score was 556 (IQR: 126-1410). Patients with higher immune scores had superior HG-RFS (hazard ratio [HR], 0.25; 95% CI; 0.07-0.85; P=0.02) with Gem/Doce as compared to BCG. At 2 years, those with higher immune scores had better HG-RFS with Gem/Doce versus BCG (90% versus 63%, HR 0.25; 95% CI; 0.07-0.85; P=0.02), whereas HG-RFS was similar in those with lower scores (86% versus 72%, HR 0.71; 95% CI; 0.26-1.95; P=0.50). Conclusions: This study represents the first molecular comparison of treatment-naïve HR-NMIBC treated with Gem/Doce versus BCG. While this HR-NMIBC cohort is predominately luminal molecular subtype, immune signature scores at diagnosis show substantial variation. Patients with high immune scores at diagnosis have particularly suboptimal response to BCG and may derive greater benefit from Gem/Doce. These findings require validation.