Five-year median follow-up update of PURE-01: A phase 2 study of neoadjuvant pembrolizumab followed by radical cystectomy in patients with muscle-invasive bladder cancer (MIBC).

4595 Background: In patients (pts) with MIBC, the PURE-01 study pioneered the use of neoadjuvant immunotherapy by administering 3 courses of pembrolizumab before radical cystectomy (RC). Multiple reports have outlined various results of this trial, but assessment of the very long-term benefit of this strategy was pending and potentially noteworthy. Methods: Pts age ≥18 y, ECOG PS 0-1, with histologically confirmed cT2-T4N0M0 MIBC, ineligible/refusing chemotherapy (CT), and scheduled for RC received 4 cycles of pembrolizumab 200 mg on D1 Q3W, followed by RC and standard-of-care management. A total of 155 pts were included in the study. Even-free survival (EFS), relapse-free survival (RFS) in the RC population and overall survival (OS) are reported. Cumulative risk of recurrence (CRR) by competing risk analyses was also reported. Updated results from transcriptome-wide profiling using the Decipher Bladder assay (Veracyte, San Diego, CA) on primary TURBT tissue are presented. Results: From 02/17 to 07/20, 155 pts were treated (12.9% females). Remaining clinical and pathological features yet have been reported. Median follow-up was 61.8 months (IQR: 53.6-68.2). In the intention-to-treat (ITT) population, 5y-EFS was 68.3% (95%CI: 61.1-76.4) and 5y-OS was 77.4% (95%CI: 70.6-84.8). Pathological response categories were significantly associated with both RFS (p < 0.001) and OS: 5y-OS for ypT0N0 responders was 89.5% vs 90.3% for ypTa/Tis/T1N0 vs 72.2% for ypT2N0 vs 58.8% for ypT3-4N0 vs 41.9% for ypTanyN+ (p < 0.001). Out of 31 total relapses, three were very late relapses > 5y post-RC, of which all were visceral relapses including isolated brain metastases in one case. The 5y-CRR was 19% (95%CI: 13-26). 7/8 pts who refused to undergo RC and received a reTURBT are alive and disease-free. Transcriptome-wide profiles were available for 102 pts. Stratification by Genomic Subtyping Classifier (GSC) allowed significant separation of RFS curves: Claudin Low subtype (N = 14) confirmed to portend the highest 5y-RFS (Log-rank Claudin-Low vs others vs NE-like p = 0.02), with 5y-OS for Claudin Low subtype being 93% (Log-rank Claudin-Low vs others vs NE-like p = 0.29). Higher immune infiltration scores quantified by Immune190 signature were associated with improved OS (HR (95% CI) per 0.1 increase = 0.60 (0.42-0.87); p = 0.006) Biologic evaluation among N = 604 MIBC GRID patients revealed highest Immune190 scores for Claudin Low subtype (p < 0.001). Conclusions: After > 5y median follow-up, PURE-01 study revealed a sustained response and survival in pts with MIBC. Important updates are the validation of pathological response as a surrogate of OS in post-IO setting, the validation of molecular subtypes in association with RFS (and possibly OS) benefit, and the need to prolong follow-up at long term due to few pts with delayed recurrences. Clinical trial information: NCT02736266 .