Neoadjuvant atezolizumab (A) with gemcitabine and cisplatin (GC) in patients (pts) with muscle-invasive bladder cancer (MIBC): A multicenter, single-arm, phase 2 trial.

4517 Background: Neoadjuvant GC is standard for pts with MIBC and can result in pathologic downstaging to non-MIBC ( < pT2N0) at radical cystectomy (RC), which correlates with improved survival. Based on the known activity of A in metastatic BC (mBC), we tested the combination of GC+A as neoadjuvant therapy for MIBC in a phase II trial (NCT02989584). Methods: Eligible pts with MIBC (cT2-T4aN0M0) received a single dose of A (1200 mg IV) and, two weeks later, began C (as either 70mg/m2 IV on D1 or 35 mg/m2 on D1,D8), G (1000 mg/m2 on D1,D8), and A (1200 mg IV on D8) every 21 days for 4 cycles followed by RC. Pts were also able to receive one additional dose of A 3 weeks after the last dose of A and prior to RC. The primary endpoint was proportion of pts with < pT2N0. Pts were considered not evaluable for the primary endpoint if they received less than 2 cycles due to withdrawal of consent or unrelated adverse events (AEs). Secondary endpoints included the proportion of pts with pT0N0, recurrence-free survival (RFS), and safety. We prespecified null and alternate < pT2N0 rates of 35% and 55%, respectively, with the null being rejected if at least 19 of 39 pts achieved < pT2N0. Pretreatment tumors underwent centralized PD-L1 staining (SP142; positive if ≥5% of immune cells). Results: Between Feb 2018 and May 2020, 44 pts were enrolled from five institutions. Five pts were not evaluable (withdrawal of consent before C3, n = 4; unrelated AEs during C1, n = 1). Of the 39 evaluable pts (cT2N0 79%, cT3N0 18%, cT4N0 3%), 1 pt refused surgery and was considered a non-responder. The primary endpoint was met, with 27 of 39 pts (69%) < pT2N0 at RC, including 15 (38%) pT0N0. All pts achieving < pT2N0 are alive and disease free. The median RFS was not reached with a median follow-up of 16.7 months (range: 7.7-33.2). The median time from last dose of chemotherapy to RC was 7.8 weeks (range 5.1 – 17). The most common grade 3-4 treatment related AEs were due to chemotherapy and were neutropenia (36%), lymphopenia (16%), and anemia (11%). Possible grade 3-4 immune related AEs included 2 pts with asymptomatic grade 3 pancreatic enzyme elevation, 1 pt with grade 3 pancreatitis, and 1 pt with hepatitis requiring steroids. Only 4 of 39 (10%) pts had PD-L1 positive tumors, which is low compared to mBC (25% positive; Powles et al. Lancet 2017) and MIBC (40% positive; Powles et al. Nature Med 2019) cohorts also tested with the SP142 clone. All 4 pts with PD-L1 positive tumors achieved < pT2N0. Twelve of 12 (100%) non-responding pts were PD-L1 negative and 23 of 27 (85%) responding pts were PD-L1 negative (p = 0.3). Conclusions: Neoadjuvant GC+A is an effective and safe regimen for the treatment of pts with MIBC. The PD-L1 positivity rate was low compared with other studies and was not predictive of pathologic downstaging. Additional interrogation of the genomic and host immune factors mediating response and resistance to GC+A is ongoing. Clinical trial information: NCT02989584.