Updated results from a pilot trial of neoadjuvant durvalumb in combination with tremelimuab for cisplatin-ineligible high-risk urothelial carcinoma.

Abstract

e16584 Background: We previously reported that neoadjuvant immune checkpoint therapy (ICT) with durvalumab (D) plus tremelimumab (T) was safe and effective in cisplatin-ineligible patients (pts) with high-risk muscle-invasive urothelial carcinoma (MIUC). Here, we present outcomes from a second cohort treated with higher-dose T plus D, updated outcomes from the first cohort, and an analysis of tertiary lymphoid structures (TLS). Methods: Cohort 2 (n = 13) pts with cisplatin-ineligible MIUC and high-risk features (lymphovascular invasion, bulky tumor, variant histology, high grade upper tract disease, and/or hydronephrosis) received D 1500mg + T 300mg, followed by D 1500mg 4 weeks later. Cohort 1 (n = 28) pts were treated with D 1500mg + T 75mg every 4 weeks for 2 cycles. Cystectomy was planned 4-6 weeks after therapy. Primary endpoint was safety; overall survival (OS) and relapse free survival (RFS) were secondary endpoints. All pts were included for OS analysis and only pts who completed cystectomy were included for RFS analysis using Kaplan Meier method. Exploratory analysis assessed TLS association with response (downstage to pT< / = 1N0) using Chi-square test. TLS association with OS/RFS were evaluated using Cox proportional hazard regression. TLS were defined as lymphocyte aggregates co-expressing CD20 and CD3 with germinal-center-like morphology. Results: In cohort 2, 9/13 pts underwent cystectomy; two declined surgery and two were unable to proceed due to treatment-related toxicities. Three pts experienced grade 3+ immune related adverse events (irAEs): colitis (n = 1), hepatitis (n = 1), and myositis/myocarditis (n = 1) myositis/myocarditis (n = 1), prompting early termination of this cohort. Among pts completing cystectomy, pathologic complete response rate was 44% (4/9) and 1-year RFS 67% (SE = 16%). 1- year OS was 62% (SE=14%) and the two pts unable to proceed to surgery due to irAEs died before receiving additional treatment while the two pts who declined surgery continue to have no evidence of disease. Updated 5-year OS for cohort 1 was 82% (SE = 7%) and RFS was 74% (SE = 9%). In combined cohorts, 69% (11/16) TLS high pts were responders vs 29% (6/21) TLS low pts (OR 0.18, 95% CI 0.04, 0.75). Median RFS for TLS high not reached (95% CI NR,NR) compared to 59 months (95%CI 8.4-NR) for TLS low (HR 0.24, 95%CI 0.05-1.1, p = 0.047); median OS for TLS high not reached (95%CI NR, NR) compared to 63 months (95%CI 14, NR) in TLS low (HR 0.32, 95% CI 0.09-1.2, p = 0.068). Conclusions: Neoadjuvant ICT remains promising for cisplatin-ineligible MIUC; however, higher T dose increased serious irAEs, underscoring the need for optimized dosing. Baseline TLS density may represent a predictive biomarker and warrants further validation with larger cohorts. Clinical trial information: NCT02812420 .