Urinary tumor DNA–guided de-escalation of systemic immunotherapy in extensive very high risk non–muscle-invasive bladder cancer.

4593 Background: Non-muscle-invasive bladder cancer (NMIBC) has entered the era of systemic immunotherapy. Large randomized trials, including CREST, POTOMAC, and ALBAN, have explored long-term systemic immunotherapy in NMIBC, yet the clinical benefit remains heterogeneous and is often accompanied by substantial toxicity. Urinary tumor DNA (utDNA) has emerged as a promising biomarker for molecular response assessment, yet its role in guiding treatment de-escalation has not been established. We evaluated whether utDNA negativity after complete response (CR) could support safe early treatment cessation of systemic therapy in patients with extensive very-high-risk NMIBC (VHR-NMIBC). Methods: Patients with extensive VHR-NMIBC who achieved CR after initial systemic therapy with the PD-1 inhibitor tislelizumab plus low-dose paclitaxel were included from a prospective cohort. After CR assessment, some patients underwent low-coverage utDNA testing as part of routine clinical decision-making following clinician–patient discussion. Patients with negative utDNA completed three additional cycles and subsequently stopped systemic therapy (utDNA-guided de-escalation group), whereas those who continued treatment beyond this point were categorized as the longer treatment-exposure group. Efficacy and safety outcomes included survival endpoints and adverse events (AEs) with exposure-adjusted incidence rates. Results: With a median follow-up of 42.8 months, 54 patients achieving CR were evaluable. Among them, 21 with negative utDNA status completed three additional cycles and underwent planned treatment cessation, while 33 continued systemic therapy with longer treatment exposure. Across key efficacy endpoints, outcomes were comparable between the utDNA-guided de-escalation and longer treatment–exposure groups, with no significant differences in recurrence-free survival (RFS; HR 1.37, 95% CI 0.21–8.87, P = 0.73), overall survival (OS; HR 2.48, 95% CI 0.12–53.7, P = 0.51), or duration of response (DOR; HR 1.33, 95% CI 0.21–8.50, P = 0.76). A difference in radical-cystectomy–free survival was observed (short vs long: HR 4.97, 95% CI 1.03–23.9, P = 0.03). Long-course treatment was associated with a higher cumulative incidence of any-grade AEs (88.1% vs 72.5%) and more AE types (median 5 vs 2, P < 0.001). Exposure-adjusted AE rates per cycle were lower in the long-course group (IRR 0.50, 95% CI 0.42–0.60, P < 0.001), while grade ≥3 AEs occurred in 2.9% and 9.5% of patients in the short- and long-course groups, respectively, predominantly during later treatment phases. Conclusions: UtDNA negativity after complete response may provide a molecular basis for individualized treatment duration, highlighting the potential role of liquid biopsy in guiding precision treatment tailoring and warranting prospective validation in larger biomarker-driven studies. Clinical trial information: NCT04730232 .