Circulating tumor DNA (ctDNA) dynamics as early biomarkers of response to enfortumab vedotin plus pembrolizumab (EVP) in advanced urothelial carcinoma (aUC).

4582 Background: Radiographic assessment of treatment response remains challenging in aUC. ctDNA testing offers a sensitive, noninvasive measure of real-time treatment response. With EVP established as a frontline standard, we hypothesized that ctDNA dynamics correlate with radiographic response and survival. Methods: We retrospectively analyzed an institutional cohort of 302 patients (pts) with aUC treated with EVP at MD Anderson; 79 pts with baseline and serial ctDNA testing using a tumor-informed assay (Signatera) (08/23–01/26) were included. ctDNA dynamics were evaluated in two ways: (1) early on-treatment change (baseline vs. first on-treatment ctDNA) for association with radiographic response, and (2) time-dependent modeling as a covariate in Cox proportional hazards analyses of progression-free survival (PFS) and overall survival (OS) to account for immortal-time bias. Results: All pts had a median of 4 ctDNA tests (range: 1-9). Baseline ctDNA was obtained before or at EVP start in 67 (84.8%). Among these patients, first ctDNA clearance occurred in 46.3%, decrease in 26.9%, and increase in 19.4%; 6% had persistently undetectable ctDNA and 1.5% was lost to follow-up. Median time to first ctDNA reduction (clearance or decrease) was 48 days IQR: 40–86. Pts without baseline ctDNA prior to EVP (12, 15.2%) were non-evaluable. The best overall response rate to EVP was 53/79 (67.1%), including complete in 13.9% and partial responses in 53.2%. Objective responses were highest among pts with first ctDNA clearance (87.1%), lower with first ctDNA decrease (55.6%), and uncommon with first ctDNA increase (38.4%) (χ² test, p =0.003). In the overall cohort (n=79), median PFS from EVP start was 20.4 months (mo) 95% CI: 8–NE and median OS from EVP start was 30.3 mo 95% CI: 26.3–34.3, with a median follow-up of 12.1 mo 95% CI: 9.1–14.9. In time-dependent Cox models accounting for immortal-time bias, increasing ctDNA levels over time were associated with inferior PFS (HR 1.49 95% CI: 1.11–2; p =0.007) and OS (HR 1.41, 95% CI: 0.96–2.08, p =0.08). Conclusions: Early ctDNA dynamics provide a clinically meaningful measure of response to EVP in aUC and correlate with imaging outcomes. ctDNA clearance identifies pts with deep and durable benefit, while lack of clearance flags early resistance. Longer follow-up will evaluate ctDNA-guided monitoring to inform treatment de-escalation or maintenance approaches. Baseline characteristics of cohort (n=79). Variable Measure Age at aUC diagnosis, median IQR 71 65 – 76 Sex, male, n (%) 56 (70.9%) Tumor Origin, n (%)BladderUpper Tract 61 (77.2%)18 (22.8%) Visceral mets, n (%) 34 (43%) Nodal-only mets, n (%) 40 (50.6%) Baseline ctDNA before/at EVP start (MTM/mL), median IQRAll (n=67)Visceral mets (n=30)Nodal-only mets (n=33) 12.9 1.1 – 105.512.8 1.3 – 140.216.9 1.4 – 96