800 Background: EVP is active in aUC; however, biomarkers predicting early benefit are lacking. ctDNA is a noninvasive marker of tumor dynamics, but its predictive value in this setting is unknown. We hypothesized that early ctDNA response (ctDNA-R) predicts long-term outcomes with EVP. Methods: Patients (pts) who received EVP and had baseline ctDNA within the UNITE study were included. Early ctDNA-R was defined as reduction in ctDNA from baseline (−4/+2 weeks relative to EVP start) to the second measurement (through ≤3 months after EVP end or before subsequent therapy), with decline evaluated as continuous percent drop, ≥50% drop, and ≥90% drop. Associations with progression-free survival (PFS), overall survival (OS), and observed response rate (ORR) were assessed. PFS and OS were analyzed using Kaplan–Meier and Cox regression, and ORR using Firth logistic regression in pts with scans after ≥1 EVP cycle. Median baseline ctDNA was also evaluated for association with outcomes. Results: Among 48 pts treated with EVP across 3 US sites, 44 had measurable ctDNA at baseline (median 18.5, range 0 – 995), and 33 of these were evaluable for ctDNA dynamics with a second timepoint. Among the evaluable pts, median age was 72 years, 76% were male, 76% had pure urothelial histology, 97% had ECOG PS 0/1, 42% had visceral metastases, 82% were treatment naive in metastatic setting, and 30% had prior immunotherapy exposure in any setting. All pts had ctDNA measured using the Signatera assay, with a median baseline ctDNA of 29.5 MTM/mL (range 0.07 - 995). With a median follow-up of 12 months, median PFS and OS were 12 mos (95% CI 6–NR) and 23 mos (95% CI 12–NR), and ORR was 52% (16/31; 95% CI 39–64). The median time from baseline to the second ctDNA read was 7 weeks (range 3–25), with a median ctDNA-R of 96% (range 0–100) over this timeframe. Early ctDNA-R ≥50% occurred in 82% (27/33), ≥90% in 58% (19/33), and complete clearance in 30% (10/33). Greater ctDNA-R was associated with improved PFS, OS, and ORR across both continuous and categorical thresholds, while baseline ctDNA levels (above vs. below median) was not (Table). Conclusions: In this multi-site retrospective analysis, early ctDNA-R was associated with improved outcomes with EVP, supporting its potential as a predictive biomarker for early treatment monitoring. Despite potential selection bias, these hypothesis-generating findings merit validation in larger prospective cohorts. ORR: OR (95% CI) PFS: HR (95% CI) OS: HR (95% CI) Percent change (continuous) 1.05 (1.01 – 1.15), p = 0.002 0.96 (0.94 – 0.98), p < 0.001 0.97 (0.95 – 0.99), p = 0.001 ≥50% drop 22.6 (2.25 - 3069) , p = 0.005 0.06 (0.02 – 0.25), p < 0.001 0.07 (0.13 – 0.37), p = 0.002 ≥90% drop 7.36 (1.67– 40.02), p = 0.008 0.19 (0.07 – 0.55), p = 0.002 0.14 (0.03 – 0.61), p = 0.009 ≥ Median baseline ctDNA 1.53 (0.43 – 5.70), p = 0.5 0.89 (0.35 – 2.23), p = 0.8 0.60 (0.18 – 2.00), p = 0.4
Jindal et al. (Sun,) studied this question.