837 Background: We published a prospective trial of Systematic Endoscopic Evaluation (SEE) prior to radical cystectomy (RC) as a predictor of pT0. The results demonstrated that 26% of patients with negative SEE harbored pT≥2 disease, and the negative predictive value (NPV) of SEE to determine pT0 disease was 48.4%. Here we present a post-hoc analysis of plasma cell free DNA collected the day of surgery to determine if circulating tumor (ctDNA) improves NPV for detecting pT0 disease to better identify patients for bladder preservation. We also present survival updates. Methods: This was a prospective study of patients with muscle-invasive and non-muscle invasive bladder cancer planned to undergo RC. Plasma collected the day of surgery was processed, stored, and sent to GRAIL for extraction of cfDNA and calculation of a ctDNA-based cancer score using a proprietary tumor-naïve methylation platform. All patients underwent SEE during surgery to predict pT stage, and patients were followed for recurrence and overall survival (RFS and OS). ctDNA status with and without SEE findings were evaluated as predictors of pT stage using Bayesian logistic regression with area under the receiver operating curve (AUROC) calculation. A grid search was used to identify a post-hoc cutoff on ctDNA score (either positive or negative) that maximized the AUROC for pT stage. RFS and OS were evaluated using Kaplan–Meier survival curves and Cox proportional hazards regression. Results: 54 of 61 patients accrued to the trial had evaluable ctDNA and pT data. A model of pT stage including SEE and continuous ctDNA-based cancer score resulted in an AUROC of 0.94. Results using the post-hoc optimized ctDNA threshold with and without SEE to predict pT stage (pT=0 vs pT>) are shown in the table. Combining SEE and ctDNA only 2 patients negative by SEE and ctDNA had pT>0 disease, correlating to a NPV of 83%. One had pTis and never recurred, while the other had fatal pT2b disease. Median OS was not reached (NR) for ctDNA negative patients vs 5.7 yrs for ctDNA positive patients (p=0.011). Median RFS was NR vs 4.7 years for ctDNA negative vs positive patients, respectively (p=0.048). Conclusions: In this post-hoc, exploratory analysis of a prospective study, ctDNA at the time of surgery combined with SEE improves predictors of pT stage and is prognostic for RFS and OS. Adding ctDNA to SEE yields a NPV of 83% for pT0 disease, compared to 48.4% for SEE alone. This analysis requires validation in additional datasets, but the high NPV is encouraging and warrants further prospective study using ctDNA with SEE to identify patients confidently able to forgo RC. ctDNA and SEE to predict for pT0 disease ctDNA+VsctDNA- ctDNA + OR SEE+ vsctDNA- AND SEE- Sensitivity 0.68 (27/40) 0.95 (38/40) Specificity 0.71 (10/14) 0.71 (10/14) Positive Predictive Value 0.87 (27/31) 0.90 (38/42) Negative Predictive Value 0.43 (10/23) 0.83 (10/12)
Zibelman et al. (Sun,) studied this question.