PURPOSE: To determine if carcinoma in situ (CIS) is a sufficient measure of minimal residual disease (MRD) in non-muscle-invasive bladder cancer (NMIBC) and to evaluate alternative measures of MRD. EXPERIMENTAL DESIGN: We evaluated concordance of CIS on TURBT and radical cystectomy (RC) to determine eradication rates of clinical CIS (cCIS) by TURBT alone and rates of occult pathologic CIS (pCIS) seen only in RC specimens. We studied a prospective cohort of patients with BCG-naive high-grade NMIBC to evaluate pretreatment urinary cytology and urinary tumor (ut)DNA as alternative biomarkers. RESULTS: Eradication of CIS was seen in 20% of patients (78/383). Positive urinary cytology, but not cCIS, was associated with pCIS. In our prospective cohort (n=173), abnormal pre-BCG urine cytology, but not pre-BCG cCIS, had worse high-grade recurrence-free survival (HG-RFS) (hazard ratio [HR] 3.56 95%, confidence interval [CI] 1.74 - 7.31, p<0.001). Median follow-up was 1.8 years (95% CI: 1.3-2.2). Among those with utDNA available, 84% (56/67) had sufficient DNA for genomic profiling. The 2-year HG-RFS rate in patients without an oncogenic alteration (n=17) in their pre-BCG urine was 100% versus 60% (95% CI: 46%-78%) in patients with detectable oncogenic alterations (n=38) (p=0.004). Area under the curve values for predicting 2-year HG-RFS were 0.52 for cCIS, 0.68 for cytology, and 0.74 for utDNA. CONCLUSIONS: We found ~20% eradication of cCIS from TURBT alone. cCIS was a poor metric of MRD, performing worse than abnormal pretreatment urinary cytology and utDNA. These urine biomarkers are more objective MRD measures than cCIS for NMIBC risk stratification and treatment assessment.
