Cell-free urinary tumor DNA to detect minimal residual disease prior to repeattransurethral resection of bladder tumor in non-muscle-invasive bladder cancer: A prospective study.

LBA445 Background: Cell-free urinary tumor DNA (utDNA) generated from tumor genomic sequencing is an emerging biomarker showing tremendous potential for detecting minimal residual disease (MRD) in muscle-invasive and metastatic bladder cancer. We performed a prospective study to assess utDNA’s ability to measure MRD at the time of standard-of-care repeat transurethral resection of bladder tumor (rTURBT) in non-muscle invasive bladder cancer (NMIBC). Methods: Patients with high-risk NMIBC were enrolled prior to rTURBT. Index tumor and rTURBT mutational profile was performed via PredicineWES whole exome sequencing across 20,000 genes. Genes with non-synonymous (NS) mutations observed in index and rTURBT specimens identified in a minimum of two samples are reported. Urine samples were taken immediately prior to r-TURBT. utDNA detection was performed by ultra-deep sequencing of urinary cfDNA using a custom panel of up to 50-baseline mutations per patient together with a fixed core panel covering hotspot regions and actionable variants via PredicineBEACON. Urinary tumor fraction was used to call utDNA positivity. The primary endpoint was the detection of utDNA to predict MRD at the time of repeat-TURBT. Results: Eleven patients underwent rTURBT for high-risk NMIBC. Residual tumor was detected at rTURBT in 8/11 (73%) patients. Fifty genes with NS mutations found in at least two samples were identified including TP53, PIK3CA, RB1, MYC, CDKN1A and ARID1A. A median of 146 (range 39-418) and 91 NS (range 2-312) mutations were identified in the index and re-TUR specimens, respectively. Concordance rates were 83% on average between index and rTURBT, highest for patients upstaged to T2 disease, and lowest for patients harboring CIS. New somatic variants not observed in the index tissue were detected in rTURBT for 4/6 patients and 7/14 were also detected in the pre-operative urine. Overall, 50% of utDNA mutations were concordant with index tumor tissue, 39% were concordant with rTURBT tissue, and 33% were concordant with both. The tumor fraction of utDNA was higher in patients with residual disease (mean 2.5% vs. 0.2%, p=0.10). Using tumor-fraction to predict MRD, the area under the receiver-operator curve was 0.85. Using an optimal threshold of ≥3.3% tumor-fraction to define utDNA positivity, the test had a sensitivity of 75% at 100% specificity. Conclusions: Urinary tumor DNA shows promise as a surrogate for minimal residual disease and may predict TURBT pathology for NMIBC. Genomic alterations between index and rTURBT tumors are highly concordant in papillary tumors even in the setting of upstaging, which may aid in targeted intravesical or systemic therapy selection. Larger cohorts and long-term follow up is needed to determine if utDNA can risk-stratify patients prior to rTURBT and predict long-term recurrence.