Introduction:

Upper tract urothelial carcinoma (UTUC) is an aggressive cancer for which use of neoadjuvant chemotherapy (NAC) is limited by suboptimal clinical staging prior to radical nephroureterectomy (RNU). Detection of circulating tumor DNA (ctDNA) is associated with locally advanced and metastatic urothelial carcinoma of the bladder and may help identify UTUC patients who would benefit from NAC. However, the genomic profiles of UTUC tumors in the context of their concordance with detectable plasma ctDNA and urinary tumor DNA (utDNA) remain poorly elucidated. Herein, we prospectively sequenced UTUC tumors, ctDNA, and utDNA at the time of RNU in order to examine the feasibility and utility of plasma ctDNA and utDNA detection in the diagnosis of muscle-invasive and non-organ confined UTUC.

Methods:

Patients with high-grade UTUC without radiographic evidence of metastases undergoing up-front RNU were prospectively accrued with IRB approval. Blood and urine were collected preoperatively on the day of surgery, and plasma, buffy coat, and urine supernatant were processed for extraction of ctDNA, genomic DNA, and utDNA, respectively. Formalin-fixed paraffin embedded (FFPE) tumor tissue from RNU was used for tumor genomic DNA extraction. Next-generation sequencing (NGS) using both shallow whole-genome sequencing and a 152-pancancer panel (PredicineCARE) profiled variants. Copy number burden (CNB) and cell-free DNA tumor fraction (TF) were calculated from genome-wide plasma and urine copy number alterations.

Results:

Of 32 patients, 16 were found to have noninvasive (<pT2 and pN0/x) and 16 muscle-invasive or NOC (≥pT2 or ≥pN1) UTUC at RNU. NGS identified variants in 30 matched preoperative FFPE and plasma ctDNA (Figure 1) and 11 matched preoperative utDNA. The most common tumor alterations were in the TERT promoter, tp53, FGFR3, myc, and ERBB2 (Figure 2a).

Seventy-seven percent of patients had at least one detectable plasma ctDNA alteration using the targeted NGS panel. Using a tumor-informed approach, two or more plasma ctDNA variants were detectable in 11/14 of muscle-invasive or NOC UTUCs, resulting in a 79% sensitivity, and 3/16 noninvasive UTUCs, resulting in a 81% specificity (AUC=0.86).  Mean CNB and TF was significantly higher for muscle-invasive or NOC UTUC in both plasma ctDNA and utDNA (Figure 2b-c).

Conclusion:

Prospective ctDNA and utDNA analysis using a targeted NGS panel is a feasible nonsurgical approach to prediction of muscle-invasive and non-organ-confined UTUC and has the potential for identification of upper tract urothelial cancer patients that may derive benefit from NAC.

Funding: Support for sequencing was provided by Predicine, Inc.