Urinary cell-free DNA analysis for tumor mutation detection in patients with oligometastatic colorectal cancer.

Abstract

e15545 Background: Cell-free DNA (cfDNA) analysis from blood is a useful liquid biopsy strategy. However, it requires phlebotomy and moderate volumes of plasma. Here, we explored the feasibility of urinary cfDNA (ucfDNA) analysis in patients with oligometastatic colorectal cancer (CRC). Methods: We applied error-corrected targeted next-generation sequencing (NGS) (AVENIO ctDNA surveillance assay) to 10 patients with oligometastatic CRC to the liver who underwent curative-intent surgery. 40 paired samples (tumor, plasma, urine supernatant, peripheral blood mononuclear cells (PBMCs)) were collected, from which DNA was extracted. Plasma and urine cfDNA samples (acquired preoperatively) were sequenced to a median de-duplicated depth of 8,065X with on-target rate of 71%. Similarly high depth targeted NGS was applied to tumor and PBMC samples. Variants were called from plasma cfDNA and tumoral DNA and queried in ucfDNA using methods derived from CAncer Personalized Profiling by deep Sequencing (CAPP-Seq) with integrated digital error suppression (iDES). Clonal hematopoiesis was accounted for by removing mutations also present in PBMCs. Results: Patients had a median of 2 tumors resected (range 1-7) with average sum of longest diameters of 8.7 cm (range 0.6-11). We used an average of 10 ml of blood and 37 ml of urine for cfDNA extraction. Circulating tumor DNA (ctDNA) was detected in all preoperative plasma samples (n = 10) at a mean mutant allele fraction (MAF) of 2.35% (range 0.09-11.78). Plasma ctDNA MAFs correlated significantly with tumor SLD (Pearson r = 0.81, P = 0.03). The average number of single nucleotide variants detected in plasma cfDNA, tumor and ucfDNA was 8 (range 1-42), 4 (range 1-7) and 2 (range 0-11), respectively. Indicative of concordance, 79% of the mutations called in tumor were detected in plasma. Tumor-derived DNA was detected in the preoperative urine of 6 patients (60%) with a MAF of 0.05% (range 0.03-0.12), ~35-fold less than in plasma (MAF of 1.78%, range 0.09-7.94) (P = 0.004). 38% of mutations identified in tumor and 26% from plasma were detected in ucfDNA for these 6 patients. In two patients, targetable mutations (BRAF p.Asp594Gly and PIK3CA p.Glu545Lys) present in tumor were also detected in ucfDNA, with the latter not detected in plasma. Conclusions: Our results indicate that ucfDNA analysis in patients with oligometastatic CRC is feasible; however, the mutational levels and detection sensitivity were much lower than from plasma. Yet, there may be situations where actionable mutations are missed in plasma but detected in urine.