Size-based Enrichment Of Urinary Cell-free DNA Compared To Plasma Cell-free DNA For Liquid Biopsy Analysis Of Oligometastatic Colorectal Cancer

Abstract

Recently groups including ours have performed noninvasive liquid biopsy analysis from urine for patients with genitourinary (GU) malignancy by adapting technology developed for plasma cell-free (cf) DNA analysis. Indeed, obtaining urine is less costly, less invasive and requires less clinical overhead than obtaining blood plasma. Suggesting such methods could also be adapted to non-GU cancers, cfDNA likely arising from blood plasma (based on size characteristics) has been described in urine cell-free (ucf) DNA. Here we develop a size-based DNA enrichment protocol to explore the feasibility of performing ucfDNA analysis of non-GU cancer, focusing on colorectal cancer (CRC), compared to plasma cfDNA analysis. Ten patients with oligometastatic CRC to the liver were included in this study. Urine, plasma, matched peripheral blood mononuclear cells (PBMCs) and tumor were collected from each patient (n = 40 samples). Using Q-Sepharose resin technique (Dudley, 2018), ucfDNA was isolated from urine. To remove high molecular weight DNA originating from urothelial cells, we tested different concentrations of AMPure magnetic beads to enrich for nucleosomal-sized DNA representative of plasma cfDNA (70-450 bp range). We then performed error-corrected targeted next-generation sequencing to all samples using the Roche AVENIO surveillance kit. 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. We collected an average of 37 ml of urine (range 19-50) and 10 ml of plasma (range 3-18) preoperatively. AMPure beads added at 0.6X concentration to ucfDNA isolated by Q-Sepharose resin technique resulted in the greatest enrichment of nucleosomal-sized DNA -- 92% in the 70-450 bp range, compared to 29% non-enriched (P = 0.03). An average of 49 nanograms (ng) of enriched ucfDNA and standard plasma cfDNA from each patient were then inputted into sequencing library preparation. Urine and plasma cfDNA samples were sequenced to a median de-duplicated depth of 4828X and 11285X with on-target rates of 66% and 71%, respectively. The average number of single nucleotide variants detected in ucfDNA vs. plasma cfDNA was 2 (range 0-11) vs. 8 (range 1-42) (P = 0.009). Tumor-derived DNA was detected in the preoperative urine of 6 patients (60%) with an average of 61 haploid genome equivalents per ml (hGE/ml) (range 15-118). In one patient, a targetable mutation (PIK3CA p.Glu545Lys) present in tumor was also detected in enriched ucfDNA, but not in plasma. Our findings demonstrate that magnetic bead enrichment significantly increases the fraction of ucfDNA representative of plasma cfDNA, enabling mutation analysis from urine in patients with oligometastatic CRC. While the overall number of mutations detected was low, a targetable PIK3CA mutation was detected in enriched ucfDNA that was missing in corresponding plasma.