Background: Liver resection can be curative for well-selected metastatic colorectal cancer (CRC) patients. Circulating tumor DNA (ctDNA) has shown promise as a biomarker for tumor dynamics and recurrence following CRC resection. This prospective pilot study investigated the use of ctDNA to predict disease outcome in resected CRC patients.
Methods: Between November 2014 and November 2015, 60 patients with CRC were identified and prospectively enrolled. During liver resection, blood was drawn from peripheral (PERIPH), portal (PV), and hepatic (HV) veins, and 3-4 weeks postoperatively from a peripheral vein (POSTOP). Kappa statistics were used to compare mutated (mt) genes in tissue and ctDNA. Disease-specific and disease-free survival (DSS and DFS) were assessed from surgery with Kaplan-Meier and Cox methods.
Results: For the 59 eligible patients, the most commonly mutated genes were TP53 (mtTP53: 47.5%) and APC (mtAPC: 50.8%). Substantial to almost-perfect agreement was seen between ctDNA from PERIPH and PV (mtTP53: 89.8%, κ = 0.73, 95% confidence interval [CI] 0.53-0.93; mtAPC: 94.9%, κ = 0.83, 95% CI 0.64-1.00), as well as HV (mtTP53: 91.5%, κ = 0.78, 95% CI 0.60-0.96; mtAPC: 91.5%, κ = 0.73, 95% CI 0.51-0.95). Tumor mutations and PERIPH ctDNA had fair-to-moderate agreement (mtTP53: 72.9%, κ = 0.44, 95% CI 0.23-0.66; mtAPC: 61.0%, κ = 0.23, 95% CI 0.04-0.42). Detection of PERIPH mtTP53 was associated with worse 2-year DSS (mt+ 79% vs. mt- 90%, P = 0.024).
Conclusions: Peripheral blood reflects the perihepatic ctDNA signature. Disagreement between tissue and ctDNA mutations may reflect the true natural history of tumor genes or an assay limitation. Peripheral ctDNA detection before liver resection is associated with worse DSS.