Cell-free KRAS in pancreatic cancer as a rapid prognostic marker.

Abstract

e15534 Background: KRAS mutations are present over 90% of patients with pancreatic cancer (PC). Liquid biopsies provide the opportunity to genotype alterations in a less invasive way and offer a possibility to monitor the molecular characteristics of a cancer through the course of treatment. We aim to establish an association between the mutational levels of KRAS detected in plasma with patient stage and evolution. Methods: Plasma samples were collected from patients with newly diagnosed PC, before any treatment and during one year. Diagnosis of any previous cancer was an exclusion criterion. Samples were frozen (-80ºC) until analysis of KRAS mutations in cell free DNA (cfDNA) by real time PCR. Specifically, Idylla TM technology was used due to the rapid capacity to report up to 21 KRAS mutations in exon 2, 3, and 4 (LOD = < 5%), without necessity of cfDNA isolation. Mutation was considered present (categorical measurement), taking into account cycle quantification values (Cq) established by the commercial test. Cut-off values for Cq were studied by the ROC curve (receiver operating characteristic) with the Youden's J statistic. Paired tissue samples were analyzed when available. Results: A total of 23 patients were enrolled, median age 65 years old, 57% male, 100% adenocarcinoma, 18 (70%) metastatic. Sixteen patients could also be tested in tissue, being 15 (93.8%) positive for KRAS mutation. Plasma basal mutations were detected in 10 patients (43.5%), all in stage IV (p = 0.046), and interestingly, all with liver involvement (p = 0.003). Moreover, these patients also had metastasis in another organ apart from the liver. The most frequent mutation was G12V (5), followed by G12D (4) and G12R (1). Patients with basal KRAS mutations detected in plasma had a survival of 6.1 months (CI 95%, 3.7-8.5), and non-mutated 11.2 months (95% CI, 7.8-14.7) (p = 0.041). The presence of these cell-free KRAS mutations was more informative for survival than stage IV disease vs III (p = 0.054). Paying attention to KRAS wild type detection in plasma samples used as an internal control of the PCR, we could establish a Cq value cut-off of < 21 Cq for mutant samples, with a sensitivity of 100% (95% CI, 75%-100%) and specificity of 50% (95% CI, 19%-81%), area under ROC curve 0.823 (95% CI, 0.649-0.997). Conclusions: Measurement of cell-free KRAS in plasma samples at diagnosis could be used to quickly predict metastasis and survival in patients with pancreatic adenocarcinoma.