Introduction Over 90% of pancreatic adenocarcinoma (PDAC) has a KRAS mutation, 98% of which are single-base missense variants occurring at codon 12 (G12), codon 13 (G13), or codon 61 (Q61), with numerous studies demonstrating their clinical utility. Mutant KRAS has been detected in circulating tumor DNA (ctDNA) at all stages of PDAC using analysis methods such as next-generation sequencing (NGS). This study aimed to investigate the prognostic potential of KRAS G12, G13, and Q61 mutations present in circulating tumor DNA (ctDNA) prior to treatment initiation in localized PDAC using both standard depth (20,000x) and ultra-deep (≥100,000) sequencing. Methods Plasma samples were collected from 45 patients with non-metastatic PDAC prior to treatment start. Baseline ctDNA was analyzed for KRAS G12/G13/Q61 mutations using the PredicineCARE liquid biopsy assay (20,000x sequencing depth) and the PredicineCare ULTRA assay (≥100,000x sequencing depth), providing full KRAS gene coverage. Results Standard depth sequencing identified pathogenic KRAS mutations (G12/G13/Q61) in 11 patients. Ultra-deep sequencing demonstrated a 46% increase in detection, identifying mutant KRAS in these same 11 patients and an additional 7 patients deemed mutation-negative by standard assay. Presence of any KRAS mutation at baseline by standard sequencing was not significantly associated with overall survival (OS) (HR 2.57, 95% CI 0.94-7.04). However, detection of any KRAS mutation by ultra-deep sequencing in baseline ctDNA was significantly predictive of worse OS (HR 3.13, 95% CI 1.18-8.29). In both univariate and multivariate analysis, detection of mutant KRAS G12, Q61, or G13 in baseline ctDNA by whole-genome sequencing was superiorly predictive of worse OS over pre-therapy serum CA 19-9. Conclusion Ultra-deep sequencing demonstrates robust detection of pertinent KRAS mutations that support improved prognostic stratification for patients with localized PDAC, demonstrating the potential value of advanced liquid biopsy technology to tailor treatment decisions and improve patient outcomes.
Cox et al. (Mon,) studied this question.