Pancreatic cancer is the fourth leading cause of cancer-related death in the Western world, with a median overall survival of less than one year for patients with advanced disease. The current standard blood-based biomarker, carbohydrate antigen 19-9 (CA19-9), suffers from limitations in sensitivity and specificity. Circulating tumor DNA (ctDNA) has emerged as a promising alternative, offering the potential to predict clinical outcome and monitor treatment response. Previous studies demonstrated the prognostic value of indirect ctDNA detection and direct KRAS mutation detection in advanced pancreatic cancer. However, these studies were limited in sample size and relied on a single marker. This study aimed to improve ctDNA detection by combining mutation detection in eight frequently mutated genes (KRAS, TP53, SMAD4, CDKN2A, ARID1A, TGFBR2, RNF43, and GNAS) using HYTEC-seq, a hybridization capture next-generation sequencing approach, with genome-wide copy-number variation (CNV) analysis. The improved methodology was used to investigate the prognostic and monitoring potential of ctDNA in a cohort of patients with advanced pancreatic cancer, comparing the results with CA19-9 measurements and radiological imaging.

The literature review highlights the limitations of CA19-9 as a biomarker for pancreatic cancer, emphasizing its poor sensitivity and susceptibility to false positives and negatives. It then establishes the growing body of evidence supporting the use of ctDNA as a superior alternative in various cancer types, including its prognostic and predictive capabilities. While some pilot studies showed promise in using ctDNA for pancreatic cancer monitoring, these were often limited by small sample sizes and reliance on detecting only KRAS mutations, overlooking patients lacking these mutations or having low ctDNA levels. This study aimed to overcome these limitations by adopting a more comprehensive approach.

This study included 56 patients with locally advanced or metastatic pancreatic cancer. Peripheral blood samples were collected before and monthly during treatment. cfDNA was isolated from plasma using the QIAamp Circulating Nucleic Acid kit. HYTEC-seq was employed for library preparation and sequencing, targeting eight genes frequently mutated in pancreatic cancer. The bioinformatic pipeline involved signal processing, base calling, quality control, alignment, and variant calling using custom scripts (TagXtractor, SSCScreator, PlasmaMutationDetector2). Copy-number aberration (CNA) analysis was performed using CNVkit. Digital droplet PCR (ddPCR) validated variants detected by HYTEC-seq and excluded clonal hematopoiesis of indeterminate potential (CHIP) variants. Statistical analysis included Kaplan-Meier estimates, log-rank test, Cox proportional hazards regression, and multivariable Cox regression. PFS and OS were defined according to RECIST criteria.

CtDNA, detected through point mutations or CNAs, showed significant prognostic value. Patients with ctDNA-positive status at baseline exhibited significantly shorter PFS and OS compared to ctDNA-negative patients (P < 0.001). Multivariable Cox regression analyses confirmed ctDNA point mutation status and CNA status as independent prognostic factors for both PFS and OS, along with ECOG performance status and first-line treatment. Dynamic ctDNA analysis revealed a significant decrease in ctDNA variant allele frequency (VAF) after one month of chemotherapy, followed by a significant increase at the time of progression. ctDNA persistence (lack of at least a 10-fold VAF reduction) after two months of therapy was significantly associated with shorter PFS and OS. Longitudinal monitoring of ctDNA in 27 patients showed that a >25% increase in ctDNA detected progression in 19/27 patients (70%), with a median lead time of 19 days compared to radiological imaging/death. This lead time was significantly longer than the 6-day median lead time observed with CA19-9 increases.

The findings support the clinical utility of ctDNA in predicting prognosis and monitoring treatment response in advanced pancreatic cancer. ctDNA persistence after chemotherapy initiation emerged as a strong predictor of poor outcomes, suggesting its potential to identify patients unlikely to respond to treatment. The ability of ctDNA to detect disease progression with a substantially longer lead time compared to CA19-9 highlights its value for early intervention. While the sensitivity of ctDNA detection was not perfect, its specificity in detecting progression was high. Discrepancies between ctDNA detected progression and radiological findings might be attributable to limitations of RECIST criteria or biological/technical limitations of ctDNA detection. The study provides robust evidence to support further investigation of ctDNA as a valuable tool for personalized medicine in advanced pancreatic cancer.

This study confirms the prognostic value of ctDNA and demonstrates its potential for disease monitoring in advanced pancreatic cancer. ctDNA levels decrease after chemotherapy initiation and subsequently increase upon progression. ctDNA persistence signifies treatment failure, and longitudinal monitoring can detect progression earlier than imaging. While sensitivity limitations exist, the high specificity of ctDNA increases warrants further prospective trials to evaluate its use in guiding treatment decisions and ultimately improving patient outcomes.

The study's limitations include the lack of tumor tissue for comparison, challenges in maintaining regular blood sample collection, relatively small patient numbers, and the use of less than the optimal amount of plasma for cfDNA isolation. These factors could have impacted the sensitivity of ctDNA detection, and a larger, prospective trial would strengthen the conclusions.