Journal: Cancer discovery
This study reports the development and validation of a non-invasive circulating cell-free DNA (cfDNA)-based model for early detection of pancreatic cancer (PC).
In a case-control cohort of 232 PC patients and 235 healthy controls, the model showed high diagnostic accuracy with an AUC of approximately 0.98 in training and 0.96 in validation.
In a prospective study of 1,926 individuals with diabetes and obesity, the model identified 75% of PC cases—including all Stage 0 cases—with a lead time of up to 298 days, outperforming the standard biomarker CA19-9.
The model also shows promise in differentiating high-risk from low-risk pancreatic cysts, supporting improved risk stratification.
These findings suggest that cfDNA-based screening could be a scalable and effective tool for early PC detection, pending further large-scale validation.