Background: Serum carbohydrate antigen 19-9 (CA19-9) correlates with response to therapy and overall survival (OS) for patients with pancreatic ductal adenocarcinoma (PDAC). This study aimed to define the chronologic relationship between CA19-9 elevation and radiographic recurrence to develop a model that can predict the risk of recurrence (RFS) and prognosis during interval surveillance for patients with resected PDAC. Methods: A retrospective review examined patients undergoing surgery for pancreatic adenocarcinoma from January 2010 to May 2016. Their CA19-9 levels were classified at diagnosis, after surgery, and at 6-month surveillance intervals. Recurrence was defined by radiographic evidence. The CA19-9 levels were correlated with RFS and OS at every time point using multivariate analysis. Results: The study examined 525 patients. Five patterns of CA19-9 were identified: normal (“nonsecretors,” 18.5%), always elevated, and high at diagnosis but normal after resection involving three patterns with varied behavior during surveillance. These five patterns had implications for RFS and OS. When elevation of CA19-9, as assessed at 6-month intervals, was analyzed relative to detection of radiographic disease, CA19-9 had poor positive predictive value (average, 35%) but high negative predictive value (average, 92%) for radiographic recurrence. Conditional RFS showed that CA19-9 elevation did not equal radiographic recurrence but predicted subsequent RFS. Additionally, conditional OS showed that CA19-9 elevation alone was predictive at each time point. Conclusion: This study showed that CA19-9 patterns beyond the post-resection period predict RFS and OS. High CA19-9 frequently is discordant with recurrence on imaging and may precede it by more than 6 months. At each surveillance interval, CA19-9 is predictive of prognosis, which may help in counseling patients and could be used to direct protocols of salvage chemotherapy.
ASJC Scopus subject areas