Background: The most widely accepted biochemical test for preoperative differentiation of mucinous from benign, nonmucinous pancreatic cysts is cyst fluid carcinoembryonic antigen. However, the diagnostic accuracy of carcinoembryonic antigen ranges from 70% to 86%. Based on previous work, we hypothesize that pancreatic cyst fluid glucose may be an attractive alternative to carcinoembryonic antigen. Methods: Pancreatic cyst fluid was collected during endoscopic or operative intervention. Diagnoses were pathologically confirmed. Glucose and carcinoembryonic antigen were measured using a patient glucometer and automated analyzer/enzyme-linked immunosorbent assay. Sensitivity, specificity, accuracy, and receiver operator characteristic analyses were performed. Results: Cyst fluid samples from 153 patients were evaluated (mucinous: 25 mucinous cystic neoplasms, 77 intraductal papillary mucinous neoplasms, 4 ductal adenocarcinomas; nonmucinous: 21 serous cystic neoplasms, 9 cystic neuroendocrine tumors, 14 pseudocysts, 3 solid pseudopapillary neoplasms). Median cyst fluid glucose was lower in mucinous versus nonmucinous cysts (19 vs 96 mg/dL; P < .0001). With a threshold of ≤ 50 mg/dL, cyst fluid glucose was 92% sensitive, 87% specific, and 90% accurate in diagnosing mucinous pancreatic cysts. In comparison, cyst fluid carcinoembryonic antigen with a threshold of >192 ng/mL was 58% sensitive, 96% specific, and 69% accurate. Area under the curve for glucose and CEA were similar at 0.91 and 0.92. Conclusion: Cyst fluid glucose has significant advantages over carcinoembryonic antigen and should be considered for use as a routine diagnostic test for pancreatic mucinous cysts.
ASJC Scopus subject areas