Background: This study was performed to develop a management algorithm that accurately predicts the necessity of operative intervention and results in significant cost savings for patients with cystic pancreatic tumors.
Methods: We reviewed 60 patients treated between 1992 and 2003. Accuracy rates of tests used to differentiate benign from premalignant and malignant cysts were calculated. A management algorithm was generated that incorporated clinical presentation, radiologic findings, and selective use of endoscopic procedures. This algorithm was tested for predictive accuracy, and savings between actual management and proposed management were compared.
Results: There was an average of 3 preoperative tests performed per patient. Endoscopic procedures included endoscopic retrograde cholangiopancreatography in 21 patients and endoscopic ultrasound in 25 patients. A cyst fluid carcinoembryonic antigen (CEA) level of 158 ng/mL or greater had an accuracy rate of 87.5% that was significantly higher than endoscopic retrograde cholangiopancreatography (72%), endoscopic ultrasound morphology (45%), or endoscopic ultrasound cytology (66.7%). There was no combination of tests that provided greater accuracy than cyst fluid CEA level alone (P < .05). The management algorithm had a positive predictive value of 81%. The average actual cost of radiographic and endoscopic procedures was dollar 8,080 per patient. The proposed cost based on the algorithm was dollar 6,677 with a savings of dollar 1,403 per patient (P = .009).
Conclusions: In patients with clinical symptoms or radiographic findings of mucinous or malignant tumors, further testing is excessive. Of endoscopic tests available, the cystic fluid CEA level most accurately predicts the presence of a mucinous neoplasm. A management algorithm based on presenting symptoms, radiographic findings, and cyst fluid CEA level provides a guideline for the evaluation of cystic lesions in the most cost-efficient manner while ensuring proper care.