Background: Several modalities currently exist for tissue confirmation of suspected pancreatic cancer prior to therapy. Since there is a paucity of cost-minimization studies comparing these different biopsy modalities, we analyzed costs and examined effectiveness of four alternative strategies for diagnosing pancreatic cancer.
Methods: A decision analysis model of patients with suspected pancreatic cancer was constructed. We analyzed costs, failure rate, testing characteristics, and complication rates of four commonly employed diagnostic modalities: 1) computerized tomography or ultrasound-guided fine-needle aspiration (CT/US-FNA), 2) endoscopic retrograde cholangiopancreatography with brushings (ERCP-B), 3) Endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNA), and 4) laparoscopic surgical biopsy. If the first attempt with a particular modality failed, a different modality was employed to identify the most preferable secondary biopsy strategy.
Results: This analysis identifies EUS-FNA as the preferred initial modality for the diagnosis of pancreatic cancer. Resultant expected costs and strategies in decreasing optimality include: 1) EUS-FNA (1,405 dollars), 2) ERCP-B (1,432 dollars), 3) CT/US-FNA (3,682 dollars), and 4) surgery (17,711 dollars). If a patient presents with obstructive jaundice, decision analysis modeling resulted in a total expected costs of 1,970 dollars if ERCP-B is successful at the time of biliary stent placement. Additional analyses to identify the preferred follow-up modality after a failed alternative method showed that EUS-FNA is the preferred secondary modality if any of the other three modalities failed first, in both the setting of and absence of obstructive jaundice. One- and two-way sensitivity analysis of the variables shows unchanged results over an acceptable range.
Conclusions: This cost-minimization study illustrates that EUS-FNA is the best initial and the preferred secondary alternative method for the diagnosis of suspected pancreatic cancer. In addition to local expertise and availability, costs and diagnostic yield should be considered when choosing an optimal diagnostic strategy.