Purpose: The optimal management of patients with potentially resectable pancreatic carcinoma is controversial.It is unknown whether preoperative staging with a combination of CT, EUS and MRI allows for better selection of patients for surgery to improve survival than does a strategy of exploratory laparotomy alone for all patients. Methods: We constructed a decision model to determine whether a strategy utilizing cross-sectional imaging to pre-select patients to receive palliative care or curative surgery results in an increased overall survival in patients with potentially resectable pancreatic adenocarcinoma. The baseline model was selected as a patient with suspected pancreatic cancer based on CT scan and clinical symptoms, with no evidence on CT of metastatic disease. Probability estimates for the baseline model were obtained from a review of the medical literature. One-way sensitivity analyses were performed on all variables included in the model. Two-way sensitivity analyses were performed on the sensitivity and specificity of the cross-sectional imaging tests (CT, EUS and MRI). Results:In the baseline model, exploratory laporotomy alone for all patients resulted in a 7% greater life expectancy than the strategy of triage using only CT, EUS and MRI. Sensitivity analyses revealed that for a combination of EUS, CT and MRI to improve mortality beyond that of exploratory laparotomy the combined sensitivity of all tests needed to exceed 92%. We also found that increasing the overall specificity of the combined imaging studies did not improve patient survival beyond that of exploratory laporotomy alone. Conciusion:Our model demonstrates a survival benefit of preoperative testing over exploratory laporotomy alone in patients with potentially resectable pancreatic adenocarcinoma if the combined sensitivity of the imaging modalities for determining disease resectability is greater than 92%.Based on the results of this model the use of preoperative testing to triage patients with potentially resectable pancreatic cancer to surgical or medical therapy should be reassessed. Further study is needed to determine whether preoperative multi-modality staging exceeds the threshold sensitivity predicted by this model.
Read full abstract