An evaluation of the accuracy of CT when determining resectability of pancreatic head adenocarcinoma after neoadjuvant treatment

BackgroundTo evaluate the accuracy of MDCT for determination of resectability R0 after neoadjuvant therapy in patients with pancreatic head adenocarcinoma locally advanced.MethodsFrom January 2005 to December 2010, 80 patients with pancreatic head adenocarcinoma underwent multidetector CT before surgery. Of these, 38 patients received neoadjuvant therapy because tumor was considered locally advanced on baseline CT scan. We retrospectively correlated imaging interpretations with operative and histological data and compared results in patients without (control group) or with (neoadjuvant group) preoperative treatment.Results41/42 patients in control group and 31/38 patients in neoadjuvant group finally had curative resection. While resection R0 is similar in both groups (83% and 81%), CT accuracy in determining resectability R0 was significantly decreased in neoadjuvant group (58% versus 83%; p = 0.039). CT scan specificity was significantly lower after neoadjuvant therapy (52% versus 88% in control group) due to an overestimation of vascular invasion: 12/31 patients with complete resection in neoadjuvant group were evaluated at high risk of incomplete resection on CT scan. Tumor size tends to be underestimated in control group (−2 mm) and overestimated in neoadjuvant group (+10 mm). T-staging accuracy was decreased in neoadjuvant group (39% versus 78% in control group; p = 0.002).ConclusionNeoadjuvant therapy significantly decreases the accuracy of CT scan in determining operability, T-staging, and resectability R0 of pancreatic head carcinoma. Overestimation of tumor size and vascular invasion significantly reduces CT scan specificity after preoperative treatment.