Preoperative MDCT assessment of resectability in borderline resectable pancreatic cancer: Effect of neoadjuvant chemoradiation therapy

OBJECTIVE. The purpose of this study is to evaluate the diagnostic performance of MDCT in assessing tumor resectability in patients with borderline resectable pancreatic cancers after receiving neoadjuvant chemoradiation therapy (CRT) in comparison with those undergoing upfront surgery. SUBJECTS AND METHODS. Thirty-seven patients with borderline resectable pancreatic cancers were randomly allocated to the neoadjuvant CRT group (arm 1; n = 18) or upfront surgery group (arm 2; n = 19). Three radiologists rated the likelihood of local resectability on a 5-point scale at preoperative MDCT in two separate sessions (session 1: post-CRT of arm 1, baseline of arm 2; session 2: using new imaging criteria reflecting the changes during CRT of arm 1). The AUC of each reviewer, as well as sensitivity, specificity, and accuracy based on consensus interpretation, were compared between arms and sessions. RESULTS. For local resectability (n = 30), AUC values at session 1 were 0.664, 0.669, and 0.588 for reviewers 1, 2, and 3, respectively, and were not significantly different between arms 1 (n = 15; 0.759, 0.713, and 0.593) and 2 (n = 15; 0.852, 0.685, and 0.722) (p > 0.05). In arm 1, MDCT sensitivity, specificity, accuracy were 22%, 100%, and 53%, respectively, at session 1 versus 78%, 67%, and 73%, respectively, at session 2 (p > 0.05). CONCLUSION. In patients with borderline resectable pancreatic cancers, neoadjuvant CRT did not significantly decrease the performance of MDCT for the prediction of local resectability. However, by considering post-CRT changes, such as nonprogression in tumor-vascular contact, MDCT may provide better sensitivity for locally resectable disease.