Preoperative tumor restaging and resectability assessment of gastric cancers after chemotherapy: diagnostic accuracy of MDCT using new staging criteria.

Abstract

To evaluate the diagnostic performance of preoperative MDCT for tumor restaging and determination of resectability in gastric cancers after chemotherapy using new staging criteria. This retrospective study was approved by our institutional review board and the requirement for informed consent was waived. Thirty-seven patients with initially unresectable gastric cancers who had received chemotherapy followed by surgery were included. Two independent radiologists reviewed preoperative MDCT images to determine the TNM staging and rate the overall likelihood of resectability using a 5-point scale (5: definitely unresectable, 1: definitely resectable). New post-chemotherapy MDCT criteria do not use non-enhancing perigastric infiltrations, non-enhancing lymph nodes (LNs), and subtle remaining infiltrations after marked decrease in the size of distant metastases for T, N, and M upstaging, respectively. Discrepancies in TNM staging were resolved by a third reviewer. The diagnostic performances of MDCT were assessed using pathologic results or operation records as reference standards. For predicting resectability, the areas under the ROC curve were 0.885 and 0.882 (95% CIs 0.737-0.966 and 0.733-0.964) in reviewers 1 and 2, respectively, with substantial inter-reader agreement (weighted κ=0.689). Sensitivities and specificities of MDCT for tumor restaging on a consensus review were 80.0% (4/5) and 100% (29/29) for T4b, 35.3% (6/17) and 81.3% (13/16) for N-positive, and 63.6% (7/11) and 100% (26/26) for M1, respectively. For gastric cancers after chemotherapy, new MDCT criteria demonstrated high specificities for T4b and M-staging and good performances to predict resectability before conversion surgery.