Diagnostic accuracy of MRI in assessing tumor regression and identifying complete response in patients with locally advanced rectal cancer after neoadjuvant treatment.

Abstract

The diagnostic accuracy of Magnetic Resonance Imaging (MRI) in restaging locally advanced rectal cancers (LARC) after neoadjuvant chemo-radio therapy (NCRT) has been under recent scrutiny. There is limited data on the accuracy of MRI and its timing in assessing tumor regression grade (TRG) and in identifying patients with complete response (CR). NCRT seems to cause tissue inflammation and oedema which renders reading the scans difficult for radiologist. This study aims to assess the accuracy of MRI at different time intervals after NCRT in staging TRG and in identifying CR. Inter-observer agreement between 2 blinded radiologists will also be assessed. In this retrospective analysis, all patients diagnosed with LARC between January 2003 and 2014, who underwent long-course NCRT, who had at least one post-treatment MRI scan, and who underwent surgery with available pathology results are included. Histopathology staging is considered the reference standard. Accuracy of MRI in T staging and in TRG staging is assessed using weighted kappa. Accuracy, sensitivity, and specificity in identifying CR are calculated from a 2 × 2 contingency table. Inter-observer agreement between two-staging blinded radiologists is calculated using weighted kappa. These are calculated at 2 different time intervals after completion of NCRT. 114 patients were identified who had a first post-treatment MRI scan at an average of 6.2weeks after completion of NCRT. A subgroup of 68 patients had a second post-treatment MRI at an average of 10.4weeks. Pathology results were available for 103 patients. By the second post-treatment scan, an additional 25% of patients experienced downstaging; accuracy in T staging increased from 43% to 57.4%; accuracy in TRG staging rose from 28.2% to 38.1%; accuracy in identifying CR rose from 83.4% to 84.1%. Inter-observer agreement in T staging rose from 0.1 for first post-treatment MRI to 0.206 for second post-treatment MRI. This study advocates that restaging should occur at 10weeks rather than the standard 6weeks. This results in higher complete response rates and higher concordance with pathological specimens. Our results also showed that it is easier for radiologists to stage the MRI scans, resulting in higher inter-rater agreements.