Diagnostic performance of magnetic resonance imaging in preoperative local staging of rectal cancer after neoadjuvant chemoradiotherapy

PURPOSE This paper aims to investigate the diagnostic performance of magnetic resonance imaging (MRI) in predicting the pathologic stage of locally advanced rectal cancer (LARC) after neoadjuvant chemo-radiotherapy (CRT) and the role of MRI in selecting patients with a pathologic complete response (ypCR). METHODS Restaging MRI (yMRI) examinations of 136 patients with LARC treated with neoadjuvant CRT fol-lowed by surgery were retrospectively analyzed by two radiologists. All examinations were per -formed on a 1.5 Tesla MRI machine with a pelvic phased-array coil. T2-weighted turbo spin-echo images and diffusion-weighted imaging were obtained. Histopathologic reports of the surgical specimens were the reference standard. The accuracy, sensitivity, specificity, positive and negative predictive values (PPV and NPV) of yMRI in predicting the pathologic T-stage (ypT), N-stage, and ypCR were calculated. The inter-observer agreement was evaluated using kappa statistics. RESULTS The yMRI results showed 67% accuracy, 59% sensitivity, 80% specificity, 81% PPV, and 56% NPV in identifying ypT (ypT0-2 versus ypT3-4). In predicting the nodal status, the yMRI results revealed 63% accuracy, 60% sensitivity, 65% specificity, 47% PPV, and 75% NPV. In predicting ypCR, the yMRI results showed 84% accuracy, 20% sensitivity, 92% specificity, 23% PPV, and 90% NPV. The kappa statistics revealed substantial agreement between the two radiologists. CONCLUSION Utilization of yMRI showed high specificity and PPV in predicting the tumor stage and high NPV in predicting the nodal stage; in addition, yMRI revealed moderate accuracy in the T and N classifica-tions, mainly due to underestimating the tumor stage and overestimating the nodal status. Finally, yMRI revealed high specificity and NPV but low sensitivity in predicting the complete response.

Automated summary

This study aims to investigate the diagnostic performance of MRI in predicting the pathologic stage of LARC after neoadjuvant CRT and the role of MRI in selecting patients with ypCR. The results showed moderate accuracy and sensitivity in predicting ypT and N stages, mainly due to underestimating tumor stage and overestimating nodal status. Furthermore, high specificity and NPV but low sensitivity were observed in predicting the complete response.