Objectives: To determine retrospectively the additional value of DWI-MRI toT2-MRI for predicting complete response (ypT0N0 = CR) after chemoradiation-therapy (CRT) in locally advanced rectal cancer.
Methods: Seventy locally advanced rectal cancer patients underwent CRT followed by restaging MRI and resection. Two readers with different experience levels independently scored T2 images for CR and, in a second reading, combined T2 and DWI. A 5-point confidence-level score was used to generate ROC curves. Areas under the ROC curves (AUC) and interobserver agreement were compared for both readings. Histology served as reference standard.
Results: The interobserver agreement increased after addition of DWI from 0.35 to 0.58 but the AUC improved only for the experienced reader (0.77 to 0.89, p = 0.005 vs. 0.74 to 0.70, p > 0.05). Sensitivity and NPV improved from 20-30 % to 40-70 %, respectively 88 % to 91-95 %. Specificity and PPV improved only for the experienced reader (87 to 93 % respectively 27 to 63 %).
Conclusion: Adding DWI to T2-MRI improves consistency between readers and has potential to improve readers' accuracy dependent on his/her experience. DWI could be of additional value, particularly in ruling out CR (high NPV), but considering the sub-optimal PPV one should be cautious about relying solely on MRI for the clinical decision to offer a wait-and-see strategy.