Abstract
PURPOSE: To assess the diagnostic performance, interobserver agreement and confidence level for determining response to neoadjuvant chemoradiotherapy (NACRT) using morphology based MR-tumour regression grade (MR TRG), diffusion weighted imaging (DWI) patterns and their combination in patients with locally advanced rectal cancer. METHODS: This was a retrospective study including patients with locally advanced rectal cancer treated with NACRT and subsequent surgery. Two independent radiologists blinded to the histopathology reviewed staging and restaging MRI. Diagnostic performance of morphology based MR-TRG, DWI patterns and their combination for determining complete (CR) and incomplete (IR) response was assessed with pathological response as the reference. Likert's scale was used to assess the radiologist's level of confidence. Interobserver agreement was determined using Kappa statistics. RESULTS: The study included 251 patients (mean age of 47.9+/-14 (range 19-86) years, M:F = 164:87). Rate of pathological CR was 14.7 % (n = 37). Pattern based interpretation of DWI and combined approach (DWI + T2-HR) had superior diagnostic performance than morphology based assessment alone with area under curve (AUC) for T2HR, DWI and their combination being 0.531, 0.887, 0.874 respectively for observer 1 and 0.558, 0.653, 0.678 respectively for observer 2, p < 0.001. Interobserver agreement was substantial (k = 0.688) for combined approach, moderate (k = 0.402) for DWI patterns and fair (k = 0.265) for T2 HR MRI with both observers exhibiting highest level of confidence for determining response with the combined approach. CONCLUSION: Complete response to neoadjuvant chemoradiotherapy can be determined with excellent accuracy, substantial interobserver agreement and high level of confidence by combined interpretation of DWI and T2 high resolution MRI.