Evaluating the diagnostic performance of MRI in tumour regression grading of rectal carcinoma: a retrospective correlation with histopathology

Magnetic resonance imaging (MRI) is widely used for response assessment after neoadjuvant chemoradiotherapy in rectal carcinoma, but post-treatment fibrosis, oedema, mucin and microscopic residual tumour can complicate interpretation. This study evaluated agreement between MRI tumour regression grade (mrTRG) and pathological tumour regression grade (pTRG), and assessed the diagnostic performance of mrTRG 1–2 for identifying pathological complete response (pCR). This retrospective diagnostic-accuracy study included 54 patients with histologically confirmed rectal adenocarcinoma who received neoadjuvant treatment, underwent post-treatment pelvic MRI and subsequently underwent surgical resection. MRI examinations were performed on a 1.5-T GE HDXT system (GE Medical Systems, Milwaukee, Wisconsin, USA) using an 8-channel pelvic phased-array surface RF coil, with high-resolution T2-weighted imaging and diffusion-weighted imaging. MRI response was assigned using the established five-category mrTRG scale, whereas pathology was graded using the five-category Dworak pTRG system. For binary diagnostic analysis, pCR (pTRG 4) was the target condition and mrTRG 1–2 was considered MRI-positive. Agreement was assessed using quadratic-weighted Cohen’s kappa after aligning the opposite directions of the two scales. Ten of 54 patients (18.5%) had pCR. For mrTRG 1–2 versus pTRG 4, there were 8 true-positive, 14 false-positive, 2 false-negative and 30 true-negative results. Sensitivity was 80.0% (95% CI 44.4–97.5%), specificity 68.2% (52.4–81.4%), positive predictive value 36.4% (17.2–59.3%), negative predictive value 93.8% (79.2–99.2%) and accuracy 70.4% (56.4–82.0%). Five-category agreement was fair (κ = 0.325; 95% CI 0.043–0.610; p = 0.014). mrTRG showed only fair agreement with histopathological regression grading. Although mrTRG 1–2 identified most pathological complete responders, its low positive predictive value indicates that a favourable MRI regression grade alone cannot reliably establish pCR. MRI should therefore be interpreted as part of multimodal response assessment rather than as a stand-alone basis for organ-preservation decisions.