Advanced MR Technique for the Prediction of a Complete Response Following Neoadjuvant Chemoradiotherapy in Patients...

Background/Objectives: From the viewpoint of personalized medicine, it is important to properly select patients with a complete response (CR) after neoadjuvant chemoradiotherapy (nCRT) because they are candidates for "watch-and-wait" management in locally advanced rectal cancer (LARC). The aims of this study were to evaluate the added value of diffusion-weighted imaging (DWI) b2000 to high-resolution T2-weighted images (HR-T2WI) and to compare the diagnostic performance of DWI b2000 and b1000 for assessing a CR in patients with LARC after nCRT. Methods: This retrospective study included 55 patients (male: 40, female: 15; mean age: 60.1) who underwent pre- and post-nCRT 3T MRI, including DWI_b1000 and DWI_b2000. Two blinded radiologists independently assessed the likelihood of a CR using a five-point scale across three image sets: T2-weighted images (T2WI), T2WI + DWI_b1000, and T2WI + DWI_b2000. T2WI were provided because DWI is not used as a standalone sequence in clinical practice when evaluating the therapeutic response to nCRT. A crossover review of DWI_b1000 and DWI_b2000 with a four-week interval was used to minimize recall bias. The reference standard of a CR was histopathologic examination or clinical remission in cases of watch-and-wait management. For each reader, diagnostic performance was compared across three image sets using pairwise comparisons of the receiver operating characteristic (ROC) curves. Results: The study population comprised 19 patients with a CR and 36 patients with residual tumor. For both readers, adding DWI_b2000 significantly enhanced diagnostic performance compared to T2WI alone (AUC for reader 1: 0.918, 0.806, p = 0.0288; AUC for reader 2: 0.868, 0.768, p = 0.0414, respectively), likewise, DWI_b1000 (for reader 1: 0.955, p = 0.0015; for reader 2: 0.863, p = 0.0162). However, no significant difference in diagnostic performance was found between DWI_b1000 and DWI_b2000 for each reader (for reader 1: p = 0.0931; for reader 2: p = 0.8936). Conclusions: Adding DWI_b2000 to T2WI improved diagnostic performance for identifying a CR after nCRT compared to T2WI alone. No statistical difference was detected between DWI_b1000 and DWI_b2000.