Integrating dynamic contrast-enhanced magnetic resonance imaging and diffusion kurtosis imaging for neoadjuvant chemotherapy assessment of nasopharyngeal carcinoma.

Background: Since neoadjuvant chemotherapy (NAC) has proven a benefit for locally advanced nasopharyngeal carcinoma (NPC), early response evaluation after chemotherapy is important to implement individualized therapy for NPC in the era of precision medicine. Purpose: To determine the combined and independent contribution between dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and diffusion kurtosis imaging (DKI) in the early monitoring of NAC response for NPC. Study Type: Prospective. Population: Fifty-three locally advanced NPC patients. Field Strength/Sequence: Four examinations before and at 4, 20, and 40 days after NAC initiation were performed at 3T MRI including DCE-MRI and DKI (b values=0, 500, 1000, 1500 s/mm(2)). Assessment: DCE-MRI parameters (K-trans [the volume transfer constant of Gd-DTPA], k(ep) [rate constant], v(e) [the extracellular volume fraction of the imaged tissue], and v(p) [the blood volume fraction]) and DKI parameters (D-app [apparent diffusion for non-Gaussian distribution] and K-app [apparent kurtosis coefficient]) were analyzed using dedicated software. Statistical Tests: MRI parameters and their corresponding changes were compared between responders and nonresponders after one or two NAC cycles treatment using independent-samples Student's t-test or Mann-Whitney U-test depending on the normality contribution test and then followed by logistic regression and receiver operating characteristic curve (ROC) analyses. Results: The responder group (RG) patients presented significantly higher mean K-trans and D-app values at baseline and larger Delta K-(0-4)(trans,) Delta v(p(0-4)), and Delta Dapp(0-4) values after either one or two NAC cycles compared with the nonresponder group (NRG) patients (all P < 0.05). ROC analyses demonstrated the higher diagnostic accuracy of combined DCE-MRI and DKI model to distinguish nonresponders from responders after two NAC cycles than using DCE-MRI (0.987 vs. 0.872, P=0.033) or DKI (0.987 vs. 0.898, P=0.047) alone. Data Conclusion: Combined DCE-MRI and DKI models had higher diagnostic accuracy for NAC assessment compared with either model used independently. Level of Evidence: 2 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2018;47:1208-1216.