Purpose:To explore the clinical value of diffusion kurtosis imaging (DKI) for predicting early treatment response to neoadjuvant chemotherapy (NAC) in patients with nasopharyngeal carcinoma (NPC).Materials and Methods:Consecutive fifty-nine patients who were pathologically and clinically diagnosed with stage Ⅲ-Ⅳb NPC were enrolled in this study from september 2013 to May 2014. All the patients underwent four MR scans:prior to, on the 4th,21st (after the first cycle of chemotherapy was completed),42nd (after the second cycle of chemotherapy was completed) days after NAC initiation. The first three examinations include T2-weighted imaging (T2WI) and Diffusion Weighted Imaging (DWI:b=0,500,1000,1500 s/mm2), and the last examinations just included T2WI. The DWI data were fitted by DKI model and mono-exponential DWI model, and the corresponding parameters (corrected diffusion coefficient, D; excess diffusion kurtosis coefficient, K; apparent diffusion coefficient, ADC) were calculated. All the patients were classified into effective group and ineffective group based on RECIST 1.1. Student’s t test or Mann-Whitney U test were used to compare the differences of all the parameters at each timepoint. Receiver operating characteristic (ROC) curve analyses was performed to calculate the area under the cure (AUC), cutoff value, sensitivity, specificity of each parameter in terms of its value for predicting the NAC treatment outcomes. Pairedx2 test was performed to compare the sensitivity and specificity of the parameters separately calculated by the two diffusion models.Results:After two NAC cycles,43 (73%) of the 59 primary lesions were categorized as effective group and 16 (27%) were categorized as ineffective group. The differences of tumor size between the two groups were not significant until the first cicle of NAC was finished (P=0.011,0.000). No significant differences were found in ADC(pre) between the RG and NRG (P= 0.360). However, the mean ADC(day4), ADC(day21) and △ADC(day4) of the effective group was significantly higher (P= 0.002,0.019,0.025) than that those of the ineffective group.was significant lower in the effective group (P= 0.020). D(day4) and △D(day4) were significantly different between the two groups (P= 0.005,0.000). There was no significant difference in either K or △K at any time point during NAC between the two groups (P= 0.519 to 0.840). For DKI, △D(day4) was the most powerful parameter with AUC of 0.895, sensibility of 95.3% and specificity of 73.3% when △D(day4)> 0.036 × 10-3 mm2/s, and for mono-exponential DWI, ADC(day4) was the most powerful parameter with AUC of 0.761, sensibility of 72.1% and specificity of 81.3% when ADC(day4)> 1.063 × 10-3 mm2/s. The differences between the two ROC curves were significant (P= 0.043), and AD(day4) was more sensitive than ADC(day4) to predict treatment response to NAC (P= 0.006).Conclusion:Both DKI and mono-exponential DWI were potential to detect the treatment response to NAC prior to morphological change. D value obtained from DKI may be more sensitive than ADC value obtained from mono-exponential DWI for predicting the early response to NAC in patients with locally advanced NPC. |