The value of diffusion kurtosis imaging in assessing pathological complete response to neoadjuvant chemoradiation therapy in rectal cancer: a comparison with conventional diffusion-weighted imaging

Abstract

Objectives: The aim of this study is to comprehensively evaluate the advantage of diffusion kurtosis imaging (DKI) in distinguishing pathological complete response (pCR) from non-pCR patients with locally advanced rectal cancer (LARC) after neoadjuvant chemoradiation therapy (CRT) in comparison to conventional diffusion-weighted imaging (DWI).

Material and Methods: Fifty-six consecutive patients diagnosed with LARC were prospectively enrolled and underwent pre- and post-CRT MRI on a 3.0 T MRI scanner. Apparent diffusion coefficient (ADC), mean diffusion (MD) and mean kurtosis (MK) values of the tumor were measured in pre- and post-CRT phases and then compared to histopathologic findings after total mesorectal excision (TME). Both Mann-Whitney U-test and Kruskal-Wallis test were used as statistical methods. Diagnostic performance was determined using receiver operating characteristic (ROC) curve analysis.

Results: For a total of 56 rectal lesions (pCR, n = 14; non-pCR, n = 42), the MKpre and MKpost values were much lower for the pCR patients (mean±SD, 0.72±0.09 and 0.56±0.06, respectively) than those for the non-pCR patients (0.89±0.11 and 0.68±0.08, respectively) (p < 0.001). The ADCpost and the change ratio of apparent diffusion coefficient (ADCratio) values was significantly higher for the pCR patients (mean±SD, 1.31±0.13 and 0.64±0.34, respectively) than for the non-pCR patients (1.12±0.16 and 0.33±0.27, respectively) (p < 0.001 and p = 0.001, respectively). In addition, the MDpost and the change ratio of mean diffusion (MDratio) (2.45±0.33 vs. 1.95±0.30, p < 0.001; 0.80±0.43 vs. 0.35±0.32, p < 0.001, respectively) also increased, whereas the ADCpre, MDpre and the change ratio of mean kurtosis (MKratio) of the pCR (0.82±0.11, 1.40±0.21, and 0.23±0.010, respectively) exhibited a neglectable difference with that of the non-pCR (p = 0.332, 0.269, and 0.678, respectively). The MKpost showed relatively high sensitivity (92.9%) and high specificity (83.3%) in comparison to other image indices. The area under the receiver operating characteristic curve (AUROC) that is available for the assessment of pCR using MKpost (0.908, cutoff value = 0.6196) were larger than other parameters and the overall accuracy of MKpost (85.7%) was the highest.

Conclusions: Both DKI and conventional DWI hold great potential in predicting treatment response to neoadjuvant chemoradiation therapy in rectal cancer. The DKI parameters, especially MKpost, showed a higher specificity than conventional DWI in assessing pCR and non-pCR in patients with LARC, but the pre-CRT ADC and MD are unreliable.