MRI prediction of pathological response in locally advanced rectal cancer: when apparent diffusion coefficient radiomics meets conventional volumetry.

AIM

To investigate the role of diffusion-weighted imaging (DWI), T2-weighted (W) imaging, and apparent diffusion coefficient (ADC) histogram analysis before, during, and after neoadjuvant chemoradiotherapy (CRT) in the prediction of pathological response in patients with locally advanced rectal cancer (LARC).

MATERIALS AND METHODS

Magnetic resonance imaging (MRI) at 1.5 T was performed in 43 patients with LARC before, during, and after CRT. Tumour volume was measured on both T2-weighted (V) and on DWI at b=1,000 images (V) at each time point, hence the tumour volume reduction rate (ΔV and ΔV) was calculated. Whole-lesion (three-dimensional [3D]) first-order texture analysis of the ADC map was performed. Imaging parameters were compared to the pathological tumour regression grade (TRG). The diagnostic performance of each parameter in the identification of complete responders (CR; TRG4), partial responders (PR; TRG3) and non-responders (NR; TRG0-2) was evaluated by multinomial regression analysis and receiver operating characteristics curves.

RESULTS

After surgery, 11 patients were CR, 22 PR, and 10 NR. Before CRT, predictions of CR resulted in an ADC value of the 75th percentile and median, with good accuracy (74% and 86%, respectively) and sensitivity (73% and 82%, respectively). During CRT, the best predictor of CR was ΔV (-58.3%) with good accuracy (81%) and excellent sensitivity (91%). After CRT, the best predictors of CR were ΔV (-82.8%) and ΔV (-86.8%), with 84% accuracy in both cases and 82% and 91% sensitivity, respectively.

CONCLUSIONS

The median ADC value at pre-treatment MRI and ΔV (from pre-to-during CRT MRI) may have a role in early and accurate prediction of response to treatment. Both ΔV and ΔV (from pre-to-post CRT) can help in the identification of CR after CRT.