We compared the parameters derived from diffusion-weighted imaging (DWI) and positron emission tomography/computed tomography (PET/CT) for treatment response evaluation and response prediction in patients with gastrointestinal stromal tumor (GIST). Seven patients with histologically proven metastatic disease were enrolled. DWI and PET/CT data were collected from all patients at diagnosis and from six at follow-up. All 37 lesions were identifiable in DWI with a sensitivity of 100%. To achieve higher accuracy, we used the apparent diffusion coefficient (ADC) of liver and background noise as thresholds for the measurement of the ADCs of lesions. Significant inverse correlations were found between ADC(mean_thr) (ADC(mean) with thresholds) and SUV(mean) (mean standardized uptake value) (R(2)=0.523, p<0.001 at diagnosis, and R(2)=0.916, p<0.001 at follow-up), between ADC(mean_thr) and SUV(max) (maximum SUV) (R(2)=0.529, p<0.001 at diagnosis, and R(2)=0.761, p<0.001 at follow-up), between ΔADC(mean_thr) (percentage change in ADC(mean_thr) ) and ΔSUV(mean) (percentage change in SUV(mean) ) (R(2)=0.384, p<0.001), and between ΔADC(mean_thr) and ΔSUV(max) (percentage change in SUV(max) ) (R(2)=0.500, p<0.001). In lesion-based analysis, pre-treatment ADC(mean_thr) outperformed SUV(mean) and SUV(max) in treatment response prediction, with an area under the receiver operating characteristic curve of 0.706. These results show that DWI can provide a quantitative assessment comparable with PET/CT in GIST lesion characterization, treatment response evaluation and response prediction.
Copyright © 2012 John Wiley & Sons, Ltd.