Contrast-enhanced dual-energy CT of gastrointestinal stromal tumors: is iodine-related attenuation a potential indicator of tumor response?

To assess the correlation of true nonenhanced (TNE) and virtually nonenhanced (VNE) images of abdominal dual-energy computed tomography (DECT) in patients with metastatic gastrointestinal stromal tumors (GIST), and further to investigate the correlation of iodine-related attenuation (IRA) of DECT with the Choi criteria.
Twenty-four consecutive patients (5 women aged 61 ± 10 years) with metastatic GIST underwent DECT of the abdomen (80 kV, 140 kV) using first-generation dual-source computed tomography (CT). All patients had at least one or more liver lesions (median, 4; maximum, 9). Image data were processed with a dedicated DECT software algorithm designed for evaluation of iodine distribution in soft tissue lesions, and VNE CT images were generated. The tumor density (according to Choi criteria) and the maximum transverse diameter of the lesions (according to Response Evaluation Criteria in Solid Tumors [RECIST]) were determined. TNE and VNE lesion attenuation and Choi criteria and IRA were correlated with each other.
A total of 291 liver lesions were evaluated, of which 220 were cystic and 71 were solid. The mean lesion size was 4.5 ± 3.2 cm (1.1-18.7 cm). The mean attenuation of all lesions was significantly higher in the TNE images than in the VNE images (P=0.0001). Pearson statistics revealed an excellent correlation of r=0.843 (P=0.0001) between IRA and Choi criteria for all lesions. DECT showed significantly higher IRA in progressive (23.3 ± 9.5 HU) lesions compared with stable or regressive (17.8 ± 9.1 HU) lesions (P=0.0185). Similarly, the Choi criteria differed significantly between progressive (39.9 ± 12.8 HU) and stable/regressive (31.1 ± 10.3 HU) lesions (P=0.0003).
DECT is a promising imaging method for the assessment of treatment response in GIST, as IRA might be a more robust response parameter than the Choi criteria. VNE CT data calculated from DECT may eliminate the need for acquisition of a separate unenhanced data set.