Predictive values of 18F-FDG PET/CT image feature and metabolic parameters for the malignant potential of gastrointestinal stromal tumor / 中华核医学与分子影像杂志

ABSTRACT

Objective:To investigate the predictive values of 18F-FDG PET/CT image feature and metabolic parameters for the malignant potential of gastrointestinal stromal tumor (GIST). Methods:From March 2014 to June 2020, the 18F-FDG PET/CT imaging and surgical pathological data of 35 patients with GIST (27 males, 8 females; age 44-84 years) from Union Hospital, Tongji Medical College, Huazhong University of Science and Technology and Zhongnan Hospital of Wuhan University were analyzed retrospectively. Patients were divided into ring-shaped uptake group and other uptake patterns group according to 18F-FDG PET/CT image feature. Fisher′s exact test was used to analyze the differences of tumor necrosis and National Institutes of Health (NIH) risk classification (short for NIH classification) between different image feature groups. Mann-Whitney U test was used to analyze the differences of SUV max , metabolic parameters at different thresholds (2.5, 40%, 50%) of SUV max (metabolic tumor volume (MTV; MTV 2.5, MTV 40%, MTV 50%) and total lesion glycolysis (TLG; TLG 2.5, TLG 40%, TLG 50%)) between different clinicopathological features (lesion location, tumor diameter, mitotic count, Ki-67, necrosis, image feature, NIH classification) groups. Spearman rank correlation analysis was used to explore the correlation between clinicopathological features and metabolic parameters. ROC curve analysis was used to distinguish NIH classification of different metabolic parameters. Delong test was used to compared differences between different AUCs. Results:Of 35 GIST patients, 11(31.4%) were ring-shaped uptake and 24(68.6%) were other uptake patterns, and the differences of necrosis (7/11 vs 12.5%(3/24); P=0.004) and NIH classification (11/11 vs 25.0%(6/24); P<0.001) between the two groups were significant. There were significant differences of metabolic parameters between different groups of tumor diameter, mitotic count, necrosis, image feature, NIH classification ( z values from -4.70 to -2.09, all P<0.05), while there were no significant differences of Ki-67 ( z values from -0.83 to -0.71, all P>0.05). Metabolic parameters were correlated with mitotic count, tumor diameter, necrosis, image feature and NIH classification ( rs values 0.36-0.81, all P<0.05), while was not correlated with Ki-67 ( rs values 0.12-0.14, all P>0.05). The differences of AUCs between SUV max and MTV 2.5, TLG 2.5, TLG 40%, TLG 50%were significant (0.752, 0.856, 0.856, 0.882, 0.886; z values 1.96-2.12, all P<0.05). Conclusions:The NIH classification of GIST with ring-shaped uptake on 18F-FDG PET/CT is higher and more prone to necrosis. The 18F-FDG PET/CT metabolic parameters based on different thresholds of SUV max have certain significance for the prediction of NIH classification of GIST, and may be superior to SUV max.