[Correlation between the CT features and malignancy risk of small (≤5 cm) gastric stromal tumors].

Objective: To evaluate the correlation between computed tomography (CT) features and malignancy risk category of small (≤5 cm) gastric stromal tumors (GST), in order to provide an image reference for preoperative assessment and intraoperative pathological diagnosis. Methods: Eighty-three patients with surgically and pathologically proven GST (≤5 cm) between January 2011 and November 2015 were recruited, and their clinical, pathological and CT data were retrospectively analyzed. According to the pathological results and malignancy risk category, the patients were divided into 2 groups, the benign biological behavior group (very low and low risk) and malignant biological behavior group (intermediate and high risk). The clinical, pathological and CT imaging findings of the two groups were analyzed. Based on the tumor diameter, the receiver operating characteristic curve (ROC) was applied to evaluate the sensitivity, specificity and the best cut-off point for distinguishing the malignancy risk between the two groups. Results: The lobulation and ulceration of the tumors presented statistically significant difference for the malignancy risk between the two groups (χ(2)=6.273 and 4.163, respectively; all P<0.05), but there was no significant difference in the sex, clinical symptoms, serum ferritin, tumor site, growth pattern, cystis degeneration and calcification (all P>0.05). No statistically significant differences were detected for the tumor CT value, arterial CT value, venous CT value, degrees of enhancement in arterial phase (DEAP), enhancement in portal venous phase (DEPP), and patient's age for distinguishing the malignancy risk between the two groups (all P>0.05). On the other hand, significant differences were found in the maximum diameter (Dmax) of tumor and the minimum diameter (Dmin) of tumor (t=-3.256 and -3.466, respectively; all P<0.05). When the cut-off point of Dmax was 1.6 cm, the area under the ROC curve, sensitivity and specificity were 0.704, 92.3% and 75.4%, respectively. When the cut-off point of Dmin was 1.5 cm, the area under the ROC curve, sensitivity and specificity were 0.713, 88.5% and 71.9%, respectively. Conclusion: CT features of the GST (≤5 cm) may predict, before surgery, the malignancy risk of small gastric stromal tumors, and provide the an image reference for preoperative assessment and intraoperative pathological diagnosis of the disease.