An ultrasound based method for predicting the malignant potential of primary gastrointestinal stromal tumors preoperatively.

Abstract

Gastrointestinal stromal tumors (GISTs) are difficult to identify the risk level accurately without surgical pathological confirmation. The purpose of our study was to propose a noninvasive prediction method for predicting the malignant potential of GISTs preoperatively by using contrast-enhanced ultrasound (CEUS) with gastric distention. We reviewed 47 GISTs who underwent CEUS from April 2017 to August 2023 retrospectively, all the lesions were certificated by pathology after surgery. The age of the patient, size of the lesion, shape, necrosis, calcification in the lesion, perfusion parameters including arrival time (AT), peak intensity (PI), time to peak (TTP), and area under the curve (AUC) of the lesion and surrounding normal tissue were analyzed. Logistic regression analyses were performed. Of the 47 GISTs, 26 were high-risk and 21 low-risk tumors respectively. Compared with low-risk GISTs, high-risk GIST had faster AT (7.7s vs. 11.5s, p < 0.05), higher PI (15.2dB vs. 12.5dB, p < 0.05), and larger size (4.4cm vs. 2.2cm, p < 0.001). In multivariate logistic regression, AT, PI, and size were significant features. The corresponding regression equation In (p/(1-p)=-5.9 + 4.5 size + 4.6 PI + 4.0 AT). The size, AT, and PI of the GISTs on CEUS can be used as parameters for a noninvasive risk level prediction model of GISTs. This model may help identify the different risk levels of GISTs before surgery.