Gastrointestinal stromal tumors (GIST): a proposal of a "CT-based predictive model of Miettinen index" in predicting the risk of malignancy.

Purpose: To identify the predictors of malignancy on CT for the evaluation of gastrointestinal stromal tumors (GIST) by correlating CT findings with the mitotic index in order to propose a "CT-based predictive model of Miettinen index."

Methods: One radiologist and one resident in radiology with 14- and 4-year experience in oncological field reviewed the CT findings of 42 patients by consensus, with respect to lesion site, size, contour, tumor growth pattern, enhancing pattern, degree of enhancement of tumor, percentage of tumor necrosis, mesenteric fat infiltration, ulceration, calcification, regional lymphadenopathy, direct invasion to adjacent organs, and distant metastasis. All parameters were correlated with the mitotic index evaluated at histopathological analysis following surgery. Normality of variables was evaluated using Shapiro-Wilk test. Pearson's correlation test was used to assess the interaction between variables. The diagnostic accuracy percentage of tumor necrosis was measured by receiver operating characteristic (ROC) analysis for detecting whether the number of mitosis per 50 high-power fields was > 5.

Results: A significant statistical correlation was found between percentage of tumor necrosis and the mitotic index (p < 0.005), dimension, and location of the tumor.

Conclusion: CT could be an accurate technique in the prediction of malignancy of GIST in a CT risk assessment system, based on the location of the tumor, its size, and the percentage of tumor necrosis.