CT Radiogenomics for Differentiating KIT Exon 11 from non-KIT Exon 11 Gene Mutations in Small Intestinal Gastrointestinal Stromal Tumors

Abstract Backgroud This study aimed to explore the non-invasive differentiation of gene mutational subtypes of KIT exon 11 from non- KIT exon 11 in small intestinal gastrointestinal stromal tumors (siGISTs) using radiogenomics based on the contrast-enhanced CT (CECT) images. Materials and Methods Patients with primary siGISTs diagnosed by surgery and pathology at our hospital from May 2010 to December 2022 were retrospectively evaluated. The patients were randomly divided into a training set and a validation set with a ratio of 7:3. The synthetic minority oversampling technique algorithm was used by screening radiomics features and constructing models. Then, CECT, radiomics, and combined models are established. The performance of the three models was evaluated by the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Results This study included 91 siGISTs, including 59 (64.8%) with KIT exon 11 mutations and 32 (35.2%) with non- KIT exon 11 mutations. The diagnostic performance of the CECT model and radiomics model were sufficient, with the AUC values of its ROC curve being 0.746 and 0.869 in the training set, and 0.676 and 0.787 in the validation set, respectively. Additionally, the combined model composed of CECT and radiomics features has good performance, with AUC values of 0.882 (95% confidence interval [CI]: 0.828–0.937) and 0.618 (95% CI: 0.373–0.862), respectively. Conclusions The combined radiogenomics model based on CECT has the value of non-invasive differentiating KIT exon 11 from non- KIT exon 11 mutations in patients with siGISTs.