Preoperative Prediction Value of Pelvic Lymph Node Metastasis of Endometrial Cancer: Combining of ADC Value and Radiomics Features of the Primary Lesion and Clinical Parameters

• Published in: Journal of oncology Vol. 2022; pp. 1 - 10
• Main Authors: Bo, Juan, Jia, Haodong, Zhang, Yu, Fu, Baoyue, Jiang, Xueyan, Chen, Yulan, Shi, Bin, Fang, Xin, Dong, Jiangning
• Format: Journal Article
• Language: English
• Published: New York Hindawi 30-06-2022; John Wiley & Sons, Inc; Hindawi Limited
• Subjects: Cancer; Data analysis; Drunk driving; Endometrial cancer; Hysterectomy; Lymphatic system; Metastasis; Morphology; Nomograms; Physicians; Radiomics; Software; Surgical outcomes; Tumors; United States > US
• ISSN: 1687-8450; 1687-8450
• DOI: 10.1155/2022/3335048

Objective. To investigate the value of apparent diffusion coefficient (ADC) value of endometrial cancer (EC) primary lesion and magnetic resonance imaging (MRI) three-dimensional (3D) radiomics features combined with clinical parameters for preoperative prediction of pelvic lymph node metastasis (PLNM). Methods. A total of 136 patients with EC confirmed by postoperative pathology were retrospectively reviewed and analyzed. Patients were randomly divided into training set (n = 95) and test set (n = 41) at a ratio of 7 : 3. Radiomics features based on T2WI, DWI, and contrast-enhanced T1WI (CE-T1WI) sequence were extracted and screened, and then radiomics score (Rads-score) was calculated. Clinical parameters and ADC value of EC primary lesion were measured and collected, and their correlation with PLNM was analyzed. Receiver operating characteristic (ROC) curve was plotted to assess the diagnostic efficacy of the model. A nomogram for PLNM was created based on the multivariate logistic regression model. Results. The ADC value of the EC primary lesion showed inverse correlation with PLNM, while CA125 and Rads-score were positively associated with PLNM. A predictive model was proposed based on ADC value, Rads-score, CA125, and MR-reported pelvic lymph node status (PLNS) for PLNM in EC. The area under the curve (AUC) of the model is 0.940; the sensitivity and specificity (87.1% and 90.6%) of the model were significantly higher than that of the MRI morphological signs. Conclusion. A combination of ADC value, MRI 3D radiomics features of the EC primary lesion, and clinical parameters generated a prediction model for PLNM in EC and had a good diagnostic performance; it was a useful supplement to MR-reported PLNS based on MRI morphological signs.