Construction and evaluation of a multifactorial clinical model for discriminating benign and malignant breast tumors using LASSO algorithm based on retrospective cohort study.

Abstract

Breast cancer is one of the malignant tumors that seriously threaten women's health, and early diagnosis and detection of breast cancer are crucial for effective treatment. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an important diagnostic tool that allows for the dynamic observation of blood flow characteristics of breast tumors, including small lesions within the affected tissue. Currently, it is widely used in clinical practice and has been shown promising prospects. This study included a total of 1,987 patients who underwent breast surgery at Huangpu Branch, Shanghai Ninth People's Hospital, Shanghai Jiaotong University School of Medicine from January 1, 2019 to December 31, 2019. Comprehensive patient information was collected, including ultrasound, mammography findings, physical examination details, age, family history, and pathological diagnoses. The least absolute shrinkage and selection operator (LASSO) algorithm was employed to assign values to the x variables, facilitating the construction and validation of the LASSO model group. Receiver operating characteristic curves were generated using support vector machines to determine the area under the curve (AUC), as well as to assess sensitivity and specificity. There were no statistically significant differences (P>0.05) in average age, body mass index, tumor location, or tumor benignity/malignancy between the training and test sets. The AUC, sensitivity, and specificity of mammography for predicting the benignity or malignancy of breast tumors were 0.83, 86.96%, and 76%, respectively. In comparison, the AUC, sensitivity, and specificity of DCE-MRI for the same predictions were 0.91, 91.3%, and 88%, respectively. The predictive performance of DCE-MRI was significantly higher than that of mammography (P<0.05). In conclusion, both mammography and DCE-MRI demonstrated high AUC, sensitivity, and specificity in predicting the benignity or malignancy of breast tumors. However, DCE-MRI showed superior predictive performance, making it a valuable tool for the early detection of clinical breast cancer with potential for broader clinical application.