Prediction of Germline BRCA Mutations in High-Risk Breast Cancer Patients Using Machine Learning with Multiparametric...

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Germline BRCA1/2 (BRCA) mutations are crucial for high-risk breast cancer treatment planning.
• Genetic testing for BRCA mutations can be expensive or inaccessible.
• Multiparametric breast MRI (mpMRI) offers noninvasive imaging biomarkers for potential BRCA mutation prediction.

Purpose of the Study:

• To investigate the ability of mpMRI features to predict BRCA mutation status in high-risk breast cancer patients.
• To identify specific mpMRI features associated with BRCA mutations.
• To evaluate machine learning models for BRCA mutation prediction using mpMRI data.

Main Methods:

• Retrospective analysis of 231 high-risk breast cancer patients (2013-2019) with BRCA testing and preoperative mpMRI.
• Inclusion of computer-aided diagnosis (CAD)-derived kinetic features, morphologic features, and apparent diffusion coefficient (ADC) values from diffusion-weighted imaging (DWI).
• Multivariate analysis to identify significant predictors, followed by evaluation of 13 machine learning (ML) models.

Main Results:

• Univariate analysis indicated associations between BRCA mutation and higher CAD-derived washout component, peak enhancement, larger tumor size, angio-volume, peritumoral edema, axillary adenopathy, and minimal/mild background parenchymal enhancement (BPE).
• Multivariate analysis identified washout component ≥ 19.5%, minimal/mild BPE, and tumor size ≥ 2.5 cm as significant predictors (ORs 3.89, 2.57, 2.41 respectively).
• The highest performing ML model achieved an Area Under the Curve (AUC) of 0.72 for BRCA mutation prediction.

Conclusions:

• Machine learning models integrating mpMRI features demonstrate good performance in predicting BRCA mutations in high-risk breast cancer patients.
• This noninvasive imaging approach may assist in personalized treatment strategies and genetic counseling.
• mpMRI features serve as valuable biomarkers for noninvasive BRCA mutation status prediction.