Luminal breast cancer is characterized by marked heterogeneity, resulting in diverse treatment responses and long-term outcomes. This project aims to integrate MRI and multiomics data to achieve non-invasive molecular typing and precise response prediction. By linking imaging phenotypes with underlying molecular and pathological characteristics, the investigators will develop predictive models for treatment resistance, recurrence, and metastasis, ultimately supporting personalized treatment strategies and precision oncology.
Luminal breast cancer represents the most common type of breast cancer, characterized by its intricate tumor heterogeneity that poses a significant challenge in clinical management due to resistance to endocrine therapy and high risk of long-term recurrence. It is significant for the accurate prediction of molecular subtypes and treatment response for luminal breast cancer. Our team has previously identified four molecular subtypes and seven pivotal molecules associated with luminal breast cancer utilizing multiomics techniques. The investigators posit that the integration of MRI-driven multiomics studies holds promise in achieving precise typing and response prediction for luminal breast cancer. This project intends to use multiomics molecular subtypes and key molecules as the gold standard to extract comprehensive quantitative features from diverse regions and levels utilizing MRI, thus facilitating non-invasive diagnosis. Additionally, our approach involves correlating MRI data with multiomics information to unveil the biological significance of imaging models at both pathological and molecular levels. Finally, the investigators aim to construct response prediction models through the fusion of multi-temporal MRI features and multiomics data across various scales, enabling precise forecasts of treatment resistance, recurrence, and metastasis. This initiative aims to enhance treatment decision-making and promote application transformation. This study will include a large-scale real-world retrospective and prospective population to validate and improve the effectiveness of model.
Study Type
OBSERVATIONAL
Enrollment
2,000
Fudan university Shanghai Cancer Center
Shanghai, Shanghai Municipality, China
Diagnostic performance of breast MRI for molecular subtyping of luminal breast cancer, with comparison to multiomics
The primary outcome is the diagnostic performance of AI-assisted analysis for molecular subtyping of luminal breast cancer on contrast-enhanced breast MRI. Quantitative radiomic features and deep learning features are extracted from DCE-MRI, followed by classification into multiomics-defined molecular subtypes. Performance metrics include sensitivity, specificity, positive predictive value, negative predictive value, accuracy, and area under the receiver operating characteristic curve (AUC). Participants must have undergone both breast MRI and multiomics profiling of tumor tissue. Performance metrics will be compared with those obtained from multiomics classification within the same participants to evaluate the relative diagnostic performance.
Time frame: 1 year
Predictive Performance of Multiomics Model for Pathological Complete Response (pCR) in Luminal Breast Cancer
The model integrates multiomics data, including breast MRI, pathological features, and other relevant molecular and clinical variables, to predict pathological complete response (ypT0/is ypN0) following neoadjuvant therapy in patients with luminal breast cancer. Performance metrics include sensitivity, specificity, positive predictive value, negative predictive value, accuracy, area under the receiver operating characteristic curve (AUC), C-index, and time-dependent AUC. Participants must have undergone neoadjuvant therapy with available pathological response assessment.
Time frame: 1 years
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.