Investigators at Massachusetts General Hospital's Cancer Center have developed the CLARITY Model, an artificial intelligence-driven tool designed to identify breast cancer patients who will benefit from targeted therapy with HER2-low antibody-drug conjugates (ADCs) like trastuzumab deruxtecan (Enhertu). This model analyzes digitized tumor slides and was presented at the 2023 San Antonio Breast Cancer Symposium (SABCS).
Currently, determining HER2-low status relies on immunohistochemistry (IHC) staining, which can suffer from reproducibility issues and interobserver variability, particularly for cases in the 1+ category. Such limitations can lead to misidentification of eligible patients for these specific therapies. The CLARITY Model utilizes a deep learning algorithm trained on data from 1,226 patients across two cohorts, including participants from a phase 2 trial (DB-04) of trastuzumab deruxtecan (T-DXd) in HER2-low metastatic breast cancer, aiming to circumvent these challenges.
The CLARITY Model demonstrated an accuracy of 0.88 in predicting HER2-low status when compared with IHC results and successfully predicted treatment benefit with trastuzumab deruxtecan (T-DXd). Dr. David Ting, clinical director at Massachusetts General Hospital Center for Innovation in Digital Health and artificial intelligence in medicine, stated that the model could expand eligibility for this therapy. Investigators are now planning to validate the model in a larger, prospective study.
