The accuracy of HER2 breast cancer scoring improved with the use of AI assistance, especially for patients with low and ultralow levels of HER2 expression, results from a multinational study showed. The findings were presented in a press briefing ahead of the 2025 ASCO Annual Meeting (Abstract 1014). With AI assistance, fewer cases were misclassified as HER2-null, which would have prevented the patient from being eligible for treatment with HER2-targeted antibody-drug conjugates.
“Roughly 65% of breast tumors once called HER2‑negative actually demonstrate some level of HER2 expression and belong to subgroups now classified as HER2-low or HER2-ultralow breast cancers. Some of these tumors could be treated with HER2-targeted drugs, but only if we detect their HER2 expression levels. Our study provides the first multinational evidence that artificial intelligence can help close a critical diagnostic gap and open the door to new therapies like antibody-drug conjugates for a majority of patients who, until recently, had not been offered these options,” stated lead study author Marina De Brot, MD, PhD, Associate Pathologist, A.C. Camargo Cancer Center, São Paolo, Brazil.
Study Methods and Results
Researchers implemented an AI-integrated digitial training platform, called ComPath Academy, to teach pathologists about HER2 scoring of breast cancer samples. The study comprised a masterclass of 105 pathologists from 10 countries who assessed 20 digital breast cancer cases for HER2 expression across three exams, with AI assistance for decision making available only during the final exam.
HER2 immunohistochemistry (IHC) scoring classifications followed ASCO/CAP 2023 guidelines, with the addition of HER2-ultralow (IHC 0 with membrane staining) and HER2-null (IHC 0 with no membrane staining) newer classifications.
The pathologists achieved an accuracy rate with reference scores of 89.1% without AI assistance in the first two exams compared with 96.1% accuracy with AI assistance in the third exam. Concordance among the pathologists was calculated at 0.506 without AI and 0.798 with AI.
Classification of each clinical category improved from 90.1% without AI to 95.0% with AI assistance. Concordance among pathologists by HER2 clinical category increased from 0.4994 without AI to 0.732 with AI. The rate of misclassification of HER2-low or HER2-ultralow cases as HER2-null was reduced by 24.4% with AI assistance.
“Accurate HER2 scoring is important to ensure that patients receive the best treatment for their breast cancer. This international study shows that an AI-assisted approach improved HER2 scoring, including in situations that would affect treatment decisions. These findings shed light on the promising role for AI in oncology, not as a replacement for the physician, but as a powerful tool to help us work smarter and faster to deliver high-quality, more personalized care,” stated Julian Hong, MD, MS, Associate Professor and Medical Director of Radiation Oncology Informatics at the University of California, San Francisco, and an ASCO expert in AI.
Disclosure: For full disclosures of the study authors, visit asco.org.
