Artificial intelligence (AI)-assisted HER2 scoring improved diagnostic accuracy and reduced turnaround time in determining the immunohistochemistry of patients with breast cancer whose HER2 status is equivocal, according to findings presented at the San Antonio Breast Cancer Symposium 2025. 

“...the visual interpretation of HER2 expression is subjective and semi-quantitative, which leads to intra- and inter-pathologist variability. One of the accepted reference standards for determining HER2 status in equivocal HER2 2+ cases is in situ hybridization (ISH) to assess gene amplification from FFPE tumor samples. However, assessment of gene amplification by ISH can significantly increase result turnaround times.”

─ Manuela Vecsler, PhD, vice president of clinical and scientific affairs at Ibex Medical Analytics in Tel Aviv, Israel, and colleagues 

The researchers performed an observational study of excisions and biopsy samples from 2300 patients in the US, EU, and the UK. Twenty-eight pathologists reviewed slides of a wide array of breast cancer samples stained with anti-HER2 antibody, with each reader examining between 50 and 200 slides. 

In one arm, pathologists worked unassisted, whereas the other arm was assisted by a proprietary AI tool that automatically detects tumors, classifies tumor cells with HER2 staining, and applies ASCO/CAP guidelines to determine a HER2 immunohistochemistry score. Researchers compared accuracy between the two arms and sought feedback from those who used the AI tool.

Unassisted pathologists had a 76.4% accuracy rate, whereas the AI-assisted accuracy rate was 82%. This improvement in accuracy extended across all categories of HER2 status, from null to positive, with an accuracy rate of 79.3% for unassisted reading and 84.4% for AI-assisted reading. 

The proportion of slides requiring ISH testing was 29% with unassisted reading vs 18.6% with assisted reading. Vecsler and colleagues wrote that this could result in a 35% shorter patient turnaround time. 

Pathologists reported in a survey that the AI tool increased their confidence in the accuracy of their HER2 scoring and helped them maintain consistency across slides; 83% said they were motivated to keep using the tool. 

“[T]he use of AI reduced equivocal (HER2 2+) classifications, enriched HER2 gene amplification status in cases classified as [equivocal], potentially enabling faster diagnostic turnaround. Pathologist feedback indicated increased confidence in scoring and strong motivation to adopt AI-assisted workflows over manual interpretation. These findings highlight the value of AI systems in biomarker interpretation, providing pathologists with enhanced decision-making tools with explainability at the individual cell level and improving diagnostic precision in HER2 [immunohistochemistry] interpretation. Improved scoring accuracy and consistency can support more reliable patient stratification and treatment selection, helping to ensure that patients receive the most appropriate HER2-targeted therapies.”

─ Manuela Vecsler, PhD, vice president of clinical and scientific affairs at Ibex Medical Analytics in Tel Aviv, Israel, and colleagues

--
By Andrew John

Reference: https://www.abstractsonline.com/pp8/#!/21392/presentation/257

Disclosures: Vecsler is an employee of Ibex Medical Analytics, which created the AI tool. Some authors declared financial ties to drugmakers. See full study for details.

Photo Credit: Getty Images