Improvements in the accuracy and reproducibility of HER2 scoring among pathologists working in the breast cancer HER2-low setting remains a largely unmet challenge. The application of artificial intelligence (AI)-powered digital image analysis (DIA) may help to meet these needs by improving the consistency of HER2 expression measurement.We used published inter-observer HER2 concordance data generated in a study where a group of 16 pathologists with recognised expertise in the area undertook the HER2 status assessment of ‘real-world’ breast cancer core biopsies, the majority of which had previously been shown to express HER2 in the low range.This data was compared with that produced by DIA of the same sample set using a commercially available application (APP) for HER2 evaluation. The aim of the study was to produce information about the level of agreement between shown between a market-leading DIA APP and a large group of expert raters as a robust measure of the feasibility of implementing DIA in the assessment of HER2 status in the clinical setting.

Andrew Dodson¹, Abeer Shaaban², Lila A Zabaglo¹, Suzanne Parry¹

1. UK National External Quality Assessment Scheme for Immunocytochemistry and In-Situ Hybridisation, London UK.
2. University Hospitals Birmingham NHS Foundation Trust Queen Elizabeth Hospital, Birmingham, UK.