Systematic Analysis of Factors Affecting Human Epidermal Growth Factor Receptor 2 Interpretation Consistency: Staining Protocols, Artificial Intelligence–Based Image Standardization, and Classification Criteria

Abstract

The efficacy of human epidermal growth factor receptor 2 (HER2)–targeting antibody-drug conjugates has underscored the critical need for precise HER2 diagnostics in breast cancer treatment. Despite the clinical importance, variability in immunohistochemical (IHC) staining protocols and interobserver inconsistencies challenge the reliability of HER2 status assessment, which is critical for guiding patient treatment strategies. To investigate the factors affecting HER2 interpretation consistency, tissue microarrays from 1063 breast carcinoma cases underwent 3 distinct IHC protocols, and a novel artificial intelligence (AI) model was developed to standardize HER2-stained images. A total of 5 sets of tissue microarrays (Nordi QC, protocol 1, protocol 2, protocol 1 AI, and protocol 2 AI) were independently reviewed by 8 pathologists. The Fleiss Kappa value and overall agreement rate measured interobserver agreement, with logistic regression analyzing the impact of variables on diagnostic accuracy. Our results showed that the Nordi QC protocol had the highest interobserver agreement (Kappa 0.754). AI-based image normalization notably enhanced consistency, particularly for HER2 low cases, aligning scores toward the Nordi QC standard. Logistic regression analysis indicated that both staining protocol and AI-based image standardization significantly influenced diagnostic accuracy (P < .001). The American Society of Clinical Oncology/College of American Pathologists 2018 binary criteria demonstrated the highest HER2 interobserver consistency (Kappa > 0.95). Compared with the American Society of Clinical Oncology/College of American Pathologists 2023 criteria, the newly proposed null, ultra-low/low, positive criteria, merging HER2 low and ultra-low categories, demonstrated improved reliability and agreement, especially in distinguishing the challenging HER2–ultra-low cases, which showed an exceedingly low interobserver agreement (Kappa < 0.20) across all protocols. Overall, variability in IHC staining protocols and HER2 classification criteria significantly affect the diagnostic consistency among pathologists. The integration of an AI model for image standardization and the adoption of the null, ultra-low/low, positive criteria may refine diagnostic precision and bolster clinical decision-making in breast cancer treatment.