ABSTRACT
CONTEXT.: Recently, a new type of antibody-drug conjugate, trastuzumab-deruxtecan (T-DXd), has been approved for the treatment of metastatic breast cancer with low level of human epidermal growth factor receptor 2 (HER2) gene expression. Thereby, eligibility relies on an accurate diagnosis of HER2-low status defined by immunohistochemistry IHC 1+/2+ with no gene amplification. OBJECTIVE.: To assess pathologists' accuracy and training efficacy in the diagnosis of HER2-low. DESIGN.: Agreement rates of HER2-low scoring in breast cancer tissue were assessed between expert consensus and real-world pathologists (n = 77 from 14 countries) before and after a specific 4-hour training for HER2-low detection. Two assays were evaluated, the Ventana Pathway 4B5 CDx and the Dako HercepTest (polyclonal). Concordance of the pathologists with consensus score and efficacy of training were measured by Cohen κ, overall rater agreement, and receiver operating characteristic (ROC) curve statistics. RESULTS.: In the Ventana 4B5 HER2-low category, baseline agreement rates were >80% but <90%. Negative percentage agreement was improved from 80.6% to 91.1% by training. In the HER2-0 category, positive percentage agreement (74.6%) was the only parameter below the 80% benchmark but was significantly improved to 89.2% after training. Training efficacy was confirmed by ROC curve analysis, which shows improvement for the identification of HER2-0 and HER2-low cases. Finally, in-depth examination of cases with discordant HER2 status disclosed specific issues of HER2-low underscoring and overscoring. CONCLUSIONS.: The ability of pathologists to achieve acceptable diagnostic accuracy in identifying patients with HER2-low breast cancer could be enhanced by short-term training. Potential routes to improve the quality of HER2-low scoring in clinical practice have been identified.