Abstract Trastuzumab deruxtecan (T-DXd) is commonly used for treating metastatic breast cancer (MBC); however, traditional HER2 immunohistochemistry has largely failed to predict T-DXd activity. We reviewed survival outcomes and tested the reliability of multiple HER2 quantitative assays in predicting T-DXd’s performance among 191 patients with MBC. We demonstrate that T-DXd’s activity varies depending on the temporal evolution of HER2 immunohistochemical expression, with the longest activity observed among patients with HER2-positive disease or maintaining HER2-low disease across primary and metastatic settings. Quantitative HER2 assessment on pre-T-DXd samples showed that time-to-next treatment progressively increased by High Sensitivity-HER2 quartiles, Reverse Phase Protein Array HER2 quartiles, HER2DX ERBB2 mRNA scores and plasma-based DNADX HER2 signature tertiles (all with log-rank p < 0.05). Conversely, HER2 immunohistochemical subtypes showed limited predictive value for clinical outcomes. Additionally, elevated TOPO1 expression was associated with worse outcomes with T-DXd in HER2-negative breast cancer, suggesting potential relevance for payload-related markers in predicting T-DXd performance.
Tarantino et al. (2026) studied this question.