Spatially-resolved single-cell HER2 tumor heterogeneity captured by biophysical modeling


Breast cancer is a highly heterogeneous disease. Prognosis and treatment are usually decided based on the expression of 3 molecular markers; one being the human epidermal growth factor receptor 2 (HER2). Classification of HER2 tumors generally follows a binary system, either HER2 positive or HER2 negative, the latter encompasses the triple negative breast cancer (TNBC) types. However, HER2 heterogeneity to a new nomenclature: HER2 positive, HER2 low, and HER2 negative. Here we present a state-of-the art technique to facilitate the identification of HER2-low patients by capturing the spatial distribution of HER2 expression across the entire tumor. This identifies TNBC patients that express HER2-low and for whom there are no targeted therapies and who could benefit from anti-HER2 drugs.

We identified the range of HER2 expression present from microarrays run on biopsies from HER2-positive (n=48) and TNBC (n=84) patients from a subset of the ISPY1 & 2 trials and characterized the overlap in both HER2 expression and expression of HER2 signaling-responsive genes. We next integrated single cell RNA-seq data with patient-specific tumor morphology using a biophysical modeling platform, SimBioSys TumorScope, and statistical matching algorithms to characterize HER2 heterogeneity within individual patient tumors. We then visualized HER2 heterogeneity within tumors using 3D maps of HER2 expression.

We identified a subset of TNBC patients (17/84) that both expressed HER2 at low levels and a previously published HER2-responsive gene expression signature. Additionally, we characterized the 3D intra-tumoral heterogeneity of HER2 expression predicted to occur within these tumors.

Biophysical modeling can integrate multiple large datasets together and capture tumor heterogeneity at high resolution and large scale. These advances enable a more comprehensive understanding of tumor biology and heterogeneity, and ultimately improve clinical decision making across solid tumors. In our study, this translates to the identification of a subset of TNBC patients that express HER2-low which could benefit from HER2-targeted therapy, like T-DXd which, unlike trastuzumab or T-DM1, has shown promising results in HER2-low patients.

View the full publication here.