Grading HER2 at the nanoscale in clinical tissue

To guide diagnosis and treatment, breast cancer biopsies are assessed for HER2 status and assigned one of four grades (0-3+). While current practices are sufficient for detection of HER2 overexpression (3+), there is a need for more sensitive methods capable of characterising lower HER2 expression in patients who may still benefit from HER2-targeted therapies. Super-resolution fluorescence microscopy techniques, such as single molecule localisation microscopy (SMLM), have reshaped the study of nanoscale molecular architecture by visualising single target molecules in a range of sample types. Here, we have developed a quantitative SMLM workflow to visualise HER2 nanoclustering in patient-derived xenografts (PDX) and clinical breast tumour tissue from eight patients spanning all disease grades. Analysis of HER2 cluster architecture revealed grade-dependent changes in size of cluster and HER2 abundance. We then applied a blinded data-driven approach to regroup samples based on this nanoscale HER2 clustering. This led to the reclassification of three samples into new groups, due to similarities in nanoscale signature. Together, these findings demonstrate that quantitative fluorescence nanoscopy can be used to identify clinical HER2 phenotypes across a range of expression levels due to its exquisite sensitivity, and this could be leveraged to stratify patients for targeted therapy.