Abstract
Intra-tumor heterogeneity is a pervasive property of human cancers that poses a major clinical challenge. Here, we describe the characterization, at the transcriptional level, of the intra-tumor topography of two prominent breast cancer biomarkers and drug targets, epidermal growth factor receptor 2 (HER2) and estrogen receptor 1 (ER) in 49 archival breast cancer samples. We developed a protocol for single-molecule RNA FISH in formalin-fixed, paraffin-embedded tissue sections (FFPE-smFISH), which enabled us to simultaneously detect and perform absolute quantification of HER2 and ER mature transcripts in single cells and multiple tumor regions. We benchmarked our method with standard diagnostic techniques, demonstrating that FFPE-smFISH is able to correctly classify breast cancers into well-established molecular subgroups. By counting transcripts in thousands of single cells, we identified different expression modes and levels of inter-cellular variability. In samples expressing both HER2 and ER, many cells co-expressed both genes, although expression levels were typically uncorrelated. Finally, we applied diversity metrics from the field of ecology to assess the intra-tumor topography of HER2 and ER gene expression, revealing that the spatial distribution of these key biomarkers can vary substantially even among breast cancers of the same subtype. Our results demonstrate that FFPE-smFISH is a reliable diagnostic assay and a powerful method for quantification of intra-tumor transcriptional heterogeneity of selected biomarkers in clinical samples.