Background Triple negative breast cancer (TNBC) accounts for 10–20% of all diagnosed breast cancer cases in the US and is characterized by loss of HER2, estrogen receptors, and progesterone receptors. TNBC is an aggressive, complex disease with a poor prognosis due to resistance to traditional therapies. Understanding the underlying biology and tumor microenvironment is critical to the development of diagnostic biomarkers and to guide the search for effective therapies. Here, we demonstrated the ability of the 10x Genomics Visium Spatial Gene Expression Solution to elucidate the immunological profile and microenvironment of TNBC samples in conjunction with standard pathological techniques.

Methods Spatial transcriptomics technology complement pathological examination by combining the benefits of histological stains with the throughput and deep biological insight of RNA-seq. We investigated serial sections of TNBC by using the 10x Genomics Visium Spatial Gene Expression Solution to spatially resolve the samples’ cellular composition and expressed microenvironment. Visium incorporates ~5000 molecularly barcoded, spatially encoded capture probes in spots over which a tissue section is placed and imaged. The samples are permeabilized and native mRNA is captured. Imaging and RNA sequencing data are processed together, resulting in whole transcriptome gene expression mapped to the tissue image.

Results We captured spatial patterns of gene expression and mapped the information back to H&E-stained images with regional annotations. Serial sections were then subject to fluorescence immunohistochemical staining for immune infiltrate paired with spatial gene expression capture. Subsequently, we combined these data with 3’ single-nuclei RNA-seq from the same tumor, generating expression profiles that were used to automatically annotate cell-types across the sections. This allowed for an understanding of the tumor microenvironment that could not be captured by image-based techniques alone. We resolved subgroups of spatially and biologically distinct immune, stem, and cancer progenitor cells. Finally, we digitally annotated tumor and normal tissue regions using expressed genetic mutations alone. Annotated tumor regions expressed more deleterious mutations than normal regions and we were able to automatically cluster regions of tumor vs. normal cells without any prior histopathological information. We also found intratumor gradients of mutational burden in oncogenes as well as non-cancer associated loci.

Conclusions Taken together, we demonstrated that Visium can provide a powerful complement to traditional histopathology, enabling both targeted panels and whole-transcriptome discovery of gene expression. This spatially resolved molecular information provides an unprecedented view into the tumor microenvironment and a powerful new tool for discovery of new biomarkers and therapeutic targets.