We developed an integrated multi-omics approach to analyze early cancer precursors. This approach enabled us to infer a richer set of cellular genomic features than would otherwise be generated from a single platform assay. This integrated approach combined conventional RNA-seq of these precursors with information from spatial transcriptomics and single cell RNA-seq. On its own, standard gene expression studies provide molecular signatures which can be associated with specific clinical outcomes such as increased risk of cancer. However, gene expression as a sole data source can neither identify individual aberrant cells, nor provide topographic disease tissue features such as mapping expression signatures to specific precancerous tissue regions. Our approach addressed all of these limitations.
RNA-seq was applied to an extended cohort of early stomach cancer precursors that are referred to as gastric intestinal metaplasia. We leveraged the sample size of RNAseq to identify a novel high-risk gene signature. We applied spatial transcriptomics (10X Visium) to an independent set of gastric intestinal metaplasia and gastric cancer samples. This spatial information enabled us to map differentially expressed high-risk genes specifically to pathologist-annotated regions of intestinal metaplasia. To improve the inference of the affected cells types in regions of metaplasia, we used single cell information to make assignment of aberrant epithelial cells. Finally, we used InferCNV on the spatial gene expression data to identify chromosome 8q amplifications, an early event associated with gastric cancers and restricted only to the aberrant epithelial cell types. Overall, our integrated multi-omics approach enabled the identification of early genomic events harbored by aberrant cell types in high-risk precancerous lesions. This approach may prove useful for discovering new clinical and diagnostic biomarkers.