Transcriptomic-based clustering of advanced atherosclerotic plaques identifies subgroups of plaques with differential underlying biology that associate with clinical presentation

Histopathological studies have revealed key processes of atherosclerotic plaque thrombosis. However, the diversity and complexity of lesion types highlight the need for improved sub- phenotyping. We hypothesized that unbiased clustering of plaques based on gene expression results in an alternative categorization of late-stage atherosclerotic lesions. We analyzed the gene expression profiles of 654 advanced human carotid plaques. The unsupervised, transcriptome-driven clustering revealed five dominant plaque types. These novel plaque phenotypes associated with clinical presentation (p<0.001) and showed differences in cellular compositions. Validation in coronary segments showed that the molecular signature of these plaques was linked to coronary ischemia. One of the plaque types with most severe clinical symptoms pointed to both inflammatory and fibrotic cell lineages. This highlighted plaque phenotype showed high expression of genes involved in active inflammatory processes, neutrophil degranulation, matrix turnover, and metabolism. For clinical translation, we did a first promising attempt to identify circulating biomarkers that mark these newly identified plaque phenotypes. In conclusion, the definition of the plaque at risk for a thrombotic event can be fine-tuned by in- depth transcriptomic based phenotyping. These differential plaque phenotypes prove clinically relevant for both carotid and coronary artery plaques and point to differential underlying biology of symptomatic lesions.