Coronary Plaque Sampling Reveals Molecular Insights Into Coronary Artery Disease (2024)

Much of our understanding of human coronary atherosclerosis is extrapolated from studies of animal models that do not develop coronary atherosclerosis and have limited evidence of plaque instability, or analyses of human coronary atherosclerotic plaques from expired patients or explanted hearts, or peripheral artery plaques.

We developed a coronary plaque sampling approach that could be applied broadly in live patients with coronary artery disease to obtain molecular and cellular insights into human coronary atherosclerosis. Our approach combined RNA retrieval directly from balloons used in percutaneous coronary intervention and inexpensive, low-input RNA-Seq using SMART-seq (cost of about $160 for library preparation and sequencing; Figure [A]). The study protocol was approved by the Medical College of Wisconsin’s Institutional Review Board.

We generated SMART-Seq libraries from coronary samples from 27 patients. The complete dataset is available in the Gene Expression Omnibus (GSE236610). More than 9 million uniquely mapped read pairs were obtained from each library. The median mapping rate was 92%. The median number of genes detected in a library was 30 375, and the interquartile range was 19 651 to 32 973. Of the 27 patients, 13 were confirmed to have stable coronary artery disease (sCAD), and 14 confirmed to have been performed on lesions causing acute coronary syndrome (ACS). Demographic and procedural information was summarized in Figure [B]. Patients with ACS met criteria for type 1 myocardial infarction (fourth universal definition). sCAD was defined as exertional anginal symptoms without severity change >30 days before percutaneous coronary intervention.

The cell type composition of a complex sample can be imputed from bulk RNA-seq data with tools such as CIBERSORTx.¹ We applied CIBERSORTx to analyze the SMART-seq data from each of the 27 samples. A single-cell RNA-seq dataset from human atherosclerotic coronary arteries from explanted hearts of transplant recipients was used as the reference.² To ensure the identification of any neutrophils, a gene expression profile of human neutrophils from 10× Genomics was added to the reference. The CIBERSORTx analysis identified up to 13 named cell types and several unnamed cell clusters in each sample (Figure [C]). Fibroblasts and fibromyoctes were enriched, while smooth muscle cells were reduced, in samples from ACS compared with patients with sCAD (Figure [C]). Smooth muscle cells of coronary arteries may undergo phenotypic modulation to fibromyocytes or inflammatory-like cells in atherosclerosis.^2,3 Other detected cell types did not differ significantly between sCAD and ACS (Figure [C]).

We calculated the coefficient of variance of each cell type across the samples as median absolute deviation divided by median. The median value of the coefficient of variance of all cell types was 30% for patients with sCAD and 28% for patients with ACS, indicating that the cell type proportions were reasonably consistent across the samples. We looked for expression signatures of neutrophils, which make up 50% to 70% of white blood cells, to assess blood contamination as red blood cells contain minor amounts of RNA. Neutrophils constituted, on average, 9.8% of the cells in our samples, with an interquartile range of 7.8% to 10.5% and a coefficient of variance of 26% in sCAD and 24% in ACS. While it is not clear how prevalent neutrophils are in coronary plaques,⁴ these data suggested that any blood contamination in our samples was minimal and relatively consistent across all samples.

A cuffdiff analysis with adjustment for multiple comparisons identified 371 genes as significantly differentially expressed (q<0.05) between patients with sCAD and ACS (Figure [D]). The ACS samples tended to cluster together in a principal component analysis, while sCAD samples were more scattered (Figure [E]). Gene set enrichment analysis using normalized count data identified apical junction and angiogenesis as the top enriched hallmark gene sets in patients with ACS compared with patients with sCAD (P<0.05), but the enrichment did not reach significance following adjustment for multiple comparisons.

These findings support the feasibility and robustness of coronary plaque sampling for transcriptome analysis and its unique potential to provide molecular and cellular insights into coronary plaques sampled from live patients and potentially to improve individualized clinical decision making. The method is broadly applicable to all patients undergoing percutaneous coronary and not restricted to more narrowly applied procedures such as rotational atherectomy.⁵ Variabilities in percutaneous coronary intervention procedures, the balloon contact with plaques, and low-input RNA-seq may introduce noise into the data. However, we were able to identify statistically significant differences between sCAD and ACS, the proportions of most cell types were reasonably consistent across all samples, and blood contamination appeared minimum, all of which support the robustness of our approach. Studies with larger sample sizes are needed to both confirm our findings and determine if coronary plaque transcriptome data are predictive of additional downstream cardiovascular events.

In conclusion, our data demonstrate the feasibility of obtaining meaningful transcriptome information from human coronary plaques in live patients broadly and using this information to identify differences between patient populations. This methodology holds promise to improve our understanding of human coronary plaque pathology and potentially to help prognosticate risk of coronary artery disease progression or ACS.

ARTICLE INFORMATION

REFERENCES