Whole-blood transcriptomics differentiates circulating gene expression between coronary artery disease and peripheral artery disease

Despite both being prevalent atherosclerotic disorders, coronary artery disease (CAD) and peripheral artery disease (PAD) exhibit distinct clinical and pathological presentations. ¹ Recent genome-wide association studies of PAD and CAD have identified genetic differences between the diseases, notably suggesting a stronger association of smoking and thrombosis with PAD compared with CAD. ² Patients with PAD are also particularly high risk for adverse cardiac events even compared with patients with prior myocardial infarction. ³ The pathophysiologic mechanisms underlying these differences remain elusive. Identifying whole-blood gene expression differences between PAD and CAD may improve understanding of risk factors and mechanisms specific to each disease type. Whole-blood RNA sequencing was performed in patients with CAD (N = 51) and patients with PAD (N = 20) to uncover circulating transcriptomic differences.

The studies were conducted in accordance with the policies of the New York University Langone Health Institutional Review Board and with the Helsinki Declaration. Informed consent was obtained from each subject. Patients with symptomatic PAD were recruited into the Platelet Activity in Vascular Surgery and Cardiovascular Events (PACE) study (ClinicalTrials.gov Identifier: NCT02106429). Patients with CAD were recruited into the cardiovascular prevention registry. Participants were recruited from New York University Langone Health, Bellevue Hospital, or the Veterans Affairs NY Harbor Healthcare System. PAD subjects included patients scheduled for lower-extremity revascularization with chronic limb-threatening ischemia (CLTI) or ankle–brachial index (ABI) < 0.6. Participants with CAD included those with a prior coronary stent, a prior coronary artery bypass, or a coronary angiogram showing > 50% stenosis. Patients with concomitant PAD and CAD based on the criteria above were excluded. Participants aged 40–65 years old were included in this analysis.

Peripheral whole-blood RNA was collected into PAXgene Blood RNA Tubes (Becton and Dickinson and Company, Franklin Lakes, NJ). RNA was isolated and analyzed using the Illumina HiSeq 4000 (Illumina, San Diego, CA). FASTQ files were processed as previously described. ⁴ Differential expression analysis was performed with DESeq2, and was adjusted for age, sex, race, and ethnicity. Differentially expressed transcripts were considered with a Benjamini–Hochberg–adjusted p-value less than 0.1. Gene set enrichment analysis was performed using a false discovery rate cutoff of 0.1 to identify relevant pathways. Cell-type fractional deconvolution of the bulk RNA-seq was performed using dtangle. ⁵

Sequencing data were analyzed from 20 subjects with PAD (mean age 59 years, 75% men, 35% White) and 51 subjects with CAD (mean age 52 years, 82% men, 78% White). In the PAD cohort, 17 subjects (85%) had CLTI; in the CAD cohort, 51 subjects (100%) had prior myocardial infarction. The PAD cohort was older and had a smaller proportion of Caucasian subjects compared with the CAD cohort. There were no differences between the groups in clinical variables including body mass index, hypertension, hyperlipidemia, smoking status, and diabetes. On complete blood count (CBC) analysis, there was no difference between PAD and CAD in the overall white blood cell count, neutrophil count, and monocyte count, but the CAD group had a higher lymphocyte count (1.9 vs 1.5 × 10⁹ cells/L, p = 0.04; Supplemental Table S1).

Using a significance cutoff of p adj. < 0.1, 106 genes were differentially expressed between PAD and CAD with 38 upregulated in PAD and 68 upregulated in CAD (Supplemental Figures S1A and S1B). Top enriched pathways in PAD included neutrophil degranulation (normalized enrichment score [NES] = 1.9, q = 6.8 × 10^–8), antimicrobial peptides (NES = 2.2, q = 0.001), and regulation of Toll-like receptor (TLR) by endogenous ligand (NES = 1.9, q = 0.07). Top enriched pathways in CAD included interferon-α/β signaling (NES = 3.0, q = 5.3 × 10^–8), interferon-γ signaling (NES = 2.2, q = 0.001), and antiviral mechanism by interferon-stimulated genes (NES = 2.0, q = 0.03) (Figure 1A). To test the influence of relevant comorbidities, we repeated pathway analyses after additional adjustment for diabetes. After adjustment, neutrophil degranulation pathways remained significantly enriched in PAD and interferon signaling pathways remained significantly enriched in CAD (Figure 1B).

OPEN IN VIEWER

Figure 1.

Circulating inflammatory pathway differences in PAD versus CAD. (A) Selected enriched Reactome pathways (q < 0.1) in PAD (n = 20) versus CAD (n = 51). Analyses were adjusted for age, sex, race, and ethnicity. (B) Top enriched pathways in PAD versus CAD after differential gene expression analysis with no adjustment, adjustment for demographic variables, and adjustment for demographic variables plus diabetes. Colors represent normalized enrichment scores. *p adj. < 0.1, **p adj. < 0.01, ***p adj. < 0.001. (C) Cell-type fractional abundance estimates generated using the dtangle cellular deconvolution method for selected immune cell types in PAD and CAD. Group comparisons were performed using the Mann–Whitney U test. (D) Neutrophil–lymphocyte ratio in PAD (n = 20) versus CAD (n = 44). *p < 0.05, using the Mann–Whitney U test.

Next, we estimated fractional abundances of immune cell subtypes between groups. Consistent with CBC analysis, patients with CAD had a higher fractional abundance of naive CD4⁺ T-cells (p = 0.0071) and CD8⁺ T-cells (p = 0.012) than patients with PAD (Figure 1C). To further validate the pathway analyses, we measured the neutrophil–lymphocyte ratio (NLR), which reflects the balance of innate and adaptive immunity, and found that patients with PAD had higher NLR than those with CAD (3.4 vs 2.3, p = 0.01; Figure 1D).

Interferon signaling and T cells are well-established contributors to atherosclerotic plaque development, with recent single-cell sequencing studies showing a high abundance of interferon-secreting T cells in human coronary plaque. ⁶ Our results suggest upregulation of these processes in CAD compared with PAD. In PAD, we found upregulation of neutrophil degranulation and related immune pathways, which have been shown to be associated with cardiovascular risk in PAD through effects on thrombus formation and vascular inflammation. ⁷ Our finding adds to a previous report of increased neutrophil degranulation in femoral compared with coronary circulation in patients with concomitant PAD and CAD. ⁸ Additionally, we found that NLR, a biomarker of inflammation associated with cardiovascular risk, ⁹ is increased in PAD compared with CAD.

There are several limitations to our study. First, not all subjects underwent diagnostic testing for both PAD and CAD. There were likely patients with PAD who had subclinical CAD and patients with CAD who had subclinical PAD. Nonetheless, our data provide novel insights into differences in the transcriptome of patients who present with clinical manifestations of isolated PAD or CAD. Second, a significant proportion of the PAD cohort had CLTI and thus represents a more severe phenotype of PAD, which may account for some of the differences observed between groups.

Overall, our data reveal differences in the circulating transcriptome of patients with isolated PAD or CAD that may underlie clinical and pathological differences between the disease states.

Supplemental Material

sj-docx-1-vmj-10.1177_1358863X251406528 – Supplemental material for Whole-blood transcriptomics differentiates circulating gene expression between coronary artery disease and peripheral artery disease

Supplemental material, sj-docx-1-vmj-10.1177_1358863X251406528 for Whole-blood transcriptomics differentiates circulating gene expression between coronary artery disease and peripheral artery disease by Richard Ni, Matthew Muller, Kelly V Ruggles, Tessa J Barrett and Jeffrey S Berger in Vascular Medicine