Machine learning based deconvolution of microarray atrial samples from atrial fibrillation patients reveals increased fractions of follicular CD4+ T lymphocytes and gamma-delta T cells

A potential relationship between T cell immunity and development of atrial fibrillation (AF) has been proposed. Historically in AF patients it has been reported that peripheral blood had elevated CD4+ T cells. However few studies have explored whether there is a direct increase of CD4+ T cells in atrial tissues with AF. In this study, public domain micro-array dataset of cardiac surgery patients with atrial tissue biopsies in AF and non-AF patients have been used to explore immune cell subsets. Machine learning based deconvolution of permanent atrial fibrillation microarray atrial samples was applied using Cibersort to enumerate the relative fractions of twenty-two different leukocyte subpopulations. Cibersort enumerated significantly increased fractions of follicular CD4+ T lymphocytes and gamma-delta T cells in the atria of permanent AF subjects. Gene expression analysis of permanent AF microarray tissue samples with elevated follicular CD4+ T cell fractions with gene pathways associated with myocardial substrate remodelling. That is both integrin and non-integrin mediated gene interactions between inflammatory cells and the extra cellular matrix, including infiltrating follicular CD4+ T cells that trafficked to the atria by virtue of the repertoire of cell surfaced expressed adhesion molecules. Additionally, IL-17 and other interleukin inflammatory gene heat maps were associated with enhanced CD4+ follicular T cell expression in our profiled atrial tissues with AF. These observations suggest that atrial structural remodelling was associated with the presence of pathogenic T cell mediated inflammation, present in AF atria but not in non-AF atria.