Abstract
BACKGROUND: Despite advances in understanding atrial fibrillation (AF) pathophysiology, there is limited agreement on the key genes driving its pathophysiology. To understand the genome-wide transcriptomic landscape, we performed a meta-analysis from studies reporting gene expression patterns in atrial heart tissue from patients with AF and controls in sinus rhythm (SR). METHODS: Bibliographic databases and data repositories were systematically searched for studies reporting gene expression patterns in atrial heart auricle tissue from patients with AF and controls in sinus rhythm. We calculated the pooled differences in individual gene expression from fourteen studies comprising 534 samples (353 AF and 181 SR) to create a consensus signature (CS), from which we identified differentially regulated pathways, estimated transcription factor activity, and evaluated its performance in classifying validation samples as AF or SR. RESULTS: Despite heterogeneity in the top differentially expressed genes across studies, the AF-CS in both chambers were robust, showing a better performance in classifying AF status than individual study signatures. Functional analysis revealed commonality in the dysregulated cellular processes between chambers, including extracellular matrix remodeling (highlighting epithelial mesenchymal transition, actin filament organization, and actin binding hallmark pathways), cardiac conduction (including cardiac muscle cell action potential, gated channel activity, and cation channel activity pathways), metabolic derangements (highlighting oxidative phosphorylation and asparagine n linked glycosylation), and innate immune system activity (mainly neutrophil degranulation, and TNFα signaling pathways). Finally, the AF-CS showed a good performance differentiating AF from controls in three validation datasets (two from peripheral blood and one from left ventricle samples). CONCLUSIONS: Despite variability in individual studies, this meta-analysis elucidated conserved molecular pathways involved in AF pathophysiology across its phenotypes and the potential of a transcriptomic signature in identifying AF from peripheral blood samples. Our work highlights the value of integrating published transcriptomics data in AF and the need for better data deposition practices.