Authors: Smith A1, McLarnon T1, Peace A2, Watterson S1, Rai TS1

Affiliations: 1 Personalised Medicine Centre, School of Medicine, Ulster University, Derry/Londonderry, Northern Ireland, BT48 7JL. 2 Western Health and Social Care Trust, Altnagelvin Area Hospital, Glenshane Road, Derry/Londonderry, Northern Ireland, BT47 6SB.

Background/ Introduction: Atrial fibrillation (AF) is the most common cardiac arrhythmia globally. Abnormal firing of electrical impulses results in an irregular heartbeat in AF patients. The mechanisms underlying AF are not fully understood; therefore, treatment options are suboptimal and no cure currently exists. Senescence, defined as the permanent growth arrest of cells, has been implicated in literature to contribute to AF through inflammation and fibrosis; however, the exact role of senescence in AF remains unknown. Improved understanding of the biological mechanisms underlying AF subtypes may support personalised treatment in AF patients and ultimately improve patient outcomes. This research aimed to characterise the biological differences underlying AF subtypes through identification of proteins capable of differentiating paroxysmal from permanent AF patients, and to explore the role of senescence through identification of senescence-associated proteins.

Material & Methods: Proteomic data retrieved from a local cohort of 98 AF participants (49 paroxysmal, 49 permanent) was analysed. Baseline demographics identified age, weight and BMI to be statistically significant within the cohort. Spearman’s correlation analysis identified weight and BMI to be highly correlated (r=0.75). Two models were implemented in the pipeline, one adjusted for age and BMI and another for all three features. Differential expression analysis was completed to identify differentially expressed proteins (DEPs) between conditions. Top DEPs were then inputted into a machine learning pipeline consisting of model selection, iterated feature selection with nested hyper-parameter optimisation and stratified five-fold cross validation.

Results: Several cardiac-specific proteins were identified as capable differentiators of paroxysmal and permanent AF, with known links to senescence. External validation was completed using UK Biobank, with similar cardiac and senescent proteins identified.

Conclusion: This research identified senescent cardiac proteins that facilitate insight into key biological mechanisms that differ AF subtypes. These results provide a foundation for future research focused on identifying prognostic proteins associated with AF progression