Background Atrial fibrillation (AF) is the most common cardiac arrhythmia, affecting up to 10% of the elderly. Ischemic stroke is the main complication of AF and cardioembolism is one of the leading causes of ischemic stroke, accounting for approximately one third of cases. Oral anticoagulant therapy (OAC) is a cornerstone in stroke prevention in patients with AF. According to randomized controlled trials of direct oral anticoagulants, a residual risk of ischemic stroke of 1-2% per year for so-called breakthrough stroke remains, despite adequate intake of OAC. The majority (\>70%) of these breakthrough strokes are cardioembolic in nature and only a minority are related to medication issues (e.g. non-compliance) or other, non-AF related etiologies. Stroke recurrence risk after such a breakthrough stroke markedly increases to 8-9% per year indicating a particularly high-risk situation. Why OAC fails in certain patients, but not in others remains as poorly understood, as does the reason why the subsequent risk of stroke is so high. Current risk stratification tools, such as the widely used CHA2DS2-VA(Sc)-score, fail to predict stroke risk in such a high-risk cohort, as they were intended to guide the initiation of OAC in low to moderate risk patients. In light of new therapeutic strategies currently being investigated, such as percutaneous left atrial appendage occlusion in patients with breakthrough strokes (ELAPSE - NCT05976685) or in AF-patients deemed high-risk (LAAOS IV - NCT05963698), improved risk stratification and characterization of high-risk AF patients is highly warranted. Several clinical factors, such as those reflected in the CHA2DS2-VA(Sc)-score, and especially a high AF-burden are associated with increased risk of cardioembolic stroke. Several cardiac serum biomarkers are thought to be surrogates not only of cardiac function, but also of cardioembolic risk. Reflecting ventricular and atrial wall tension, myocardial injury, oxidative stress and thrombogenicity, elevated NT-proBNP, MR-proANP, high-sensitive Troponin T and D-Dimers have all been associated with cardioembolic stroke in different AF and non-AF populations. As the main location of thrombus formation, the left atrium (LA) and more specifically its appendage (LAA) are of particular interest in the pathogenesis of cardioembolism. Pronounced LA-enlargement, compared to a normal-sized LA, correlates with an increased risk of cardioembolism in AF-patients. As over 80% of thrombi form within the LAA, several LAA-characteristics, such as slower LAA-flow velocity and larger LAA-orifice area have also been demonstrated to be associated with higher stroke risk. Although there is data on each one of these factors, they have only been investigated in low to moderate risk populations, such as AF-patients without prior stroke, OAC-naïve patients, or even within the general population as a whole. Their role in high-risk AF-patients and in breakthrough stroke is unknown. Hypothesis The investigators hypothesize that specific clinical factors, serum cardiac biomarkers and markers of LA- and LAA-morphology and function are associated with breakthrough stroke / OAC-failure and may improve risk stratification. Methods CARE-AF is a single-center, prospective cohort study conducted at the Stroke Center of the Inselspital, University Hospital Bern, Switzerland. Patients with an index ischemic stroke and AF (breakthrough and non-breakthrough cases) will be enrolled. The investigators will collect clinical data, serum cardiac biomarkers and echocardiographic indices of the LA and LAA. All patients will receive standardized annual follow-ups until the end of the study, defined as 12 months after the inclusion of the last participant. The primary endpoint is ischemic stroke or systemic embolism during follow-up. First, in a cross-sectional design, the study will assess the association between serum cardiac biomarkers and echocardiographic indices among patients with breakthrough vs. non-breakthrough stroke as index event, applying multivariate regression models. Second, the investigators will perform a longitudinal analysis assessing the association between the variables mentioned above and breakthrough stroke as index event with the primary endpoint, using multivariate Cox regression models. The study aims to enroll a minimum of 500 patients, which provides sufficient power to detect a clinically meaningful adjusted hazard ratio for recurrent stroke of 1.5 with 80% power at an alpha level of 5%. Conclusion The results of this project will enhance understanding of the role of specific clinical factors, cardiac serum biomarkers and echocardiographic indices in the residual risk of stroke in patients with AF on anticoagulation therapy. They may improve current risk stratification and have the potential to help guide therapeutic decisions in high-risk situations considering evolving therapeutic possibilities.