Clinical evaluation and implications of left atrial remodeling in atrial fibrillation : From silent cerebral lesions and atrial stunning to novel electrocardiographic tools for prediction of arrhythmia outcome

Sammanfattning: Atrial fibrillation (AF) is the most common cardiac arrhythmia. Left atrial (LA) remodeling and reverse remodeling are associated with cerebral involvement and cognitive function (CF) changes. Risk stratification for AF related outcomes is essential in the management of patients with AF. This thesis aimed to 1) explore the effects of AF in a prospective cohort of anticoagulant-naïve patients, who underwent cardioversion (CV) within 48 hours after debut (Studies I and II) on i) occurrence of new silent thromboembolic events using brain magnetic resonance imaging, CF, cerebral biomarker ii) atrial remodeling and thrombogenicity using echocardiography, and hypercoagulability biomarkers; 2) identify novel electrocardiographic (ECG) predictors of 12-months AF recurrence, (Study III), in patients with non-permanent AF after CV or pulmonary vein isolation and study its effect on reverse atrial electrical remodeling (RAER) and 3) to evaluate traditional and novel ECG- and clinical predictors of new-onset AF (new-o-AF) on hospitalized Covid-19 patients (Study IV)  and explore the impact of AF on clinical outcomes.In Papers I and II, acute silent cerebral lesions could not be identified. A higher incidence of white matter hyperintensities was associated with higher CHA2DS2-VASc-score. A transient increase in cerebral damage biomarker was observed. Persistent AF patients had inferior CF test results. LA stunning resolved within ten days. The reverse functional remodeling was incomplete in patients with AF history. Higher levels of hypercoagulability-related biomarkers were observed prior to CV. In Paper III, the novel Peq-time>33ms, from P-wave onset to the peak positive deflection, independently predicted 12-months AF recurrence. The P-leftward-area, from peak positive deflection to the offset of P-wave, showed the largest change during follow-up, describing RAER. Machine-learning predictive model including variables from the novel P-wave partitioning showed the best predictive performance.In Paper IV, the novel Peq-time>33ms, PR-interval>190ms and P-wave-duration>115ms were independent predictors of n-o-AF. Admission to the intensive care unit (ICU), need for respiratory support, advanced age, males and increased body mass index (BMI) independently predicted new-o-AF. Logistic regression predictive models including age, sex, BMI, ICU admission and Peq-time or PR-interval had the best balanced accuracy.In conclusion, our findings in Studies I and II might suggest an enhanced thrombogenicity, even in patients with low stroke risk, supporting the concept of anticoagulation pericardioversion. We introduced the novel Peq-time, independently predicting AF recurrence in Study III and, along with PR-interval, new-o-AF in Study IV. Predictive models of arrhythmia outcome could be implemented in individually-tailored AF management and surveillance.