Abstract
Background and aims
Post-operative atrial fibrillation (POAF) is one of the most common complications following open heart surgery, and correlates with increased morbidity, mortality, and financial burden. Previous research has examined possible underlying mechanisms for this common clinical problem, and developed models to predict POAF. With the introduction of the concept of atrial cardiomyopathy and advances in cardiac deformation imaging, a more comprehensive approach was made feasible. In this thesis, we had the overarching aim of predicting POAF using a model that combined histopathological findings in atrial tissue along with strain deformation parameters of both left and right atria, in addition to previously studied clinical and demographic variables.
Methods
Right atrial appendage tissue samples were collected during clinically indicated coronary artery bypass graft and/or valve surgeries. Tissue analysis was completed using a dedicated automated tissue analysis machine learning algorithm developed as part of this project. Echocardiographic parameters were measured including speckle-tracking deformation parameters. Characterising the correlates of tissue variables and imaging variables in our cohort was performed using regression analysis. Models to explain the incidence of POAF were developed using logistic regression.
Results
The study included a total cohort of 294 patients; 250 were included in the final analysis for POAF incidence. 178 patients had tissue staining and analysis. The mean age for the whole cohort was 67.6 (SD 10.5) years, 77% were males. Patients’ age, hypertension, higher degrees of mitral regurgitation, and left ventricular global strain correlated with left atrial global strain (LA GLS) (B -0.2%, P <0.05; B 4.6, P <0.05; B -11.2, P < 0.05; B -0.5, P <0.05, respectively), while patients’ age, right ventricular function and higher degree of tricuspid regurgitation correlated with right atrial global strain (B -0.5, P < 0.05; B -9.8, P < 0.05; B -12.7, P < 0.05, respectively); other clinical and demographic parameters did not show significant correlation with atrial deformation parameters. Tissue variables did not significantly correlate with right atrial deformation. Patients’ age (OR 1.05 (95%CI 1.01, 1.09), P < 0.05), and LA GLS (OR 0.94 (95%CI 0.90, 0.98), P <0.01) but not left ventricular ejection fraction, bypass time, or tissue fibrosis, were predictive of incident POAF.
Conclusions
Underlying histological changes did not significantly correlate with either atrial deformation parameters or the incidence of POAF. Both right and left atrial strain become more impaired with age, and atrial function correlates with the associated ventricular function. Only patients’ age and left atrial strain were predictors of POAF in our cohort; left atrial strain remained significant even after adjusting for age. These findings highlight the potential benefit of LA GLS in addition to patient’s age in predicting POAF. These findings will require further validation prospectively.