Electrocardiographic P-Wave Indices for Risk Stratification in Cardiovascular Diseases

Abstract

Atrial cardiomyopathy was examined in different cardiac conditions, including Brugada syndrome, mitral stenosis, mitral regurgitation, and heart failure. Systematic review and meta-analysis of published studies found that inter-atrial block (IAB; P-wave duration (PWD) > 120 ms by electrocardiography) significantly predicted new onset AF or recurrent AF after ablation. PWD, maximum P-wave area and P-wave terminal force in V1 (PTFV1) predicted incident ischemic stroke. Analysis of registry data found that Brugada patients had a higher proportion of IAB, mean PWD, maximum PWD and PTFV1 compared to age- and sex-matched controls. Mean PWD, PR interval and QT interval predicted incident AF. In mitral stenosis, age, systolic blood pressure and P-wave area (lead V3) predicted incident AF. A decision tree learning model with individual and nonlinear interaction variables with age achieved the best performance for outcome prediction. In mitral regurgitation, age, hypertension and mean PWD were significant predictors of new onset AF. Low left ventricular ejection fraction (LVEF) and abnormal PTFV1 predicted TIA/stroke. Age, smoking, hypertension, diabetes mellitus, hypercholesterolaemia, ischemic heart disease, urea, creatinine, neutrophil-to-lymphocyte ratio (NLR), prognostic nutritional index (PNI), left atrial diameter (LAD), LVEF, IAB and baseline AF predicted all-cause mortality. In heart failure, age, left atrial reservoir strain and contractile strain were significant predictors of new onset AF. Age and smoking predicted incident stroke. Age, hypertension, type 2 diabetes mellitus, chronic kidney disease, PTFV1, the presence of partial IAB, left atrial diameter, ejection fraction, global longitudinal left ventricular strain, serum creatinine and albumin, high NLR, low PNI, left atrial reservoir strain and contractile strain predicted all-cause mortality. Multi-task learning achieved a better performance for outcome prediction compared to logistic regression approach.