Incremental Predictive Value of Left Atrial Parameters Over Clinical Risk Scores for Subsequent Atrial Fibrillation

Function Beyond Size

Corresponding Author

• Download figure
• Open in new tab
• Download powerpoint

• atrial fibrillation
• echocardiography
• left atrium

Assessment of left atrial (LA) anatomy and function is useful for cardiovascular (CV) risk stratification, including assessing risk of atrial fibrillation (AF). Increased LA size mainly reflects elevated wall tension as a consequence of chronically increased LA pressure and/or volume overload, as well as impairment in LA function secondary to an atrial myopathy (1). Initially, LA size was assessed as a diameter (LAD) by M-mode echocardiography. However, the weight of evidence confirms the superiority of LA volume (LAV) to LAD (2), reflecting a nonlinear relationship between them (3). Age-related LA enlargement is a reflection of the pathophysiologic perturbations which often accompany advancing age rather than chronologic aging (4). This is corroborated by the finding that the effect of age on the indexed LAV is minimal among subjects free of CV disease (5). Consequently, the current guidelines recommend LAV over LAD for cardiac chamber quantification (6) and diastolic function assessment (7).

The vast majority of LAV assessment, however, has concerned maximum LAV (LAV_max) (2) with only a few reports addressing the significance of minimum LAV (LAV_min), which reflects LA function. In the early 1990s, Appleton et al. (8) reported that LAV_min had stronger associations with pulmonary wedge pressure, left ventricular (LV) pressure at pre-A wave, and LV end-diastolic pressure than LAD or LAV_max (8). Recently, LAV_min by 3-dimensional (3D) echocardiography has been shown to have incremental prognostic value over LAV_max in the prediction of major adverse CV events in patients with clinically indicated echocardiography (9) as well as in a community-based cohort (10). Nevertheless, little is known about the AF-specific predictive value of LAV_min in a community-dwelling population.

In this issue of iJACC, Olsena et al. (11) report the incremental prognostic value of LA emptying fraction (EF) and LAV_min over LAV_max in 1,951 community-dwelling people without AF at baseline. The population was from an echocardiographic subcohort of the 4th Copenhagen City Heart Study. During a median follow-up period of 11.0 years, 9.4% (n = 184) developed new AF, defined by an International Classification of Diseases and Related Health Problems, 10th revision (ICD-10) code (I48.9), which includes all types of AF (paroxysmal, persistent, permanent, and atrial flutter). Of the 3 2-dimensional methods for LAV, they used the area-length method, rather than the method of discs or the prolate-ellipsoid method (12). In this cohort, those who developed AF were older and had higher prevalence of hypertension than the remainder, and had greater LV mass and LA sizes (LAD, LAV_max, and LAV_min) with smaller LAEF. With regard to diastolic function, the AF group had smaller e’ and greater E/e’, although both groups shared a normal LV ejection fraction of 60%.

There are 6 major findings. 1) Both the CHADS₂ and CHARGE-AF risk stratification schemes were significant predictors of AF in the population (hazard ratio [HR]: >2). 2) All LA measures (LAV_max, LAV_min, and LAEF) were significant univariable predictors of AF, and were independent predictors of AF, incremental to clinical risk stratification — with the exception of LAEF, which was not incremental to CHARGE-AF. These findings were also confirmed in the competing risk analysis. 3) After adjusting for 18 clinical and echocardiographic variables in a multivariable Cox proportional hazards model, only the LA volumes remained significant predictors of the outcome. However, in a further competing risk analysis, only LAV_min remained significant. 4) An interaction analysis revealed that hypertension modified the relationship between LAV_min or LAEF and subsequent AF, which was not the case for LAV_max. The same interactions were confirmed in the competing risk analysis. 5) They further tested these relationships after stratifying the entire population into hypertensive and normotensive groups. Although they confirmed independent and incremental prognostic values of the above 3 LA measures over risk stratification schemes, all of them lost significance after adjusting for the 18 variables. 6) In a normotensive subgroup, all LA measures were independent predictors when adjusting for the CHADS₂ or CHARGE-AF risk scores, and all increased Harrell’s concordance (C) statistics in comparison to the CHADS₂, but not to the CHARGE-AF score. They concluded that functional measures of the LA, the LAV_min and LAEF, are independent predictors of incident AF in the general population, which was not the case for the LAV_max. This particularly applies to individuals without hypertension. Even in people without enlarged LA, these measures indicate an increased risk of AF, which goes undetected by the LAV_max.

The strengths of this study are that it is a large-scale community study that shows independent and incremental predictive value of the measures of interest over clinical risk stratification schemes; and it provides a sophisticated statistical analysis including competing risk models and appropriate evaluation of interaction. Now that AF is recognized as a common disease, further risk stratification with noninvasive imaging in a community-based cohort is crucial and could be readily applicable to clinical practice. Effect modification by hypertension was observed in this study, emphasizing the critical importance of hypertension in the pathogenesis of AF, as well as confirming that management of hypertension is key for AF prevention.

The authors should be congratulated on their completion of LAVs and EF analysis in addition to clinical risk assessments in more than 1,900 cases, with a sound statistical analysis. Nevertheless, the study has some limitations. The use of an ICD code as the endpoint permits the inclusion of only clinically diagnosed AF. Although the authors validated this approach by random sampling (13), subclinical AF could have been missed. The measurement of LAV index by 2D echocardiography rather than 3D may have introduced some inaccuracy in the assessment of LAV (9).

The work of Olsena et al. (11) emphasizes the importance of LA functional assessment in the prediction of incident AF. Recent advances in atrial imaging including 3D echocardiography (9) and strain (14) have also shown incremental predictive value over clinical risk stratification for the prediction of incident AF (15). It is possible that a comprehensive approach incorporating 3D and strain might better delineate patient risk. In addition, the LA is not a static chamber but constantly remodels anatomically and functionally (16,17). Because the effect of the LA remodeling on AF incidence remains unknown, further head-to-head comparisons to determine the best biological signal among these parameters would be intriguing and warranted.

• 2018 American College of Cardiology Foundation