Results
Overall Study Population
The ENGAGE AF-TIMI 48 echocardiographic substudy enrolled 1120 subjects, of whom 971 (87%) had technically adequate images. These results were obtained prior to the unblinding of subject's study drug allocation. When compared with the 971 subjects included in the echocardiographic analysis, the 149 subjects not included were slightly younger (median 71 vs. 73 years, P = 0.020), had comparable CHADS2 scores (median 3 in both groups), and the proportions with persistent or permanent AF were similar. When compared with the other 20 134 subjects in the parent ENGAGE AF-TIMI 48 trial, the 971 subjects included in the final echocardiographic analysis were older (median 73 vs. 72 years, P < 0.001), but were less likely to be female (34 vs. 38%, P = 0.015), and more frequently Caucasian (92 vs. 80%, P < 0.001). Paroxysmal AF was more common in the echocardiographic cohort (33 vs. 25%, P < 0.001); while permanent AF was less frequent (46 vs. 52%, P = 0.001). However, the CHADS2 score was similar between the echocardiographic substudy and parent trial.
In the echocardiographic substudy, LA enlargement (defined as LAVI ≥29 mL/m) was present in nearly two-thirds of all subjects and there was a significant inverse relationship between LA function, measured by LAEF, and LAVI according to type of AF (Figure 1A). This relationship was similar when stratified by rhythm at the time echocardiography as well (Figure 1B). However, even among AF subjects with normal LAVI, approximately half had impaired LA EF.
(Enlarge Image)
Figure 1.
The relationship between left atrial emptying fraction and left atrial volume index. (A) Left atrial emptying fraction was inversely correlated with left atrial size. Left atrial size was normal (LAVI <29 mL/m) in 36% of all atrial fibrillation patients and 17% had both normal left atrial size and left atrial emptying fraction (>45%). (B) The inverse relationship between left atrial emptying fraction and left atrial size was similar between subjects in sinus rhythm (n = 321) and atrial fibrillation (n = 650) at the time of echocardiography.
Paroxysmal, Persistent, or Permanent Atrial Fibrillation
Paroxysmal, persistent, and permanent AF were present in 33, 21, and 46% of subjects, respectively (Table 1). Approximately 25% of those with permanent AF had a CHADS2 score of 4–6 compared with 18–19% of those with paroxysmal or persistent AF (P = 0.09).
Left atrial size, as determined by LAVIs, increased across AF types (Figure 2A) with 48, 60, and 77% of paroxysmal, persistent, and permanent AF having enlarged LA, respectively (Table 2). Left atrial function, assessed by EF, significantly declined across the spectrum of AF types (Figure 2B). There was a corresponding decrease in LA compliance or reservoir function as assessed by the expansion index across AF types (Figure 2C). Left atrial conduit function, as determined by transmitral E-wave velocities, increased according to type of AF (Table 2). In the subset of subjects who were not in AF at the time of echocardiography, LA contractile function, determined by transmitral spectral Doppler A-wave velocities and mitral annular tissue Doppler a′ velocities, declined with progression from paroxysmal to persistent to permanent AF (Figure 2D).
(Enlarge Image)
Figure 2.
Left atrial size and function according to type of atrial fibrillation. (A) Left atrial size measured by the volume index significantly increased. (B) Left atrial emptying fraction significantly declined. (C) The left atrial expansion index, a measure of reservoir function and left atrial compliance, significantly declined. (D) Among those in sinus rhythm at the time of echocardiography, left atrial contractile function, assessed by tissue Doppler imaging A′ velocities, was lowest in those with permanent atrial fibrillation.
Left ventricular and ejection fraction was preserved in most subjects (78%) regardless of AF type, although the prevalence of reduced LVEF (<50%) was twice as high in permanent when compared with paroxysmal AF. There was a corresponding trend towards a larger LV size and a higher LVMI in those with permanent AF, although the prevalence of abnormal LV geometry, defined as hypertrophy or concentric remodelling, did not differ. Left ventricular filling pressures (E/e′) were also similar across groups (Table 2).
CHADS2 Score
CHADS2 scores of 2, 3, and 4–6 were present in 49, 29, and 22% of subjects, respectively (Table 3). The frequency of reduced LVEF (<50%) was similar regardless of the CHADS2 score (Table 4). Abnormal LV geometry increased in frequency with higher CHADS2 scores. There was an inverse relationship between LV early diastolic relaxation (e′) velocities and CHADS2 score with a corresponding increase in LV filling pressures (E/e′) across CHADS2 categories. Similarly, LA size increased with higher CHADS2 scores, with a parallel decline in LA EF. In forward stepwise multivariate-ordered logistic regression analyses, higher LV filling pressures (E/e′), larger LAVI, and the presence of abnormal LV geometry were significantly associated with higher CHADS2 score as a measure of stroke risk (Table 5). While both the burden of AF and the CHADS2 score were related to LA size and function (Figure 3A and B), the electrical burden of AF appeared to be a stronger contributor than the CHADS2 score. Lastly, as the CHA2DS2-VASc score is a validated method for assessing stroke risk in AF, we also evaluated LA structure and function according to this system and found similar results as with CHADS2, namely that LAVI increased, LAEF decreased, and LA TDI A′ velocities declined with a greater CHA2DS2-VASc score (Supplementary material online, Table S1).
(Enlarge Image)
Figure 3.
Relationship between left atrial size and function according to type of atrial fibrillation and stroke risk expressed in the CHADS2 score. (A) Increasing left atrial size, measured by left atrial volume index, appears more strongly related to type of atrial fibrillation (P < 0.001) than to stroke risk expressed in the CHADS2 score (P = 0.007). (B) Worsening left atrial emptying fraction appears more strongly related to type of atrial fibrillation (P < 0.001), than to the CHADS2 score (P = 0.041).