Study design

A total of 37 patients with chronic atrial fibrillation (CAF; 28 men and 9 women) were evaluated at baseline (within 4 h after cardioversion) and at 1 and 6 months after cardioversion in sinus rhythm. They were compared with a cohort of 37 healthy people (10 men and 27 women). None of the controls had a history of ischaemic heart disease or marked valvular abnormalities, peripheral vascular disease, cerebrovascular disease, hypertension or diabetes. None of the controls were receiving cardioactive drugs.

The 37 patients with CAF had atrial fibrillation for a mean duration of 35.8 weeks. All enrolled patients had sinus rhythm restored by electrical cardioversion and had not undergone catheter ablation or the Maze procedure. None of the enrolled patients had greater than mild to moderate valvular regurgitation. Left ventricular function was mildly impaired in the 7 patients who had a history of myocardial infarction, and 13 were hypertensive. At baseline and during the follow-up period of 6 months, 11 patients were taking digoxin, 9 were on sotalol and 11 were on amiodarone. The study was approved by the Committee for Human Research at Westmead Hospital (Sydney, New South Wales, Australia).

Standard transthoracic echocardiography

Doppler M-mode and two-dimensional echocardiography were carried out according to established clinical laboratory practice using commercially available equipment (Vivid 5, General Electric Medical Systems, Horten, Norway) equipped with 3.5 MHz variable-frequency harmonic-phased array transducers. The left atrial volume was calculated from the apical four-chamber and apical two-chamber zoomed views, using the biplane method of discs^10,11 to obtain left atrial end systolic volume (LAESV, the maximum left atrial volume in ventricular systole) and left atrial end diastolic volume (the minimum volume in ventricular diastole). Left atrial stroke volume was estimated as the difference between the LAESV and the left atrial end diastolic volume. The left atrial ejection fraction (LAEF) was calculated as (left atrial stroke volume/LAESV)×100.¹¹ The left ventricular ejection fraction was measured in all patients by the biplane method of discs from the apical two-chamber and four-chamber views.¹²

Mitral inflow velocity was obtained by pulsed-wave Doppler sampling at the tips of the mitral leaflets using a sample volume of 5 mm axial length from the apical four-chamber view at a sweep speed of 100 mm/s. The peak velocity of atrial contraction in late diastole (A-wave velocity) was measured.^13,14 The velocity time integral (VTI) of the mitral A wave was measured and the atrial fraction was estimated as A VTI divided by the total VTI of mitral inflow,^13,14 and an average of three beats was analysed.

Pulsed-wave Doppler tissue imaging

Peak velocity in late diastole secondary to atrial contraction (A′) was measured in all subjects using pulsed-wave Doppler at a sweep speed of 100 mm/s. The pulsed-wave Doppler sample volume was placed on the atrial side of the mitral annulus at the basal interatrial septum in the apical four-chamber view.^4,15 Special care was taken to align the Doppler beam to the interatrial septum to optimise Doppler measurements obtained during end expiration. An average of three beats was measured.

Atrial strain and strain rate imaging

Images were obtained using a narrow sector (frame rate >110 fps), and attempts were made to align the atrial wall parallel to the Doppler beam. Because of the thin atrial walls, a narrow (10×2 mm) sample volume was selected and placed in the basal septal, inferior, lateral and anterior walls of the atrium from the apical four-chamber and two-chamber views. The sample volume was placed in the middle of the respective basal segment and the image was tracked frame by frame, ensuring in each frame that the sample volume was moved to its original location in the middle of the segment using dedicated software available on an offline measuring station (EchoPac PC, GE-Vingmed, Horten, Norway). The peak strain rate in early diastole and that in late diastole (A-sr) corresponding to atrial contraction were measured (fig 1⇓). Gaussian smoothing (60) was applied before peak strain rate was measured. For each subject, measurements were obtained as an average of two consecutive cycles. If the angle of interrogation exceeded 30°, the patient was excluded from final analysis. The four basal atrial segments (septal, lateral, inferior and anterior) were analysed at baseline. However, in several subjects, the angle of interrogation for the basal anterior and basal lateral segments exceeded 30°. For the serial follow-up, A-sr was studied only in the basal septal and basal inferior walls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

 (A) Correlation of septal atrial strain rate (A-sr) to peak (Pk) A velocity (VEL). (B) Correlation of septal A-sr to A wave velocity time integral (VTI). (C) Correlation of septal A-sr to atrial fraction. (D) Correlation of septal A-sr to left atrial ejection fraction (LAEF). (E) Correlation of septal A-sr to A′ velocity (Doppler tissue imaging).

In addition to amplitude parameters, the timing parameter of the time from aortic valve closure (AVC) to tA-sr was measured. The AVC was obtained from Doppler tissue images as the point where the descent of the S velocity intersected the zero line. The tA-sr was measured as the duration from AVC to the peak A-sr.

Observer agreement

In 10 randomly selected studies from each cohort, two readers independently estimated the peak A′ velocity, the A-sr, the peak A velocity, and the A VTI and atrial fraction. One observer evaluated the same 10 studies on a separate occasion to determine the intraobserver variability.

Analysis

All values are expressed as mean (standard deviation (SD)) unless otherwise stated. Differences between groups were examined by two-sample Student’s t test. Repeated-measures analysis of variance was carried out to estimate within-patient changes for the group enrolled in long-term follow-up. Spearman’s rank correlation was used to examine the pairwise associations between A-sr and LAEF, peak A velocity, A wave VTI, atrial fraction and A′ velocity. Bland–Altman analysis¹⁶ was used to assess reproducibility within and between observers. Data were analysed using Statview Student V.4.0 and SPSS for Windows V.12.0.

Clinical and echocardiographic characteristics

Table 1A⇓ summarises the demographic and echocardiographic variables of the normal and CAF cohorts. As shown in table 1A⇓, a small yet significant difference in age was observed between the two groups, with the normal cohort being younger than the CAF cohort. Additionally, the body surface area (BSA) was also significantly higher in the CAF group. Traditional parameters of atrial function—namely, the A velocity, A wave VTI, atrial fraction and LAEF—were significantly higher in the controls than those in the atrial fibrillation group at baseline (table 1A⇓). Doppler-derived A-sr and peak positive atrial strain were also lower in the CAF group, and the tA-sr was prolonged.

As atrial size and function are thought to be altered with ageing and also with varying BSA, a separate statistical analysis was carried out to compare the CAF and control groups, adjusting for age and BSA as covariates. However, significant differences between the two groups in left atrial size (LAESV) and in parameters of atrial function (including LAEF, peak A velocity, A wave VTI, atrial fraction, A′ velocity and A-sr) remained even after correcting for the effects of age and BSA (table 1B⇑).

Correlation of A-sr to traditional parameters of atrial function

A-sr showed a moderate correlation to traditional parameters of atrial function, including the peak A velocity (Spearman’s r = 0.56; p = 0.001), the A wave VTI (Spearman’s r = 0.54; p = 0.001) and the atrial fraction (Spearman’s rho  =  0.53; p = 0.001; fig 1A⇑–C). The A-sr also correlated with the LAEF (Spearman’s r = 0.65; p = 0.001) and the A′ velocity obtained from Doppler tissue imaging (DTI) measurements (Spearman’s r = 0.63; p = 0.001; fig 1⇑ D,E).

A-sr in atrial fibrillation and controls

No significant differences were noted between the contractility of the basal septal and basal inferior segments in subjects in each group at baseline immediately after cardioversion (table 2⇓). In patients, velocities in the basal lateral and anterior walls were lower than those noted in the basal septal and inferior walls. A paired t test showed no significant difference between the basal septal and basal inferior segments (mean (SD) septal A-sr: −1.6 (0.72) v −1.8 (0.96) s⁻¹; p = 0.72) and between the basal lateral and basal anterior segments (−0.95 (0.64) v −0.9 (0.6) s⁻¹; p = 0.8). Both the basal lateral and basal anterior A-sr were significantly lower than the basal septal A-sr (p = 0.01 and 0.08, respectively). This may relate to the angle of interrogation in these segments. A significantly lower A-sr was noted in the CAF group than in the controls.

The tA-sr was significantly longer in the CAF group than in the normal cohort at baseline. A significant difference was noted in the tA-sr between the groups even after adjusting for heart rate (controls v CAF: mean (SD) 0.49 (0.19) v 0.54 (0.19) s). There was no significant correlation noted between left atrial size and the tA-sr.

Temporal improvement in atrial function with maintenance of sinus rhythm

An increase in A-sr was noted in the atrial fibrillation cohort when followed up temporally after cardioversion (table 3⇓). The maximum within-patient improvement in A-sr was noted in the first 4 weeks after cardioversion, similar to other parameters of atrial function. Interestingly, although the amplitude parameters showed a temporal improvement, timing parameters of tA-sr remained unaltered in the CAF cohort.

Observer variability

Ten subjects in each group were randomly selected for interobserver and intraobserver variability analysis. Peak strain rate in early diastole and A-sr were remeasured by the same observer and by a second independent observer from the digital data. Bland–Altman analysis for these parameters showed no evidence of any systematic difference between the two sets of measurements. Table 4⇓ shows the mean difference and confidence intervals of interobserver and intraobserver variabilities.

Atrial tissue velocity versus strain rate parameters

DTI can evaluate regional myocardial function.^3,17 We previously showed that DTI can be used to evaluate global and segmental atrial functions.⁴ However, a major limitation of DTI is that it cannot distinguish between intrinsic myocardial motion and that produced by passive translatory motion due to the effects of tethering. More recently, DTI-derived strain and strain rate imaging have been evaluated in ventricular myocardium.^5,6 Strain and strain rate imaging of the ventricular myocardium have been extensively reported in normal^18,19 and in diseased states.^20–,²² However, there is a paucity of data on strain and strain rate imaging of the atria.

The difference in atrial and ventricular emptying and filling occurs in a complementary fashion. Thus, in systole the ventricle contracts while the atrium fills, and conversely in diastole the ventricle fills while the atrium undergoes passive and active emptying. We would therefore assume that opposing directions of longitudinal contraction should be obtained from atrial and ventricular sampling. However, with DTI, concordant motion of the atrium to that of the ventricle is observed, presumably reflecting the inability of tissue Doppler imaging to distinguish atrial contraction from mitral annular and ventricular motion. By contrast, the longitudinal shortening and lengthening of the atrium are discordant with ventricular longitudinal motion, because the atrium fills during ventricular systole and empties during ventricular diastole (fig 2⇓). The recognition of this discordance is possible because A-sr, unlike DTI-derived A′ velocity, shows a site-specific directional difference.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

 (A) Atrial strain rate trace obtained with sample volume placed in the basal atrial septum. The vertical red line denotes aortic valve closure (AVC). The horizontal arrow denotes the time from AVC to peak atrial strain rate (tA-sr). (B) Ventricular strain rate trace obtained with sample volume placed in the basal ventricular septum.

A-sr as a marker of atrial function

Evidence shows atrial dysfunction and stunning occur immediately after cardioversion. Both invasive^23,24 and non-invasive^13,25,26 techniques have evaluated the recovery of atrial function, with restoration of sinus rhythm. These studies showed a dissociation between electrical and mechanical recovery after successful cardioversion to sinus rhythm. They also showed a temporal improvement in atrial function in the first 3–4 weeks after cardioversion.^13,26 Similar to traditional parameters, such as the peak A velocity and A VTI, A-sr shows reduced atrial contractility immediately after cardioversion. A significant temporal improvement was noted, as has been described previously. The main improvement in atrial function occurs within the first 4 weeks after sinus rhythm is restored.

The peak A velocity has previously been shown to normalise 4 weeks after cardioversion.^13,26 However, this study of A-sr shows a persisting degree of atrial dysfunction despite the maintenance of sinus rhythm for a 6-month period. This dysfunction could warrant the longer-term use of anti-arrhythmic treatment in the CAF cohort. Likewise, similar studies of the A-sr in paroxysmal atrial fibrillation may show differences in the A-sr compared with the CAF group.

Timing of atrial contraction

Delayed atrial contraction was also observed in the patients with CAF at baseline after cardioversion. However, with temporal follow-up over 6 months, no decrease was observed in the delayed atrial activation, suggesting that atrial myocardium is perhaps intrinsically altered in atrial fibrillation. Thus, although contractile function is restored in some measure, atrial conduction remains prolonged. This difference in atrial myocardium could explain the increased propensity for these patients to develop subsequent atrial fibrillation, and perhaps may represent an atrial myopathic process that persists despite the restoration of sinus rhythm. An alternative explanation is that the left atrial size is considerably larger in the CAF group and this may account for the prolongation in the tA-sr. However, in this preliminary cohort, no correlation was observed between atrial size and tA-sr.

Limitations

The reference standards used for comparison were traditional echocardiographic parameters. Invasive tests such as cardiac catheterisation to evaluate left atrial pressure were not considered feasible. Finally, the A-sr is Doppler derived and remains angle dependent, making measurements from the basal anterior and lateral walls difficult.

Conclusion

A-sr is a site-specific measure of intrinsic atrial contractility that is reduced in CAF after cardioversion to sinus rhythm. A temporal improvement is noted in the A-sr, with maintenance of sinus rhythm. However, atrial contraction is delayed in the CAF cohort and the tA-sr showed no marked improvement over time. Our data also show that A-sr occurs with differences in regional deformation to left ventricular strain, suggesting that this parameter is a more accurate measure of atrial contractility.