Left atrial (LA) remodeling and dysfunction reflect chronic exposure to elevated left ventricular (LV) filling pressures. The aim of this longitudinal cohort study was to define the effect of reducing LV filling pressures on reverse remodeling of LA volume (LAV) and function.
This retrospective cohort included 195 patients (52% men; mean age, 64 ± 14 years) in sinus rhythm with LA dilatation and sequential echocardiograms (median interval, 1 year; interquartile range, 0.5–2.0 years). One hundred seventy-four patients underwent medical therapy (82 with reduced E/e′ ratios), and 21 underwent surgery for valvular heart disease. Biplane LAV (normal value, ≤68 mL for men, ≤62 mL for women), LA strain (ε) (normal value, >32%) and LV filling pressures (assessed as E/e′ ratio; normal value, <13) were measured.
Although LAV at baseline and follow-up were 88 ± 27 and 81 ± 24 mL, LA ε and E/e′ ratio remained stable at 26 ± 11% and 14 ± 7, respectively. Changes in E/e′ ratio were associated with changes in LAV ( r = 0.37, P < .001) and LA ε ( r = −0.51 P < .001). Although reduced E/e′ ratio or improved LA ε at follow-up occurred in about 50% of the patients, only 26% (51 of 195) had normalized LAV. Compared with surgery, successful reduction of E/e′ with medical therapy was less effective in reducing LAV ( P < .001) but produced similar improvement in LA ε. Having normal or improved E/e′ ratio at follow-up was not associated with normalization of LAV (relative risk, 1.29 [ P = .326] and 1.22 [ P = .421], respectively) but was associated with normalized LA ε (relative risk, 2.04 [ P = .011] and 1.86 [ P = .017], respectively) independently of LAV.
Reduction in LV filling pressures reduces but rarely normalizes LAV. The strong association of reduced LV filling pressure with improved LA function indicated by LA longitudinal ε supports the increasing interest of LA ε measurement.
There is convincing evidence that left atrial (LA) enlargement, as determined by echocardiography, strongly and independently predicts many cardiovascular outcomes, including atrial fibrillation, stroke, heart failure, and mortality. Although LA size may be influenced by many factors, LA enlargement correlates well with the severity of diastolic dysfunction and usually reflects chronic exposure to elevated left ventricular (LV) filling pressure. LA remodeling is accompanied by deterioration in LA function, which may be important, as the left atrium maintains cardiac performance by modulating LV filling. Impaired LA function is therefore associated with many cardiovascular adverse events, including atrial fibrillation, stroke, and mortality.
Fortunately, reversal of LA remodeling is possible, particularly during the early stages of LA structural and functional alteration. Restoring sinus rhythm from atrial fibrillation and repairing the mitral valve among patients with severe mitral regurgitation have been reported to reduce LA size. The reduction in LV filling pressures with medical therapy may also lead to reverse structural and functional remodeling of the left atrium. In this study, we hypothesized that reduction in LV filling pressures by medical and surgical management would lead to reduced LA volume (LAV) and improved LA function. All parameters were assessed noninvasively, using LA strain (ε) to assess LA function and E/e′ ratio to estimate LV filling pressures.
This retrospective cohort study included 195 patients (52% men; mean age, 64 ± 13 years) with LA dilatation (>68 mL for men, >62 mL for women) who were admitted to the Royal Hobart Hospital (Australia) and underwent sequential transthoracic echocardiographic examinations as routine follow-up from 2005 through 2013. All subjects were >18 years of age, were in sinus rhythm, had no histories of heart transplantation at baseline, and had an interval of ≥4 weeks between echocardiographic studies. Those who had severe valvular heart disease at both baseline and follow-up without any surgical intervention were also excluded from the study. Of the total of 195 patients included in this study, 93 (48%) were diagnosed with heart failure with New York Heart Association functional classification I ( n = 23), II ( n = 29), III ( n = 33), or IV ( n = 8). Seven patients had moderate to severe mitral regurgitation, 11 had moderate to severe aortic stenosis, and three had moderate mitral regurgitation and moderate to severe aortic regurgitation or stenosis. All these patients underwent surgery (not including a maze procedure) at some time within the follow-up period of this study. Because E/e′ ratio may not be a reliable measurement of LV filling pressure in patients with mitral valve disease or surgically repaired mitral valves, surgical patients were excluded from the main analyses of this study. The results of comparing LA structure and function between these surgical patients and those who underwent medical management are reported in the supplemental information. The median follow-up period was 1 year (interquartile range, 0.5–2 years). The study was approved by the Tasmanian Human Research Ethics Committee.
All echocardiographic studies were performed in a single echocardiographic laboratory by experienced clinical sonographers. Two-dimensional and Doppler measurements were obtained using standard techniques and procedures following the American Society of Echocardiography guideline. End-systolic LAV was measured using the biplane area-length method. Normal LAV was defined as ≤68 mL (men) or ≤62 mL (women), as recommended in the guidelines. These cutoffs were used to distinguish between “mildly abnormal” and “moderately abnormal” in the 2005 guidelines but align with the new cutoff to define normal LAV in the 2015 updated guidelines. Alternative cutoffs for normalization of LAV were defined by a 15% reduction in LAV from baseline, as previously suggested.
Peak LA longitudinal ε was measured using two-dimensional speckle-tracking software (Research Arena; TomTec Medical Imaging, Unterschleissheim, Germany) by a single rater who was blinded to clinical information, following standard methodologies for speckle-tracking reported in the guidelines. After the observer manually traced the LA endocardial surface in the four-chamber view using a point-and-click approach, the software automatically tracked the LA endocardial borders throughout the cardiac cycle. We took the onset of the QRS complex on the electrocardiogram as a reference point and measured positive peak LA longitudinal ε. The region of interest was divided into six segments. Peak LA longitudinal ε was obtained by averaging the peak values of all segments. The normal range of LA ε was defined as >32%, as previously recommended. LA contractile function was assessed by averaging the increment in ε after the onset of the P wave. LA conduit function was calculated as the difference between global LA ε and LA contractile function.
Pulsed-wave Doppler derived mitral inflow early (E) and late (A) velocities and medial and lateral tissue Doppler derived early (e′) velocities were obtained, and the mean E/e′ ratio was used for analysis. Normal LV filling pressure was defined as an E/e′ ratio < 13. LV ejection fraction (LVEF) was calculated using the biplane modified Simpson method. An index of LA wall stiffness was calculated as the ratio of E/e′ to LA ε, as previously suggested.
Ten patients were randomly chosen for 20 repeated measurements (10 at baseline and 10 at follow-up) of LA ε and E/e′ ratio to calculate intrarater variability. These measurements were repeated again by another expert to calculate interrater variability. These measurements were conducted ≥4 weeks apart. The raters were blinded to patients’ clinical information and the original measurements and were allowed to pick the best cardiac cycle each time of remeasurement.
Other Clinical Factors
Body weight, blood pressure, and medication use were obtained from medical records at both baseline and follow-up.
Paired t tests were used to compare data at baseline and follow-up, and standard t tests were used to compare the mean values of clinical and echocardiographic parameters between groups. Log-binomial regression was used to calculate relative risks. Pearson correlation and linear regression were used to estimate the strengths and effect sizes of linear relationships. Changes in E/e′ ratio (ΔE/e’), LAV (ΔLAV), and LA ε (Δε) were calculated by taking the differences between baseline and follow-up values. Intraclass correlation and absolute mean and SD of the differences between repeated measurements were used to quantify intra- and interrater variability of LA ε and E/e′ ratio. Stata versuin 12.0 (StataCorp LP, College Station, TX) was used for analyses.
Clinical characteristics at baseline and follow-up are shown in Table 1 . Most patients had preserved LVEFs and elevated LV filling pressures at both baseline and follow-up. Compared with baseline, patients had increased body weight and LVEFs and reduced LAVs but maintained their blood pressure, E/A ratios, E/e′ ratios, LA ε, and medication use at follow-up. Two patients (1%) developed paroxysmal atrial fibrillation during follow-up.
|Weight (kg)||84.8 ± 22.5||86.8 ± 24.3||.028|
|Systolic pressure (mm Hg)||140.1 ± 18.7||138.5 ± 17.1||.723|
|Diastolic pressure (mm Hg)||81.3 ± 11.1||81.8 ± 13.5||.684|
|LVEF (%)||52 ± 12||54 ± 11||.004|
|E/A ratio||1.3 ± 0.8||1.3 ± 0.8||.594|
|E/e′ ratio||13.5 ± 6.6||13.8 ± 6.8||.613|
|E (m/sec)||0.87 ± 0.26||0.85 ± 0.29||.397|
|e′ (m/sec)||0.07 ± 0.02||0.07 ± 0.03||.882|
|LAV (mL)||88.2 ± 27.3||81.6 ± 24.0||.001|
|Total LA ε (%)||26.4 ± 11.1||26.3 ± 11.0||.466|
|Contraction proportion (%)||44.7 ± 20.0||44.5 ± 19.2||.619|
|Conduit proportion (%)||55.3 ± 20.2||55.5 ± 19.2||.424|
|LA wall stiffness||0.70 ± 0.66||0.77 ± 0.93||.234|
|β-blocker use||40% (78)||42% (82)||.564|
|ACE inhibitor/ARB use||72% (140)||75% (146)||.231|
|Diuretic use||60% (117)||65% (127)||.100|
|Aldosterone use||25% (49)||25% (49)||.889|
|Statin use||35% (68)||34% (67)||.870|
The intrarater intraclass correlations were 0.89 (95% CI, 0.76–0.97), 0.92 (95% CI, 0.80–0.98), and 0.83 (95% CI, 0.73–0.96) for global LA ε, conduit function, and contractile function, respectively. The interrater intraclass correlations were 0.86 (95% CI, 0.68–0.94), 0.84 (95% CI, 0.64–0.93), and 0.76 (95% CI, 0.47–0.90) for global LA ε, conduit function, and contractile function, respectively. The intra- and interrater intraclass correlations for E/e′ were 0.98 (95% CI, 0.95–0.99) and 0.96 (95% CI, 0.90–0.98), respectively. The interrater absolute difference was −0.94 ± 2.23% (LA ε) and −0.05 ± 1.55 (E/e′ ratio).
Approximately 50% of the patients had improved E/e′ ratios (92 of 174) or LA ε (97 of 174) during the follow-up period, but only 25% (47 of 174) showed reverse remodeling resulting in normalization of LAV. The changes in E/e′ ratio were due to changes in both E and e′. Among patients with ΔE/e′ < 0, their E reduced from 0.92 ± 0.24 m/sec at baseline to 0.79 ± 0.28 m/sec at follow-up ( P < .001), and their e′ increased from 0.07 ± 0.03 m/sec at baseline to 0.09 ± 0.04 m/sec at follow-up ( P < .001). Similarly, among those with ΔE/e′ ≥ 0, E increased from 0.80 ± 0.24 m/sec at baseline to 0.87 ± 0.27 m/sec at follow-up ( P = .007) and their e′ decreased from 0.07 ± 0.02 m/sec at baseline to 0.06 ± 0.02 m/sec at follow-up ( P = .016).
Compared with patients in whom the left atrium remained enlarged ( n = 141), those whose LAVs normalized ( n = 47) included a similar proportion undergoing surgery (10% vs 11%, P = .87) and had similar length of follow-up (median, 1.0 vs 1.1 years; P = .701) but had lower LAVs at baseline (77 vs 92 mL, P < .001) and greater magnitude of changes in E/e′ (−3.5 vs 0.5, P = .04).
Baseline LAV was negatively associated with changes in LAV ( r = −0.43 P < .001). That is, patients with very large LAVs to begin with were likely to reduce greater absolute values of LAV. Enlargement in LAV was moderately associated with reduction in LA ε; this negative association ( r = −0.30, P < .001) between changes in LA structure (LAV) and function (ε) is shown in Figure 1 . Similarly, the moderate associations of contrary directions between changes in LV filling pressures (ΔE/e′) and LAV ( r = 0.33 P < .001) and LA ε ( r = −0.52, P < .001) are shown in Figure 2 . Changes in filling pressures were moderately associated with both the conduit ( r = −0.48, P < .001) and the contractile ( r = −0.35, P < .001) function of the left atrium and strongly associated with changes in LA wall stiffness ( r = 0.71, P < .001). These results were adjusted for age, sex, length of follow-up, body weight at baseline, changes in body weight, blood pressure, and respective outcome (LAV or LA ε) at baseline.
Changes of Filling Pressure and LA Changes
Table 2 shows the differences in echocardiographic parameters between patients who showed reduced LV filling pressure (ΔE/e′ < 0) and those with stable or increased LV filling pressures (ΔE/e′ ≥ 0) during follow-up. Patients whose LAVs normalized at follow-up had lower LAVs at baseline and had greater ΔLAV and ΔE/e′ than those whose LAVs remained enlarged. However, 15% (21 of 141) of the patients in whom LAV remained enlarged showed reductions of LAV (by −23 mL) and E/e′ (by −3.8), to a similar degree as those with normalized LAVs. Although patients with ΔE/e′ < 0 had reduced LAVs and improved LA ε and LVEF at follow-up, those with ΔE/e′ ≥ 0 maintained similar LVEFs and LAVs from baseline but demonstrated impairment of LA ε.
|Variable||ΔE/e′ ≥ 0 ( n = 90)||ΔE/e′ < 0 ( n = 84)||P|
|Change in LAV (mL)||0.5 ± 26.4||−9.0 ± 21.8||.002|
|Change in LA ε (%)||−3.9 ± 9.5||5.5 ± 11.7||<.001|
|Change in LVEF (%)||0 ± 6||4 ± 9||.006|
Table 3 summarizes the changes in E/e′ on the basis of the changes in LAV and LA ε between baseline and follow-up. Although there were only minimal changes in LV filling pressures among patients who maintained similar LA ε from baseline to follow-up, LV filling pressures increased among patients whose LA ε became abnormal and decreased among patients whose LA ε normalized. These results were regardless of LAV and independent of whether LAV normalized or remained enlarged and were consistent with the stronger association of ΔE/e′ with Δε than with ΔLAV, as shown in Figure 2 .
|Change||LAV remained enlarged||LAV normalized|
|LA ε remained abnormal ( n = 89)||71||1.0 ± 4.5||18||0.1 ± 3.1|
|LA ε became abnormal ( n = 21)||18||6.9 ± 9.7||3||5.4 ± 4.0|
|LA ε normalized ( n = 32)||18||−4.3 ± 6.8||14||−5.5 ± 5.9|
|LA ε remained normal ( n = 32)||21||−0.2 ± 3.0||11||0.2 ± 2.1|