METHODS

Patients

One hundred and twelve patients (75% men) who had been admitted to hospital for clinical signs and symptoms of HF and had impaired systolic function by echocardiography (ejection fraction < 50%) were recruited. They were stabilised by anti-HF medications and underwent elective tissue Doppler imaging 4–6 weeks later. Of these, 71 had coronary artery disease, 32 had dilated cardiomyopathy, 6 had hypertensive heart disease, 2 had alcoholic cardiomyopathy, and 1 had aortic valve heart disease. Patients with coronary artery disease had either previous evidence of myocardial infarction (n = 52) diagnosed by standard criteria or angiographic evidence of significant disease with or without previous percutaneous coronary intervention (n = 19). Patients with atrial fibrillation were excluded from the study. Among the participating patients, QRS complexes were narrow (> 120 ms; narrow QRS group) in 67 (60%) and prolonged (< 120 ms; wide QRS group) in 45 (40%) (table 1). The results of echocardiography were compared with those of 88 normal healthy volunteers (62% men, χ² = 1.31, NS v HF group). The volunteers had no history of cardiovascular or systemic diseases and had normal physical examination, and electrocardiographic and echocardiographic findings.

Echocardiography

Standard echocardiography with Doppler studies were performed (System 5, Vingmed-General Electric, Horten, Norway). LV dimension and ejection fraction were measured by two dimensional guided M mode echocardiography according to the guidelines of the American Society of Echocardiography.¹³ Tissue Doppler imaging was performed in the apical views (four chamber, two chamber, and long axis) for the long axis motion of the LV as previously described.^12,14 In brief, two dimensional echocardiography with tissue Doppler colour imaging was performed with a 2.5 or 3.5 MHz phase array transducer. The system was set by bypassing the high pass filter, while the low frequency Doppler shifts were input directly into an autocorrelator.¹⁵ Gain settings, filters, and pulse repetitive frequency were adjusted to optimise colour saturation, and a colour Doppler frame scanning rate of 100–140 Hz was used. At least three consecutive beats were stored and the images were digitised and analysed off line by a computer (EchoPac 6.3, Vingmed-General Electric). Myocardial regional velocity curves were constructed from the digitised images.¹⁶ For detail assessment of regional myocardial function, the sampling window was placed at the myocardial segment of interest. In each view, both the basal and mid segments were assessed. In this way, the following segments were interrogated: septal, anteroseptal, anterior, lateral, inferior, and posterior segments at both basal and middle levels. For the measurement of timing, the beginning of the QRS complex was used as the reference point, where the time to peak myocardial sustained systolic (T_S) and early diastolic velocities (T_E) was quantified.⁹ For the assessment of synchronicity, the standard deviation of T_S (T_S-SD) and T_E (T_E-SD) of all 12 LV segments and the maximal difference in T_S and T_E between any two of the LV segments were calculated. To assess global cardiac function, the mean myocardial sustained systolic (mean S_M) and early diastolic (mean E_M) velocities from the six basal segments were calculated.¹⁴

Statistical analysis

Data were analysed using a statistical software program (SPSS for Windows, version 10.0.7, SPSS Inc, Chicago, Illinois, USA). For comparison of parametric variables between the three groups, analysis of covariance was used to examine the effect of age and heart rate on the dependent variables, followed by one way analysis of variance with the Scheffé correction for significance. Linear regression analysis was performed to investigate the correlation between parametric variables. Categorical data between two or more groups were compared by the Pearson χ² test. Stepwise multiple regression analysis was performed to assess potential independent covariates on systolic and diastolic asynchrony. The results are expressed as mean (SD). A probability value of p < 0.05 was considered to be significant.

RESULTS

There was no difference in age between the patient group as a whole and normal controls (66 (12) v 64 (10) years, NS). The heart rate was observed to be faster in the patient group than in controls (73 (15) v 65 (11) beats/min, p < 0.001). However, age and heart rate were not confounders of systolic and diastolic asynchrony according to analysis of covariance, and hence unadjusted p values are presented. Table 1 shows the comparison of baseline clinical characteristics between the narrow and wide QRS groups. The age, sex, and aetiology of HF were not different between the two groups. The prescription of medications was also similar except for a slightly greater use of diuretics in the wide QRS group. In the wide QRS group, 23 had a left bundle branch block pattern, 10 had a right bundle branch block pattern, and 12 had intraventricular conduction delay. The mean LV ejection fraction was lower in the patient group than in controls (37.6 (9.6)% v 78.7 (5.7)%, p < 0.001). The wide QRS group had significantly lower LV ejection fraction (33.2 (10.8)% v 40.6 (7.2)%, p < 0.001) and larger LV end diastolic (6.7 (1.1) v 6.1 (0.9) cm, p = 0.001) and end systolic (5.9 (1.1) v 5.1 (0.9) cm, p < 0.001) diameters than the narrow QRS group.

Systolic asynchrony

The systolic synchronicity was impaired in patients with HF. The maximal difference in T_S was prolonged in the narrow QRS group compared with normal controls (p < 0.001) and was the longest in the wide QRS group (table 2, fig 1). Similarly, the T_S-SD was significantly prolonged in the narrow QRS group (p < 0.001), though it was further increased in the wide QRS group (table 2, fig 2A). When a maximal difference of T_S of > 100 ms was used to define significant systolic asynchrony, it was not found in the control group but was present in 34 (51%) patients in the narrow QRS group and in 33 (73%) in the wide QRS group (χ² = 83.2, p < 0.001). When a T_S-SD of > 32.6 ms (+ 2 SD of normal controls) was used to define significant systolic asynchrony, it was present in only 3 (3%) control subjects but in 29 (43%) patients in the narrow QRS group and in 29 (64%) in the wide QRS group (χ² = 58.3, p < 0.001). In addition, systolic asynchrony was more prevalent in patients with wide than in those with narrow QRS complexes by both maximal difference in T_S (χ² = 5.7, p = 0.02) and in T_S-SD (χ² = 4.8, p = 0.03).

Table 2 shows the regional T_S in individual LV segments. It was significantly delayed in all 12 LV segments in both narrow and wide QRS groups compared with normal controls. In addition, it was further prolonged in some of the LV segments in the wide QRS group compared with the narrow QRS group.

Diastolic asynchrony

Similar to systole, diastolic asynchrony was also evident in patients with HF. The maximal difference in T_E was prolonged in the narrow QRS group (p < 0.001) and was the longest in the wide QRS group (table 3, fig 1). In addition, T_E-SD was significantly longer in the narrow QRS group than in the normal controls (p < 0.001) and was further prolonged in the wide QRS group (table 3, fig 2B). When a maximal difference of T_E of > 100 ms was used to define significant diastolic asynchrony, it was not found in the normal controls but was present in 31 patients (46%) in the narrow QRS group and 31 (69%) in the wide QRS group (χ² = 71.5, p < 0.001). When a T_E-SD of > 30.1 ms (+ 2 SD of normal controls) was used to define significant diastolic asynchrony, it was present in only 2 (2%) of control subjects but in 29 (43%) patients in the narrow QRS group and in 33 (73%) in the wide QRS group (χ² = 73.0, p < 0.001). In addition, diastolic asynchrony was significantly more prevalent in the wide than in the narrow QRS group by both maximal difference in T_E (χ² = 5.6, p = 0.02) and in T_E-SD (χ² = 9.8, p = 0.002). There was only a modest correlation between systolic and diastolic synchronicity as shown by T_S-SD and T_E-SD (r = 0.40, p < 0.001). In addition, in patients with narrow QRS complexes, the prevalence of merely abnormal T_S-SD was 27%, of merely abnormal T_E-SD was 27%, and of abnormal T_S-SD and T_E-SD was 16%, which was different from those with wide QRS complexes with a prevalence of 16%, 24%, and 49%, respectively (χ² = 15.5, p = 0.001).

Table 3 shows the regional T_E in individual LV segments. Unlike T_S, T_E was delayed in only 3 of 12 LV segments in the narrow QRS group and in 5 of 12 LV segments in the wide QRS group.

Predictors of systolic and diastolic asynchrony

Clinical and echocardiographic predictors of systolic asynchrony were sought. In a univariate model there was no relation between maximal difference in T_S or T_S-SD and the duration of the QRS complex. However, low LV ejection fraction, mean S_M, and mean E_M, as well as an enlarged LV significantly correlated with more severe systolic asynchrony (all p < 0.001) (table 4). In the stepwise multiple regression model, only a low mean S_M (p = 0.02) and a large LV end systolic diameter (p = 0.009) predicted the severity of systolic asynchrony (table 5).

For diastolic asynchrony, univariate analysis found that all the tested parameters significantly correlated with diastolic asynchrony. However, the stepwise multiple regression model found that only a low mean E_M (p < 0.001) and prolonged QRS complex duration (p < 0.001) were independent predictors of diastolic asynchrony (table 5).

DISCUSSION

The present study illustrates the changes in systolic and diastolic synchronicity in patients with HF. Irrespective of QRS duration, patients with HF could develop mechanical asynchrony in both systole and diastole. Although the condition was more prevalent in the wide QRS group, it was not uncommon in patients with narrow QRS complexes. Among various clinical and echocardiographic predictors of asynchrony, it was observed that poor systolic function and enlarged LV predicted systolic asynchrony, while poor diastolic function and prolonged QRS complex duration predicted diastolic asynchrony.

Systolic asynchrony in patients with HF

Systolic asynchrony is characteristic of patients with HF who have a wide QRS complex, which signifies electromechanical delay.^4,9,17 In this highly selected group of patients, biventricular pacing has been shown to alleviate symptoms effectively, improve cardiac function,¹ and reverse LV remodelling.^2,3,9,18 Despite stringent patient selection, some did not respond to biventricular pacing, even though they had wide QRS complexes.^4,5,18 Furthermore, neither baseline^6–8 nor shortening of QRS complex duration⁴ were consistent predictors of haemodynamic, clinical, or echocardiographic improvement after biventricular pacing. On the other hand, the mechanism of benefit of biventricular pacing is attributed to the improvement of intraventricular systolic synchrony, which was shown recently by means of tissue Doppler imaging, where the T_S-SD was shortened and the intersegmental difference in T_S was minimised after biventricular pacing.^7,9,10 Also, in patients undergoing biventricular pacing, systolic asynchrony, but not QRS complex duration, proved to be a better predictor of acute haemodynamic,⁴ echocardiographic,⁸ or clinical response.⁷ Despite improving knowledge of systolic synchronicity in HF patients with wide QRS complexes, it remains unclear whether cardiac diseases may result in systolic asynchrony even in patients with narrow QRS complexes. This study illustrates that systolic asynchrony is a common feature in these patients. Two criteria (maximal intersegmental difference in T_S of > 100 ms and T_S-SD of > 32.6 ms) consistently found that significant systolic asynchrony occurred in > 40% of patients with narrow QRS complexes and at least three quarters of those with wide QRS complexes. Intriguingly, both univariate and multivariate analyses found that the degree of LV asynchrony did not correlate with the duration of QRS complexes. Therefore, electrocardiography is not a good measure of mechanical asynchrony. This may be explained by the following reasons. Firstly, electrocardiography may not be sensitive enough to detect the presence of an electromechanical delay in all the regions of the LV, as it provides only a relatively crude representation of regional electrical conduction. Secondly, some of these patients may have mechanical asynchrony without significant electrical delay in the presence of myocardial disease as a result of deposition of extracellular matrix, myocyte loss, and pathological hypertrophy.^19,20 Thirdly, ultrastructural changes within the myocytes secondary to cardiac diseases may adversely affect myocardial contractility, such as a decreased amount of mitochondria and contractile proteins, as well as functional enzymes for oxidative phosphorylation.²¹ Tissue Doppler imaging has also been used to describe mild systolic asynchrony in patients with LV hypertrophy caused by hypertension, although the QRS duration was not specified in that study.²⁰ Radionuclide ventriculography showed regional systolic asynchrony in patients with coronary artery disease and preserved ejection fraction by using the variation of time to minimal volume.²² More recently, mechanical asynchrony was also described by phase imaging radionuclide ventriculography in four patients with HF and narrow QRS complexes.²³ In that study systolic asynchrony was less severe than in those with wide QRS complexes caused by bundle branch block, though no control group was used for comparison.²³

Predictors of systolic asynchrony were also sought in the present study. A low mean S_M and a large LV end systolic diameter were independent predictors of systolic asynchrony. QRS complex duration was found not to correlate with systolic asynchrony. The mean S_M has recently been shown to be a good index of global systolic function and is more sensitive than ejection fraction for detecting LV systolic dysfunction.¹⁴ Therefore, it appeared that more severe systolic dysfunction and LV dilatation were associated with more severe systolic asynchrony, independent of QRS complex duration.

Diastolic asynchrony in patients with HF

Similar to systolic asynchrony, diastolic asynchrony also occurred in > 40% of patients with HF and narrow QRS complexes, and in about 70% of those with wide QRS complexes. Although diastolic asynchrony has been described in patients with coronary heart disease and preserved LV function by radionuclide ventriculography,^22,24 and recently in patients with LV hypertrophy by tissue Doppler imaging,²⁰ it has not been shown in patients with HF. We observed that diastolic asynchrony occurred as commonly as systolic asynchrony in patients with HF and correlated modestly with QRS complex duration. Therefore, with increasing electromechanical delay resulting in prolongation of QRS duration, diastolic asynchrony worsened. The other predictor of diastolic asynchrony is the degree of diastolic dysfunction, as illustrated by the negative correlation with mean E_M. The latter parameter has been reported to be a good index of global diastolic function, which decreased as diastolic dysfunction worsened.¹⁴ Interestingly, coexistence of systolic and diastolic asynchrony is more common in patients with wide QRS complexes than in those with narrow ones (49% v 16%), and the correlation between the two conditions is only modest.

Clinical implication

LV systolic and diastolic asynchrony resulted in ineffective contraction and relaxation, respectively. As cardiac output is dependent not only on systolic emptying but also on diastolic filling, systolic and diastolic asynchrony may cause additive haemodynamic compromise in the failing heart. Cardiac resynchronisation has proved to be effective in improving symptoms and systolic function and in reducing LV size in patients with wide QRS complexes,^1,7–9,25 as a result of improved LV systolic synchronicity.^7–9 However, accurate selection of those who will respond to the treatment is vital and will help to ensure that the treatment is cost effective. This has not been satisfactorily achieved on the basis of the current guidelines, where QRS complex duration is the only surrogate determinant of cardiac asynchrony.^1,2,5,26 In patients with wide QRS complexes, systolic asynchrony assessed by magnetic resonance⁴ or tissue Doppler imaging^7,8 was superior to QRS complex duration to predict acute haemodynamic, clinical, or echocardiographic responses. Therefore, direct assessment of systolic synchronicity is probably a better guide for patient selection than the duration of QRS complexes. On the basis of the results of our study, it is imperative to study the efficacy of biventricular pacing in HF patients with narrow QRS complexes who have coexisting mechanical asynchrony. A recent a pilot study reported that biventricular pacing improved functional status of patients with HF with normal QRS duration who had echocardiographic evidence of LV systolic asynchrony.²⁷ However, this result needs to be confirmed by clinical trials with more patients and a longer follow up period.

In conclusion, LV systolic and diastolic asynchrony is not uncommon in patients with systolic HF with normal QRS duration, although it is less prevalent than in those with wide QRS complexes. Therefore, these patients may potentially benefit from cardiac resynchronisation. Selection for such treatment should also be based on information about cardiac synchronicity.