Effects of cardiac resynchronization therapy on coronary blood flow: Evaluation by transthoracic Doppler echocardiography☆

2.1. Study population

Twenty-two consecutive patients with idiopathic dilated cardiomyopathy, successfully treated with CRT for at least 3 months, were recruited for the study. For each patient, significant coronary artery disease had been ruled out by angiography. The initial indication for CRT was drug-refractory NYHA class III or IV heart failure, wide QRS complexes (≥120 ms) and LV ejection fraction (LVEF) ≤35%. All patients had left bundle branch block (LBBB) on the baseline ECG. Positive clinical response to CRT was defined as an improvement in functional class, by at least one class, and any improvement in exercise tolerance after 3 months of CRT. The study was approved by the Ethics Committee of the Karolinska University Hospital (Stockholm, Sweden) and all patients provided written informed consent to participation.

2.2. Pacing systems

Fourteen patients (64%) were implanted with a pacemaker for CRT (CRT-P system), with bipolar atrial and right ventricular (RV) leads. The RV lead was routinely placed in a septal position and the LV lead was placed either laterally or postero-laterally. Eight patients (36%) were implanted with a defibrillator for CRT (CRT-D system), with a similar pacing lead configuration to CRT-P systems, except for the RV lead, which was positioned in the RV apex. Immediately after implantation the atrioventricular and interventricular delays were individually optimized according to Ritter [19] formula and maximum aortic flow [20], respectively. During TTE measurements of CBF the CRT devices were randomly programmed to: intrinsic conduction (IC), RV pacing, simultaneous biventricular pacing (BVP), BVP with 40 ms LV pre-activation. The echocardiographic evaluation was performed in the spontaneous atrial-synchronized pacing configuration, in order to maintain a similar heart rate during different pacing modes. In the 3 patients without IC, heart rate was programmed to 60 bpm and was kept constant throughout the examination. Each pacing mode was maintained for 15 min and an equilibrium period of 10 min was allowed between echocardiographic recordings. Blood pressure was measured at rest and at the end of 15 min during each pacing modality.

2.3. Echocardiography

Patients were examined in the supine lateral decubital position. TTE was performed using a Vivid 7 system (Vingmed-General Electric, Horten, Norway) equipped with a phased array 3.5 MHz transducer (Doppler frequency 5.0–3.5 Mhz). Images were digitally stored to a dedicated server and post-processed using a workstation (GE EchoPAC sw only, version 5.1.0, Horten Norway). All data analysis was performed off-line on the EchoPAC workstation. A standard evaluation of LV volumes and LVEF was performed for each patient.

2.4. Statistics

Continuous data are presented as mean±one standard deviation in text and tables. The data were analysed using procedure Mixed Model Analyses in SAS®. A one-way repeated measures ANOVA was performed with pacing mode as the within-subjects variable. Patients with missing data (absent IC in 3 patients) were included in the model. The Mixed procedure can accommodate missing data under the assumption that data are missing at random. Three different covariance pattern models were tested: compound symmetry, unstructured and autoregressive. The covariance structure with the smallest value of Akaike's Information Criterion was considered most desirable. If the F-ratio for the factor “pacing mode” was significant, Dunnett's post-hoc contrast between pacing modes was performed. Since the distribution of some variables was positively skewed, log-transformation was performed before the analyses. Relations between variables were assessed using Pearson correlation coefficient. A Cochran Q test was performed to test whether there was a systematic difference between the pacing modes in the variable “synchrony” (yes/no). P<0.05 was considered statistically significant.

3.1. Coronary flow variables

LAD flow variables during different pacing modalities are presented in Table 2. Coronary flow assessment could be performed during each pacing modality in all the enrolled patients (measurements during intrinsic conduction were missing only in the 3 patients without spontaneous rhythm). Mean CFV was increased both by simultaneous BVP (p=0.0063) and by BVP with LV pre-activation (p<0.0001) vs. IC, and was also higher during BVP with LV pre-activation than during simultaneous BVP (p=0.027). Accordingly, an improvement in mean CFG was observed during simultaneous BVP (p=0.0026 vs. IC) and BVP with LV pre-activation (p<0.0001 vs. IC; p=0.010 vs. simultaneous BVP). Peak CFV and CFG were higher during BVP with LV pre-activation than during the other programming modes. An increase in LAD flow VTI was observed during simultaneous BVP and BVP with LV pre-activation, but reached statistical significance vs. IC only in the latter programming modality (p=0.0005). Except for a slight increase in peak CFV and CFG, no significant changes in mean CFV, mean CFG and LAD flow VTI were detectable during RV pacing compared to IC (p=0.149, p=0.064 and p=1.0, respectively). There was a trend towards an increase in coronary flow slope from IC to BVP with LV pre-activation (p=0.07), whereas no significant changes in time of coronary flow and time of coronary flow/RR were observed. The intra-observer variability for coronary flow variables is presented in Table 3. There were no acute changes in systolic and diastolic blood pressure and heart rate after switching to different pacing modalities.

3.2. TVI variables

In the three patients without spontaneous rhythm and in two patients with suboptimal TVI images dyssynchrony could not be evaluated. As reported in Table 2, intraventricular dyssynchrony, expressed by septal-to-lateral delay, showed a significant decrease during simultaneous BVP and BVP with LV pre-activation vs. RV pacing and intrinsic conduction (p<0.0001). No significant changes in intraventricular dyssynchrony were observed during RV pacing compared to IC. By using a septal-to-lateral delay ≥65 ms as an indicator of intraventricular dyssynchrony [21], 16 patients (94%) were dyssynchronous during IC, 15 (88%) during RV pacing and 3 (18%) both during simultaneous BVP and BVP with LV pre-activation (p<0.05 at Cochran Q test). There was a significant inverse correlation between percentage increase in coronary VTI and percentage decrease in septal-to-lateral delay during BVP with LV pre-activation (r=−0.48; p=0.04). A similar trend, although not significant, was found during simultaneous BVP (r=−0.38; p=0.12). Peak systolic velocity in the basal septum was significantly higher during simultaneous BVP and BVP with LV pre-activation than during IC (p=0.026 and p=0.001, respectively). A post-systolic velocity in the basal septum was observed in 9 patients (53%) during IC, and significantly decreased after switching to simultaneous BVP and BVP with LV pre-activation (p=0.0018).

4.1. Mechanisms of increased coronary flow

CBF is regulated by several mechanisms, depending on the driving pressure and the resistance offered by the vascular bed. LV filling pressures and wall stress are further factors which may affect myocardial perfusion through an extravascular compression of the coronary vasculature. CRT has been shown to reduce LV filling pressures [22], but under resting conditions the autoregulation of the vascular tone may prevent global CBF from being affected by acute haemodynamic changes [4]. In the present study no significant differences were observed in systolic and diastolic blood pressure, or in heart rate, by acutely changing pacing modality, suggesting that variations in LAD blood flow variables were independent of these parameters. Myocardial perfusion is also affected by variations in the diastolic variables. An increase in diastolic filling time, which may prolong the diastolic perfusion time, has been observed during CRT [23]. However, in line with data from Knaapen et al. [4], we found no significant differences in the diastolic perfusion time between different CRT pacing modes. Other pathophysiological mechanisms should therefore be considered to explain the regional changes in myocardial perfusion.

Our results suggest that the increase in regional LAD flow velocity during simultaneous BVP and BVP with LV-pre-activation may be dependent on a more homogeneous ventricular activation and contraction pattern. In heart failure patients with LBBB, the relatively inefficient work of the dyssynchronized LV is probably an effect of segments contracting in an untimely manner, with contraction in one segment resulting in pre-stretching of another segment [24]. Late activated regions start to contract at higher wall stress and also stretch early activated regions as they enter relaxation. This asynchronous electro-mechanical activation most likely reduces metabolic demands in the septum and may explain the reduced septal perfusion observed by PET investigations in patients with severe dilated cardiomyopathy and LBBB [4,7,25]. CRT seems to restore a balance in regional perfusion inhomogeneities without altering global CBF, by reducing the unfavourable LBBB associated ventricular activation and contraction pattern. This has been documented by PET as an increase of CBF and myocardial oxygen consumption in the interventricular septum, with a simultaneous decrease in the lateral wall [4,7,11]. The results of the present study not only confirm an increase of CBF in LAD during CRT, but also suggest that these changes are associated with a more synchronous activation pattern, expressed by a shortening in septal-to-lateral delay. Almost all patients became synchronous during simultaneous BVP and LV pre-activation, and the shortening in septal-to-lateral delay was correlated to the improvement in CBF variables. These findings are in line with a recent TTE study showing that CBF velocities in the distal LAD and LV functional parameters are improved by CRT, and decrease by acutely withholding the active biventricular stimulation [18].

The transition towards a more synchronous LV contraction during simultaneous BVP and LV pre-activation was also reflected by a decrease in post-systolic motion in the basal septum. Post-systolic motion has been described as a manifestation of LV intraventricular dyssynchrony, reflecting regional disturbances in myocardial function due to prolonged contraction, delayed relaxation or recoil after systolic bulging [26]. A post-systolic contraction may lead to increased vascular resistance and reduced regional perfusion. Furthermore, it may elevate LV filling pressures. In this perspective, the decrease in post-systolic motion observed in a subgroup of patients in our study may be one of the mechanisms responsible for the improvement of LAD flow in these patients.

4.2. Myocardial perfusion and function

In a series of studies evaluating the acute and medium-term haemodynamic effects of BVP, LV and RV pacing [27 28 29], shifts between different pacing modalities were associated with a significant and immediate improvement in LV systolic parameters only during LV or BVP stimulation, whereas RV pacing was associated with an acute pronounced worsening of cardiac index [27,29]. Our results suggest that the increase in LAD flow velocity during simultaneous BVP pacing and LV pre-activation may be one factor supporting the haemodynamic and functional benefits of BVP vs. RV pacing. In fact, the improvement in LAD flow was associated with an increase in regional longitudinal myocardial contraction, expressed by peak tissue velocity in septum, during simultaneous BVP and LV pre-activation, but not during RV pacing. The adverse haemodynamic effects induced by RV pacing have been usually related to ventricular remodelling and induction of increased dyssynchrony [30]. More recently, an experimental animal study showed that myocardial perfusion of the septum is acutely impaired by RV pacing, while it is improved by LV pacing and unaffected by BVP [31]. In our study LAD flow and intraventricular dyssynchrony were overall not significantly impaired by RV pacing in comparison with IC, and this may be due to the RV lead positioning. Traditionally RV pacing is performed from the RV apex but in this study a supposedly more physiological septal pacing site was used.

4.3. Comparison with other methods

Different techniques have been used to assess coronary flow in CRT patients. In a recent follow-up study evaluating proximal LAD blood flow by transoesophageal echocardiography [13], CRT induced an increase, although not significant, in resting CBF and a significant improvement in hyperaemic CBF after 3 months of active biventricular pacing. No comparison between pacing modes was performed. The characteristics of the study design, performed under sedation, in a population including both ischaemic and non-ischaemic patients, may seemingly explain the differences in resting CBF assessment, as compared to our results.

In an acute haemodynamic study, Nelson et al. [3] found no significant modifications in proximal LAD blood flow, measured by an intracoronary Doppler catheter, before and during CRT. The lack of increased LAD blood flow during CRT, differently from our findings, could perhaps be explained by the characteristics of the study protocol, performed under sedation, and by different LV pacing sites compared to the chronic CRT systems evaluated in our study.

As compared to transoesophageal echocardiography or intracoronary Doppler flow wire CBF assessment, TTE allows a non-invasive, reproducible and more feasible assessment of CBF [14 15 16], and it is also more available and cheaper than PET. The easy applicability therefore makes TTE a useful methodology for routine assessment of CBF and CBF velocity reserve. In patients with non-ischaemic dilated cardiomyopathy, the assessment of an abnormal CBF reserve by TTE has been recently demonstrated to be an independent prognostic marker of bad prognosis [32].