Imaging in cardiac resynchronisation therapy

Abstract

A substantial number of heart failure (HF) patients do not respond after cardiac resynchronisation therapy (CRT). Recent studies observed that assessment of intraventricular (LV) dyssynchrony may allow identification of potential responders to CRT. In addition, presence of scar tissue and venous anatomy may play a role in the selection of candidates. In this review, an extensive overview of the available LV dyssynchrony measurements is provided using different echocardiographic modalities. In addition, the value of other noninvasive techniques such as magnetic resonance imaging, nuclear imaging and computed tomography for the selection of potential candidates for CRT will be discussed. (Neth Heart J 2008;16(Suppl1):S36-S40.)

Cardiac resynchronisation therapy (CRT) is an effective treatment for patients with advanced heart failure (HF) (New York Heart Association class (NYHA) III or IV), depressed left ventricular ejection fraction (LVEF, <35%) and wide QRS complexes (>120 ms) as demonstrated in various large multicentre trials. The beneficial effects include improvement in HF symptoms, exercise capacity, and LV function, as well as less HF hospitalisations and lower mortality rates.¹ Despite these impressive results, a consistent percentage of patients show no benefit after CRT, the so-called ‘non-responders’. The prevalence of non-responders is around 30% when clinical endpoints are used (e.g. improvement in NYHA class or exercise capacity), but can be much higher, up to 50%, when echocardiographic endpoints, such as improvement in LV function or reduction in LV end-systolic volume (LVESV), are used.¹

The presence of LV dyssynchrony prior to implantation and its subsequent reduction after implantation are proposed as the key mechanism for response to CRT.² Initially, QRS duration was used as a marker of LV dyssynchrony. However, QRS duration appeared to have a low predictive value for response to CRT. Furthermore, several studies demonstrated a weak relation between QRS duration and LV dyssynchrony.³ Interestingly, good relations were reported between QRS duration and inter (right versus left) ventricular dyssynchrony.³ Recently, various studies have demonstrated that patients with extensive baseline LV dyssynchrony have a high likelihood of responding after CRT implantation, whereas patients without baseline LV dyssynchrony fail to benefit.² A variety of techniques have been proposed to quantify LV dyssynchrony in HF patients, ranging from simple M-mode echocardiog-raphy to more complicated techniques such as realtime three-dimensional echocardiography (RT3DE) and strain imaging. Moreover, non-echocardiographic imaging techniques such as magnetic resonance imaging (MRI) and nuclear imaging have also been advocated to assess LV dyssynchrony.

Besides lack of baseline LV dyssynchrony, other factors may influence response after CRT.² Several noninvasive imaging techniques may provide additional information on scarred and viable myocardium, and venous anatomy.

Echocardiography

Echocardiographic techniques provide the quickest and most practical approach to evaluate LV dyssynchrony in CRT candidates. Table 1 summarises the main echocardiographic parameters and techniques used to predict response after CRT.

Table 1.

Main echocardiographic parameters, techniques, and cut-off values for assessment of LV dyssynchrony for prediction of response to CRT.

Measurement	Description	Technique	Cut-off value
SPWMD	Septal-to-posterior wall motion delay	M-mode	≥130 ms
Sum asynchrony	Delay in Ts of 3 basal LV segments (septal, lateral, posterior) and basal RV segment	Pulsed-wave TDI	>102 ms
S-L delay	Septal-to-lateral delay: delay in Ts between basal septal and lateral wall	Colour-coded TDI	≥60 ms
LV dyssynchrony	Maximum delay in Ts between 4 basal LV segments (anterior, posterior, septal and lateral)	Colour-coded TDI	≥65 ms
Ts-SD-12	SD of Ts of 12 LV segments	Colour-coded TDI	≥33 ms
Ts-Diff-12	Maximum delay in Ts between 12 LV segments	Colour-coded TDI	≥100 ms
PVD	Peak velocity difference: maximum delay in Ts between 6 basal LV segments	Colour-coded TDI	≥110 ms
S-L delay	Septal-to-lateral delay: delay in Ts between basal septal and lateral wall	TSI	≥65 ms
AS-P delay	Anteroseptal-to-posterior delay: delay in Ts between basal anteroseptal and posterior wall	TSI	≥65 ms
Ts-SD-12	SD of Ts of 12 LV segments TSI ≥34 ms
Radial dyssynchrony	Delay in Tε between basal septum and posterior wall	TDI strain (radial)	≥130 ms
AS-P delay	Delay in Tε between anteroseptal and posterior wall	2D-strain (radial)	≥130 ms
SDI	Systolic dyssynchrony index: SD of Tv for 16 LV segments	RT3DE	≥5.6%

2D=two-dimensional, LV=left ventricular, RT3DE=real-time three-dimensional echocardiography, SD=standard deviation, RV=right ventricular, Tε=time from onset of QRS to peak systolic strain, Ts=time from onset of QRS to peak systolic velocity, Tv=time from onset of QRS to minimum systolic volume, TDI=tissue Doppler imaging, TSI=tissue synchronisation imaging.

M-mode echocardiography

This simple technique measures the shortest delay between the maximal inward displacement of the septal wall and the posterior wall in the parasternal short-axis view of the LV at the level of the papillary muscles. Pitzalis et al. measured this septal-to-posterior wall motion delay (SPWMD) in 20 HF patients with LVEF ≤35% and QRS ≥140 ms.⁴ An SPWMD of ≥130 ms appeared to be predictive for LV reverse remodelling (defined as a reduction of≥15% in LVESV) after one month of CRT. However, recent data in 98 patients reported poor feasibility of SPWMD due to the absence of a clear systolic displacement (53% akinesia of the septum, 12% akinesia of the posterior wall, or 3% both) or a poor acoustic window in the parasternal view (32%).⁵ Consequently, SPWMD yielded only limited predictive value for CRT response (sensitivity 66%, specificity 50%). Therefore, this technique has limited value in daily clinical practice.

Tissue Doppler imaging

Tissue Doppler imaging (TDI) is the most popular technique for the evaluation of LV dyssynchrony. TDI includes assessment of myocardial velocity in different myocardial regions using the apical views. The timing of myocardial velocity is related to the QRS complex, providing electro-mechanical delays. Data can be acquired on-line using pulsed-wave TDI or reconstructed off-line using colour-coded TDI.

Pulsed-wave TDI allows quick on-line evaluation of regional synchronicity by measuring the time to onset of mechanical contraction by placing a sample in the region of interest. This technique was studied by Penicka et al., who measured the time intervals in the four basal LV segments and in the basal segment of the free wall of the right ventricle (RV).⁶ The authors reported a cut-off value of >102 ms for sum asynchrony (defined as the sum of LV and interventricular delay) for the prediction of improved LV function after CRT, yielding a sensitivity of 96% and a specificity of 77%. The assessment of only one sample area at a time is an important limitation of pulsed-wave TDI; colour-coded TDI allows simultaneous examination of multiple myocardial segments, thereby avoiding potential errors from differences in probe position, respiration and cardiac frequency.

Using colour-coded TDI, several parameters have been proposed to quantify LV dyssynchrony, ranging from 2- to 12-segmental models. For example, Bax et al. demonstrated that a delay of ≥65 ms between four basal segments (septal, lateral, inferior and anterior using four- and two-chamber apical views, respectively, see figure 1) was predictive for both clinical (sensitivity/ specificity 80%) and echocardiographic (sensitivity/ specificity 92%) improvement after six months of CRT.⁷ Yu et al. proposed calculating a dyssynchrony index (Ts-SD) by using the standard deviation of 12 time intervals of the six basal and six mid myocardial segments (two-, three- and four-chamber views). A recent large study in 256 CRT patients showed that LV reverse remodelling after CRT (defined as a reduction of ≥15% in LVESV after three to six months) could be predicted by Ts-SD ≥33 ms with a sensitivity of 93% and specificity of 78%.⁸ The major limitations of TDI are that it does not differentiate between active and passive motion and its angle-dependency.

Figure 1.

Colour-coded tissue Doppler imaging. Example of LV dyssynchrony assessment using colour-coded tissue Doppler imaging. Two samples are placed in the basal septum (S) and in the basal lateral wall (L) in the apical four-chamber view (upper left). Postprocessing yields velocity tracings related to the QRS complex (right); in this example severe LV dyssynchrony of 110 ms is present as indicated by the delay in the peak systolic velocity of the septum (yellow curve) as compared with the lateral wall (green curve). Adapted from Bax et al. ²

Tissue synchronisation imaging

Tissue synchrony imaging (TSI) colour codes the myocardium according to semi-automatic calculation of time delays between peak systolic velocities. Green represents early contraction and yellow to red later contraction. In addition, quantitative assessment is possible using myocardial velocity curves (similar to TDI). Van de Veire et al. compared the manually and automatically calculated values for LV dyssynchrony (time delay between basal septum and lateral wall) and found an excellent correlation.⁹ In addition, TSI was able to predict LV reverse remodelling after six months of CRT (sensitivity 81%, specificity 89%) using a cutoff value of 65 ms (similar to TDI). The main advantage of this technique is the quick visualisation of the most delayed segment.

TDI-derived strain

Strain imaging can be performed by off-line analysis of colour-coded TDI images. In contrast to TDI, which only measures myocardial velocities, strain imaging is able to measure the percentage of myocardial deformation during systole using the Doppler velocity gradients. LV dyssynchrony can be calculated by measuring the time delays of time-to-peak systolic strain (comparable with TDI). When applied to the apical views, low reproducibility has been reported due to the relatively high operator and angle dependency.¹⁰

Consequently, longitudinal strain appears to be a poor predictor for response to CRT.¹⁰ Despite its ability to differentiate between active and passive motion, the use of TDI-derived strain in clinical practice is limited because of the high image quality needed, and the significant post-processing time and operator experience.

2D-derived strain imaging

A more promising echocardiographic technique for quantification of myocardial deformation or strain is speckle tracking, which is based on conventional 2D greyscale images. The main advantage of this technique over TDI-derived strain is its lack of angle-dependency. Similar to MRI tagging, it is possible to study three forms of deformation: radial, circumferential and longitudinal. Suffoletto et al. calculated time to peak radial strain in six LV segments (anterior, anteroseptal, lateral, posterior, inferior and septal) in 48 CRT candidates. A cutoff value of ≥130 ms for anteroseptal to posterior wall delay was highly sensitive in predicting both acute and long-term response to CRT.¹¹ Importantly, speckle tracking analysis was possible in 96% of the patients with high reproducibility.

Real-time 3D echocardiography

RT3DE can capture the full volume of the LV within one breath-hold. Assessment of LV dyssynchrony can be reconstructed off-line by quantifying regional function and change in volumes for each of the 16 LV segments in systole and diastole. The systolic dyssynchrony index (SDI) is used as a marker for global LV dyssynchrony and is defined as the standard deviation of the time to minimal volume for each segment. In addition, the area of latest activation can be identified (figure 2). However, the use of this technique is limited by specialised equipment, time-consuming post-processing and lack of validated data. Only one study has addressed the predictive value of RT3DE for acute response after CRT;¹² Ajmone Marsan et al. found that an SDI of 5.6% was predictive for an immediate decrease in LVESV of ≥15% (sensitivity 88% and pecificity 86%).

Figure 2.

Real-time 3D echocardiography. Example of LV dyssynchrony analysis from an RT3DE dataset using parametric images. The global time to mean systolic volume is used as timing reference; early segments are coded in blue, whereas late segments are coded in red. In this example, the posterolateral segments are activated last (indicated in red). Substantial dyssynchrony is present as indicated by a systolic dyssynchrony index (SDI) of 13.6%. Adapted from Ajmone Marsan et al. ¹²

Other noninvasive imaging modalities

Magnetic resonance imaging

The use of MRI for selecting patients for CRT is expanding. Cardiac MRI can not only provide information on size, shape and function of the LV (of interest in patients with a suboptimal acoustic window), but also on LV dyssynchrony and the presence of scar tissue. Westenberg et al. compared LV dyssynchrony by TDI with LV dyssynchrony by velocity-encoded MRI in 20 HF patients and found an excellent agreement between both modalities (r=0.97, p<0.01).¹³ With contrast-enhancement MRI, precise visualisation of scar tissue is possible. Bleeker et al. applied this technique in 40 patients with ischaemic HF and investigated the relationship between response to CRT and the presence of scar tissue in the posterolateral wall (the preferred region for the LV pacing lead).¹⁴ After six months of CRT, 21 of the 26 (81%) patients without transmural scar tissue in the posterolateral region responded to CRT compared with only two of the 14 (14%) patients with transmural scar tissue in the posterolateral area (p<0.01). Moreover, patients with transmural posterolateral scarring exhibited no significant reduction in LV dyssynchrony the day after CRT implantation, suggesting absence of LV resynchronisation, explaining the lack of response.

Nuclear imaging

Similar to MRI, nuclear imaging is well suited for the assessment of LV dyssynchrony, and viability and scar tissue. Only a few small studies used radionuclide angiography with phase image analysis for the assessment of LV and interventricular dyssynchrony in CRT candidates. However, these studies demonstrate greatest benefit of CRT in patients with LV rather than interventricular dyssynchrony, which is in contrast with larger TDI studies, and further studies including comparisons with TDI are required. Recent studies showed that gated SPECT imaging can also be used for the assessment of LV dyssynchrony. Henneman et al. analysed four quantitative indices of phase analysis in 75 HF patients and compared them with conventional TDI. The variables histogram bandwidth and phase SD showed a good correlation with TDI. In addition, cut-off values of 135° for histogram bandwidth and 43° for phase standard deviation were proposed to predict an improvement in NYHA class after CRT.¹⁵ In addition, nuclear imaging is well suited for assessment of viability and scar tissue. Sciagra et al. were the first to demonstrate that patients with severe resting defects on 99mTc-sestamibi SPECT at baseline showed lack of response after CRT.¹⁶

Computed tomography

Computed tomography (CT) techniques cannot be used to determine LV dyssynchrony, but do allow non-invasive visualisation of the coronary venous system. Van de Veire et al. retrospectively investigated the cardiac venous anatomy in 34 patients with a previous myocardial infarction, 38 patients with significant coronary artery disease and 28 control patients using a 64-slice MSCT.¹⁷ The coronary sinus, anterior interventricular vein, and posterior interventricular vein could be visualised in nearly all these patients. However, in patients with a history of myocardial infarction, the left marginal vein was detected significantly less often (27%). At present, MSCT is not routinely used prior to CRT implantation, mainly because of the radiation dose and lack of information on LV dyssynchrony and site of latest activation. However, these findings suggest that patients with previous infarction may have less suitable venous anatomy for LV lead placement, and surgical (epicardial) LV lead positioning may be considered.

Clinical implications and future perspectives

Several noninvasive imaging modalities have been proposed for the quantification of LV dyssynchrony in HF patients; e.g. TDI using myocardial velocities, strain using myocardial deformation and RT3DE using volume changes. However, to date there is no agreement on which technique best predicts response to CRT and most performed studies are small, single-centre, non-randomised studies with short-term follow-up. Furthermore, feasibility in daily practice is an important issue. For now, most experience is gained with TDI and it remains the technique of choice. Major consensus is the more LV dyssynchrony prior to implantation, the higher the likelihood of significant LV inverse remodelling during follow-up. Available evidence is limited on the value of non-echocardiographic imaging methods, MRI and nuclear imaging, to assess LV dyssynchrony. However, these techniques can provide other information, for instance the presence of scarred and viable myocardium, and venous anatomy, potentially important for the selection of CRT candidates.

Conflicts of interest

Dr Bax received research grants from GE Healthcare, BMS medical imaging, Boston Scientific, Medtronic and St Jude. Dr Schalij received research grants from Biotronik, Medtronic and Boston Scientific.