The value of left ventricular strain–volume loops in predicting response to cardiac resynchronization therapy

Cardiac resynchronization therapy (CRT) is an established treatment for patients with dyssynchronous heart failure (DHF) [1], meanwhile growing evidence supports that the secondary changes of CRT in molecular and cellular play an important role in reversing left ventricular (LV) remodeling. Previous study [2] has demonstrated that molecular polarization within the dyssynchronous LV was consistent with segmental heterogeneity of myocardial load distribution. Strain describes exactly myocardial deformation under the action of hemodynamic load. However, most of segmental strain analysis proposed previously [3] neglected the dynamic relationship of strain with volume load changing across the cardiac cycle. As suggested in Carasso’s study [4], LV segmental strain–time curve morphology is actually highly predictive of response to CRT, while the same values of strain peaks or timings could be observed with different strain curve morphologies [5].

In our study, we hypothesized that the new approach, based on the simultaneous strain–volume analysis which combining temporal changing data from function and structure, could provide unique information for predicting CRT response. Three-dimensional (3D) speckle tracking imaging (STI) allows the simultaneous evaluation of cardiac principal strain (PS) and volume changes frame-by-frame. Since 3D PS overcoming limitations of geometry-dependent reference directions (i.e. longitudinal, radial, and circumferential) [6, 7], it has been proven that 3D PS correlated well with cardiac muscle fiber arrangements and could more accurately detect regional myocardial mechanical alterations [8, 9]. To validate our hypothesis, data of each patient across one cardiac cycle were plotted in a Cartesian system to develop the PS–volume loop which was analyzed by two characteristics of the linear fitting curve: the slope and the coefficient of determination which reflecting the degree of systolic–diastolic coupling (R²-S/D coupling). Slope reflects dynamic relationship between strain responses alongside volume changing. As to the R²-S/D coupling, the smaller its value is, the severer the degree of systolic–diastolic uncoupling is, which indicating that (global or segmental) myocardial efficiency reduced in dyssynchronous LV because myocardial systolic shortening and diastolic lengthening doesn’t synchronize with chamber volume decreasing and increasing.

The aim of our study was to explore whether the LV strain–volume loops, a novel approach linking the structural changes to functional alterations, could provide a new perspective for assessing LV remodeling in HF patients and reverse remodeling following CRT, in the hope of offering additional information for predicting response to CRT and exploiting additive value of this approach in selection of CRT candidates.

It has been demonstrated in our study that the absolute value of GLS, GCS increased and LV mechanical dyssynchrony indices (TpPS-SD% and Tmsv-SD%) reduced after CRT. However, such echocardiographic parameters at baseline didn’t show no significant differences between responders and non-responders. Similar results were found in previous studies [15]. Methods of dyssynchrony index assessment were studied extensively in the past but with incomplete success to predict outcome. Early enthusiasm towards such methods [16, 17] was dampened by the multicenter Predictors of Response to Cardiac Resynchronization Therapy (PROSPECT) trial [18] that manifested the dyssynchrony parameters had marked variability in prediction of the clinical composite score response (sensitivity 6–74%, specificity 35–91%) and prediction of LV end-systolic volume response (≥15% reduction; sensitivity 9–77%, specificity 31–93%). SDI, which was a concept of wasted work introduced by Lim et al. [14] and has been proven to have a strong predictive value for predicting response to CRT [19], was comparable between responders and non-responders at baseline and even had no significant change after CRT in our study. All these parameters have limiting ability to predict CRT response probably because none of them incorporates dynamic changing myocardial loading and stress state. As shown in our study, Midlateral peak PS was higher in responders than in non-responders, but Midseptal peak PS didn’t differ at baseline so that we couldn’t judge whether septal was working effective or not. Besides, in view of the difference of peak PS between Midseptal and Midlateral did not reach statistical significance, we couldn’t infer that the heterogeneity of segmental strain distribution in dyssynchrony LV. Therefore, Russell et al. [20] proposed the LV pressure-strain loop analysis which allows the estimation of regional and global LV work (quantified by strain rate multiplying by instantaneous LV-pressure) and furthermore introduced the wasted work ratio as an index to predict response to CRT [21]. Vecera et al. [22] also demonstrated the wasted work in the septum calculated by similar methods was a strong predictor of response to CRT. However, clinical use of the LV pressure-strain analysis is limited by the challenge to noninvasive acquire instantaneous LV pressure.

The new method based on simultaneous strain–load (volume or area) analysis has been applied in certain physiological state [23, 24] and pathological LV remodeling [25, 26]. Besides, it has been recently validated the same method with tagging cardiac magnetic resonance [27]. It took about 3–5 min to perform strain–volume loop analysis off-line for each patient after practice in our study. So, it would be more generally applicable in clinical use by virtue of its noninvasive and convenient. Considering LV mapping data [28] have shown that the mid-septal region was electrically activated first and U-shaped conduction pattern through the apical regions was imposed on the LV activation sequence by a transmural functional line of block located between the LV septum and the lateral wall with a prolonged activation time, our study focused on septal and lateral strain–volume relationship analysis at the mid-ventricular level, which not only enhancing reproducibility since segmental heterogeneity was most evident between septum and lateral wall but also enhancing operability by simplifying evaluation and reducing data loss. Firstly, our study explored characteristics of the LV PS–volume loops in healthy subjects to verify the reliability of this new method and provide reference foundation to the establishment of evaluation criteria of LV remodeling such as in HF. As expected, the healthy subjects showed the normal myocardial shortening and lengthening was well coupled to the simultaneous progressive reduction and expansion of volume. In healthy subjects, all kinds of PS–volume loops showed steep slope, indicating a good systolic and diastolic performance. Besides, a similar strain value could be observed for any given LV volume during systole and diastole in healthy individuals, which suggesting the presence of strong systolic–diastolic coupling. This result was concordant with previous study [23].

At baseline evaluation of HF patients, all kinds of PS–volume loops showed a lower slope when compared with that in healthy subjects. As a result of the simultaneous presence of chamber dilatation (rightward shift of the loop) and strain reduction (upward shift of the loop), the two parameters were moving in the opposite direction from the normal loops in the Cartesian system and we could observe significant alteration of the slope. Furthermore, dissociation occurred between systolic and diastolic strain at the same volume, which indicating the presence of uncoupling in the relationship between strain and volume in HF patients. In dyssynchronous LV, systolic–diastolic uncoupling of strain–volume loop reflects myocardial efficiency reducing because myocardial systolic shortening and diastolic lengthening doesn’t synchronize with chamber volume decreasing and increasing, which resulting in much wasted work globally or segmentally. In addition, strain–volume loop provides an intuitive visual representation to detect the response to CRT. As shown in our study, if CRT effective, a simultaneous volume reducing (leftward shift of the loop) and PS increasing (downward shift of the loop) would make the loop steeper and better systolic–diastolic coupling.

At baseline, the R²-S/D coupling of the Midseptal PS–Global volume loop in CRT responders was significantly lower than that in non-responders, besides, the R²-S/D coupling of the Midseptal PS–Global volume loop at baseline was found to be an independent predictor of CRT response in multivariate analysis whereas dyssynchrony parameters were not. This result indicated that much wasted myocardial work was located to the septum at baseline in CRT responders. The amount of septal wasted myocardial work at baseline was related to the magnitude of benefit following CRT. The more the septal wasted at baseline, the higher probability of response to CRT was achieved. Nevertheless, the slope of the Midseptal PS–Global volume loop didn’t show significant difference between CRT responders and non-responders. While the synchronism of global and segmental myocardium maintains higher level with good systolic-diastolic coupling, myocardial contractility could be reflected by the slope of strain-volume loop. But if higher degree dyssynchrony of global and segmental myocardium is presented with systolic-diastolic uncoupling, the slope of strain-volume loop would underestimate the myocardial contractility. After CRT, the slope and R²-S/D coupling of the Midseptal PS–Global volume loop as well as Midseptal peak PS were increased significantly, accompanied by a significant increase in the slope of Global PS–Global volume loop as well as GLS and GRS, which suggesting an improvement in septal efficiency following CRT could contribute to the improvement in LV global function.

Midlateral peak PS as well as the slope of Midlateral PS–Global volume loop and Midlateral PS–Midlateral volume loop in CRT responders were higher than in non-responders at baseline. Therefore, we speculate that the compensatory maintenance of LV lateral wall function at a certain level was associated with a favorable response to CRT, which might be inconsistent with the view of Zweerink [29]. In Zweerink’s study, absolute values of lateral wall strain, strain rate, and work were significantly higher in responders, but this finding is not getting the attention it deserved for further investigation and deeper analysis.

In addition, our results also indicated that applying CRT to patients with abnormal segmental heterogeneity between septum and LV lateral wall, which contrary to that in healthy subjects, may be more likely to response to CRT. This heterogeneous distribution of myocardial work can be rebalanced by restoring normal electrical activation following CRT. It is helpful to improve patient selection for CRT to identify this abnormal but reversible segmental heterogeneity. Electric activation delay generates contractile dyssynchrony, with early-activated septum earlier onset of shortening during isovolumetric contraction phase against a low afterload while late-activated lateral wall shortening against increased wall stress and loading at late systole into early relaxation [30]. The early-developed septal force is dissipated in generating sufficient energy to open the aortic valve and in stretching the late-activated lateral wall. The latter event represents wasted energy during early ejection. Passive stretch might influence regional myocardial contractility because changes of effective preload triggered local Frank-Starling mechanism. The passive pre-stretch would enable the late-activated segments to contract to a greater extent in order to compensate for the increased loading conditions. Thus, septal work efficiency being reduced and lateral function compensatory being improved together generate the abnormal segmental heterogeneity. CRT could improve septal work efficiency and normalize lateral function in responders, which making heterogeneous distribution of myocardial work significantly decrease significantly or even eliminate. As observed in our study, responders showed improvements in the slope and R²-S/D coupling of the Midseptal PS–Global volume loop after CRT, whereas no significant change was observed in characteristics of the Midlateral PS–Global volume loop.

Most probably as the result of an optimized local loading state by CRT, the slope of both the Midseptal PS–Midseptal volume loop and the Midlateral PS–Midlateral volume loop were significantly increased in responders after CRT.

Analysis of strain–volume loops could provide unique information for predicting response to CRT and may become a potential new tool to detect the beneficial effects of CRT on LV function. Our study demonstrated that assessment of wasted work in septum at baseline would be helpful to improve patient effective selection for CRT. R²-S/D coupling of Midseptal PS–Global volume loop at baseline is proved to have predictive value for predicting response to CRT.