OUP user menu

Haemodynamic vector personalization of a quadripolar left ventricular lead used for cardiac resynchronization therapy: use of surface electrocardiogram and interventricular time delays

Luca Trolese, Juergen Biermann, Maximilian Hartmann, Fabienne Schluermann, Thomas S. Faber, Christoph Bode, Stefan Asbach
DOI: http://dx.doi.org/10.1093/europace/euu136 1476-1481 First published online: 3 June 2014


Aims The choice of left ventricular pacing configurations (LVPCs) of quadripolar leads used for cardiac resynchronization therapy (CRT) affects haemodynamic response and thus may be a tool for device optimization. The value of surface electrocardiograms and interventricular time delays (IVDs) for optimization is unknown.

Methods and results Sixteen patients implanted with a CRT device with a quadripolar LV lead underwent invasive testing of LV dP/dt. QRS durations at baseline (bl) and during biventricular pacing (biv) were measured using different LVPCs (total of 141 LVPCs; 8.8 per patient). Variations in QRS duration during biv were calculated for each patient (ΔQRS) and, when compared with intrinsic QRS duration, for all LVPCs (ΔQRSLVPC). Interventricular time delays between the poles of the LV lead were obtained from intracardiac electrograms. ΔIVD was calculated as IVDmax − IVDmin. Parameters were correlated with LV dP/dt. ΔQRS and ΔQRSLVPC both significantly correlated with LV dP/dt (P < 0.01). Correlation was found for patients with ischaemic (P < 0.001) and non-ischaemic cardiomyopathy (P < 0.05), and for patients with bl QRS duration >168 ms (P < 0.001), but not <168 ms (P = ns). The LVPC with shortest QRS duration also yielded maximal LV dP/dt in 6 of 16 patients (37.5%), and was equal or better in LV dP/dt in 12 of 16 patients (75%). ΔIVD neither correlated with ΔQRS nor ΔLV dP/dt.

Conclusion ΔQRS predicts the maximal value of vector personalization in the individual. Reductions in QRS width, but not IVDs, correlate with acute haemodynamic response. Intraindividually, in 75% of patients, the LVPC with the shortest QRS duration gives equal or superior haemodynamic results when compared with the LVPC with longest QRS duration.

  • Cardiac resynchronization therapy
  • Quadripolar lead
  • Haemodynamics
  • QRS duration
  • Interventricular time delays

What's new?

  • The selection of a pacing configuration of a quadripolar left ventricular lead affects haemodynamics and QRS width.

  • Variations in QRS width and acute haemodynamic response (according to the selection of the pacing vector) correlate. QRS variation therefore can be used to screen for the individual benefit of vector personalization.

  • Programming the quadripolar lead to the vector that provokes the shortest QRS duration gives equal or superior haemodynamic response compared with other vectors in 75% of patients.

  • Sensed interventricular time delays do not correlate with acute haemodynamics.


Cardiac resynchronization therapy (CRT) has been shown to improve symptoms, quality of life, exercise capacity, cardiac function, and to reduce all-cause and heart failure morbidity and mortality in selected heart failure patients16 and is therefore indicated for patients with depressed left ventricular (LV) function, New York Heart Association Class II–IV heart failure, and a wide QRS complex.7

However, the rate of non-responsiveness remains high. While the reasons for non-response are complex and manifold, optimal pacing site and -settings may increase treatment success. A recently introduced quadripolar LV lead (Quartet™, St Jude Medical) offers 10 LV pacing configurations (LVPCs) and was mainly developed to overcome technical issues with pacing thresholds and phrenic nerve stimulation.810 The distance of 4.7 cm between the distal and the proximal pole has evoked investigations of the haemodynamic response of stimulations at these sites. It has recently been shown by echocardiography, that the choice of the LVPC affects cardiac output and various cardiac electromechanical parameters.11,12 Our group has shown, by invasive measurement of LV dP/dt, that vector selection affects acute haemodynamic response, yielding an additional average 10% increase in LV dP/dt when comparing best and worst LVPCs,13 with significant interindividual differences. Invasive testing in clinical practice however does not seem to be adequate for every patient, therefore, we sought to investigate whether QRS duration, obtained from the surface electrocardiogram (ECG), and intracardiac time delays, obtained by measurements of the timing of local electrograms, are suitable to predict acute haemodynamic response.


We have previously published the results of a clinical trial (German Clinical Trials Registry DRKS00000573) on acute haemodynamics during biventricular pacing (biv) using a quadripolar LV lead (Quartet™, St Jude Medical), which offers a maximum of 10 LVPCs.13 While in these we could for the first time show that the choice of the LVPC affects acute haemodynamic response, we now present the data on the utility of surface ECG and intracardiac time delays to predict changes in acute haemodynamics.

During invasive testing, surface 12-lead ECG was continuously recorded (GE Medical Systems, CardioLab v6.5) during baseline (bl) without pacing and during biv in all possible LVPCs for analysis of QRS duration. QRS duration was measured by an electrophysiologist, who was unaware of the results of the haemodynamic measurements. QRS duration was measured from the first deflection until return to bl in any lead of the 12-lead surface ECG (Figure 1). Left ventricular dP/dt measurement has been described in our initial publication.13 Briefly, LV dP/dt was measured at bl and during biv in all possible LVPCs, and the percentage change in LV dP/dt was calculated, beat-by-beat, as compared with preceding bl without pacing. These changes were averaged over a 2 min pacing period for each LVPC (ΔLV dP/dtLVPC).

Figure 1

Example of the changes in QRS duration with respect to the selected LVPC in the patient with the minimal (A) and the maximal (B) ΔQRS.

QRS duration was measured during bl and during simultaneous (VV delay = 0 ms) biv at an individually optimized atrioventricular (AV) delay in all possible LVPCs. QRS complexes with maximal QRS duration during biv (i.e. the least decrease of QRS duration during biv) and minimal QRS duration during biv (i.e. the maximal decrease in QRS duration during biv) were subtracted (ΔQRS). ΔQRS thus expresses, in the individual, the maximal variation in QRS duration during biv in all possible LVPCs. ΔQRSLVPC was calculated for every LVPC as the difference of QRS duration during biv and bl, and thus expresses the decline in QRS duration for the corresponding LVPC.

ΔLV dP/dt was calculated in every patient as maximal minus minimal changes in LV dP/dt during biv, thus expresses the maximal intraindividual variation in LV dP/dt during biv with respect to the programmed LVPC. ΔLV dP/dt was correlated to ΔQRS. ΔLV dP/dtLVPC was calculated as specified above and was accordingly correlated to ΔQRSLVPC.

In addition, via the programmer (Merlin, St Jude Medical), we manually measured time delays from the local bipolar atrial-sensed electrogram (Asens) to the bipolar electrograms obtained from the right ventricle (RV) and four poles from the LV electrode (sensing programmed to D1-M2, M2-P4, M3-M2, and P4-M2, respectively). These measurements were obtained during bl without pacing. Interventricular time delays (IVDs) of the four poles of the quadripolar lead were calculated as the time period (Asens-LVPC) − (Asens-RV). ΔIVD was calculated as IVDmax – IVDmin and thus denotes the activation time along the quadripolar LV electrode.

Statistical analysis

Statistical analyses were performed using SAS Version 9.2, ©SAS Institute Inc., for linear regression and GraphPad Prism Version 5.0c, ©GraphPad Software, Inc., otherwise. Paired observations were compared using Student's t-test and repeated-measure analysis of variance with Bonferroni's correction, where appropriate. For correlation, we calculated Pearson's correlation coefficient r and the 95% confidence interval (CI). A P-value < 0.05 (two-sided) was considered statistically significant. Linear regression analysis accounting for constant within-patient correlation (compound symmetry) was performed to find a best-fit line and to compare the slopes of different regression lines.


As previously described, a CRT device (Promote Quadra™, St Jude Medical) with a quadripolar LV electrode (Quartet™, St Jude Medical) was implanted in 20 patients; in 16 of these, a total of 145 LVPCs were invasively tested by the measurement of LV dP/dt. In three patients, ECGs of four additional LVPCs could not be adequately measured due to missing data (n = 2) or low quality (n = 2). Therefore, ECGs of 141 paced QRS complexes and 16 bl QRS complexes were measured. Mean QRS duration at bl was 171.1 ± 20.6 ms. Biventricular pacing reduced the QRS duration significantly in all LVPCs (P < 0.0001). Averaged minimal (123.1 ± 16.9 ms) and maximal (140.0 ± 17.7 ms) QRS duration were significantly different (P < 0.0001).

Longer bl QRS duration by trend predicted lower bl LV dP/dt (P = 0.06).

ΔQRS, as the maximal variation in QRS duration during biv in all possible pacing configurations, correlates significantly with ΔLV dP/dt (Figure 2; slope = 0.69 ± 0.23, P < 0.01). Thus, with larger variations in QRS duration during biv at all LVPCs (ΔQRS), larger variations in acute haemodynamic response (ΔLV dP/dt) with respect to the choice of the LVPC can be expected.

Figure 2

Linear regression of ΔQRS and ΔLV dP/dt.

Over all patients, ΔQRSLVPC significantly correlates with ΔLV dP/dtLVPC (Figure 3; slope = −0.22 ± 0.05, P < 0.0001). This correlation was found both for patients with ischaemic (Figure 4A; slope = −0.41 ± 0.11, P < 0.001) and non-ischaemic cardiomyopathy (Figure 4A; slope = −0.12 ± 0.05, P < 0.05). A correlation was found for patients with wide (>168 ms) QRS complex (Figure 4B; slope = −0.29 ± 0.08, P < 0.001), but not for those with a bl QRS duration <168 ms (Figure 4B; slope = −0.12 ± 0.06, P = ns).

Figure 3

Linear regression of ΔQRSLVPC and ΔLV dP/dtLVPC.

Figure 4

(A) Linear regression of ΔQRSLVPC and ΔLV dP/dtLVPC. Patients divided based on type of cardiomyopathy: ischaemic (iCMP) and non-ischaemic (niCMP). (B) Linear regression of ΔQRSLVPC and ΔLV dP/dtLVPC. Patients divided based on QRS duration.

Intraindividually, in 6 of 16 patients (37.5%), the LVPC with greatest reduction in QRS width (ΔQRSLVPC) also offered maximal increase in ΔLV dP/dtLVPC. In 12 of 16 patients (75%), the LVPC with maximal reduction in QRS duration (i.e. the shortest QRS duration; ΔQRSLVPC) resulted in better (11 of 16) or equal (1 of 16) increase in ΔLV dP/dtLVPC than the LVPC that resulted in the least increase (the longest QRS duration during biv) in ΔLV dP/dtLVPC.

The mean time delay of the local signal in RV and LV (averaged over the four poles) was 84.0 ± 36.2 ms. Latest activation occurred in the distal pole in four, in M2 in six, in M3 in four, and in the proximal pole P4 in five patients (simultaneous activation of two poles in three patients). The time delay was maximal (latest activation) in a pole that also offered maximal haemodynamic benefit in only 3 of 16 patients.

The averaged ΔIVD in all patients was 20.4 ± 12.3 ms (time delay along the quadripolar LV electrode). ΔIVD neither correlated with ΔQRS (r = −0.27, 95% CI: −0.67 to 0.26, P = 0.32) nor with ΔLV dP/dt (r = −0.38, 95% CI: −0.74 to 0.15, P = 0.15).


A large amount of evidence shows that CRT has beneficial effects in patients with heart failure and electrocardiographic signs of asynchrony.16 However, in the aforementioned large clinical trials, a persistently consistent number of patients remain non-responders. While no universal definition of non-response exists,14 some clinical features, such as the amount of myocardial scar,15 the extent of mitral regurgitation,16 and the bl QRS duration and morphology17 may be predictive. One means of treating non-responders is optimization of device settings, which usually includes optimization of the AV- and VV-intervals,18 guided by electrocardiography, echocardiography, and/or device-based algorithms.19 As cumulative evidence suggests that the choice of the LVPC of a quadripolar LV lead affects haemodynamics and electromechanical parameters,1113 adequate selection of the LVPC arises as an additional option to individually optimize device settings. While a growing number of programming opportunities may allow an improved individual optimization, it must be acknowledged that, in daily practice, such methods must be easily accessible, reliable, and promptly accomplishable. In our initial study,13 we have chosen invasive assessment of LV dP/dt with a LV pressure wire. This certainly is not feasible for all patients in a routine clinical setting. Others have chosen echocardiography, which is time-consuming, operator-dependent, and dependent on imaging quality. Electrocardiography is easily accessible, standard for every CRT follow-up visit, and thus might be a tool to optimize the patient's device according to the optimal LVPC. Our results show that ΔQRS, as the maximal difference in the QRS duration between the LVPCs, is useful to screen whether the choice of LVPCs in the individual gives rise to large changes in haemodynamics. With larger changes in the QRS duration, larger effects on acute haemodynamics can be expected. The correlation of ΔQRSLVPC and ΔLV dP/dtLVPC gives evidence that greater reduction in QRS duration is linked to a more pronounced acute haemodynamic effect. These results are in line with previous studies that have examined the role of bl and paced QRS duration on response rates in patients undergoing CRT with a standard LV lead.20 Effects were more pronounced in patients with ischaemic heart disease, which may be related to the more heterogeneous activation pattern in these patients,21 and more pronounced in patients with longer bl QRS duration (here defined as >168 ms). The latter finding most likely reflects the more homogeneous activation pattern in patients with shorter bl QRS duration, and is in line with subgroup analyses from large multicentre trials, which have shown a more pronounced CRT effect in patients with longer bl QRS duration.46 Intraindividually, selection of the optimal LVPC guided by ΔQRSLVPC gave optimal results in 37.5% of patients and equal or better results than the worst LVPC in 75% of patients.

Measurement of IVDs seems to be a comprehensive approach: pacing from the latest activated site should yield the optimal effect. Indeed, the time interval from the Q-wave in the surface ECG to the local LV signal (QLV interval) has been shown to be predictive for response to CRT in a large cohort of patients.22 It has however not been shown that this is true for the individual. Nevertheless, this concept is implemented in current programmer software (St Jude Medical, Merlin PCS 16.2.1), which offers an algorithm which proposes the optimal LVPC based on automatic measurement of RV-sensed or RV-paced time delays. We could however not find a correlation of IVDs with QRS duration or LV dP/dt in our measurements of RV-sensed local electrograms. Our results therefore do not support the use of the algorithm that uses RV-sensed electrograms to optimize CRT devices with respect to the selected LVPC. A reason for this discrepancy might be the electromechanical delay, which has been shown to be pronounced especially in late activated myocardium.23


Due to the invasive nature and the lengthy protocol, the number of included patients is small. Our evaluation of measurements of IVDs to predict haemodynamic response is negative; however, measurements of RV-paced electrograms might be more reasonable, since CRT usually is delivered during biv. It is unclear whether acute haemodynamic response translates into chronic beneficial effects. Therefore, it remains to be examined whether it is beneficial with respect to clinical endpoints to individually optimize CRT patients according to the LVPC.


ΔQRS predicts the maximal value of vector personalization in the individual and might serve as a screening tool: Larger variations in QRS duration predict larger variations in acute haemodynamics. Overall, reduction in QRS duration correlates with acute haemodynamic response. Intraindividually, in 75% of patients, the LVPC with the shortest QRS duration gives equal or superior haemodynamic results when compared with the LVPC with longest QRS duration.

Conflict of interest: J.B., M.H., F.S., T.F., and S.A. are investigators in clinical trials sponsored by St Jude Medical. T.F. and S.A. have received a research grant and lecture fees from St Jude Medical not pertaining to this project.


View Abstract