Acute haemodynamic comparison of multisite and biventricular pacing with a quadripolar left ventricular lead

Pacing from multiple sites in the left ventricle (LV) may bring about further resynchronization of the diseased heart compared with biventricular (BiV) pacing. We compared acute haemodynamic response (LV d P /d t max ) of multisite and BiV pacing using a quadripolar LV lead. the most distal and proximal electrodes of the quadripolar LV lead most commonly yielded greatest LV d P /d t max . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Keywords A 56-year-old man with a history of persistent and refractory atrial ﬁbrilla-tion (AF) presented with a cerebrovascular accident. Cardioversion was attempted previously, however his AF recurred. A cardiac magnetic resonance imaging (MRI) was done (panels A , B ) and he underwent extensive AF ablation which improved exercise tolerance. Owing to inconsistent anticoa-gulation with warfarin and intolerance to dabigatran he was offered percu-taneous exclusion of the left atrial appendage (LAA) via an epicardial approach with the Lariat w . A left atrial angiogram and an intraoperative transesophageal echocardiogram were performed to conﬁrm occlusion of the appendage (panels C , D ). Three months after Lariat w closure of the LAA, a repeat AF ablation was planned and the patient underwent a second MRI which allowed assessment of LAA closure (panels E , F ). The LAA is widely recognized as the likely source of thrombi in patients with non-valvular AF. Exclusion of the LAA with the Lariat w demonstrates complete occlusion of the appendage by magnetic resonance as well as angio-graphic imaging and may be useful in patients who are poor candidates for chronic anticoagulation.


Introduction
Cardiac resynchronization therapy (CRT) reverses remodelling of the dilated left ventricle (LV) and reduces morbidity and mortality in patients with symptomatic heart failure, prolonged electrical delay, and impaired systolic function receiving optimal pharmacological therapy. 1 -8 The immediate effect of biventricular (BiV) pacing may be observed by invasive haemodynamic measurement of peak increase in LV pressure (LV dP/dt max ), reflecting cardiac contractility. 9 -11 Prior studies have demonstrated beneficial effects of CRT on LV dP/dt max , 12 -14 and have used this index to determine optimal atrioventricular (AV) delay and interventricular paced timing intervals, explore alternative pacing sites, and guide lead placement. 15 -19 Although there is an evolving understanding of the relationship between acute haemodynamic changes from CRT and clinical response to therapy, it is well recognized that the outcome of CRT varies among patients. In a sizeable fraction, there is little functional or clinical improvement. 20,21 Potential reasons may include suboptimal LV pacing site, non-uniform electrical wavefront propagation, and inability to produce coordinated mechanical activation. 22,23 Biventricular pacing is performed with leads implanted in the right ventricle (RV) and along the lateral wall of the LV through a venous tributary of the coronary sinus (CS). The search for improvements to CRT has led several groups to examine the effect of additional ventricular stimulation sites on acute cardiac function and longer-term outcomes, but many of these therapies require placement of another pacing lead. 24 -29 However, variation in haemodynamic response may be achieved by changing LV pacing site even within a vein, although no common optimal site has been identified for all patients. 13,16 A quadripolar LV lead (Quartet model 1458Q, St Jude Medical, Inc.) has recently been developed with three ring electrodes located 20, 30, and 47 mm from the tip electrode. With compatible pacing system, it is possible to deliver independent pacing pulses to multiple electrodes of the lead, potentially capturing a larger area and engaging multiple zones in the long axis of the LV. Our objective, therefore, was to determine whether pacing from more than one electrode positioned within a tributary of the CS could improve systolic function. We compared LV dP/ dt max during simultaneous pacing from two or more LV sites and the RV with BiV pacing using only the tip electrode of the LV lead.

Study population
The protocol for this prospective acute haemodynamic comparison of BiV and multisite pacing at Montreal Heart Institute was approved by the local institutional review board. The study (clinicaltrials.gov identifier NCT00964938) enrolled patients with standard CRT indication providing written informed consent. The study was conducted in accordance with the Declaration of Helsinki.

Implant procedure
Procedures were performed under conscious sedation, with a steady rate of intravenous propofol and remifentanil. The environment was kept as quiet as possible to minimize changes in underlying physiological state of the patient. The quadripolar LV and conventional right atrial (RA) and RV leads were implanted using standard techniques. Left ventricular lead implantation was targeted to a CS tributary along the left lateral wall. The leads were connected to an external multichannel stimulation system that delivered pacing to electrodes of the LV, RV, and RA leads. The naming convention for the four LV electrodes, from distal tip electrode to proximal ring electrode, was D1, M2, M3, and P4. Each of the four LV electrodes was connected for pacing in extended bipolar configuration with the cathode in the LV and the RV coil used as anode. Capture thresholds from all electrodes were recorded, and the proximal P4 electrode was replaced by M3 in the pacing protocol if P4 did not capture the ventricle.

Haemodynamic assessment protocol
Acute haemodynamic measurements were recorded using a 0.014-inch guidewire pressure sensor and recording system (Pressure-Wire Certus, RadiAnalyzer and PhysioMon software, St Jude Medical, Inc.,). The guidewire was introduced at the femoral artery and inserted retrograde into the LV for continuous measurement of pressure and dP/dt. A bolus of intravenous unfractionated heparin was administered to reduce risk of thrombo-embolic complication. To avoid ratedependent variability, all pacing was delivered at a fixed rate 10% above the patient's intrinsic heart rate, using a single AV delay that produced the highest LV dP/dt max for BiV pacing.
Test configurations included BiV pacing, using the LV lead's distal D1 electrode in extended bipolar configuration, and simultaneous multisite pacing using two or more LV electrodes as cathodes in extended bipolar configuration. Configurations were delivered simultaneously with RV bipolar simulation. Test configurations were performed in a pre-determined randomized sequence for 30 s each during haemodynamic recording. A measurement protocol was specifically developed to minimize effects of noise, physiological drift over time, and beat-to-beat variations. At the same time, the measurements needed to be completed within a short time to avoid clinical deterioration and minimize the possibility of adverse haemodynamic consequences.
Four multisite pacing configurations were tested three times each in randomized order, with return to baseline recording using standard BiV pacing after each test configuration. The multisite pacing configurations used extended bipolar pacing from all LV electrodes: from distal and most proximal electrodes, from distal and middle electrodes, and from middle and most proximal electrodes. After the pacing and haemodynamic measurement protocol was completed, the quadripolar LV lead was removed and replaced by a standard bipolar LV lead as part of a permanent CRT system.

Haemodynamic data
Cardiac cycles included in LV dP/dt max calculations were selected in post-processing analysis performed while blinded to the pacing configuration. Ectopic ventricular beats and the two subsequent beats were excluded, as were the first few beats after activating a new pacing configuration. Each 30 s LV dP/dt max recording spanned multiple complete respiratory cycles. To minimize impact of respiration and physiological variation, LV dP/dt max was measured during three separate recordings for each test configuration.
Left ventricular pressure data for individual beats were analysed using Excel (Microsoft Corp.) and Matlab (The MathWorks, Inc.). Relative change for each multisite pacing configuration was computed with reference to baseline BiV measurements immediately before and after the test configuration. Atrial-only pacing was also performed at the same rate, and compared with adjacent BiV recordings. The acute haemodynamic response of a test configuration was reported as the mean of relative change among all recordings for that configuration. Figure 1 contains an example of LV dP/dt max data for a patient. Each bar represents the mean value for a single recording. In the calculation for pacing configuration T2 (multisite pacing using LV electrodes D1 What's new? † Comparison of haemodynamics of biventricular (BiV) and multisite pacing with a multipolar left ventricular (LV) lead. † Rigorous method with invasive haemodynamic measurement, performing each test configuration three times, compared with baseline BiV measurements before and after. † Median increase in LV dP/dt max over atrial pacing: 8.2% using BiV, 10.2% using multisite with most distal and proximal electrodes, 10.3% using best multisite pacing configuration. † Of all tested configurations of multisite pacing, 72% produced greater LV dP/dt max than obtained with BiV pacing. † Distal/proximal multisite pacing is the most common optimal configuration, superior to BiV in 74% of patients. and M3), the relative change is calculated with reference to the average of standard BiV recordings before and after pacing configuration T2. The overall effect of the pacing configuration is reported as the mean among the three readings.

Statistical analysis
Change in LV dP/dt max was also calculated for each configuration with respect to atrial pacing. Since the Shapiro-Wilk statistic showed nonnormal distribution across patients, median and interquartile range of changes in LV dP/dt max were reported with respect to atrial pacing (along with mean and standard deviation for historical comparison). Wilcoxon signed-rank tests were used to compare differences in LV dP/dt max , compared with atrial pacing, obtained from the various pacing configurations. For each multisite pacing configuration, patients were ranked in order of improvement in LV dP/dt max over standard BiV pacing. Spearman correlation of the rankings was used to determine whether the same patients tended to achieve greater improvement, regardless of pacing configuration. Customized patient groupings were developed based on empirical categorization of acute haemodynamic response. A P value below 0.05 was considered to be statistically significant. Analyses were performed using SAS version 9.2 (SAS Institute Inc.).

Results
The protocol was successfully completed in 21 of the 26 patients enrolled. Three subjects were excluded due to difficulty in placing the guidewire pressure sensor. The procedure was aborted in two additional subjects, one without a suitable distal vein for LV lead placement and the second due to communication failure in the haemodynamic measurement system. Characteristics of the remaining 21 patients are listed in Table 1. The quadripolar LV lead was placed in a lateral branch in 10 patients; in a posterolateral branch in 6; and in an anterolateral branch in the remaining 5 patients. The proximal P4 electrode reached a location that could stimulate the ventricle in 16 of 21 patients (76%).

Haemodynamic data
Duration of the pacing procedure and haemodynamic recordings was 34 + 8 min. In two subjects, pressure and dP/dt data contained extracardiac noise that generated erroneous dP/dt max readings; these two were excluded from haemodynamic data analysis.
Pacing was performed at a rate of 89 + 16 beats per minute in the 19 analysed patients. In 17 of the 19 (89%) patients, at least one multisite pacing configuration produced an increase in LV dP/dt max when compared with standard BiV pacing. In 16 of 19 (84%) patients, two or more of the four configurations were improved compared with BiV pacing. In 9 of 19 (47%) patients, all four multisite pacing configurations increased LV dP/dt max , compared with BiV pacing. In total, 72% of all tested multisite pacing configurations yielded improvement in LV dP/dt max . Although the best pacing configuration varied among patients, the configuration using most distal and proximal electrodes improved LV dP/dt max in  Figure 1 Example of LV dP/dt max calculation for a single pacing configuration. Bars, shown from left to right in order of recording, represent mean dP/dt max during each pacing configuration. Standard BiV pacing is repeated after each multisite pacing configuration. In the example calculation shown for multisite pacing using LV electrodes D1 and M3 as cathodes (marked D + M), each DdP/dt max is calculated as the relative increase during the test recording with reference to average of preceding and following recordings of BiV pacing. Relative change for each recording of the test configuration is used in calculation of mean DdP/dt max . In the example shown, relative change for the three measurements of multisite pacing with D1 and M3 is 10.1%, 5.8% and 6.2%; mean DdP/dt max is 7.4%.
14 of 19 (74%) patients when compared with BiV pacing. This configuration yielded the highest LV dP/dt max of all tested configurations in 8 of 19 (42%) patients. The combination with middle and proximal electrodes was the only multisite pacing configuration that was less likely than BiV pacing to be the best choice. All pairwise correlations of the ranking of patients by change in LV dP/dt max compared with BiV pacing for any two multisite pacing configurations reached statistical significance, with correlations ranging from 0.70 (P ¼ 0.0009) to 0.84 (P , 0.0001). Table 3 summarizes the percent change in LV dP/dt max achieved by BiV pacing and multisite pacing compared with atrial pacing, in all patients. Biventricular pacing improved LV dP/dt max by a median of 8.2% (interquartile range 2.3%, 15.7%) compared with atrial-only pacing. The best of the four multisite pacing configurations yielded a median of 10.3% (7.0, 28.6%) improvement. Left ventricular dP/ dt max increased by 10.2% (6.1%, 25.6%) over atrial-only pacing with the distal and proximal electrode configuration. The best multisite pacing configuration selected for each patient had a significantly higher LV dP/dt max compared with BiV pacing (W ¼ 68, P ¼ 0.005). Similarly, the multisite pacing configuration using distal and proximal electrodes for all patients generated a significantly higher LV dP/dt max compared with BiV pacing (W ¼ 60, P ¼ 0.014).
Prior evaluations of CRT using LV dP/dt max have reported the response among patients with parametric statistics. Mean improvement with BiV pacing was 13.3% over atrial-only pacing. The best of four multisite pacing configurations increased LV dP/dt max by 17.5%, and the benefit achieved by using distal and proximal electrodes in all patients was 16.5%. Incremental improvement using best multisite pacing configuration was 4.3% in absolute terms, amounting to a relative increase of 32.2% above BiV pacing.
Change in LV dP/dt max for each of these patient groupings is shown in Figure 3. Overall, the acute response rate to CRT further improved in 12 of 19 (63%) patients with multisite pacing compared with BiV pacing.

Main findings
The rigorous protocol for recording LV dP/dt max provided a sensitive method of identifying true acute haemodynamic effects associated with additional LV pacing sites with a quadripolar LV lead for CRT. In the vast majority of patients, multisite pacing using  two or more LV electrodes increased cardiac contractility beyond values achieved by BiV pacing. An increase in contractility was observed in all but 2 of the 19 patients from at least 1 of the 4 tested multisite pacing configurations, and from nearly 75% of all configurations tested. Although the magnitude of response was highly variable, the proof of concept that pacing from multiple electrodes may acutely augment contractile coordination in diseased hearts was demonstrated, compared with standard BiV pacing using the tip electrode. Synchronization of additional myocardium carries the potential of further augmenting systolic function. The optimal multisite pacing configuration differed among patients, even if choice of pacing configuration did not markedly change which patients received greatest benefit. Pacing from the most distal and proximal electrodes most frequently yielded the highest LV dP/dt max values and provided results that closely approximated individualized optimal configurations. Such a setting, therefore, appears reasonable as an empirical or nominal configuration. The noted improvement from a pacing configuration including the most proximal electrode suggests an advantage of pacing from multiple zones in the long-axis of the LV. The potential benefit of an added pacing site several centimetres from the lead tip is also not inconsistent with recent observations linking apical positioning of a conventional LV lead with worse outcome from CRT. 20,30 However, selecting two LV pacing electrodes with shorter distance, either in the distal or proximal direction, appeared to reduce the benefit of multisite pacing.

Relationship of acute haemodynamic response to outcome
Although our study was limited to the acute setting, there is a growing body of literature suggesting that acute increases in LV dP/dt are associated with longer-term benefits. A recent study found that acute haemodynamic improvement, defined as a 10% increase in LV dP/dt max with pacing, predicted reverse remodelling of the dilated LV at 6-month follow-up, with 94% sensitivity and 64% specificity. 19 Another study evaluated the relationship between echocardiography-based LV dP/dt and clinical events in 53 patients. 31 Patients were classified as high responders (over 25% increase in dP/dt with CRT), low responders (0-25% increase in dP/dt with CRT), and non-responders (decrease in dP/dt with CRT). Overall, 89% of high responders, 59% of low responders, and 38% of non-responders were free of events at 12-month follow-up. 31 Prior investigations, such as the study by Duckett et al. 19 that used a 10% threshold increase in LV dP/dt max to predict remodelling, have largely relied on a single measurement of each pacing configuration. Each test was performed once and compared with an individual baseline measurement. Reanalysing our data using such traditional methodology, 9 of 19 (47%) patients experienced at least 10% increase in LV dP/dt max in the first recording of BiV pacing. With the best multisite pacing configuration, 15 of 19 (79%) patients had at least 10% increase in LV dP/dt max . Similarly, the group of patients that would qualify as high responders (over 25% increase in dP/dt) would increase from 3 of 19 (16%) with BiV pacing to 9 of 19 (47%) with multisite pacing. Taken together, these data suggest a potential for multisite pacing to increase the response rate to CRT when compared with BiV pacing. Table 4 summarizes the number and percentage of patients from the present study fitting into the various categories of response using the historical dP/dt analysis comparing each individual recording with an initial baseline measurement, and with the method developed for the present study comparing each recording with baseline BiV pacing immediately before and after, and taking an average from three recordings.

Limitations
While multisite pacing appeared to confer incremental haemodynamic benefit, the strength of this study's results is limited by the small number of patients. The study was not designed to address the question of lead positioning or potential adverse effects of apical pacing. All single-site BiV pacing used the tip LV electrode, rather than identifying the site of greatest haemodynamic response or selecting the most basal electrode. A small number of pre-selected multisite pacing configurations were tested to quantify precisely the acute haemodynamic effect, instead of identifying optimal electrode configuration and timing for each patient.
The unique protocol of this study makes it difficult to compare results with prior reports on LV dP/dt max . Finally, as the quadripolar LV lead was used for an acute pacing protocol and replaced with a permanent LV lead, no information is available on clinical outcomes or follow-up with multisite pacing.

Conclusions
A rigorous protocol with multiple recordings of each pacing configuration and repeated baseline measurements demonstrated that multisite pacing using two or more pacing sites in a quadripolar LV lead was superior to BiV pacing from the tip electrode in improving acute systolic function in 89% of patients. Although the optimal configuration varied among patients, pacing from the most distal and proximal electrodes most commonly yielded the maximum improvement in LV dP/dt. Multisite pacing, therefore, carries the potential to improve outcomes with CRT. Prospective follow-up studies are required to demonstrate clinical benefit.