Cardiac resynchronization using fusion pacing during exercise

Fusion pacing requires correct timing of left ventricular pacing to right ventricular activation, although it is unclear whether this is maintained when atrioventricular (AV) conduction changes during exercise. We used cardiopulmonary exercise testing (CPET) to compare cardiac resynchronization therapy (CRT) using fusion pacing or fixed AV delays (AVD).


| INTRODUCTION
Cardiac resynchronization therapy (CRT) provides electrical and mechanical resynchronization in heart failure with a reduced ejection fraction with a widened QRS duration (QRSd), improving morbidity and mortality, 1 and having a key role in heart failure guidelines. 2 However, achieving maximal electrical resynchronization via optimized CRT entails more than a simple "on or off" function of biventricular (BiV) pacing and can depend on multiple patient and device factors, [3][4][5][6][7][8][9][10] one of which is selection of appropriate CRT pacing algorithms along with optimized individual programming.
Fusion pacing in CRT utilizes timing of the left ventricular (LV) pacing impulse to allow fusion of its wavefront with intrinsic conduction or the wavefront from right ventricular (RV) pacing. 11,12For this to occur effectively requires sinus rhythm and relatively preserved atrioventricular (AV) conduction, as well as optimization of the AV delay (AVD) to allow LV pacing to time correctly with either an atrial sensed or paced impulse.Manufacturer specific algorithms therefore allow dynamic assessment of intrinsic AV conduction, with subsequent adjustment of the AVD.
One such algorithm is SyncAV™ (Abbott, Illinois, USA).However, it is unknown to what extent such algorithms maintain fusion pacing during exercise and whether this could therefore influence exercise capacity.Intrinsic AV conduction commonly changes with exercise, normally shortening with increasing heart rate (HR), 13 but in heart failure patients, increasing HR can result in a greater degree of shortening than in normal controls 14 or conversely even in lengthening of conduction. 15This lack of a uniform dynamic response combined with the adjustment of the AVD by a fixed delta could result in loss of fusion and lack of effective CRT.
Cardiopulmonary exercise testing (CPET) is the established method of measuring exercise capacity via measurement of ventilation, oxygen consumption (V ̇O2 ) and carbon dioxide production.It is well validated in the heart failure population, 16,17 where quality of life and exercise capacity are inextricably linked, making CPET an ideal method to assess this.CRT is known to improve exercise capacity in responders, as assessed by CPET. 18,19To assess whether fusion pacing is maintained during exercise and whether this influences exercise capacity, we used CPET with 12 lead electrocardiography (ECG) monitoring to compare fusion pacing with conventional BiV pacing, in a doubleblinded, randomized, crossover study.

| METHODS
Local research ethics committee (REC 18/SC/0611) and institutional review board committee approval was granted, and the study complied with the Declaration of Helsinki.All study participants gave written informed consent.The study was registered on ClinicalTrials.gov(NCT03768804).

| Participants
Patients were recruited from those with a suitable CRT device under current follow up at Oxford University Hospitals NHS Foundation Trust.
Inclusion criteria were as follows: • Age ≥ 18 years and able to give informed consent.
• Able to exercise to perform CPET.
• Evidence of response to CRT, defined as functional improvement or LV remodeling on imaging at least 6 months post-CRT implant.
Exclusion criteria were as follows: • Any form of atrial arrhythmia.
• PR interval ≥ 250 ms or second-or third-degree heart block.
• Chronotropic incompetence, defined as use of a rate-response algorithm or ≥80% atrial pacing.
• Any concurrent condition contraindicating use of CPET.
• Pregnancy or breast feeding.
A power calculation estimated a sample size of 18 participants, assuming an α-value of .05 and a β-value of .8, for an estimated difference in mean QRSd between groups of 10 ms and an SD of 15 ms, and with paired samples.To allow for a 10% dropout rate, we aimed to recruit 20 participants.

| CRT programming
CRT programming was optimized via the Oxford method as previously reported. 20In brief, Q-LV was first tested on each pole, with the pole with the longest Q-LV used for all subsequent testing, unless there was phrenic nerve stimulation near the threshold, in which case next-best Q-LVs were used sequentially instead.AVDs were set to 120/120 ms, as this has previously been shown to be noninferior to a fusion-based AV optimization algorithm at rest in the SmartAV study. 21For ventriculo-ventricular (VV) optimization, the VV delay (VVD) (i.e., delay between RV and LV pacing) was altered in 10 ms intervals to achieve the narrowest QRSd and this delay was used for both programming groups (SyncAV™ only allows for BiV fusion pacing rather than LV only pacing).Other settings (e.g., multipoint pacing in n = 1) were left unchanged from pre-existing settings.Participants were then randomized to either fixed AVD or SyncAV™.For the fixed AVD setting, AVDs were also set to 120/ 120 ms, unless there was evidence of fusion (as assessed as an intrinsic sensed AV interval shorter than the programmed 120 ms AVD), in which case they were shortened in 10 ms intervals until there was no fusion (n = 1, AVD used 120/100 ms).

GREEN ET AL. | 147
SyncAV™ measures intrinsic conduction on the third of three intrinsic beats, which are conducted every 256 beats.A delta offset is then programmed (this is adjustable but nominally −50 ms).The final AVD is intrinsic conduction adjusted by the delta offset.For the SyncAV™ setting, the delta was adjusted in 10 ms intervals to give the narrowest QRSd.

| CPET
Calculation of peak oxygen consumption on exercise (peak V ̇O2 , or V ̇O2 PEAK ), first and second ventilatory thresholds (VT1 and VT2), peak respiratory exchange ratio (RER) and other CPET variables such as workload (Watts) and blood lactate level allows an objective and quantitative assessment of the peak exercise capacity and anaerobic threshold to be made.Furthermore, the combination of heart rate and oxygen uptake (O 2 pulse) provides a surrogate for stroke volume.
In addition, the BORG rating of perceived exertion (BORG-RPE) scale is a recognized method by which patients can report perceived exercise intensity. 22,23rticipants underwent CPET testing with 12 lead ECG monitoring at each visit programmed with either fixed AVD or SyncAV™.Visits were at least 1 week apart.Cardiopulmonary exercise test operators and participants were blinded to randomization.All tests were performed using a static bicycle ergometer (Ergoselect™, Ergoline) and a spiroergometry system (Metalyzer™ 3B, Cortex Medical), in conjunction with an oronasal V2 reusable facemask (Hans Rudolph).The system was calibrated to atmospheric pressure and with a calibration gas of known oxygen and carbon dioxide concentration before each test.A 15 watt step protocol was used, with 3 min stages and a 15 min recovery period.Data were recorded using commercial software (MetaSoft™ Studio Version 5.13.0,Cortex Medical) and simultaneous 12 lead ECG recordings were obtained using separate commercial software (Custo Med).
CPET data were smoothed using a rolling average of breath-bybreath data on the basis of a "rolling period" of 10 s.V ̇O2 at peak F I G U R E 1 Study protocol.AVD, atrioventricular delay; BiV, biventricular; CPET, cardiopulmonary exercise testing; CRT, cardiac resynchronization therapy.

exercise (V ̇O2
PEAK ) was defined as the maximum oxygen uptake achieved over a rolling 20 s period.The V ̇O2 at VT1 (V ̇O2

VT1
) was defined using the "V-slope" method of analyzing the slope of the function V ̇CO 2 versus V ̇O2 .This method recognizes that the slope of V ̇CO 2 /V ̇O2 (the RER) will have a value less than 1.0 at intensities below VT1, and that the slope will steepen to values of ≥1.0 when there is a contribution to the RER of CO 2 , which is derived from the buffering of metabolic acid by bicarbonate.The V ̇O2 at VT2 (V ̇O2

| ECG measurement
Assessment of exercise ECG intervals was performed automatically by the recording software (Custo-Med), with those beats suitable for analysis in the last 10 s of each exercise stage automatically selected.QRSd was calculated from the earliest and latest markers on any lead (global QRSd) as demonstrated in Figure 2. The placement of automatic ECG interval markers was manually checked for accuracy and all ECG markers were checked by a second experienced clinician, blinded to randomization.significance was taken at a p ≤ .05.Statistical analysis was performed with Graphpad™ Prism Version 9 (Graphpad Holdings).

| Baseline clinical characteristics
Twenty participants were recruited in total and baseline demographics and clinical details are given in Table 1.All were well optimized on standard heart failure medication and had evidence of either symptomatic improvement since CRT implant, and/or significant LV reverse remodeling with an increase in LV ejection fraction (LVEF) and decrease in LV internal diameter in diastole.

| CPET performance: Fixed AVD versus fusion pacing
Optimized fixed AVD and fusion pacing resulted in similar narrowing of QRSd from intrinsic rhythm at rest (

| Comparison of programming giving the narrowest versus widest QRSd at rest
The cohort was also analyzed by comparing whichever programming gave the narrowest versus widest QRSd at rest.Eight (40%) were narrower with fixed AVD, eight (40%) with fusion pacing, and in four patients (20%) there was no discernible difference.On this basis, median QRSd could be significantly shorter by 15 ms at rest and by 2 ms during exercise (Table 3 and Figure 3A,B).Peak cardiac efficiency and O 2 Pulse PEAK were improved in the narrowest QRSd group, although there was no significant difference in any other CPET parameters (Table 3 and Figure 3C,D).

| DISCUSSION
The main findings of this study are as follows: 1. Fusion pacing is maintained during exercise with no difference in exercise capacity between optimized BiV pacing using fixed AVD or fusion pacing.
2. Choosing whichever programming gives the narrowest QRSd at rest (which in 40% of patients required fusion pacing) also gives a narrower QRSd during exercise, higher O 2 Pulse PEAK , and higher peak cardiac efficiency.

| Use of CPET to evaluate pacing algorithms
CPET has been used previously to attempt to identify those patients most likely to respond to CRT and so aid patient selection. 24,25In addition, it has been used to assess changes in exercise capacity as a result of rate responsive pacing algorithms 26 and rate adaptive AVDs, 27 and the use of noninvasive hemodynamic monitoring to optimize AV and VVDs. 28However, to our knowledge, this is the first time that 12 lead ECG recordings and measures of exercise capacity have been combined by CPET in the evaluation of fusion pacing algorithms.This combination allows for the assessment of maintenance of effective fusion as determined by QRSd, as well as exercise capacity, and the finding in our study of increased O 2 Pulse PEAK and higher peak cardiac efficiency along with narrower QRSd at stress in those with narrowest QRSd at rest shows the utility of combining these parameters.This may therefore provide a novel method of initially assessing new dynamic pacing algorithms, before conducting large, expensive, randomized controlled trials.| 151 Several measures obtained during CPET, such as peak V ̇O2 , V ̇O2 at VT1, and exercise duration, 17,[29][30][31] have been shown to be predictors of mortality in heart failure.The optimum peak V ̇O2 cutoff to predict survival continues to be debated. 30In addition, heart rate recovery in the first minute after exercise has been shown to be a predictor of overall mortality. 32Therefore, it would seem reasonable that such parameters might be useful to assess dynamic pacing protocols, especially given that the key function of such protocols is to optimize CRT during changes in HR, as during exercise.

| Comparison between fusion pacing and fixed AVD
The finding that there is no significant overall difference in terms of exercise duration or QRSd during exercise between programming arms suggests that fusion pacing during exercise can be maintained.
This provides reassurance for its use in active individuals, who are likely to achieve significant increases in their HR.Using the programming available which gives the narrowest QRSd at rest, is associated with a narrower QRSd at stress, as well as higher O 2 Pulse PEAK (a surrogate of stroke volume) and higher peak cardiac efficiency.This is in keeping with the breadth of evidence that narrower QRSd in response to CRT improves hemodynamics and responder rates.Of note, 40% of our study participants have a narrower QRSd at rest with fusion pacing and so we would advocate its use where this can be achieved.This is the first study to show that a narrower QRSd during exercise correlates to better CPET performance and gives reassurance that CPET can detect the impact of a narrower QRSd on performance.
The newer SyncAV Plus™ algorithm from Abbott utilizes a delta which is a percentage of intrinsic AV conduction, rather than a fixed value.Although our study has not shown evidence of loss of fusion on exercise with SyncAV™, use of a percentage narrowing in this way with SyncAV Plus™ would intuitively seem to be more physiological and maintain better AV optimization.For example, in a patient with short intrinsic AV conduction at rest and then further shortening on exercise, it would help prevent AVDs being too short, leading to impairment of diastolic filling. 33In addition, as well as SyncAV™, the other main fusion pacing algorithm available is AdaptivCRT™ (Medtronic) as SmartAV™ (Boston Scientific) does not provide dynamic optimization.AdaptivCRT™ paces at ~70% of intrinsic conduction, but importantly can switch between LV-only pacing and BiV pacing (still with adaptive AVD, adjusted every minute).If intrinsic AV conduction is prolonged or if HR goes above 100 bpm, then it will automatically switch to optimized BiV pacing, reducing uncertainty over potential loss of fusion at higher heart rates.

| Limitations
This is a relatively small study for comparison of acute CRT programming changes.However, the crossover design and paired F I G U R E 3 Using whichever programming gave the narrowest QRS duration (QRSd) at rest (A) was also associated with a narrower QRSd during exercise over all stress stages (B), improved peak efficiency (C), and higher peak O 2 pulse (D), a marker for stroke volume.
nature of the data, with the same participants undergoing repeated testing in a double-blinded manner, adds considerable statistical power to the analyses and is a significant strength.The cohort size is therefore sufficient to draw the above conclusions based on our a priori power calculation.
Given that SyncAV™ only reassesses every 256 beats, it could be that the length of CPET test does not allow enough time for sufficient reassessments and readjustments to take place.However, where it was possible to identify the adjustments based on visualization of the intrinsic beats, a median of 7 were carried out per test during the exercise stages.This should be sufficient for any adjustments to influence overall QRSd.Furthermore, a greater proportion of these would be at stages with higher HRs, the subgroup in whom there would presumably be the greatest benefit to reassessment and readjustment.
Finally, recruited participants had to have either subjective or objective evidence of response to CRT, which could limit the applicability of these findings to patients at initial implant.However, responders are more likely to be undergoing exercise with resultant significant changes in their HR and so are likely to be the cohort for whom concerns about lack of fusion during exercise are greatest.

| Clinical perspectives
We provide evidence that fusion pacing is maintained during exercise without significantly impairing exercise capacity compared to fixed AVD.Using whichever programming gave the narrowest QRS duration at rest was associated with a narrower QRS duration during exercise, higher peak stroke volume and improved cardiac efficiency.
This supports the use of fusion pacing in the 40% of patients in whom this achieves the narrowest QRSd at rest.CPET is a useful tool for assessing new dynamic pacing algorithms in the heart failure population before large scale clinical trials.
a combination of the upward inflection in the relationship of pulmonary ventilation (VE) to V ̇CO 2 and the start of a sustained fall in end-tidal CO 2 .A maximal test is defined in this laboratory by a V ̇O2 , which does not rise over a 20 s period, despite a continued rise in workload.Venous blood was withdrawn from a cannula at baseline, end of 15 W stage, peak exercise and after 15 min of a recovery stage, to measure lactate concentration.Participants also indicated their BORG-RPE rating at the end of each stage.Tests were terminated when participants were unable to maintain increasing workload at a pedaling cadence of ≥60 rpm.Participants crossed over to perform CPET at the alternative programmed setting at the second visit.At the end of both visits, CRT settings were reset to the participant's pre-existing settings.A summary flow diagram of the study is shown in Figure 1.

2. 5
| Statistics Continuous data are presented as median (interquartile range [IQR]) and discrete data as "n" or %.Wilcoxon signed-rank tests (paired data) or two-way analysis of variance were performed.Statistical F I G U R E 2 Example of automated QRS duration analysis (Custo-Med), showing placing of auto-markers on the electrocardiography (ECG) (with the start and end of the QRS highlighted with black arrows) in the same participant with both SyncAV TM (left) and fixed atroventricular delay (AVD) (right) programming.
).The median SyncAV™ delta used was −50 (IQR: −50 to −40) ms, with a range of −30 to T A B L E 1 PEAK ), or in the BORG-RPE score at the median stage or peak of exercise.
T A B L E 3 CPET measurements based on programming with the narrowest QRSd at rest.