Multicenter Randomized Controlled Crossover Trial Comparing Hemodynamic Optimization Against Echocardiographic Optimization of AV and VV Delay of Cardiac Resynchronization Therapy

Objectives BRAVO (British Randomized Controlled Trial of AV and VV Optimization) is a multicenter, randomized, crossover, noninferiority trial comparing echocardiographic optimization of atrioventricular (AV) and interventricular delay with a noninvasive blood pressure method. Background Cardiac resynchronization therapy including AV delay optimization confers clinical benefit, but the optimization requires time and expertise to perform. Methods This study randomized patients to echocardiographic optimization or hemodynamic optimization using multiple-replicate beat-by-beat noninvasive blood pressure at baseline; after 6 months, participants were crossed over to the other optimization arm of the trial. The primary outcome was exercise capacity, quantified as peak exercise oxygen uptake. Secondary outcome measures were echocardiographic left ventricular (LV) remodeling, quality-of-life scores, and N-terminal pro–B-type natriuretic peptide. Results A total of 401 patients were enrolled, the median age was 69 years, 78% of patients were men, and the New York Heart Association functional class was II in 84% and III in 16%. The primary endpoint, peak oxygen uptake, met the criterion for noninferiority (pnoninferiority = 0.0001), with no significant difference between the hemodynamically optimized arm and echocardiographically optimized arm of the trial (mean difference 0.1 ml/kg/min). Secondary endpoints for noninferiority were also met for symptoms (mean difference in Minnesota score 1; pnoninferiority = 0.002) and hormonal changes (mean change in N-terminal pro–B-type natriuretic peptide -10 pg/ml; pnoninferiority = 0.002). There was no significant difference in LV size (mean change in LV systolic dimension 1 mm; pnoninferiority < 0.001; LV diastolic dimension 0 mm; pnoninferiority <0.001). In 30% of patients the AV delay identified as optimal was more than 20 ms from the nominal setting of 120 ms. Conclusions Optimization of cardiac resynchronization therapy devices by using noninvasive blood pressure is noninferior to echocardiographic optimization. Therefore, noninvasive hemodynamic optimization is an acceptable alternative that has the potential to be automated and thus more easily implemented. (British Randomized Controlled Trial of AV and VV Optimization [BRAVO]; NCT01258829)

The beneficial effects of CRT stem ultimately from the changes in timing of cardiac activation.
The landmark CARE-HF (CArdiac REsynchronisation in Heart Failure) trial performed atrioventricular (AV) delay optimization after device implantation by using echocardiography. Echocardiography remains the most commonly recommended method for optimization (13,14). In this process the AV delay is set to maximize separation of the E and A waves on transmitral Doppler imaging. The precise AV and ventriculoventricular (VV) delays that maximize hemodynamic measurements vary among patients, perhaps because of the complexity of the disease and anatomic variations in lead position (15,16).
In current clinical practice, however, many patients do not undergo an echocardiographic optimization process, partly because of shortage of skilled staff time. Only 40% of physicians perform any form of optimization (13). There was initially doubt over the benefit of optimizing CRT (17), but there have been recent encouraging results on clinical outcomes using the AdaptivCRT algorithm (Medtronic, Minneapolis, Minnesota) (18) and methods dependent on implanted hemodynamic sensors (SonR, LivaNova, London, United Kingdom) (19). However, each of these algorithms is limited to a single manufacturer. The ideal alternative to the time-and labor-intensive echocardiographic method, with reproducibility that can be challenging (20), would be a manufacturerindependent and fully automatable method.
Instead of Doppler findings, an alternative target for optimization is blood pressure, which does not require expert judgment and therefore can be automated to accelerate analysis and save resources. It is important to take steps to minimize noise, and we have therefore developed an acquisition protocol that involves taking multiple repeated measurements between the tested setting and a reference setting (16). We have previously shown that when systolic blood pressure is used, it is most efficient to sample immediately after the change in pacemaker setting.

A B B R E V I A T I O N S A N D A C R O N Y M S
This is because even if no changes are made, the pressure tends to change away from the starting value with the passage of time in response to spontaneous physiological processes. In addition, when settings are changed to improve cardiac output there is a resulting reflex fall in peripheral resistance that returns the blood pressure toward the mean, even though improvements in cardiac output remain (15).
It is therefore better to sample blood pressure before this occurs. Patients were recruited from 19 centers in the United Kingdom. Patients were randomly allocated to an optimization method using an online system. They were followed up for 6 months and then crossed over to the other optimization method for a further 6 months of follow-up ( Figure 1).

INCLUSION AND EXCLUSION
CRITERIA. Study inclusion and exclusion criteria are shown in Table 1. OPTIMIZATION OF AV AND VV DELAY. We performed echocardiographic optimization of the AV delay using Doppler echocardiography of transmitral flow by using the iterative method as used in the CARE-HF trial (22). VV delay optimization was performed by maximizing aortic outflow tract aortic Doppler measurements.
Hemodynamic optimization of AV and VV delay was performed using multibeat averages acquired through noninvasive blood pressure measured using the Finometer device (Finapres Medical Systems, Amsterdam, the Netherlands). To obtain a narrow confidence interval (CI) we used a specific algorithm (23). This performs multiple alternations between a tested and reference AV delay and calculates the mean relative change in systolic blood pressure. It closely mirrors invasive optimization (24). We first calculated the AV optimum, and then we determined the VV optimum at that AV delay. Some previous studies have used LV dP/ dt max as a target for maximization. The BRAVO trial used systolic blood pressure because this can be acquired noninvasively or invasively with equal precision, and it reflects the external consequences of cardiac function. We have previously shown that this method is highly reproducible (25). Its noninvasive nature permits large numbers of replicates, which narrow the CI of the estimated optimum (26)       Step 1. Multiple replicate measurements of BP of tested setting against reference setting Step 2. Average change in BP relative to reference setting plotted against VV delay Step 3. Peak of parabolic curve selected as optimum Peak of parabola selected as AV optimum

Relative Systolic BP (mm Hg)
This same process is also performed with AV-delays to plot an AV optimization curve: Two replicate measurements of one tested setting... Continuous noninvasive beat-to-beat measurements are made through the Finometer (Finapres Medical Systems, Amsterdam, the Netherlands).

Reference
Multiple alternations are carried out between a tested atrioventricular (AV) or ventriculoventricular (VV) delay and reference AV or VV delay.
Blood pressures (BPs) before and after a transition in pacing state are measured as an average of 8 to 10 beats, as previously described (16). The average change in BP is plotted against AV or VV delay to fit a curve. The peak of the curve is used to select the optimum. LV ¼ left ventricular.

British RCT of AV and VV Optimization (BRAVO)
A U G U S T 2 0 1 9 : 1 4 0 7 -1 6 Declaration of Helsinki. The study was approved by the South West London Research Ethics Committee (3), and site-specific assessments were performed for each participating hospital. All patients gave prior written informed consent. The trial was registered with ClinicalTrials.gov (NCT01258829). STATISTICS. Distributions are described by their mean AE SD. NT-proBNP is expressed as log 10 NT-proBNP because it has a positive skew. Comparison between arms of the trial was performed by paired Student's t-test. Analysis was restricted to patients with before and after data for that variable. Differences between arms of the study are expressed as mean and 95% CI. The noninferiority margin for peak oxygen uptake (primary endpoint) was 0.75 ml/kg/min, for the Minnesota Living with Heart Failure score it was 4 points, for the 36-Item Short Form Health Survey version 2 physical component score it was 8.5, for NT-proBNP it was a fall of 0.062 log units (i.e., approximately a 13% decrease), for LV end-diastolic dimension it was 2 mm, and for LV end-systolic volume it was 2 mm. p noninferiority was calculated for these variables against their respective noninferiority margins.
The study sample size was chosen to have 90% power to detect a margin of equivalence of 0.75 ml/kg/min at the 5% significance level, on the basis of a published reproducibility of 2.4 ml/kg/min (30). On this basis, 177 participants per arm of the trial were required.

RESULTS
A total of 401 patients met the enrollment criteria and gave informed consent to participate in the BRAVO trial. Baseline characteristics are displayed in Table 2.
Patients' flow and study withdrawals are illustrated in The results met the primary pre-specified noninferiority criteria (p noninferiority ¼ 0.0001). There was no significant difference in peak oxygen uptake with noninvasive hemodynamic optimization compared with echocardiographic optimization, with a mean difference of 0.1 ml/kg/min (95% CI: À0.25 to þ0.41 ml/kg/min) (Figure 4, Table 3).
No significant difference was observed in the minute ventilation/carbon dioxide production slope, with a mean difference of 0.3 (95% CI: À0.6 to þ1.2; p ¼ 0.20).
There was also no difference in exercise duration, with a mean difference of À0.02 min (95% CI: À0.26 to þ0.21 min; p ¼ 0.76). See also Supplemental Table 2 for the statistics for order effects.
Left ventricular dimension. Differences in end-systolic and end-diastolic dimensions between the 2 arms Values are mean AE SD or %.
ACE ¼ angiotensin-converting enzyme; CRT-D ¼ cardiac resynchronization therapy defibrillator; CRT-P ¼ cardiac resynchronization therapy pacemaker.   Patients were randomized to either optimization method for 6 months before crossing over to the other arm of the trial for a further 6 months.
Investigations performed at each stage are listed. HF ¼ heart failure.
Whinnett et al.  (23). This is reflected in the BRAVO trial. Figure 5 shows that the distribution in AV optima between the 2 methods is different, but in both cases w70% of the optima are within AE20 ms of 120 ms. It is the w30% of patients whose optimal AV delay is more than 20 ms from optimum who are likely to have the most to gain from patient individualized optimization.
The modal AV optima are slightly shorter with echocardiographic than with hemodynamic optimization. However, the difference is small, at 40 ms. Because this occurs at the shallow portion of the AV delay optimization curve, this difference will produce only relatively small changes in cardiac output. These small differences in cardiac output are unlikely to be detected with the clinical outcome measures we used in this study. Hemodynamic optimization using beat-to-beat noninvasive blood pressure was noninferior to the conventional established method of echocardiographic optimization. DVO 2max ¼ change in peak oxygen uptake. Log 10 NT-proBNP, log 10 pg/ml 2.8 0.6 2.8 0.6 0 262 *Scores (with SD) are listed for all the primary and secondary outcomes measures for the study following 6 months of randomization in each arm. Hemodynamic optimizations were performed at a higher heart rate using atrial pacing rather than using atrial sensing, to improve signal-to-noise ratio. Rather than performing an optimization during atrial sensing, we programmed the sensed AV delay 60 ms shorter than the AV delay identified as optimal during atrial pacing. It is possible that we would have identified a different optimal-sensed AV delay had we performed optimization during atrial sensing. However, using this protocol we found hemodynamic optimization to be noninferior to echocardiographic optimization. In this study, there was a relatively high dropout rate (24% of those randomized); 250 patients completed both exercise tests. Of those participants completing the study, 13% failed to have both exercise tests performed. The numbers of dropouts do, however, appear to be balanced between the groups.
The relatively long duration of participation in the study, compared with other CRT studies, may explain the high frequency of dropout. The most common reason for dropout was an inability to carry out exercise testing for noncardiac reasons. Despite the high dropout rate, the study still met the predefined noninferiority threshold. This was because the midpoint value of the result was so close to neutral that the outer limit of the CI was well away from the noninferiority threshold.
This was a heterogeneous, "real-world" group of patients with heart failure who came from a In approximately one-third of patients, the optimal atrioventricular (AV) delay was found to be more than 40 ms longer or shorter than the commonly used nominal setting of 120 ms. These patients are likely to have the most to gain from AV delay optimization. KEY WORDS biventricular pacing, cardiac resynchronization therapy, echocardiographic optimization, heart failure, hemodynamic optimization, optimization APPENDIX For a supplemental figure and tables, please see the online version of this paper.