- Email: [email protected]
The atrioventricular delay of cardiac resynchronization can be optimized hemodynamically during exercise and predicted from resting measurements
The atrioventricular delay of cardiac resynchronization can be optimized hemodynamically during exercise and predicted from resting measurements Zachary I. Whinnett, BM BS, B Med Sci, Cathy Briscoe, BSC, MSC, Justin E. R. Davies, MBBS, MRCP, Keith Willson, MSc, MIPEM, Charlotte H. Manisty, MA, MRCP, D. Wyn Davies, MD, FRCP, FHRS, Nicholas S. Peters, MD, FRCP, FHRS, Prapa Kanagaratnam, PhD, MRCP, Alun D. Hughes, PhD, MBMS, Jamil Mayet, MD, MBA, FESC, FRCP, Darrel P. Francis, MD, MRCP From the International Centre for Circulatory Health, St. Mary’s Hospital and Imperial College, London, United Kingdom. BACKGROUND Atrioventricular (AV) optimization of cardiac resynchronization therapy (CRT) is typically calculated at rest. However, patients often become symptomatic during exercise. OBJECTIVE In this study, we use acute noninvasive hemodynamics to optimize the AV delay of CRT during exercise and investigate whether this exercise optimum can be predicted from a threephase resting model. METHODS In 20 patients with CRT, we adjusted the sensed AV delay while the patient exercised on a treadmill up to a heart rate of 100 bpm to identify the hemodynamically optimal value. Separately, at rest, by pacing with three different configurations and calculating the sensed-paced difference, we calculated an “expected” value for the exercise optimum.
made at rest during atrial-sensed pacing showed a poorer correlation with exercise (r ⫽ 0.64, mean difference ⫾ SD ⫽ 2.2 ⫾ 24 ms). The three-phase resting model allows improved exercise hemodynamics to be achieved. Programming according to the three-phase resting model yields an exercise blood pressure of only 0.5 mmHg (⫾1.4 mmHg; P ⫽ NS) less than the true exercise optimum, whereas programming the resting sensed optimum yields an exercise blood pressure of 1.4 mmHg (⫾2.2 mmHg, P ⫽ .02) less than the true optimum. CONCLUSIONS Using acute noninvasive hemodynamics and a protocol of alternations, it is possible to optimize the AV delay of cardiac resynchronization devices even while a patient exercises. In clinical practice, the exercise optimum AV delay could be determined from three phases of resting measurements, without performing exercise.
RESULTS It was possible to perform AV delay optimization while a patient exercised. The resting three-phase model correlated well with the actual exercise optimal AV delay (r ⫽ 0.85, mean difference ⫾ standard deviation [SD] ⫽ 3.7 ⫾ 17 ms). Simply using measurements
KEYWORDS Atrioventricular delay optimization; Exercise; Optimization of cardiac resynchronization therapy; Biventricular pacing
Introduction
standard clinical methods. As a result, most centers do not routinely optimize these settings, and those that do usually limit the process to resting heart rates while the patient remains inactive. It is not known whether the optimal AV delay determined at rest corresponds to the optimal delay determined during exercise.7 It is during exercise that it is likely to be the most important to program the AV delay to the most efficient setting for an individual patient. One technique, which has potential to permit optimization during exercise, is beat-by-beat noninvasive blood pressure, measured using a calibrated digital photoplethysmograph such as the Finapres or Finometer. We have previously used this method to identify the AV delay that corresponds to peak hemodynamics in the resting state, with the heart rate either left in the normal resting state or elevated by atrial pacing.4 In this study we set out to investigate the following questions:
In heart failure, inadequate cardiac output response to exercise is a key driver of symptoms. In patients with badly coordinated ventricular and atrioventricular (AV) timings, cardiac resynchronization therapy (CRT) has been shown to improve these timings and thereby improve acute hemodynamics1– 4 and consequently symptoms, morbidity, and ultimately mortality.5,6 Exactly which timings of AV and interventricular delay are best for an individual patient is difficult to establish with This study was supported by the British Heart Foundation, Medtronic, and the Coronary Flow Trust. ZW (FS/05/068), JD (FS/05/006), and DF (FS/04/079) are British Heart Foundation fellows. CM (077049/Z/05/Z) is funded by the Welcome Trust. KW received support from the Foundation for Circulatory Health. Our institution has filed a patent on some of the methods described in this manuscript. Address reprint requests and correspondence: Dr. Zachary Whinnett, International Centre for Circulatory Health, St. Mary’s Hospital and Imperial College, 59-61 North Wharf Road, W2 1LA, UK. E-mail address: [email protected]. (Received July 19, 2007; accepted November 18, 2007.)
(Heart Rhythm 2008;5:378 –386) © 2008 Heart Rhythm Society. All rights reserved.
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2007.11.019
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
First, is it possible to perform optimization of the AV delay of CRT during exercise by using acute noninvasive hemodynamics? Second, how different is the AV delay that is determined during exercise from that determined at rest with a resting heart rate? Third, does measurement of the individual sensed-paced difference at resting heart rate and atrial pacing at higher heart rates allow improved determination of exercise optimum AV delay from resting measurements?
Methods Subjects Twenty outpatients with biventricular pacemakers or biventricular defibrillators implanted for clinical indications (New York Heart Association [NYHA] III or IV heart failure, QRS ⬎120 ms, maximal medical therapy) were enrolled into this study. Patients were excluded if they were not in sinus rhythm, were pacing dependent, or were not able to walk on a treadmill. Patients gave informed consent for this study, which was approved by the local ethics committee.
Measurements Data acquisition Noninvasive finger arterial pressure measurements were made using a Finometer (Finapres Medical Systems, Amsterdam, Holland). This technique, developed by Penˇáz8 and Wesseling et al9 uses a cuff that is placed around the finger, a built-in photoelectric plethysmograph, and a volumeclamp circuit that dynamically follows arterial pressure. This technique is well validated for measuring instantaneous changes in blood pressure.10 –14 An electrocardiogram signal was recorded using the Hewlett-Packard 78351A monitor (Hewlett-Packard Co, Andower, Massachusetts, USA). Analog output feeds of these signals were taken via a National Instruments DAQCard AI-16E-4 (National Instruments, Austin, TX) and acquired in digital form using Labview (National Instruments). They were analyzed offline with custom software based on the Matlab platform (MathWorks, Natick, MA).15
Measurement of relative change in blood pressure across different AV delays Beat-to-beat blood pressure was recorded during adjustment of the AV delay of the subject’s biventricular pacemaker. As previously described, the effect of background noise in the blood pressure trace was reduced by calculating the relative change in systolic blood pressure (SBPrel).4 This was done by comparing the tested AV delay to a reference AV delay and performing repeated alternations to obtain multiple transitions and yield at least six replicate measurements for each SBPrel. We calculated the relative change in blood pressure by subtracting the mean of the 10 beats immediately after the transition from the mean of the 10 beats immediately before the transition.4,17 We combined these to obtain, for each AV-tested delay, a mean SBPrel.
379
Our selection of systolic blood pressure was based on the reasoning that it was composed of two fundamental components: mean blood pressure (dependent on cardiac output and total peripheral resistance) and a fraction of pulse pressure (dependent on stroke volume and arterial compliance). Because each of these components would increase if cardiac function increased, we reasoned that there would be two reasons for systolic blood pressure to increase, and therefore it may give a particularly clear signal. We have studied the signal-to-noise ratios of a series of noninvasively measured hemodynamic parameters and have indeed confirmed that systolic blood pressure provided the best combination of efficiency and reproducibility when compared with other noninvasively measured hemodynamic parameters.16 We used mean rather than median because the type of noise we were trying to attenuate was likely to be randomly distributed (e.g., baseline wander due to respiration, physical movement, or other such processes), and the statistical principle of the central limits theorem ensures that the variability of the mean would be reduced in proportion to the square root of the number of transition measurements made. The advantage of the mean over the median is that every single transition measurement contributes something to the final value, and its statistical properties on increases in sample size are well understood and desirable. SBPrel was measured in the manner described above for a range of AV delays, which were 40, 80, 120, 140, 160, 200, 240, 280, 320, and 350 ms. In practice, for each subject we did not study AV delays beyond the point at which conduction became purely intrinsic. The interventricular delay was left at 0 ms or as close to this as the pacemaker allowed. We used parabolic interpolation to select the hemodynamic optimal AV delay for each optimization phase. We have previously demonstrated that the effects on acute hemodynamics of changing AV delay are curvilinear rather than linear and fit closely to a parabola.17
AV delay optimization phases Optimization was initially performed at rest while the patient sat on a couch. We used a three-phase resting model to predict the optimal exercise AV delay. This incorporated hemodynamic optimization of AV delay under three different conditions: Atrial-sensed biventricular pacing at resting heart rate, Atrial pacing at 5 bpm above the patient’s resting rate, Atrial pacing at a heart rate of 100 bpm. The patient then underwent direct measurement of optimal AV delay during exercise. The patient exercised on a treadmill, and workload was adjusted to maintain his or her heart rate between 100 and 110 bpm. The exercise AV delay optimum (with atrial sensing) was then determined by a process similar to that carried out at rest.
Differences between atrial-sensed and atrial-paced biventricular pacing The three-phase resting model must take into account the differences between atrial-sensed (P-synchronous) and atri-
380
Heart Rhythm, Vol 5, No 3, March 2008
Figure 1 Cartoon explaining why the paced AV optimum is longer than the sensed AV optimum. The left-sided AV delay has the predominant effect on cardiac output and systemic hemodynamics. A: In this example, the left-sided AV delay yielding optimal systemic hemodynamics is 110 ms. If this is achieved by atrial sensing, we need to program to allow for the atrial sensing delay (here 30 ms) and the interatrial delay (here 70 ms). This hypothetical patient therefore required a sensed AV delay of 110 ⫹ 70 ⫺ 30 ⫽ 150 ms. B: If the same AV delay is programmed for the paced AV delay, then this results in a nonoptimal left-sided AV delay (in this example, 40 ms). This difference occurs as a result of the atrial pacing latency (here 20 ms) and longer time taken for the atrial signal to travel between the right and left atria manifest as a longer interatrial delay (here 90 ms). Here the leftsided AV delay is 150 ⫺ 20 ⫺ 90 ⫽ 40 ms. C: Therefore, to obtain an optimal leftsided AV delay, the paced AV delay should be programmed to a longer value to account for the atrial pacing latency and longer interatrial delay. This patient requires a programmed paced AV delay of 110 ⫹ 90 ⫹ 20 ⫽ 220 ms.
al-paced (AV sequential pacing) AV delays. First, it is important to remember that when the AV delay is programmed on the pacemaker, this represents the right-sided AV delay. It is, however, the mechanical contraction delay between the left atrium and left ventricle that has the predominant effect on systemic hemodynamics.18 –21 There are three principal differences between atrialsensed pacing and AV sequential pacing that need to be taken into account: the atrial sensing delay, atrial pacing latency, and differences in interatrial delay (Figure 1).
Atrial sensing delay During atrial-sensed pacing, the programmed right-sided AV delay is shorter than the actual right-sided AV delay. This is because the pacemaker only detects the onset of atrial depolarization once the atrial electrical activity has propagated from the site of origin (sinoatrial node) to the atrial electrodes where it can be detected as a near-field signal. This results in a time lag in the detection of the onset of atrial depolarisation: the atrial sensing delay (Figure 1A).
Atrial pacing latency When pacing from the right atrium, there is a delay from the time of delivery of the pacing stimulus to the onset of atrial
depolarization, the so-called atrial pacing latency.22 This means that the actual right-sided AV delay is slightly shorter than the programmed AV delay (Figure 1B).
Interatrial delay The time taken for activation of the left atrium differs depending on whether activation is initiated by intrinsic atrial activity or atrial pacing. Typically, the interatrial delay is longer for an atrial-paced beat than for intrinsic activation. The longer interatrial delay may be because of less efficient interatrial conduction. The difference in intra-atrial delay between paced and sensed activation appears to be patient specific,18 –20 which is likely to reflect differences in underlying conduction, lead position, atrial size, and function. Intra-atrial conduction time appears to remain relatively constant when heart rate is altered.23
Three-phase resting model for predicting the optimal AV delay during exercise by pacing at different heart rates at rest We hypothesized that in individual patients the optimal AV delay identified during AV delay optimization may differ depending on whether optimization is performed at rest or during exercise. We believed it might be possible to identify
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
the exercise optimal sensed AV delay at rest if the heart rate was increased to the exercise level by atrial pacing and an adjustment was made for the difference between the atrialpaced AV delay and the atrial-sensed AV delay, which we call (for simplicity) the “sensed-paced difference.” We calculated the sensed-paced difference by determining the difference between the optimal AV delay determined while atrially pacing at 5 bpm above the patient’s resting heart rate and that determined during atrial-sensed pacing at a resting heart rate. We then subtracted this sensed-paced difference from the optimal AV delay determined while atrially pacing at a heart rate of 100 bpm. This gave us our estimate of optimal AV delay determined during exercise using the three-phase resting exercise model. Estimated exercise optimal AV delay A-paced A-paced ⫺ 共AVoptimumresting ⫽ AVoptimum100bpm rate A-sensed ⫺ AVoptimumresting rate) A⫺paced AV optimum100bpm : optimal AVD delay determined while atrially pacing at a rate of 100 bpm. A⫺paced AV optimumresting rate: optimal AV delay determined while atrially pacing at a rate 5 bpm above the patient’s intrinsic rate. A⫺paced AV optimumresting rate: optimal AV delay determined during atrial-sensed pacing at intrinsic rate.
Statistics The SBPrel value was determined for each tested combination of AV delay in relation to a reference AV delay (120 ms) by taking the mean of observed blood pressure changes from at least six individual transitions. Data are given as means ⫾ standard deviation (SD) of the series of replicate transitions. Comparisons were made on the basis of mean differences ⫾ SD and by Pearson correlation coefficients. P-values for comparison of means were calculated using a Student’s paired t-test, and comparisons of proportions were made using Fisher’s exact test. P ⬍.05 was considered statistically significant. The statistical package Statview 5.0 (SAS Institute, Cary, NC) was used for analysis.
Results Patient characteristics Twenty patients with biventricular pacemakers or biventricular ICDs were enrolled into this study a mean of 10 months after implantation. Ten were male and 10 were female, with an age range of 46 – 82 years (mean 68.3 years). The cause of heart failure was ischemic in 11 and idiopathic dilated in nine. Mean systolic blood pressure by sphygmomanometer was 118 ⫾ 18 mmHg. The mean ejection fraction of the patients at the time of the study was 30% ⫾ 5%. At the time of the study, one patient was in NYHA class I, 13 were in NYHA II, and six were in NYHA III.
Identification of a hemodynamic peak Despite patient movement during exercise, all patients showed a clear hemodynamic peak, with the use of the
381
protocol of repeated alternations. All patients were able to walk on the treadmill and to increase their heart rate to between 100 and 110 bpm, although some patients undergoing exercise optimization needed rest periods between different tested AV delays. The AV delay identified as optimal was specific to the individual patient. The optimal atrial-paced AV delay was on average longer than the optimal atrial-sensed AV delay. This occurred both at a resting rate (185 ms, range 150 –240 ms; versus 120 ms, range 78 –163 ms; P ⬍.0001) and at a higher rate (186 ms, range 150 –240 ms; versus 118 ms range 79 –200 ms; P ⬍.0001). All data from individual subjects are shown in Table 1. The AV delay identified as optimal is displayed for each individual patient, for the four different pacing configurations: resting rate atrial-pacing (paced just above the resting heart rate), resting rate atrial-sensing, higher rate atrial-pacing, and during exercise with atrial-sensing. The sensed-paced difference was calculated at rest and used to provide a prediction of the exercise optimal AV delay by subtracting this from the higher rate atrial-paced optimal AV delay (estimated exercise optimal AV delay (Estimated A⫺paced ⫺ (AV exercise optimal AV delay ⫽ AV optimum100bpm A⫺paced A⫺sensed optimumresting rate ⫺ AV optimumresting rate)). The means and standard deviations are also shown.
Effect of exercise on optimal AV delay In 11 patients, the optimal AV delay was shorter during exercise than at rest, while in eight patients it was longer. In one patient, the optimal AV delay remained the same. There was no significant difference in resting PR interval between the patients in whom AV delay lengthened and those in whom it shortened (mean intrinsic AV delay 201 and 200 ms, respectively; P ⫽ NS). In addition, there was no difference in etiology of the heart failure or the patients’ functional class (P ⫽ NS by Fisher’s exact test and 2-test, respectively).
Assessment of signal-to-noise ratio To make an assessment of the amount of noise within the signal, we calculated the signal-to-noise ratio and displayed the maximum standard error for each of the four different pacing configurations (Table 2). We calculated the signalto-noise ratio by dividing the range of values obtained for different pacing settings by the standard error of the measurements at each pacing setting. The signal-to-noise ratio was similar for atrial sensing at rest and during exercise and was higher when measurements were made with atrial pacing at both resting and higher heart rates. Exercise conditions generated the highest values of standard error, which is likely to be related to the physical movement of the patient. The resting, atrial-sensed, state also had a fairly high standard error, presumably due to the presence of variation (however small) in heart rate from beat to beat. The two paced states (resting rate and higher rate) had lower standard errors, which may reflect the greater regularity of the cardiac rhythm.
382
Heart Rhythm, Vol 5, No 3, March 2008
Table 1 The AV delay in milliseconds identified as optimal, for each individual patient for each of the four different pacing configurations Patient no.
Resting rate atrial-paced optimum
Resting rate atrial-sensed optimum
Higher rate atrial-paced optimum
Predicted exercise optimum
Actual exercise optimum
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean SD
160 160 167 200 197 168 240 169 160 171 150 196 200 177 200 171 200 240 169 200 185 25
163 78 120 133 145 80 120 108 82 140 105 120 134 134 111 140 118 156 97 124 120 24
200 162 200 213 189 150 240 178 164 193 164 160 189 181 192 154 180 209 200 198 186 23
203 79 153 146 137 62 120 117 86 162 119 93 123 138 103 123 98 125 128 122 121 31
200 90 112 147 129 80 138 140 90 158 79 93 120 133 107 99 86 116 127 120 118 30
Comparison of three-phase resting model with actual exercise optimum
Comparison of sensed paced difference between resting and higher heart rate
Prediction of the exercise optimum by the three-phase approach (not requiring exercise) showed good agreement: mean difference ⫾ SD ⫽ 3.7 ⫾ 17 ms: r ⫽ 0.85 (R2 ⫽ 0.73; Figure 3, left panel). The alternative of simply using the atrial-sensed resting-rate optimum as an estimate of the likely exercise optimum was less reliable: mean difference ⫾ SD ⫽ 2.2 ⫾ 24 ms: r ⫽ 0.64 (R2 ⫽ 0.41, Figure 3, right panel). The Bland-Altman plots are shown in Figure 4.
The sensed paced difference at resting heart rate and at higher heart rate was calculated by measuring the difference between the AV delay identified as optimal at rest and higher heart rates. The mean sensed paced difference at rest was 64 (⫾27) ms, and at higher heart rates this was 68 (⫾24) ms (R2 ⫽ 0.62 and a 95% limit of agreement of ⫾36 ms between the resting and high-rate sensed-paced differences).
Hemodynamic differences between using the three-phase resting model and using atrial-sensed pacing at resting rate for selection of the programmed AV delay during exercise We assessed whether programming the AV delay based on the three-phase resting optimization model or instead simply based on the atrial-sensed resting optimization results in changes in hemodynamics during exercise. We calculated the change in acute blood pressure, from the hemodynamic response curve during exercise, which would occur as a result of programming the AV delay not to the true exercise hemodynamic optimum but rather to the three-phase predicted optimum or simply to the optimum determined with resting heart rate atrial-sensed pacing. Simply using the measurements made at resting heart rate with atrial-sensed biventricular pacing yielded an SBPrel on exercise that was 1.4 mmHg (⫾2.2 mmHg, P ⫽ .02) lower than using the actual exercise optimum. Using the three-phase resting prediction yielded an SBPrel on exercise that was not significantly lower (0.5 ⫾ 1.4 mmHg; P ⫽ NS) than the actual exercise optimum.
Discussion In this study, we demonstrate a method for optimization of the AV delay of CRT devices that can be performed during exercise. Simply optimizing at resting heart rates with atrialsensed biventricular pacing selects an AV delay that correlates reasonably well with the exercise optimum. But agreement with the true exercise optimum can be improved significantly by using a predictive process, which involves three phases conducted with the patient at rest. Using this three-phase resting prediction, there appears to be a significant hemodynamic improvement over standard optimization with atrial-sensed pacing at resting heart rates.
AV delay optimization during exercise With an algorithm of multiple alternations to and from a reference delay, it is possible to reliably perform optimization of the AV delay of CRT during exercise by using noninvasive hemodynamics. Exercise represents the state in which most patients with heart failure become symptomatic. As heart rate increases, the time available for filling within the cardiac cycle becomes progressively more limited. This makes the balance
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
383
Table 2 The highest standard error and signal-to-noise ratio are displayed for each patient for each of the four different pacing configurations
during echocardiographic optimization has previously been investigated.24 Other potential methods for optimization during exercise include standard finger plethysmography.25
Patient No.
Feasibility of programming optimal AV delay
Resting rate atrial sensed
Highest standard error 1 2.6 2 1.0 3 1.9 4 0.6 5 1.3 6 0.9 7 3.0 8 8.1 9 2.3 10 1.6 11 3.1 12 1.6 13 1.6 14 2.6 15 3.5 16 2.7 17 2.4 18 2.4 19 2.0 20 2.8 Median 2.3 Mean 2.4 Signal to noise ratio 1 9.7 2 1.9 3 6.5 4 16.4 5 8.8 6 7.2 7 2.7 8 2.2 9 5.8 10 2.1 11 3.3 12 5.7 13 2.3 14 15.1 15 1.5 16 9.2 17 2.4 18 9.7 19 5.4 20 2.7 Median 5.5 Mean 6.0
Resting rate atrial paced
Higher rate atrial paced
Exercise
3.2 2.0 2.8 1.9 1.5 1.0 1.2 4.1 1.9 2.7 2.9 2.4 3.5 3.5 3.4 2.2 0.9 2.4 2.6 1.4 2.4 2.4
2.7 2.6 2.4 2.9 1.3 0.8 2.4 2.9 2.1 2.3 2.7 2.2 4.1 1.7 3.8 3.8 2.0 2.2 1.4 1.4 2.3 2.4
5.0 1.6 2.3 2.9 2.2 1.9 7.0 6.8 3.4 1.9 5.6 1.8 3.9 1.6 6.7 6.3 0.8 4.5 2.0 1.6 2.6 3.5
8.9 13.9 10.9 10.4 11.5 13.6 26.2 10.7 6.2 7.8 10.7 7.4 8.2 14.2 2.9 14.6 13.4 20.0 15.0 8.7 10.8 11.8
26.5 19.7 19.1 9.5 15.8 24.4 14.6 17.1 14.5 17.5 15.6 10.7 8.5 33.1 10.2 14.4 16.6 24.3 19.6 17.6 16.9 17.5
8.3 1.6 9.7 9.4 6.5 3.7 4.1 3.1 3.6 16.8 3.0 4.8 5.4 2.3 1.4 5.1 5.1 4.5 6.1 9.9 4.9 5.7
between the timings of the different phases of the cardiac cycle even more critical. To determine the most efficient AV delay in the state when it is most valuable for the patient, there needs to be a method for determining the optimum during exercise. The effect of posture during optimization is also an important consideration. Venous return and filling of the heart is significantly affected between the supine and upright positions. This is not taken into account during standard echo optimization, which is typically performed at rest in a nonupright position. The effect of posture and exercise
Current pacemakers have limited flexibility for programming the AV delay at different heart rates. Most algorithms allow only for shortening of the AV delay with exercise. It would be useful if pacemaker algorithms were developed to allow the AV delay to be programmed for a range of different heart rates during both atrial sensing and atrial pacing. Overall, in the group of patients studied, there was a shortening in the AV delay at which intrinsic conduction occurred with atrial-sensed pacing during exercise compared with rest. This observation did not, however, impact on the feasibility of programming the predicted optimal AV delay. In all patients in our study the predicted optimal exercise AV delay was shorter than the AV delay at which intrinsic conduction occurred during exercise.
Effect of exercise on AV delay optimum Our data demonstrate that optimal AV delay is dependent on heart rate, with the optimum identified at resting rate rarely corresponding with that identified during exercise at higher heart rates. A previous study using echocardiographic AV delay optimization (aortic outflow tract velocity time integral) immediately after exercise also showed a change in optimal AV with change in heart rate.24 In that study, across all patients, there was a lengthening of optimal AV delay with exercise, although data for response in individual patients were not available.24 In our patients, the effect of exercise on the optimal value of AV delay appeared to be specific to the individual patient. In some patients, the optimal AV delay was shorter on exercise, while in other patients the optimal AV delay lengthened. The response does not appear to be dependent on the individuals’ intrinsic PR interval or their severity or etiology of heart failure. We used a different method that is highly sensitive to even small changes in acute hemodynamics as a result of changing AV delay and is highly reproducible. The echocardiographic measurements had to be made after exercise when patients had a variable heart rate. Overall, it is likely that the response is dependent on a number of different factors, for example, the conduction properties of the left atrium, filling pressure, atrial size, and function. Patients are likely to be best served by individual optimization.
Comparison of the three-phase resting prediction at rest and exercise The three-phase resting prediction assumes that the sensed-paced difference in optimal AV delay is the same for resting heart rate as it is for higher rates. Our data suggest that this is a valid extrapolation, and therefore it may not be necessary in clinical practice to optimize a patient during exercise. Being able to accurately predict the exercise optimal AV delay from resting measures makes it clinically more
384
Heart Rhythm, Vol 5, No 3, March 2008
Figure 2 The sensed-paced difference, calculated at higher rate and at resting rate in an individual patient. Optimal paced AV delay is longer than optimal sensed AV delay, at both resting and higher rates. The sensed-paced difference is similar at the two rates. The exercise AV optimum (i.e., higher rate sensed) can therefore be predicted from the other three measurements, which do not require exercise to be undertaken. Predicted exercise optimal AV delay ⫽ 196 ⫺ (169 ⫺ 97) ⫽ 123 ms.
feasible to program the exercise optimal AV delays in most patients. Optimization during actual exercise may remain a theoretical gold standard, but the level of patient cooperation
Figure 3 Comparison of the optimal AV delay identified during exercise with the threephase resting prediction of exercise optimum and optimization at resting heart rate only. Data are derived from separate individuals. The dotted line shows the line of identity. There is a good agreement between the actual optimal AV delay determined by performing optimization during exercise and the predicted optimal AV delay determined using the threephase resting prediction (left panel). This means it is potentially possible to predict the optimal exercise AV delay without exercising the patient. In comparison, optimization performed at a resting heart rate only does not agree as well with the exercise-determined optimal AV delay (right panel).
and the number of skilled staff required mean that it may be difficult to implement into routine clinical practice in all centers.
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
385
Cross-sectional observational data suggest a 4% (95% CI ⫽ 3%–5%) relative increase in mortality per 1 mmHg decline in systolic blood pressure in outpatients with chronic heart failure.26 The lesson from these studies is that CRT improves cardiac output without deleterious effects on myocardial oxygen consumption, that is, it improves efficiency in cardiac function. Even 1- or 2-percentage-point differences in cardiac function (when achieved without cost in myocardial work) can improve outcomes.
Comparison of sensed-paced difference between resting and higher heart rate We compared the sensed-paced difference at resting and higher heart rates as we had postulated that the difference would remain relatively constant. The level of agreement we identified (correlation coefficient of 0.8) is as close as can reasonably be expected, given the fact that each of the two elements is itself a difference and is therefore subject to two sources of error: such errors necessarily attenuate the observed correlation coefficient. Figure 4 Bland Altman plots showing the comparison of the optimal AV delay identified during exercise with the three-phase resting prediction of exercise optimum (top) and optimization at resting heart rate only (bottom). Data are derived from separate individuals. The solid line shows the line of identity. The 95% limit of agreement for the comparison between the predicted and actual exercise optimum is ⫾36 ms. For the comparison between actual exercise and optimization at resting heart rate, the 95% limits of agreement are ⫺48 to 44 ms.
Hemodynamic consequence of using the three-phase resting prediction or atrial-sensed pacing at resting rate to select the programmed AV delay during exercise While initially a difference of 1.4 mmHg in acute blood pressure change may seem small, it is not negligible in relation to the size of the change seen with CRT itself, an intervention that was ultimately proven to improve not only symptoms but also survival. Onset of biventricular pacing resulted in a mean increase in acute aortic systolic pulse pressure of 2 mmHg in a previous invasive study.1 This is consistent with the landmark trials with CRT that showed an increase in blood pressure, paralleling the improvements in symptoms and survival, for patients entered into the device arm. For example, in the COMPANION trial, patients in the resynchronization arm initially gained approximately 2 mmHg (confidence interval [CI] not published) in systolic blood pressure (in comparison with the control arm) and went on to have an 18% relative reduction (95% CI ⫽ 1% increase to 42% reduction) in the combined endpoint of morbidity and mortality.5 Similarly, the CARE-HF trial showed, at 3 months, that the increment in blood pressure attributable to being in the device arm was 5.8 mmHg (95% CI ⫽ 3.5– 8.2 mmHg) and the mortality reduction was 37% (95% CI ⫽ 23%– 49%).6
Potential for automation The intention was to provide a protocol in which analysis could later be made entirely automatic. To do this, we placed the burden of establishing data quality at the time of acquisition (that is, before all stages of numerical analysis). During data acquisition, if it became evident that the Finometer signal was poor or subject to large artifact, the recording was terminated and restarted afresh after checking equipment positioning. We estimate that ⬍10% of hemodynamic recordings during exercise, once begun, had to be terminated due to motion-related artifact. In some patients, ectopy was a significant source of artifact. This was also addressed at the time of data collection by avoiding performing transitions in the few seconds immediately after any ectopics and by adding an extra transition on every occasion when there were ectopics in the few seconds immediately after a transition. (In the latter situation, both the original, noisy transition and the subsequent additional transition were included in the data analysis, so that no inadvertent bias during analysis could have affected the numerical results.) The increase in number of transitions reduces the effect of the noise on the average. In Table 2, we present the signal-to-noise ratio on exercise and rest. All recordings that were thought, at the time of acquisition, to be of adequate quality were put through the analysis and used in entirety, without any manual editing.
Study limitations In this study, we only assessed one heart rate during exercise. We do not have data on how the optimal AV delay varies within individual patients at multiple different exercise heart rates. Further studies would be required to clarify this. However, our finding that the three-phase resting prediction is a suitable proxy for actual exercise could in
386 principle be extended to a range of exercise heart rates. This study has a size adequate to answer its primary question, but subanalyses to look at trends in lengthening or shortening of AV interval during exercise may not be sufficiently powered to provide reliable information. A potential limitation to our study is that we had a relatively high prevalence of class I and II patients. This occurred because these patients were willing to undergo a protocol that required prolonged walking on a treadmill for the experiment of optimization during exercise. In more symptomatic patients there may be a higher prevalence of optimization curves where there is a relatively flat hemodynamic response over a wide range of AV delays. It could be argued that any of the AV delays in these flat regions could be considered optimal, as they all yield the same hemodynamic response. Intentional inclusion of a wide range of AV delays, including extreme ones unlikely to be optimal, allowed the parabolic interpolation to detect declines at those extremes and thereby identify a central optimal value among the flat areas. We have previously found that at resting heart rates the hemodynamic response curves tend to be flatter compared with higher heart rates.17 In developing our three-phase pacing model to predict the exercise optimal AV delay, we did not take into account the effect of posture. Measurements at rest were made with the patients in the sitting position. The systemic resistance seen by the heart pump is probably different when atrial pacing at high rates while sitting compared with the resistance seen when standing and actually exercising. This different arterial impedance may cause the readings in absolute systolic blood pressure to be different, which may affect the accuracy of the model. However, to minimize the effect of changes in absolute pressure on the reliability of the optimization method, we calculated the relative change in systolic blood pressure.
Conclusions Using acute noninvasive hemodynamics to calculate mean relative change in blood pressure, it is possible to optimize the AV delay of cardiac resynchronization devices while a patient exercises. Optimization at a resting heart rate during atrial-sensed biventricular pacing correlates with the optimum identified during exercise. However, using a threephase resting model, a closer prediction of the actual exercise optimum can be made, which has significant acute hemodynamic benefits for individual patients. (Figure 2).
References 1. Auricchio A, Stellbrink C, Block M, et al. Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. Circulation 1999;99:2993–3001. 2. Perego GB, Chianca R, Facchini M, et al. Simultaneous vs sequential biventricular pacing in dilated cardiomyopathy: an acute hemodynamic study. Eur J Heart Fail;2003:305–313.
Heart Rhythm, Vol 5, No 3, March 2008 3. Van Gelder BM, Bracke FA, Meijer A, et al. Effect of optimizing the VV interval on left ventricular contractility in cardiac resynchronisation therapy. Am J Cardiol 2004;93:1500 –1503. 4. Whinnett ZI, Davies JER, Willson K, et al. Determination of optimal atrioventricular delay for cardiac resynchronization therapy using acute non-invasive blood pressure. Europace 2006;8:358 –366. 5. Bristow MR, Saxon LA, Boehmer J, et al. Cardiac-resynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350:2140 –2150. 6. Cleland JG, Daubert JC, Erdmann E, et al. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med 2005;352: 1539 –1549. 7. Flaker G, Weachter R. In cardiac resynchronization therapy: should we opt for AV optimization? J Cardiovasc Electrophysiol 2007;18:296 –297. 8. Penˇáz J. Photoelectric measurement of blood pressure, volume and flow in the finger. Digest of the international Conference on Medicine and Biological Engineering, Dresden, 1973:104 [abstract]. 9. Wesseling KH, De Wit B, Van der Hoeven GM, et al. Physiocal calibrating finger vascular physiology for Finapres. Homeostasis 1995;36:67– 82. 10. Smith NT, Wesseling KH, deWit B. Evaluation of two prototype devices producing non-invasive, pulsatile, calibrated blood pressure measurement from a finger. J Clin Monitor 1985;1:17–29. 11. Van Egmond J, Hasenbos M, Crul JF. Invasive versus non-invasive measurement of arterial pressure. Comparison of two automatic methods and simultaneously measures direct intraarterial pressure. Br J Anaesth 1985;57:434 – 444. 12. Petersen ME, Williams TR, Sutton R. A comparison of non-invasive continuous finger blood pressure measurement (Finapres) with intra-arterial pressure during prolonged head-up tilt. Eur Heart J 1995;16:1647–1654. 13. Jellema WT, Imholz PM, van Goudoever J, et al. Finger arterial versus intrabrachial pressure and continuous cardiac output during head-up tilt testing in healthy subjects. Clin Sci 1996;91:193–200. 14. Imholz BPM, Weiling W, Montfrans GA, et al. Fifteen years experience with finger arterial pressure monitoring: assessment of the technology. Cardiovasc Res 1998;38:605– 616. 15. Davies LC, Francis DP, Jurak P, et al. Reproducibility of methods for assessing baroreflex sensitivity in normal controls and in patients with chronic heart failure. Clin Sci(Lond) 1999;97:515–522. 16. Whinnett ZI, Nott G, Davies JER, et al. Efficiency, reproducibility and agreement of five different hemodynamic measures for optimization of cardiac resynchronization therapy. Int J Cardiol 2007. E-pub ahead of print doi: 10.1016/ j.ijcard.2007.08.004. 17. Whinnett ZI, Davies JER, Willson K, et al. Hemodynamic effects of changes in AV and VV delay in cardiac resynchronisation therapy show a consistent pattern: analysis of shape, magnitude and relative importance of AV and VV delay. Heart 2006;92:1628 –1634. 18. Cha YM, Nishimura RA, Hayes DL. Differences in mechanical atrioventricular delay between atrial sensing and atrial pacing modes in patients with hypertrophic and dilated cardiomyopathy: an electrical hemodynamic catheterization study. J Interv Card Electrophysiol 2002;6:133–140. 19. Wish M, Fletcher RD, Gottdiener JR, et al. Importance of left atrial timing in the programming of dual-chamber pacemakers. Am J Cardiol 1987;60: 566 –571. 20. Janosik DL, Pearson AC, Buckingham TA, et al. The hemodynamic benefit of differential atrioventricular delay intervals for sensed and paced atrial events during physiologic pacing. J Am Coll Cardiol 1989;14:499 –507. 21. Chirife R, Tentori MC, Mazzetti H, et al. Automatic beat-to-beat left heart AV normalization: is it possible? PACE 2003;26:2103–2110. 22. Catonia SL, Maue-Dickson W. AV delay latency compensation. J Electrophysiol 1987;1:242–249. 23. Ausubel K, Klementowicz P, Furman S. Interatrial conduction during cardiac pacing. PACE 1986;9:1026 –1031. 24. Scharf C, Li P, Muntwyler J, et al. Rate-dependent AV delay optimization in cardiac resynchronization therapy. Pacing Clin Electrophysiol 2005;28:279 –284. 25. Belalcazar A, Butter C, Stellbrink C, et al. Feasibility of optimizing AV delay in cardiac resynchronization during exercise using finger plethysmography. J Card Fail 2005:11.S136 [abstract]. 26. Shamim W, Francis DP, Yousufuddin M, et al. Intraventricular conduction delay: a prognostic marker in chronic heart failure. Int J Cardiol 1999;70: 171–178.
made at rest during atrial-sensed pacing showed a poorer correlation with exercise (r ⫽ 0.64, mean difference ⫾ SD ⫽ 2.2 ⫾ 24 ms). The three-phase resting model allows improved exercise hemodynamics to be achieved. Programming according to the three-phase resting model yields an exercise blood pressure of only 0.5 mmHg (⫾1.4 mmHg; P ⫽ NS) less than the true exercise optimum, whereas programming the resting sensed optimum yields an exercise blood pressure of 1.4 mmHg (⫾2.2 mmHg, P ⫽ .02) less than the true optimum. CONCLUSIONS Using acute noninvasive hemodynamics and a protocol of alternations, it is possible to optimize the AV delay of cardiac resynchronization devices even while a patient exercises. In clinical practice, the exercise optimum AV delay could be determined from three phases of resting measurements, without performing exercise.
RESULTS It was possible to perform AV delay optimization while a patient exercised. The resting three-phase model correlated well with the actual exercise optimal AV delay (r ⫽ 0.85, mean difference ⫾ standard deviation [SD] ⫽ 3.7 ⫾ 17 ms). Simply using measurements
KEYWORDS Atrioventricular delay optimization; Exercise; Optimization of cardiac resynchronization therapy; Biventricular pacing
Introduction
standard clinical methods. As a result, most centers do not routinely optimize these settings, and those that do usually limit the process to resting heart rates while the patient remains inactive. It is not known whether the optimal AV delay determined at rest corresponds to the optimal delay determined during exercise.7 It is during exercise that it is likely to be the most important to program the AV delay to the most efficient setting for an individual patient. One technique, which has potential to permit optimization during exercise, is beat-by-beat noninvasive blood pressure, measured using a calibrated digital photoplethysmograph such as the Finapres or Finometer. We have previously used this method to identify the AV delay that corresponds to peak hemodynamics in the resting state, with the heart rate either left in the normal resting state or elevated by atrial pacing.4 In this study we set out to investigate the following questions:
In heart failure, inadequate cardiac output response to exercise is a key driver of symptoms. In patients with badly coordinated ventricular and atrioventricular (AV) timings, cardiac resynchronization therapy (CRT) has been shown to improve these timings and thereby improve acute hemodynamics1– 4 and consequently symptoms, morbidity, and ultimately mortality.5,6 Exactly which timings of AV and interventricular delay are best for an individual patient is difficult to establish with This study was supported by the British Heart Foundation, Medtronic, and the Coronary Flow Trust. ZW (FS/05/068), JD (FS/05/006), and DF (FS/04/079) are British Heart Foundation fellows. CM (077049/Z/05/Z) is funded by the Welcome Trust. KW received support from the Foundation for Circulatory Health. Our institution has filed a patent on some of the methods described in this manuscript. Address reprint requests and correspondence: Dr. Zachary Whinnett, International Centre for Circulatory Health, St. Mary’s Hospital and Imperial College, 59-61 North Wharf Road, W2 1LA, UK. E-mail address: [email protected]. (Received July 19, 2007; accepted November 18, 2007.)
(Heart Rhythm 2008;5:378 –386) © 2008 Heart Rhythm Society. All rights reserved.
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2007.11.019
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
First, is it possible to perform optimization of the AV delay of CRT during exercise by using acute noninvasive hemodynamics? Second, how different is the AV delay that is determined during exercise from that determined at rest with a resting heart rate? Third, does measurement of the individual sensed-paced difference at resting heart rate and atrial pacing at higher heart rates allow improved determination of exercise optimum AV delay from resting measurements?
Methods Subjects Twenty outpatients with biventricular pacemakers or biventricular defibrillators implanted for clinical indications (New York Heart Association [NYHA] III or IV heart failure, QRS ⬎120 ms, maximal medical therapy) were enrolled into this study. Patients were excluded if they were not in sinus rhythm, were pacing dependent, or were not able to walk on a treadmill. Patients gave informed consent for this study, which was approved by the local ethics committee.
Measurements Data acquisition Noninvasive finger arterial pressure measurements were made using a Finometer (Finapres Medical Systems, Amsterdam, Holland). This technique, developed by Penˇáz8 and Wesseling et al9 uses a cuff that is placed around the finger, a built-in photoelectric plethysmograph, and a volumeclamp circuit that dynamically follows arterial pressure. This technique is well validated for measuring instantaneous changes in blood pressure.10 –14 An electrocardiogram signal was recorded using the Hewlett-Packard 78351A monitor (Hewlett-Packard Co, Andower, Massachusetts, USA). Analog output feeds of these signals were taken via a National Instruments DAQCard AI-16E-4 (National Instruments, Austin, TX) and acquired in digital form using Labview (National Instruments). They were analyzed offline with custom software based on the Matlab platform (MathWorks, Natick, MA).15
Measurement of relative change in blood pressure across different AV delays Beat-to-beat blood pressure was recorded during adjustment of the AV delay of the subject’s biventricular pacemaker. As previously described, the effect of background noise in the blood pressure trace was reduced by calculating the relative change in systolic blood pressure (SBPrel).4 This was done by comparing the tested AV delay to a reference AV delay and performing repeated alternations to obtain multiple transitions and yield at least six replicate measurements for each SBPrel. We calculated the relative change in blood pressure by subtracting the mean of the 10 beats immediately after the transition from the mean of the 10 beats immediately before the transition.4,17 We combined these to obtain, for each AV-tested delay, a mean SBPrel.
379
Our selection of systolic blood pressure was based on the reasoning that it was composed of two fundamental components: mean blood pressure (dependent on cardiac output and total peripheral resistance) and a fraction of pulse pressure (dependent on stroke volume and arterial compliance). Because each of these components would increase if cardiac function increased, we reasoned that there would be two reasons for systolic blood pressure to increase, and therefore it may give a particularly clear signal. We have studied the signal-to-noise ratios of a series of noninvasively measured hemodynamic parameters and have indeed confirmed that systolic blood pressure provided the best combination of efficiency and reproducibility when compared with other noninvasively measured hemodynamic parameters.16 We used mean rather than median because the type of noise we were trying to attenuate was likely to be randomly distributed (e.g., baseline wander due to respiration, physical movement, or other such processes), and the statistical principle of the central limits theorem ensures that the variability of the mean would be reduced in proportion to the square root of the number of transition measurements made. The advantage of the mean over the median is that every single transition measurement contributes something to the final value, and its statistical properties on increases in sample size are well understood and desirable. SBPrel was measured in the manner described above for a range of AV delays, which were 40, 80, 120, 140, 160, 200, 240, 280, 320, and 350 ms. In practice, for each subject we did not study AV delays beyond the point at which conduction became purely intrinsic. The interventricular delay was left at 0 ms or as close to this as the pacemaker allowed. We used parabolic interpolation to select the hemodynamic optimal AV delay for each optimization phase. We have previously demonstrated that the effects on acute hemodynamics of changing AV delay are curvilinear rather than linear and fit closely to a parabola.17
AV delay optimization phases Optimization was initially performed at rest while the patient sat on a couch. We used a three-phase resting model to predict the optimal exercise AV delay. This incorporated hemodynamic optimization of AV delay under three different conditions: Atrial-sensed biventricular pacing at resting heart rate, Atrial pacing at 5 bpm above the patient’s resting rate, Atrial pacing at a heart rate of 100 bpm. The patient then underwent direct measurement of optimal AV delay during exercise. The patient exercised on a treadmill, and workload was adjusted to maintain his or her heart rate between 100 and 110 bpm. The exercise AV delay optimum (with atrial sensing) was then determined by a process similar to that carried out at rest.
Differences between atrial-sensed and atrial-paced biventricular pacing The three-phase resting model must take into account the differences between atrial-sensed (P-synchronous) and atri-
380
Heart Rhythm, Vol 5, No 3, March 2008
Figure 1 Cartoon explaining why the paced AV optimum is longer than the sensed AV optimum. The left-sided AV delay has the predominant effect on cardiac output and systemic hemodynamics. A: In this example, the left-sided AV delay yielding optimal systemic hemodynamics is 110 ms. If this is achieved by atrial sensing, we need to program to allow for the atrial sensing delay (here 30 ms) and the interatrial delay (here 70 ms). This hypothetical patient therefore required a sensed AV delay of 110 ⫹ 70 ⫺ 30 ⫽ 150 ms. B: If the same AV delay is programmed for the paced AV delay, then this results in a nonoptimal left-sided AV delay (in this example, 40 ms). This difference occurs as a result of the atrial pacing latency (here 20 ms) and longer time taken for the atrial signal to travel between the right and left atria manifest as a longer interatrial delay (here 90 ms). Here the leftsided AV delay is 150 ⫺ 20 ⫺ 90 ⫽ 40 ms. C: Therefore, to obtain an optimal leftsided AV delay, the paced AV delay should be programmed to a longer value to account for the atrial pacing latency and longer interatrial delay. This patient requires a programmed paced AV delay of 110 ⫹ 90 ⫹ 20 ⫽ 220 ms.
al-paced (AV sequential pacing) AV delays. First, it is important to remember that when the AV delay is programmed on the pacemaker, this represents the right-sided AV delay. It is, however, the mechanical contraction delay between the left atrium and left ventricle that has the predominant effect on systemic hemodynamics.18 –21 There are three principal differences between atrialsensed pacing and AV sequential pacing that need to be taken into account: the atrial sensing delay, atrial pacing latency, and differences in interatrial delay (Figure 1).
Atrial sensing delay During atrial-sensed pacing, the programmed right-sided AV delay is shorter than the actual right-sided AV delay. This is because the pacemaker only detects the onset of atrial depolarization once the atrial electrical activity has propagated from the site of origin (sinoatrial node) to the atrial electrodes where it can be detected as a near-field signal. This results in a time lag in the detection of the onset of atrial depolarisation: the atrial sensing delay (Figure 1A).
Atrial pacing latency When pacing from the right atrium, there is a delay from the time of delivery of the pacing stimulus to the onset of atrial
depolarization, the so-called atrial pacing latency.22 This means that the actual right-sided AV delay is slightly shorter than the programmed AV delay (Figure 1B).
Interatrial delay The time taken for activation of the left atrium differs depending on whether activation is initiated by intrinsic atrial activity or atrial pacing. Typically, the interatrial delay is longer for an atrial-paced beat than for intrinsic activation. The longer interatrial delay may be because of less efficient interatrial conduction. The difference in intra-atrial delay between paced and sensed activation appears to be patient specific,18 –20 which is likely to reflect differences in underlying conduction, lead position, atrial size, and function. Intra-atrial conduction time appears to remain relatively constant when heart rate is altered.23
Three-phase resting model for predicting the optimal AV delay during exercise by pacing at different heart rates at rest We hypothesized that in individual patients the optimal AV delay identified during AV delay optimization may differ depending on whether optimization is performed at rest or during exercise. We believed it might be possible to identify
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
the exercise optimal sensed AV delay at rest if the heart rate was increased to the exercise level by atrial pacing and an adjustment was made for the difference between the atrialpaced AV delay and the atrial-sensed AV delay, which we call (for simplicity) the “sensed-paced difference.” We calculated the sensed-paced difference by determining the difference between the optimal AV delay determined while atrially pacing at 5 bpm above the patient’s resting heart rate and that determined during atrial-sensed pacing at a resting heart rate. We then subtracted this sensed-paced difference from the optimal AV delay determined while atrially pacing at a heart rate of 100 bpm. This gave us our estimate of optimal AV delay determined during exercise using the three-phase resting exercise model. Estimated exercise optimal AV delay A-paced A-paced ⫺ 共AVoptimumresting ⫽ AVoptimum100bpm rate A-sensed ⫺ AVoptimumresting rate) A⫺paced AV optimum100bpm : optimal AVD delay determined while atrially pacing at a rate of 100 bpm. A⫺paced AV optimumresting rate: optimal AV delay determined while atrially pacing at a rate 5 bpm above the patient’s intrinsic rate. A⫺paced AV optimumresting rate: optimal AV delay determined during atrial-sensed pacing at intrinsic rate.
Statistics The SBPrel value was determined for each tested combination of AV delay in relation to a reference AV delay (120 ms) by taking the mean of observed blood pressure changes from at least six individual transitions. Data are given as means ⫾ standard deviation (SD) of the series of replicate transitions. Comparisons were made on the basis of mean differences ⫾ SD and by Pearson correlation coefficients. P-values for comparison of means were calculated using a Student’s paired t-test, and comparisons of proportions were made using Fisher’s exact test. P ⬍.05 was considered statistically significant. The statistical package Statview 5.0 (SAS Institute, Cary, NC) was used for analysis.
Results Patient characteristics Twenty patients with biventricular pacemakers or biventricular ICDs were enrolled into this study a mean of 10 months after implantation. Ten were male and 10 were female, with an age range of 46 – 82 years (mean 68.3 years). The cause of heart failure was ischemic in 11 and idiopathic dilated in nine. Mean systolic blood pressure by sphygmomanometer was 118 ⫾ 18 mmHg. The mean ejection fraction of the patients at the time of the study was 30% ⫾ 5%. At the time of the study, one patient was in NYHA class I, 13 were in NYHA II, and six were in NYHA III.
Identification of a hemodynamic peak Despite patient movement during exercise, all patients showed a clear hemodynamic peak, with the use of the
381
protocol of repeated alternations. All patients were able to walk on the treadmill and to increase their heart rate to between 100 and 110 bpm, although some patients undergoing exercise optimization needed rest periods between different tested AV delays. The AV delay identified as optimal was specific to the individual patient. The optimal atrial-paced AV delay was on average longer than the optimal atrial-sensed AV delay. This occurred both at a resting rate (185 ms, range 150 –240 ms; versus 120 ms, range 78 –163 ms; P ⬍.0001) and at a higher rate (186 ms, range 150 –240 ms; versus 118 ms range 79 –200 ms; P ⬍.0001). All data from individual subjects are shown in Table 1. The AV delay identified as optimal is displayed for each individual patient, for the four different pacing configurations: resting rate atrial-pacing (paced just above the resting heart rate), resting rate atrial-sensing, higher rate atrial-pacing, and during exercise with atrial-sensing. The sensed-paced difference was calculated at rest and used to provide a prediction of the exercise optimal AV delay by subtracting this from the higher rate atrial-paced optimal AV delay (estimated exercise optimal AV delay (Estimated A⫺paced ⫺ (AV exercise optimal AV delay ⫽ AV optimum100bpm A⫺paced A⫺sensed optimumresting rate ⫺ AV optimumresting rate)). The means and standard deviations are also shown.
Effect of exercise on optimal AV delay In 11 patients, the optimal AV delay was shorter during exercise than at rest, while in eight patients it was longer. In one patient, the optimal AV delay remained the same. There was no significant difference in resting PR interval between the patients in whom AV delay lengthened and those in whom it shortened (mean intrinsic AV delay 201 and 200 ms, respectively; P ⫽ NS). In addition, there was no difference in etiology of the heart failure or the patients’ functional class (P ⫽ NS by Fisher’s exact test and 2-test, respectively).
Assessment of signal-to-noise ratio To make an assessment of the amount of noise within the signal, we calculated the signal-to-noise ratio and displayed the maximum standard error for each of the four different pacing configurations (Table 2). We calculated the signalto-noise ratio by dividing the range of values obtained for different pacing settings by the standard error of the measurements at each pacing setting. The signal-to-noise ratio was similar for atrial sensing at rest and during exercise and was higher when measurements were made with atrial pacing at both resting and higher heart rates. Exercise conditions generated the highest values of standard error, which is likely to be related to the physical movement of the patient. The resting, atrial-sensed, state also had a fairly high standard error, presumably due to the presence of variation (however small) in heart rate from beat to beat. The two paced states (resting rate and higher rate) had lower standard errors, which may reflect the greater regularity of the cardiac rhythm.
382
Heart Rhythm, Vol 5, No 3, March 2008
Table 1 The AV delay in milliseconds identified as optimal, for each individual patient for each of the four different pacing configurations Patient no.
Resting rate atrial-paced optimum
Resting rate atrial-sensed optimum
Higher rate atrial-paced optimum
Predicted exercise optimum
Actual exercise optimum
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Mean SD
160 160 167 200 197 168 240 169 160 171 150 196 200 177 200 171 200 240 169 200 185 25
163 78 120 133 145 80 120 108 82 140 105 120 134 134 111 140 118 156 97 124 120 24
200 162 200 213 189 150 240 178 164 193 164 160 189 181 192 154 180 209 200 198 186 23
203 79 153 146 137 62 120 117 86 162 119 93 123 138 103 123 98 125 128 122 121 31
200 90 112 147 129 80 138 140 90 158 79 93 120 133 107 99 86 116 127 120 118 30
Comparison of three-phase resting model with actual exercise optimum
Comparison of sensed paced difference between resting and higher heart rate
Prediction of the exercise optimum by the three-phase approach (not requiring exercise) showed good agreement: mean difference ⫾ SD ⫽ 3.7 ⫾ 17 ms: r ⫽ 0.85 (R2 ⫽ 0.73; Figure 3, left panel). The alternative of simply using the atrial-sensed resting-rate optimum as an estimate of the likely exercise optimum was less reliable: mean difference ⫾ SD ⫽ 2.2 ⫾ 24 ms: r ⫽ 0.64 (R2 ⫽ 0.41, Figure 3, right panel). The Bland-Altman plots are shown in Figure 4.
The sensed paced difference at resting heart rate and at higher heart rate was calculated by measuring the difference between the AV delay identified as optimal at rest and higher heart rates. The mean sensed paced difference at rest was 64 (⫾27) ms, and at higher heart rates this was 68 (⫾24) ms (R2 ⫽ 0.62 and a 95% limit of agreement of ⫾36 ms between the resting and high-rate sensed-paced differences).
Hemodynamic differences between using the three-phase resting model and using atrial-sensed pacing at resting rate for selection of the programmed AV delay during exercise We assessed whether programming the AV delay based on the three-phase resting optimization model or instead simply based on the atrial-sensed resting optimization results in changes in hemodynamics during exercise. We calculated the change in acute blood pressure, from the hemodynamic response curve during exercise, which would occur as a result of programming the AV delay not to the true exercise hemodynamic optimum but rather to the three-phase predicted optimum or simply to the optimum determined with resting heart rate atrial-sensed pacing. Simply using the measurements made at resting heart rate with atrial-sensed biventricular pacing yielded an SBPrel on exercise that was 1.4 mmHg (⫾2.2 mmHg, P ⫽ .02) lower than using the actual exercise optimum. Using the three-phase resting prediction yielded an SBPrel on exercise that was not significantly lower (0.5 ⫾ 1.4 mmHg; P ⫽ NS) than the actual exercise optimum.
Discussion In this study, we demonstrate a method for optimization of the AV delay of CRT devices that can be performed during exercise. Simply optimizing at resting heart rates with atrialsensed biventricular pacing selects an AV delay that correlates reasonably well with the exercise optimum. But agreement with the true exercise optimum can be improved significantly by using a predictive process, which involves three phases conducted with the patient at rest. Using this three-phase resting prediction, there appears to be a significant hemodynamic improvement over standard optimization with atrial-sensed pacing at resting heart rates.
AV delay optimization during exercise With an algorithm of multiple alternations to and from a reference delay, it is possible to reliably perform optimization of the AV delay of CRT during exercise by using noninvasive hemodynamics. Exercise represents the state in which most patients with heart failure become symptomatic. As heart rate increases, the time available for filling within the cardiac cycle becomes progressively more limited. This makes the balance
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
383
Table 2 The highest standard error and signal-to-noise ratio are displayed for each patient for each of the four different pacing configurations
during echocardiographic optimization has previously been investigated.24 Other potential methods for optimization during exercise include standard finger plethysmography.25
Patient No.
Feasibility of programming optimal AV delay
Resting rate atrial sensed
Highest standard error 1 2.6 2 1.0 3 1.9 4 0.6 5 1.3 6 0.9 7 3.0 8 8.1 9 2.3 10 1.6 11 3.1 12 1.6 13 1.6 14 2.6 15 3.5 16 2.7 17 2.4 18 2.4 19 2.0 20 2.8 Median 2.3 Mean 2.4 Signal to noise ratio 1 9.7 2 1.9 3 6.5 4 16.4 5 8.8 6 7.2 7 2.7 8 2.2 9 5.8 10 2.1 11 3.3 12 5.7 13 2.3 14 15.1 15 1.5 16 9.2 17 2.4 18 9.7 19 5.4 20 2.7 Median 5.5 Mean 6.0
Resting rate atrial paced
Higher rate atrial paced
Exercise
3.2 2.0 2.8 1.9 1.5 1.0 1.2 4.1 1.9 2.7 2.9 2.4 3.5 3.5 3.4 2.2 0.9 2.4 2.6 1.4 2.4 2.4
2.7 2.6 2.4 2.9 1.3 0.8 2.4 2.9 2.1 2.3 2.7 2.2 4.1 1.7 3.8 3.8 2.0 2.2 1.4 1.4 2.3 2.4
5.0 1.6 2.3 2.9 2.2 1.9 7.0 6.8 3.4 1.9 5.6 1.8 3.9 1.6 6.7 6.3 0.8 4.5 2.0 1.6 2.6 3.5
8.9 13.9 10.9 10.4 11.5 13.6 26.2 10.7 6.2 7.8 10.7 7.4 8.2 14.2 2.9 14.6 13.4 20.0 15.0 8.7 10.8 11.8
26.5 19.7 19.1 9.5 15.8 24.4 14.6 17.1 14.5 17.5 15.6 10.7 8.5 33.1 10.2 14.4 16.6 24.3 19.6 17.6 16.9 17.5
8.3 1.6 9.7 9.4 6.5 3.7 4.1 3.1 3.6 16.8 3.0 4.8 5.4 2.3 1.4 5.1 5.1 4.5 6.1 9.9 4.9 5.7
between the timings of the different phases of the cardiac cycle even more critical. To determine the most efficient AV delay in the state when it is most valuable for the patient, there needs to be a method for determining the optimum during exercise. The effect of posture during optimization is also an important consideration. Venous return and filling of the heart is significantly affected between the supine and upright positions. This is not taken into account during standard echo optimization, which is typically performed at rest in a nonupright position. The effect of posture and exercise
Current pacemakers have limited flexibility for programming the AV delay at different heart rates. Most algorithms allow only for shortening of the AV delay with exercise. It would be useful if pacemaker algorithms were developed to allow the AV delay to be programmed for a range of different heart rates during both atrial sensing and atrial pacing. Overall, in the group of patients studied, there was a shortening in the AV delay at which intrinsic conduction occurred with atrial-sensed pacing during exercise compared with rest. This observation did not, however, impact on the feasibility of programming the predicted optimal AV delay. In all patients in our study the predicted optimal exercise AV delay was shorter than the AV delay at which intrinsic conduction occurred during exercise.
Effect of exercise on AV delay optimum Our data demonstrate that optimal AV delay is dependent on heart rate, with the optimum identified at resting rate rarely corresponding with that identified during exercise at higher heart rates. A previous study using echocardiographic AV delay optimization (aortic outflow tract velocity time integral) immediately after exercise also showed a change in optimal AV with change in heart rate.24 In that study, across all patients, there was a lengthening of optimal AV delay with exercise, although data for response in individual patients were not available.24 In our patients, the effect of exercise on the optimal value of AV delay appeared to be specific to the individual patient. In some patients, the optimal AV delay was shorter on exercise, while in other patients the optimal AV delay lengthened. The response does not appear to be dependent on the individuals’ intrinsic PR interval or their severity or etiology of heart failure. We used a different method that is highly sensitive to even small changes in acute hemodynamics as a result of changing AV delay and is highly reproducible. The echocardiographic measurements had to be made after exercise when patients had a variable heart rate. Overall, it is likely that the response is dependent on a number of different factors, for example, the conduction properties of the left atrium, filling pressure, atrial size, and function. Patients are likely to be best served by individual optimization.
Comparison of the three-phase resting prediction at rest and exercise The three-phase resting prediction assumes that the sensed-paced difference in optimal AV delay is the same for resting heart rate as it is for higher rates. Our data suggest that this is a valid extrapolation, and therefore it may not be necessary in clinical practice to optimize a patient during exercise. Being able to accurately predict the exercise optimal AV delay from resting measures makes it clinically more
384
Heart Rhythm, Vol 5, No 3, March 2008
Figure 2 The sensed-paced difference, calculated at higher rate and at resting rate in an individual patient. Optimal paced AV delay is longer than optimal sensed AV delay, at both resting and higher rates. The sensed-paced difference is similar at the two rates. The exercise AV optimum (i.e., higher rate sensed) can therefore be predicted from the other three measurements, which do not require exercise to be undertaken. Predicted exercise optimal AV delay ⫽ 196 ⫺ (169 ⫺ 97) ⫽ 123 ms.
feasible to program the exercise optimal AV delays in most patients. Optimization during actual exercise may remain a theoretical gold standard, but the level of patient cooperation
Figure 3 Comparison of the optimal AV delay identified during exercise with the threephase resting prediction of exercise optimum and optimization at resting heart rate only. Data are derived from separate individuals. The dotted line shows the line of identity. There is a good agreement between the actual optimal AV delay determined by performing optimization during exercise and the predicted optimal AV delay determined using the threephase resting prediction (left panel). This means it is potentially possible to predict the optimal exercise AV delay without exercising the patient. In comparison, optimization performed at a resting heart rate only does not agree as well with the exercise-determined optimal AV delay (right panel).
and the number of skilled staff required mean that it may be difficult to implement into routine clinical practice in all centers.
Whinnett et al
Optimization of the AV Delay of CRT in Exercise
385
Cross-sectional observational data suggest a 4% (95% CI ⫽ 3%–5%) relative increase in mortality per 1 mmHg decline in systolic blood pressure in outpatients with chronic heart failure.26 The lesson from these studies is that CRT improves cardiac output without deleterious effects on myocardial oxygen consumption, that is, it improves efficiency in cardiac function. Even 1- or 2-percentage-point differences in cardiac function (when achieved without cost in myocardial work) can improve outcomes.
Comparison of sensed-paced difference between resting and higher heart rate We compared the sensed-paced difference at resting and higher heart rates as we had postulated that the difference would remain relatively constant. The level of agreement we identified (correlation coefficient of 0.8) is as close as can reasonably be expected, given the fact that each of the two elements is itself a difference and is therefore subject to two sources of error: such errors necessarily attenuate the observed correlation coefficient. Figure 4 Bland Altman plots showing the comparison of the optimal AV delay identified during exercise with the three-phase resting prediction of exercise optimum (top) and optimization at resting heart rate only (bottom). Data are derived from separate individuals. The solid line shows the line of identity. The 95% limit of agreement for the comparison between the predicted and actual exercise optimum is ⫾36 ms. For the comparison between actual exercise and optimization at resting heart rate, the 95% limits of agreement are ⫺48 to 44 ms.
Hemodynamic consequence of using the three-phase resting prediction or atrial-sensed pacing at resting rate to select the programmed AV delay during exercise While initially a difference of 1.4 mmHg in acute blood pressure change may seem small, it is not negligible in relation to the size of the change seen with CRT itself, an intervention that was ultimately proven to improve not only symptoms but also survival. Onset of biventricular pacing resulted in a mean increase in acute aortic systolic pulse pressure of 2 mmHg in a previous invasive study.1 This is consistent with the landmark trials with CRT that showed an increase in blood pressure, paralleling the improvements in symptoms and survival, for patients entered into the device arm. For example, in the COMPANION trial, patients in the resynchronization arm initially gained approximately 2 mmHg (confidence interval [CI] not published) in systolic blood pressure (in comparison with the control arm) and went on to have an 18% relative reduction (95% CI ⫽ 1% increase to 42% reduction) in the combined endpoint of morbidity and mortality.5 Similarly, the CARE-HF trial showed, at 3 months, that the increment in blood pressure attributable to being in the device arm was 5.8 mmHg (95% CI ⫽ 3.5– 8.2 mmHg) and the mortality reduction was 37% (95% CI ⫽ 23%– 49%).6
Potential for automation The intention was to provide a protocol in which analysis could later be made entirely automatic. To do this, we placed the burden of establishing data quality at the time of acquisition (that is, before all stages of numerical analysis). During data acquisition, if it became evident that the Finometer signal was poor or subject to large artifact, the recording was terminated and restarted afresh after checking equipment positioning. We estimate that ⬍10% of hemodynamic recordings during exercise, once begun, had to be terminated due to motion-related artifact. In some patients, ectopy was a significant source of artifact. This was also addressed at the time of data collection by avoiding performing transitions in the few seconds immediately after any ectopics and by adding an extra transition on every occasion when there were ectopics in the few seconds immediately after a transition. (In the latter situation, both the original, noisy transition and the subsequent additional transition were included in the data analysis, so that no inadvertent bias during analysis could have affected the numerical results.) The increase in number of transitions reduces the effect of the noise on the average. In Table 2, we present the signal-to-noise ratio on exercise and rest. All recordings that were thought, at the time of acquisition, to be of adequate quality were put through the analysis and used in entirety, without any manual editing.
Study limitations In this study, we only assessed one heart rate during exercise. We do not have data on how the optimal AV delay varies within individual patients at multiple different exercise heart rates. Further studies would be required to clarify this. However, our finding that the three-phase resting prediction is a suitable proxy for actual exercise could in
386 principle be extended to a range of exercise heart rates. This study has a size adequate to answer its primary question, but subanalyses to look at trends in lengthening or shortening of AV interval during exercise may not be sufficiently powered to provide reliable information. A potential limitation to our study is that we had a relatively high prevalence of class I and II patients. This occurred because these patients were willing to undergo a protocol that required prolonged walking on a treadmill for the experiment of optimization during exercise. In more symptomatic patients there may be a higher prevalence of optimization curves where there is a relatively flat hemodynamic response over a wide range of AV delays. It could be argued that any of the AV delays in these flat regions could be considered optimal, as they all yield the same hemodynamic response. Intentional inclusion of a wide range of AV delays, including extreme ones unlikely to be optimal, allowed the parabolic interpolation to detect declines at those extremes and thereby identify a central optimal value among the flat areas. We have previously found that at resting heart rates the hemodynamic response curves tend to be flatter compared with higher heart rates.17 In developing our three-phase pacing model to predict the exercise optimal AV delay, we did not take into account the effect of posture. Measurements at rest were made with the patients in the sitting position. The systemic resistance seen by the heart pump is probably different when atrial pacing at high rates while sitting compared with the resistance seen when standing and actually exercising. This different arterial impedance may cause the readings in absolute systolic blood pressure to be different, which may affect the accuracy of the model. However, to minimize the effect of changes in absolute pressure on the reliability of the optimization method, we calculated the relative change in systolic blood pressure.
Conclusions Using acute noninvasive hemodynamics to calculate mean relative change in blood pressure, it is possible to optimize the AV delay of cardiac resynchronization devices while a patient exercises. Optimization at a resting heart rate during atrial-sensed biventricular pacing correlates with the optimum identified during exercise. However, using a threephase resting model, a closer prediction of the actual exercise optimum can be made, which has significant acute hemodynamic benefits for individual patients. (Figure 2).
References 1. Auricchio A, Stellbrink C, Block M, et al. Effect of pacing chamber and atrioventricular delay on acute systolic function of paced patients with congestive heart failure. Circulation 1999;99:2993–3001. 2. Perego GB, Chianca R, Facchini M, et al. Simultaneous vs sequential biventricular pacing in dilated cardiomyopathy: an acute hemodynamic study. Eur J Heart Fail;2003:305–313.
Heart Rhythm, Vol 5, No 3, March 2008 3. Van Gelder BM, Bracke FA, Meijer A, et al. Effect of optimizing the VV interval on left ventricular contractility in cardiac resynchronisation therapy. Am J Cardiol 2004;93:1500 –1503. 4. Whinnett ZI, Davies JER, Willson K, et al. Determination of optimal atrioventricular delay for cardiac resynchronization therapy using acute non-invasive blood pressure. Europace 2006;8:358 –366. 5. Bristow MR, Saxon LA, Boehmer J, et al. Cardiac-resynchronization therapy with or without an implantable defibrillator in advanced chronic heart failure. N Engl J Med 2004;350:2140 –2150. 6. Cleland JG, Daubert JC, Erdmann E, et al. The effect of cardiac resynchronization on morbidity and mortality in heart failure. N Engl J Med 2005;352: 1539 –1549. 7. Flaker G, Weachter R. In cardiac resynchronization therapy: should we opt for AV optimization? J Cardiovasc Electrophysiol 2007;18:296 –297. 8. Penˇáz J. Photoelectric measurement of blood pressure, volume and flow in the finger. Digest of the international Conference on Medicine and Biological Engineering, Dresden, 1973:104 [abstract]. 9. Wesseling KH, De Wit B, Van der Hoeven GM, et al. Physiocal calibrating finger vascular physiology for Finapres. Homeostasis 1995;36:67– 82. 10. Smith NT, Wesseling KH, deWit B. Evaluation of two prototype devices producing non-invasive, pulsatile, calibrated blood pressure measurement from a finger. J Clin Monitor 1985;1:17–29. 11. Van Egmond J, Hasenbos M, Crul JF. Invasive versus non-invasive measurement of arterial pressure. Comparison of two automatic methods and simultaneously measures direct intraarterial pressure. Br J Anaesth 1985;57:434 – 444. 12. Petersen ME, Williams TR, Sutton R. A comparison of non-invasive continuous finger blood pressure measurement (Finapres) with intra-arterial pressure during prolonged head-up tilt. Eur Heart J 1995;16:1647–1654. 13. Jellema WT, Imholz PM, van Goudoever J, et al. Finger arterial versus intrabrachial pressure and continuous cardiac output during head-up tilt testing in healthy subjects. Clin Sci 1996;91:193–200. 14. Imholz BPM, Weiling W, Montfrans GA, et al. Fifteen years experience with finger arterial pressure monitoring: assessment of the technology. Cardiovasc Res 1998;38:605– 616. 15. Davies LC, Francis DP, Jurak P, et al. Reproducibility of methods for assessing baroreflex sensitivity in normal controls and in patients with chronic heart failure. Clin Sci(Lond) 1999;97:515–522. 16. Whinnett ZI, Nott G, Davies JER, et al. Efficiency, reproducibility and agreement of five different hemodynamic measures for optimization of cardiac resynchronization therapy. Int J Cardiol 2007. E-pub ahead of print doi: 10.1016/ j.ijcard.2007.08.004. 17. Whinnett ZI, Davies JER, Willson K, et al. Hemodynamic effects of changes in AV and VV delay in cardiac resynchronisation therapy show a consistent pattern: analysis of shape, magnitude and relative importance of AV and VV delay. Heart 2006;92:1628 –1634. 18. Cha YM, Nishimura RA, Hayes DL. Differences in mechanical atrioventricular delay between atrial sensing and atrial pacing modes in patients with hypertrophic and dilated cardiomyopathy: an electrical hemodynamic catheterization study. J Interv Card Electrophysiol 2002;6:133–140. 19. Wish M, Fletcher RD, Gottdiener JR, et al. Importance of left atrial timing in the programming of dual-chamber pacemakers. Am J Cardiol 1987;60: 566 –571. 20. Janosik DL, Pearson AC, Buckingham TA, et al. The hemodynamic benefit of differential atrioventricular delay intervals for sensed and paced atrial events during physiologic pacing. J Am Coll Cardiol 1989;14:499 –507. 21. Chirife R, Tentori MC, Mazzetti H, et al. Automatic beat-to-beat left heart AV normalization: is it possible? PACE 2003;26:2103–2110. 22. Catonia SL, Maue-Dickson W. AV delay latency compensation. J Electrophysiol 1987;1:242–249. 23. Ausubel K, Klementowicz P, Furman S. Interatrial conduction during cardiac pacing. PACE 1986;9:1026 –1031. 24. Scharf C, Li P, Muntwyler J, et al. Rate-dependent AV delay optimization in cardiac resynchronization therapy. Pacing Clin Electrophysiol 2005;28:279 –284. 25. Belalcazar A, Butter C, Stellbrink C, et al. Feasibility of optimizing AV delay in cardiac resynchronization during exercise using finger plethysmography. J Card Fail 2005:11.S136 [abstract]. 26. Shamim W, Francis DP, Yousufuddin M, et al. Intraventricular conduction delay: a prognostic marker in chronic heart failure. Int J Cardiol 1999;70: 171–178.