Optimal Cardiac Resynchronization Therapy Pacing Rate in Non-Ischemic Heart Failure Patients: A Randomized Crossover Pilot Trial

Background The optimal pacing rate during cardiac resynchronization therapy (CRT) is unknown. Therefore, we investigated the impact of changing basal pacing frequencies on autonomic nerve function, cardiopulmonary exercise capacity and self-perceived quality of life (QoL). Methods Twelve CRT patients with non-ischemic heart failure (NYHA class II–III) were enrolled in a randomized, double-blind, crossover trial, in which the basal pacing rate was set at DDD-60 and DDD-80 for 3 months (DDD-R for 2 patients). At baseline, 3 months and 6 months, we assessed sympathetic nerve activity by microneurography (MSNA), peak oxygen consumption (pVO2), N-terminal pro-brain natriuretic peptide (p-NT-proBNP), echocardiography and QoL. Results DDD-80 pacing for 3 months increased the mean heart rate from 77.3 to 86.1 (p = 0.001) and reduced sympathetic activity compared to DDD-60 (51±14 bursts/100 cardiac cycles vs. 64±14 bursts/100 cardiac cycles, p<0.05). The mean pVO2 increased non-significantly from 15.6±6 mL/min/kg during DDD-60 to 16.7±6 mL/min/kg during DDD-80, and p-NT-proBNP remained unchanged. The QoL score indicated that DDD-60 was better tolerated. Conclusion In CRT patients with non-ischemic heart failure, 3 months of DDD-80 pacing decreased sympathetic outflow (burst incidence only) compared to DDD-60 pacing. However, Qol scores were better during the lower pacing rate. Further and larger scale investigations are indicated. Trial Registration ClinicalTrials.gov NCT02258061


Introduction
Cardiac resynchronization therapy (CRT) has become an important treatment strategy for a select group of heart failure (HF) patients with electrical dyssynchrony, and several studies have documented the beneficial effects of CRT on mortality and morbidity in such patients. [1,2] The prevalence of HF is projected to increase by 25% by 2030 compared to 2013 [3], and approximately 10% of an unselected group of HF patients could be eligible for biventricular pacing (BiV), [4] thus indicating a likely future increase in CRT utilization.
Although the majority of correctly selected patients respond favorably to CRT, 25-30% show little or no improvement after device implantation. To increase response rates, resources have focused on programming optimization, particularly atrioventricular (AV) and interventricular (VV) timing intervals. [5][6][7] However, few studies have examined the optimal basal atrial pacing rate and its impact on long-term outcome in CRT patients. Increasing pacing rates in BiV mode have demonstrated positive acute hemodynamic effects (e.g., decreased filling pressure and increased cardiac output). [6,[8][9][10] Furthermore, an increased atrial basal pacing rate and HR (heart rate) could prove to be favorable in HF patients with chronotropic incompetence (attenuated HR response to exercise), [11,12] which is associated with increased cardiac and all-cause mortality. [13,14] In contrast, beta-blocking therapy appears to be associated with better outcomes in terms of death and transplantation in CRT patients, [15] which aligns with the beneficial effect of a lower HR in HF patients without CRT. [16] Moreover, the "Dual Chamber and VVI Implantable Defibrillator Trial" [17] tested whether the dual-chamber pacing (DDDR-70) mode could improve hemodynamics and reduce major cardiac adverse events (MACE) compared to ventricular back-up pacing . Surprisingly, the trial showed an increase in the composite endpoint in the pacing group; however, the results were not directly comparable with the CRT population due to the lack of left ventricular pacing. Therefore, the optimal basal atrial pacing rate in CRT patients remains poorly defined.
The aim of this randomized, double-blind, crossover pilot trial was to investigate the impact of 60-bpm (DDD-60) to 80-bpm (DDD-80) basal atrial pacing rates on autonomic nerve function in CRT patients, as assessed by (i) microneurography and NT-proBNP, (ii) cardiopulmonary exercise tests and (iii) self-perceived quality of life (QoL).

Study population
Fifteen patients were included in the study from the 18 th of August 2011-5 th of January 2013; 3 patients withdrew (1 due to throat cancer, 2 due to long commuting distances), leaving 12 patients with idiopathic dilated cardiomyopathy who had had a CRT device implanted (the initial indications were in accordance with European guidelines) [18] at least 6 months prior to study inclusion. The steering committee decided to restrict inclusion to non-ischemic CRT patients to assess any potential risk before decision on inclusion of ischemic CRT patients. The patients were in New York Heart Association (NYHA) functional class II (n = 7) or III (n = 5), with mean a left ventricular ejection fraction (LVEF) of 38%±10. The patients were required to be in sinus rhythm and hemodynamically stable as well as to demonstrate biventricular pacing > 90% of the time and to be undergoing optimal medical treatment for HF, with no changes in medications during the past 3 months. All patients were considered to be CRT responders based on either NYHA or LVEF improvements. Patients were excluded if they were NYHA I, unable to perform the exercise test (e.g., diagnosed with chronic obstructive pulmonary disease, severe arthritis), diagnosed with cancer, had plasma creatinine levels > 200 micromoles per liter, or were admitted for decompensated HF or acute coronary syndrome in the preceding 3 months.

Ethics
All participating patients provided informed written consent before study participation. All relevant approvals were obtained from Capital Region of Denmark ethical committees (H-3-2011-034) the 16 th of May 2011, and the study was performed in compliance with the Declaration of Helsinki. The study was not registered in a public database before patient enrollment due to its nature as a pilot study. The authors confirm that all ongoing and related trials for this intervention are registered.

Study protocol
This double-blind, randomized, crossover pilot study tested two basal atrial pacing rates: DDD-60 versus DDD-80. Each patient was paced for 2 90-day periods, with DDD-60 and DDD-80 in a randomly assigned sequence. Complete patient histories were obtained, and physical examinations were performed for all patients at baseline, 3 months after phase 1, and 6 months after phase 2; see Fig 1 for the study protocol.
After concluding each phase, the patients underwent echocardiography, and blood samples were taken. In addition, the patients completed two separate self-perceived QoL questionnaires and performed cardiopulmonary bicycle tests; sympathetic nerve activity was assessed using microneurography (MSNA), and the CRT device was interrogated and finally reprogrammed to the opposite pacing rate or to baseline settings at the final visit. MSNA and bicycle tests were performed on two separate days at each phase to avoid exercise-induced elevated sympathetic activity. The CRT settings were blinded to the attending physician, and the patients were unaware of the study assignment sequence.
QoL questionnaire HF patients are compromised by an array of symptoms, including dyspnea, limited physical activity, and fatigue. Furthermore, increased medication intake and the possible side effects of those medications, as well as the progressive loss of self-reliance and recurrent hospitalization, all affect and impact QoL. [19] We administered two distinct questionnaires (the Minnesota Living with HF [LIfHE] questionnaire and Medical Outcome Survey Short Form SF-36]) [20] to fully comprehend changes in the signs and symptoms of HF, physical and sexual activity, and psychological and socioeconomic conditions. LIfHE is a well-validated and popular tool used to assess QoL in HF patients; it contains 21 questions designed to measure the abovementioned factors. [21] SF-36 is comprised of 36 items, which are subdivided into 8 health dimensions (vitality, physical functioning, bodily pain, general health perceptions, physical role functioning, emotional role functioning, social role functioning, mental health). The scores are transformed into a 0-100 scale, with each question equally weighted. High scores indicate less restraint, and lower scores indicate more restraint.

Device settings and programming
In this study, the time between implantation of the CRT device and randomization averaged 20.8 ± 10.7 months. Eleven patients had St. Jude devices, and one patient had a Medtronic device. The majority of patients were in DDD (n = 10) and DDD-R (n = 2) mode. Two highly experienced technicians managed the device settings and ensured optimal programming, but they were not otherwise involved in this study. The device was randomly programmed to a basal pacing rate of 60 bmp in the first phase and 80 bmp in the second phase or vice versa. Atrioventricular and interventricular delays were programmed according to the local electrophysiologist's discretion, and these settings were not changed during the study.
Device transcripts were obtained at each phase to register the levels of atrial and ventricular sensing, pacing, BiV pacing (%), and HR histograms.

Exercise protocol
A symptom-limiting bicycle (Ergoline, Baden-Württemberg, Germany) exercise test was performed to measure maximum work capacity and oxygen consumption at each phase. The HR and 12-lead electrocardiogram were recorded continuously, and blood pressure was recorded every 2 minutes. The initial workload was set to 25 Watts (W) and increased by 25 W every 2 minutes until exhaustion, at a rate of 60-70 revolutions per minute. During the test, the patients inspired room air through a low-resistance mouthpiece, with the nostrils tightly blocked. The expired oxygen and carbon dioxide partial pressures were measured (using the breath-by-breath method) with a gas analyzer (Innocor, Innovision Aps, Glamsbjerg, Denmark). Calibration procedures were performed according to the manufacturer's specification immediately before the tests, and the following parameters were measured: peak oxygen consumption (pVO 2 ), which was defined as the highest VO 2 (averaged over a 20-30-second period) achieved at maximal effort; respiratory exchange ratio (peak VCO 2 /VO 2 = RER); peak minute ventilation-carbon dioxide relationship (pVE/VCO 2 slope), which was derived using least-squares linear regression (y = mx + b; m = slope) after entering the data into a spreadsheet (Excel; Microsoft); [22] and maximal workload (Watt max ), which was defined as the highest workload maintained for a minimum of 45 seconds.

Muscle Sympathetic Nerve Activity measured with microneurography
Resting muscle sympathetic nerve activity (MSNA) recordings were obtained for all patients at both pacing rates; all patients were studied using the same set-up and were placed in the semirecumbent position after they had rested for a minimum of 30 minutes. Tungsten microelectrodes (Iowa Bioengineering) were inserted into the common peroneal nerve just below the knee (typically the left peroneal nerve). The correct position of the microelectrode tip is within close proximity to the post-ganglionic sympathetic nerve fiber bundles that innervate the vascular smooth muscle of the peroneal muscle or anterior tibial muscle vascular beds. These bundles are typically located close to the motor neurons innervating the same muscle group and close to the afferent nerves from these muscle groups. Intraneural stimulation was used to locate the microelectrode position by searching for muscle twitching and by testing muscle activation by tapping the muscle belly to engage the muscle stretch receptors. The firing of the sympathetic fibers was then recorded in the resting leg by pre-amplifying the signals (x1000) and by filtering (bandwidth 700-2000 Hz, half cut-off filters), rectifying and integrating the signals with a time-constant of 0.1 second and then further amplifying the signals (x95.5) (Iowa Bioengineering). The resulting integrated neurograms were then studied, and only the recordings with bursts that reached a signal:noise ratio of at least 3:1 were accepted for analysis. A minimum of 10 minutes was allowed for the recording and for the patient to settle after manipulating the needle; the subsequent 10 minutes of MSNA were used for the analysis. All data were analog-digital converted and saved for later analysis (ADInstruments, LabChart 7.0). The analysis was performed using macro-programming in LabChart in a manner in which all cardiac cycles were tested for significant sympathetic activity, and sympathetic bursts were counted if the amplitude and area under the curve (method: Integral relative to Baseline) [23] reached certain predetermined levels that were normalized to the noise level. The validity of these semi-automated analyses were tested by an experienced microneurographer (MS) who was blinded to the macro-result and who compared the results to manual scoring; on average, differences of less than 10% were found for the 6 conditions (3 patients) tested.
MSNA bursts are strictly synchronized to the cardiac cycle by the arterial baroreflex. Thus, arterial baroreceptors are unloaded to the extent to which they allow sympathetic outflow from the brainstem only during diastole. For every heart beat (and every diastole), only one integrated sympathetic burst can occur, but MSNA bursts do not occur with all heart beats (the cycles in which the diastolic pressure is too high to unload the baroreceptors are not followed by a sympathetic burst). In the present study, we manipulated the HR and therefore also manipulated the maximum possible MSNA burst frequency. For this reason, the MSNA burst frequency (bursts/minute) and the MSNA burst incidence were analyzed after being normalized to the HR (bursts/100 RR).

Echocardiography
Standard echocardiography was performed according to the European Association of Echocardiography guidelines. [24] The LVEF, left ventricular end diastolic dimension (LVEDD), and end systolic dimensions (LVEDS) and tricuspid annular plane systolic excursion (TAPSE) were measured.

Blood samples
NT-proBNP plasma levels were analyzed at baseline and at the end of each phase using electrochemiluminescence immunoassays. The patients were allowed to rest for 15 minutes before the samples were taken.

Study endpoints
The primary study endpoints were changes in NT-proBNP and sympathetic activity, as assessed by MSNA. Additional endpoints were exercise capacity in the form of pVO 2 and pVE/ VCO 2 as well as self-perceived QoL.

Safety
An adjudication committee oversaw the study. If the baseline NT-proBNP increased more than 50% or if the NYHA class deteriorated in more than 5 patients during the fast-pacing period, the study would have been terminated early.

Statistical Analysis
Continuous variables are presented as the means ± standard deviations (SDs), and categorical variables are presented as frequencies or percentages (%). All variables were tested for normality using the Kolmogorov-Smirnov test and normality plots. NT-proBNP underwent logarithmic transformation to obtain a normal distribution.
To rule out carryover and time effects in a crossover trial, the sum of the variables measured in the two periods were calculated for each subject and compared to the two sequence group means using an unpaired t-test. [25] There are arguments for [26,27] and against [28] the use of baseline measurements in crossover designs; therefore, a two-way repeated measures analysis of variance (ANOVA) of baseline values and a paired Student's t-test (excluding baseline values when comparing the two pacing rates for endpoints) were performed. Furthermore, a sensitivity analysis comparing baseline MSNA values to DDD-60 study period was performed, along with an intra-class correlation to estimate the MSNA variability. Wilcoxon signed-rank tests were used for non-parametric statistics, as appropriate. To enhance the understanding of the magnitude of an observed effect, we calculated the effect size (parametric r = ffiffiffiffiffiffiffi ffi , t 2 = t-score, df = degrees of freedom; non-parametric r = z ffiffi ffi N p , z = z-score, N = number of total observations), as suggested by Fritz et al. al. [29], and r = 0.10, r = 0.3, and r = 0.5 constituted small, medium, and large effects, respectively. A P value of <0.05 was considered significant. All statistical analyses were performed using SPSS 1 (version 19) statistical software (Chicago, IL, USA).

Power calculations
These estimates are linked with high uncertainties since few studies have conducted MSNA measurements on CRT patients (hence, this pilot trial). Previous studies [30,31] show a standard deviation for MSNA of 9 bursts/min and 12 bursts/100RR, and a relevant difference of 20 and 14, respectively. Sample size calculation is based on the paired difference model. With a power (1-beta) of 80% and a two-sided alpha of 5%, the null hypothesis can be rejected with a paired sample size of 8. [32]

Results
The baseline characteristics of the patients are shown in Table 1. The cohort comprised 6 males and 6 females, the mean patient age was 66 ± 9 years, and BiV pacing was present in 97±4% of the patients. All patients were receiving beta-blocker and either an angiotensin-converting enzyme inhibitor or an angiotensin receptor blocker. The pre-tests showed no carry-over or time effects for the endpoints in the sequence groups. There were no significant changes in the levels of serum hemoglobin, potassium, sodium or creatinine during the study, and the data from the ventricular histograms revealed a significant difference in the mean HR between DDD-60 versus DDD-80 pacing (the mean difference was 8.7 ± 6 bmp, p = 0.001). The atrial pacing level was elevated during the 80-bpm period compared to the 60-bpm period (82±14% vs. 8±8%, p = 0.001), Fig 2. The crude median NT-proBNP after the DDD-60 study period was 59 pmol/L [interquartile range (IQ), 12.9-106.8] compared to 75 pmol/L [IQ, 13.1-132-8] after the DDD-80 study period. However, no significant difference in log-NT-proBNP was observed between the two pacing rates by either the paired t-test (p = 0.3) or ANOVA (p = 0.44) , Fig 3. Sympathetic activities were available for 10 patients (it was not possible to locate the peroneal nerve in 2 patients), and they dropped significantly during the DDD-80 study period compared to the DDD-60 period (51±14 bursts/100 RR vs. 64±14 bursts/100 RR, respectively; p<0.05; effect size, r = 0.73). With the exception of one patient (no. 6), all patients experienced decreased sympathetic activity during the DDD-80 study period compared to the DDD-60 study period. Between the two study periods, there was no significant change in the bursts per minute (41±11 burst/min in DDD-80 period vs. 43±7.7 burst/min in the DDD-60 period, p>0.52). A sensitivity analysis was performed where baseline measurements were compared to DDD-60 period (only patients with DDD/DDD-R = 60 at baseline, n = 9). Baseline MSNA values (57 burst/100 RR) were not significantly different from the DDD-60 period, p>0.11 (results not shown). A high degree of reliability was found between the two measurements. The average intra-class correlation was 0.781 (CI:0.144-0.949, p<0.01), Figs 4-5.
All patients completed the exercise tests, and they were limited due to breathlessness or fatigue. The majority of participants did not achieve RER>1.0; and at both pacing rates, the mean RER equaled 0.94±0.11. After the DDD-80 study period, the mean pVO 2 increased nonsignificantly to 16.7±6 mL/min/kg, compared with 15.6±6 mL/min/kg after the DDD-60 study period (p = 0.43). The pVE/CO 2 slope dropped to 30.2±5.9, at 80 bpm, from 32.3±6.1, at 60 bpm, (p = 0.17). The maximal workload marginally improved in the 80-bpm study period (88 ±31 W vs. 85±40 W during the 60-bpm period, p = 0.83), and the exercise duration remained unchanged during the study.
The SF-36 Health Survey and LIhFE were completed by all patients (except for one patient in the 80-bpm study period) to assess QoL. SF-36, which included the 8 dimensions of health perception, revealed a significant difference in the "Bodily Pain" score (a high score indicates the absence of pain): 67±23 in the 60-bpm study period versus 59±30 in the 80-bpm study period (p<0.03, r = 0.45). Furthermore, the component score of Mental Health was significantly higher in the 60-bpm study period compared to the 80-bpm study period (53±11 vs. 49 ±11, respectively, p<0.04, r = 0.43). There were no significant differences in the scores of the dimensions of health perception or physical health. The LIHFE survey did not reveal any significant differences.
No significant changes were observed in the echocardiography or blood pressure measurements (systolic, diastolic, and mean arterial pressure) between the two study periods.

Discussion
The present pilot trial examined the effects of a basal pacing rate of 80 bpm compared to 60 bpm in CRT patients in an array of physiological and biometrical parameters. We found no beneficial or detrimental effects of 80 bpm in terms of the NT-proBNP levels, peak oxygen uptake, or exercise duration compared to 60 bpm. However, we found a significant drop in sympathetic activity, as measured by bursts/100 RR, in the 80-bpm study period, which represented a large decrease in the levels of sympathetic activity, as measured by the effect size. Yet, we failed to demonstrate any significant decrease in bursts/min. Interestingly, the patients reported higher scores (indicating less limitation) in the QoL surveys in some health dimensions during the 60-bpm study period. To the best of our knowledge, this study is the first to conduct a series of direct measurements of the sympathetic nervous system in CRT patients, with varying atrial pacing rates in an intermediate 3-month time period. Ståhlberg et al. observed that short-term increased atrial pacing (80 bpm vs. 60 bpm), significantly decreases the average estimated pulmonary artery diastolic, right ventricular systolic and diastolic pressures, and increases cardiac output. However, the same parameters appeared to increase during the 2-week, 80-bpm period, raising concerns for the long-term effects of HR elevation. Furthermore, in accord with our study, no significant changes were observed in the BNP levels or in the 6-minute walk test results between the 80-bpm and 60-bpm study periods. [8] The absence of clinical effects may be attributed to subtle changes that occur but cannot be  detected by usual clinical tests. In a newly published paper, by the same author group, on the acute effects of increasing CRT pacing rates on hemodynamic control and MSNA a very similar combination of changes were reported. During 10 minutes of increased pacing rates (from 50 to 90-bpm) a decreased MSNA burst incidence of 44% was reported (about 35% decrease with the step from 70 to 90-bpm). [33] Recently, in an elegantly designed trial (PEGASUS) [34], 1,309 heart failure patients with newly implanted CRT devices were randomized to a DDD-40, DDDR-40 or DDD-70 group and followed for 10.5 months. No significant differences were observed in the composite clinical endpoint of all-cause mortality, HF events, NYHA functional class, or patient global selfassessments between the different basal pacing rates. A direct comparison between our study and PEGASUS is difficult because of dissimilar endpoints, and data regarding BNP and working capacity (pVO 2 ) is yet to be published from PEGASUS. However, in contrast to our study, the mean HR difference in the PEGASUS trial was only 2-3 bpm. Moreover, the percent atrial pacing was only 43% during DDD-70, while it was greater than 82% during our 80-bpm study period. It could be speculated that a greater difference in HR and atrial pacing, and perhaps a longer follow-up period, are needed to generate a difference in clinical outcome.
It has been well established that elevated sympathetic activity is typical [30] and is well correlated to symptoms, progression and poor prognosis for heart failure, [35,36] and elevated values have been associated with heart failure mortality by Barretto et al. [30] Previously, Kuniyoshi et al. [37] and Hamdan et al. [38] have demonstrated, using MSNA, that sympathetic activity decreases after CRT device implantation. With the lower sympathetic tone at 80 bpm, our data suggest that a further increase in the basal atrial pacing rate could prove to be beneficial for heart failure patients undergoing CRT therapy.
The exact mechanisms behind the lower sympathetic tone during a higher atrial basal pacing rate are not known. Nonetheless, results from previous short-term biventricular pacing studies [39,40] indicate that an arterial baroreceptor-reflex-mediated sympathoinhibition is associated with higher pacing rates. This finding is partly derived from the decreased pulmonary arterial resistance and total peripheral resistance, thus facilitating an afterload reduction. We did not encounter any changes in blood pressure, which could indicate that any changes Optimal CRT Frequency were subtle and not measurable by our equipment and could indicate a high risk of statistical type II error due to the small number of patients. Other studies have confirmed that reducing cardiac filling pressure lowers norepinephrine spill over [41] in HF patients, which is then mirrored by decreased sympathetic activity, as measured by MSNA. [42] Thus, the diminished sympathetic activity during the 80-bpm study period could also be the result of lower cardiac filling pressure. A decrease in cardiac filling pressures would be expected to result in lower NT-proBNP values, which was not observed in the current study, thereby questioning lower filling pressures as an explanation for the decrease in MSNA.
DeMazumder et al. recently demonstrated that cholinergic signaling is highly up-regulated in the failing LV, and moreover, differential remodeling of cholinergic signaling is important for the return of sympathovagal stability by CRT. [43] Whether this novel mechanism can influence muscle sympathetic activity is unknown, but nonetheless suggests that future studies evaluating CRT treatment should implement parasympathetic measurements.

Optimal CRT Frequency
We emphasize that we did not achieve significant difference between the two study periods in burst frequency (MSNA per minute). However, this was anticipated due to manipulation of HR and the fact that MSNA is strictly linked to the cardiac cycle by the arterial baroreflex. The incidence of MSNA bursts is strongly related to the afferent input from the baroreceptors. Thus, sympathetic outflow from the brainstem only occurs in conjunction with diastole and only with those diastoles, which do not activate baroreceptors. This has led to the prevailing view that MSNA burst incidence is an indication of the level of central sympathetic outflow. Burst frequency is usually considered to illustrate the degree of activity reaching the effector organs, whereas Burst incidence reflects central control in the sense that it illustrates the relative amount of cardiac cycles filled with MSNA bursts [44][45][46] Another major concern is falling stroke volumes (SV) with elevated pacing rates. However, the hemodynamic changes suggest that cardiac output is maintained and, in some instances, increased (likely due to increased contractility and decreased filling pressures). These mechanisms seem to be in place up to at least 100-120 bpm, whereas there is a shift toward sympathoexcitation at higher pacing rates. [39,40] BNP is a valuable parameter for monitoring the effects of CRT in HF patients, [47] and prior studies have shown the importance of post-CRT BNP levels in predicting HF events and death. [48,49] Furthermore, Brenyo et al. established that the pattern of BNP change from baseline to 1 year is an independent predictor of subsequent clinical response. [49] We did not encounter a significant increase in BNP levels during the two study periods, which indicates that a basal rate of 80 bpm is not harmful. However, the long-term effects of a higher basal rate beyond three months are unknown.
It was anticipated that many of our patients would not achieve a RER>1.1 (an indicator of an adequately performed test), because evidence suggests that nearly 50% of HF patients are unable to reach this target. [22] Recent data suggest that results below the target RER do not invalidate the key parameters of pVO 2 and the pVE/CO 2 slope; in contrast, these measurements remain as strong prognostic variables, irrespective of RER. Furthermore, the pVE/CO 2 slope is considered to be an effort-independent variable and seems to be superior to pVO 2 as a prognostic variable. [22] We observed an increase in pVO 2 and a decrease in the pVE/CO 2 slope from 60 bpm to 80 bpm; however, these observations were marginal and not significant. The measured range of pVO 2 in our study is comparable to those of previous studies [50] performed in similar patient populations. Furthermore, the baseline values of pVO 2 and of the pVE/CO 2 slope indicate improved exercise tolerance in our study; thus, it is not surprising that exercise capacity did not improve. Likewise, larger trials [51,52] conducted in CRT patients with NYHA II, with integrated power to detect changes in exercise capacity (pVO 2 , workload, etc.), have also been unable to demonstrate a significant improvement in these parameters despite clear improvements in cardiac structure and function. Additionally, a follow-up period of 3 months may not be sufficient to detect exercise improvement. However, the lack of statistical power cannot be ruled out as an explanation for the absence of a significant increase in pVO 2 or decrease in the pVE/CO 2 slope.
The LIfHE questionnaire did not reveal any differences between the 60-and 80-bpm study periods, thus indicating no adverse or beneficial effect of increased atrial pacing on QoL. As stated earlier, SF-36 is comprised of 36 items, which are subdivided into 8 health dimensions. [20] The study population reported higher (better) scores in only two of the 8 health dimensions during the 60-bmp compared to the 80-bmp study period: "bodily pain" and "component score of mental health". This result is surprising because no verbal notification of increased bodily pain or lower mental health were conveyed during or after the 80-bpm period examinations. Moreover, no patient reported any discomfort that resulted in premature discontinuation during the 80-bpm period, and the calculated effect sizes revealed only a medium-level change in these parameters. One possible explanation for less "bodily pain" at 60 bpm could be a lower sensation of palpitation and a frequency closer to the physiological level. In addition, constantly increased HRs might manifest as enhanced bodily excitement/tension/stress and irritation, which might result in lower "mental health" comfort. A recent meta-analysis revealed that patients with NYHA I-II did not enjoy the same improvements in QoL as patients with NYHA III-IV when comparing CRT only with an implantable cardioverter defibrillator (ICD). [19] The majority of our patients were in NYHA class II, and perhaps they already had relatively high baseline scores that were offset by manipulating their pacemaker settings.

Limitations
This pilot study has several limitations. First, we only obtained a modest increase in the mean HR (i.e., 9 bpm) during the "fast" pacing period. This result was a consequence of a relatively high intrinsic HR, which increased the mean HR during DDD-60 pacing. One may think that more separation in the HR measurements would have resulted in a more pronounced difference between the two intervals.
Second, our data are based on constant AV delays and might not apply to different AV delays. However, Scharf et al. demonstrated that longer AV delays, unlike traditional pacemakers, were required to obtain optimal cardiac output with increasing HR when using CRT devices. Thus, we may have observed greater decreased sympathetic activity with dynamic or prolonged AV delays. However, a prolonged AV-delay carries the risk of loss of BiV pacing due to intrinsic conduction. Third, this study lacked norepinephrine measurements; however, a strong correlation with MSNA exists. [53] MSNA provides a more direct and accurate measurement of central sympathetic activity compared to catecholamine dosage or HR variability (not possible to perform due to dependence on intrinsic and normal sinus rhythm). In addition, the small number of participants in this study prevents firm conclusions, particularly when no differences were observed; however, the prospective randomized crossover design and 3-month follow-up provides greater validity and power to this study. Despite the intention to blind patients to their pacing rates, the possibility of detecting the actual pacing rate by sensation or home monitoring cannot be excluded. Moreover, our results apply only to the rates of 60 bpm and 80 bpm and cannot simply be extracted to higher basal atrial pacing rates. Furthermore, no NYHA IV patients were included in this study. Therefore, the results may not be applicable to populations with more severe disease. Yet, with the expansion of CRT to patients with mild HF, [51] this study is still relevant.

Conclusion
A modest increase in the atrial pacing rate for 3 months in non-ischemic CRT patients appears to decrease sympathetic tone (burst incidence only), as directly measured by MSNA. However, Qol (SF-36) revealed less "bodily pain" and better "mental health" during the DDD-60 study period, which is of clinical importance.
These findings may have possible implications for clinical decision-making when programming CRT devices.
Further studies are needed to determine whether increased atrial pacing leads to better long-term clinical outcomes and survival in patients with mild HF symptoms and whether the decreased sympathetic activity benefits seen at 3 months are also present in the long term.