Pacing-induced heart failure: should we avoid right ventricular pacing or not?

This editorial refers to ‘Reduction in unnecessary ventricular pacing fails to affect hard clinical outcomes in patients with preserved LV function: a meta-analysis’ by M. Shurrab et al., on pages 282–288.

Two studies took away the ‘innocence’ from right ventricular pacing (RVP). The DAVID (Dual-Chamber and VVI Implantable Defibrillator) trial has assessed whether prevention of bradycardia prevents bradycardia-dependent ventricular tachyarrhythmia, initiated by long–short cycles. This trial reported that more patients died or developed heart failure with prevention of bradycardia by dual-chamber implantable cardiverter defibrillators (ICDs) than with single-chamber ICDs that basically did not pace.¹ This result prompted calls for ‘the simpler the better’ sounding like a win–win situation: the device that is simpler for the physician to implant and programme is also better for the patient! However, the culprits for worse outcomes in DAVID were not the devices but the physicians who did not reprogramme dual-chamber ICDs and thus caused unnecessary RVP. Death or heart failure hospitalization developed during 18 months in 32% of patients if DDDR resulted in RVP for >40% of the time compared with only 8% if DDDR resulted in RVP for ≤ 40% of the time.² By definition, RVP was unnecessary in the DAVID trial because patients with an indication for pacing were excluded. The results from DAVID were likely so striking because all patients had heart failure at baseline.

These data suggest that RVP accelerates progression of heart failure by inducing ventricular dyssynchrony and should therefore be avoided.

The second study that created doubt about the ‘innocence’ of RVP was a post hoc analysis of the MOST (MOde Selection Trial) study. It showed that in patients with sinus node disease and narrow QRS at baseline, unnecessary RVP increased the risk of heart failure hospitalization and atrial fibrillation.³ Of note, patients in MOST had a normal left ventricular ejection fraction (LVEF).

These two analyses and trials such as SAVE-PACe (Search AV Extension and Managed Ventricular Pacing for Promoting Atrioventricular Conduction)⁴ formed the scientific basis for the solid opinion that RVP should be avoided by all means. As a result, dual-chamber devices were either programmed to long atrioventricular (AV) delays (in my practice routinely to 350 ms in sinus node disease) to allow spontaneous AV conduction, to automatic prolongation by AV delay hysteresis functions (up to 600 ms in some devices), or to automatically switching the pacing mode to AAI(R) during intrinsic AV conduction.

Just prolonging the AV delay still causes RVP after loss of normal AV conduction, e.g. after 350 ms, and triggers rhythm disturbances such as endless loop tachycardia, repetitive non-reentrant ventriculoatrial (VA) synchrony,⁵ or retrograde VA conduction with cannon waves. Algorithms switching from AAI(R) to DDD(R) have the advantage of not applying RVP with a long AV interval if loss of normal AV conduction occurs for a single beat. Therefore, algorithms such as MVP^® or SafeR^® that operate in AAI(R) mode during spontaneous PR conduction may have advantages in preventing unnecessary RVP.

However, the subsequent MVP (Managed Ventricular pacing vs. VVI 40 Pacing) trial failed to demonstrate non-inferiority of the MVP^® algorithm.⁶ In 1030 patients with an ICD and no indication for antibradycardia pacing, atrial pacing at 60 bpm did not reach the statistical significance to demonstrate non-inferiority to VVI, 40 bpm. A subgroup analysis showed that patients with an unpaced baseline PR interval ≥230 ms had significantly more events (death, heart failure hospitalization, and heart failure urgent care) if randomized to AAIR pacing at 60 bpm with the MVP^® algorithm. Similarly, the PreFER MVP (Prefer for Elective Replacement Managed Ventricular Pacing) trial showed a 3.4-fold increased risk for the development of persistent atrial fibrillation in patients randomized to MVP^® who had a baseline PR of >230 ms.⁷

Also for the SafeR^® algorithm, randomized trials with clinical events such as all-cause mortality, development of persistent or permanent atrial fibrillation, or heart failure did not demonstrate a benefit. The ANSWER (EvaluAtioN of the SafeR mode in patients With a dual chambER pacemaker indication) trial (including also patients with permanent AV block) showed a reduction in RVP from 94% to 12% with the SafeR^® algorithm.^8,9 However, there was no difference in the rate of death, syncope, hospitalization for heart failure, atrial fibrillation, or cardioversion. In the CAN-SAVE R (The CANadian Multi-Centre Randomized Study-Spontaneous AV Conduction pReservation) trial in which approximately half the patients had AV block as a pacing indication,¹⁰ the SafeR^® algorithm achieved a reduction of RVP to basically 0%. However, this randomized trial also did not show any impact on clinical endpoints.

Therefore, it is disappointing but not entirely surprising that the meta-analysis of seven randomized trials on prevention of RVP did not find any impact on clinical endpoints.¹¹ But we face a problem: on one hand, we have abundant evidence for deleterious effects of RVP such as dyssynchrony on echocardiography, deterioration of haemodynamic measurements, pathological alterations in histology, and posthoc analyses with increased risk of developing heart failure and atrial fibrillation. On the other hand, we have a number of randomized controlled trials that all show no clinical advantage of prevention of RVP. How can we solve this contradiction?

My favourite 10 excuses if prospective studies do not show what I want to see

Results from randomized controlled trials contradicting previously published results that have already established a firm scientific opinion are difficult to bear as they seem to contradict the common sense, and ask for an explanation. The meta-analysis by Shurrab et al.¹¹ published in this issue of EP—Europace is hard to accept: prevention of RVP should not be better than permitting unnecessary RVP? We have learned the hard way in DAVID and MOST that RVP is bad, how can it be that suddenly in 2016 it does not matter anymore?

My favourite 10 ‘cheap’ excuses why this meta-analysis does not show what I want to see are summarized in Table 1. A mean follow-up duration of 2.5 years (only 1 year in two out of seven studies!) is too short to see the effect in patients with normal LVEF. The effect may easily take 5 years to become evident! Even a sample of 2000 patients with RVP prevention is too small to detect its influence on mortality. If the development of atrial fibrillation is multifactorial, how much more multifactorial is all-cause mortality?

Looking at the studies included in this meta-analysis:¹¹ what sense does it make to apply algorithms that prevent RVP in patients with permanent third-degree AV block? These data may address the safety of RVP prevention algorithms but render a trial on RVP prevention futile. The authors admit that cross-over, upgrades to cardiac resynchronization therapy, and the use of different algorithms for RVP prevention have polluted the results to an extent that makes it difficult to draw any strong conclusions. And analyses are incomplete: the amount of atrial or ventricular pacing, ventricular pacing sites, programmed lower rate limit, etc. are all not taken into account. And wouldn’t it be more convincing to see the New York Heart Association (NYHA) functional class, brain natriuretic peptide (BNP/NTproBNP), LVEF, and the exact atrial fibrillation burden?

No attempt has been made to identify subgroups of patients where prevention of RVP is advantageous or disadvantageous. This would have been important for patients with sinus node disease and a baseline PR interval >230 ms. Implantation of atrial and ventricular leads, programming and co-medication have not been standardized which renders the interpretation of results difficult.

Interpretation of statistics may be misleading: Although no comparison was statistically significant, we can clearly see for all endpoints an odds ratio in favour of prevention of RVP, it just always missed the ‘magic’ P < 0.05. For heart failure, the P-value of 0.09 indicates that with more patients and a longer follow-up, this comparison could have been statistically significant. This meta-analysis shows that atrial fibrillation, heart failure, and all-cause mortality are influenced by so many factors that it is difficult to assess the effect of preventing one single factor.

A good study poses new questions. The criticism provocatively outlined above questions the meaning of the results of the meta-analysis by Shurrab et al.¹⁰ And doubt is a root of scientific research.

What can we learn from results that contradict widespread opinion?

The meta-analysis by Shurrab et al.¹¹ certainly leaves a number of question marks—which is good! Pacing therapy does not follow a ‘one-size-fits-all’ pattern. Potential factors of influence, listed in Table 2, emphasize the complex interaction between intrinsic rhythm, underlying heart disease, pacing sites, and programmed pacing parameters.

We have learned that RVP can cause or worsen heart failure and therefore recommend cardiac resynchronization therapy in patients with reduced LVEF and high-degree AV block in need of ventricular pacing.¹² We still believe that it is better to avoid RVP in patients with reduced LVEF if possible (e.g. in sinus node disease). We are not so sure about patients with preserved LVEF. In a recent long-term follow-up study of 991 patients with normal or only moderately reduced LVEF, only 6% developed significant deterioration of LVEF during 44 months.¹³ In patients with normal LVEF, the intrinsic and the paced PR interval may decide whether it is better to pace or not to pace the ventricle. There is evidence that also first-degree AV block may worsen heart failure¹⁴ and that therefore RVP should not be decreased to 0% by all means.¹⁵ First-degree AV block can cause mitral regurgitation, shorten diastolic filling, and cause pacemaker syndrome¹⁶ (Figure 1). There may be a turnaround point: if the intrinsic PR is >230 ms, RVP may be better than prevention of RVP. If the atrium is paced, an interval from the stimulus to the intrinsic R wave of >270 ms may be worse than RVP at optimal AV delay.¹⁷ This has not been investigated in the randomized trials on RVP prevention and also not in the meta-analysis by Shurrab et al.¹¹ and may be the key to understanding the conflicting results on the benefit of prevention of RVP.

Figure 1

Why prevention of RVP may not always be the best option. Paper speed 25 mm/s. In a patient with a dual-chamber pacemaker, atrial pacing is followed by an intrinsic QRS complex after an AV delay of 480 ms, permitted by a special algorithm. This causes the P wave almost to overlap with the preceding T wave. Haemodynamically, this can be equivalent to VVI pacing with retrograde VA conduction causing pacemaker syndrome and a reduction in cardiac output.

Open in new tabDownload slide

An important factor is the amount of atrial pacing. This depends on the programmed lower rate limit and on the indication for pacing. Prevention of RVP may be useful after an intrinsic P wave with normal AV conduction. However, in the same patient, pacing from the right atrial appendage may cause prolongation of the P wave and of AV conduction to such an extent that prevention of RVP may be haemodynamically worse than RVP. The occurrence of pacing-induced first-degree AV block is frequent in atrial pacing from the lateral wall and the right atrial appendage. Pacing from the atrial septum or roof near the insertion of Bachmann’s bundle can avoid pacing-induced first-degree AV block and frequently even shorten the PR interval after atrial pacing. More scientific efforts should be undertaken to assess whether atrial septal pacing in combination with prevention of RVP can improve the clinical results of permanent pacing.

Conclusions

This meta-analysis is disturbing and thus important: looking at the biggest randomized trials, there is no prospective evidence that prevention of RVP accomplishes clinical improvements suggested by posthoc analyses. Likely, prevention of RVP is not beneficial if forced by all means. Negative effects of pacing-induced first-degree AV block may counteract and outweigh the benefits of RVP prevention. Means to prevent pacing-induced first-degree AV block by implanting atrial leads at electrophysiologically more suitable sites may improve clinical results of pacing, particularly in patients with sinus node disease. To be assessed in prospective randomized trials after almost 50 years of cardiac pacing, there are still very basic questions such as where to place the atrial and ventricular leads.

Conflict of interest: Honoraria, travel grants (all modest, < 10,000 €) and Speaker’s bureau include Biotronik, Boston-Scientific, Medtronic, Sorin/LivaNova, and St. Jude Medical. Advisory board: Medtronic (European CRM Advisory Board) and St. Jude Medical (International Lead Advisory Board); Participants of sponsored trials/research: Medtronic (OptiLink study) and Sorin/LivaNova (OSCAR registry).

References

Author notes