Right ventricular pacing for hypertrophic obstructive cardiomyopathy: meta-analysis and meta-regression of clinical trials

Abstract Aims Right ventricular pacing for left ventricular outflow tract gradient reduction in hypertrophic obstructive cardiomyopathy remains controversial. We undertook a meta-analysis for echocardiographic and functional outcomes. Methods and results Thirty-four studies comprising 1135 patients met eligibility criteria. In the four blinded randomized controlled trials (RCTs), pacing reduced gradient by 35% [95% confidence interval (CI) 23.2–46.9, P < 0.0001], but there was only a trend towards improved New York Heart Association (NYHA) class [odds ratio (OR) 1.82, CI 0.96–3.44; P = 0.066]. The unblinded observational studies reported a 54.3% (CI 44.1–64.6, P < 0.0001) reduction in gradient, which was a 18.6% greater reduction than the RCTs (P = 0.0351 for difference between study designs). Observational studies reported an effect on unblinded NYHA class at an OR of 8.39 (CI 4.39–16.04, P < 0.0001), 450% larger than the OR in RCTs (P = 0.0042 for difference between study designs). Across all studies, the gradient progressively decreased at longer follow durations, by 5.2% per month (CI 2.5–7.9, P = 0.0001). Conclusion Right ventricular pacing reduces gradient in blinded RCTs. There is a non-significant trend to reduction in NYHA class. The bias in assessment of NYHA class in observational studies appears to be more than twice as large as any genuine treatment effect.


Introduction
Hypertrophic cardiomyopathy (HCM) is the most common cause of sudden death in young adults with a prevalence of 0.2%. 1 When asymmetrical septal hypertrophy leads to left ventricular outflow tract obstruction (LVOTO) with associated systolic anterior motion of the anterior mitral valve leaflet, this confers a diagnosis of the subphenotype hypertrophic obstructive cardiomyopathy (HOCM, also known as 'obstructed HCM'), which occurs in around 70% of HCM sufferers. 1 Alongside contributing to mortality through multiple mechanisms, including heart failure and malignant arrhythmia, LVOTO produces significant morbidity, inducing symptoms of chest pain, breathlessness, exertion intolerance, light-headedness, and syncope.
Management of symptomatic LVOTO is initially pharmacological but interventions are available in the form of surgical septal myectomy or percutaneous alcohol septal ablation, both carrying risks of complication. Failure, intolerance and reluctance with pharmacological and interventional treatment of LVOTO led to . . . . interest in the use of right ventricular pacing as an alternative method for gradient reduction and resultant symptomatic improvement. 2,3 A number of studies, both randomized and observational, have investigated the effect of dual chamber pacing in HOCM. We systematically analysed these to quantify the effect on LVOTO, left ventricular systolic function, symptoms and functional status.

Methods
We carried out a meta-analysis of studies that evaluated right ventricular pacing in HOCM in accordance with PRISMA guidelines. 4 We prospectively registered this meta-analysis on the PROSPERO international register of systematic reviews (registration number CRD42017062165). 5

Search strategy
We searched PubMed and the Cochrane Central Register of Controlled Trials for any studies [randomized controlled trials (RCTs), non-RCTs, and uncontrolled observational studies] published in the English language where adults with HOCM underwent atrial-synchronous right ventricular pacing. Studies were included if prospectively determined clinically relevant outcomes were reported: left ventricular outflow tract gradient (LVOTg), New York Heart Association (NYHA) functional status, left ventricular ejection fraction, exercise duration, and peak oxygen uptake during cardiopulmonary exercise testing. Systematic reviews were examined for references to relevant studies. Any discrepancies were resolved by consensus. The full search strategy is reported in the Supplementary material online.
The study protocol was drafted by A.D.A. and revised by all coauthors. Preliminary search and eligibility analysis was performed by K.C. and Y.A. F.d.V. and A.D.A. performed an independent preliminary search.

Inclusion and exclusion criteria
We considered all studies of right ventricular pacing in HOCM. Studies were eligible if they recruited patients with an elevated gradient (>30 mmHg). 1 We excluded case reports, studies of biventricular pacing, or studies where pacing occurred was delivered in combination with other invasive gradient reduction interventions such as septal ablation or myomectomy.

Endpoints
The primary efficacy endpoint was change in gradient (DLVOTg). The secondary efficacy outcomes were change in symptomatic and functional status measured by NYHA class, exercise time, or peak oxygen uptake, from baseline to follow-up. Detailed inclusion criteria for endpoints are found in the Supplementary material online.

Follow-up
It was anticipated that the search would reveal trials of varying follow-up duration. We prospectively determined that DLVOTg would be analysed according to the following groups of mean follow-up durations: immediate (<12 h), short-term (12 h to 6 months), medium-term (>6 months to <2 years), and long-term (>2 years). This would allow testing for progressive change in LVOTg due to remodelling over time. 6 Randomized studies were pre-specified to be analysed separately from observational studies with comparison between RCTs and non-randomized studies of similar follow-up duration. The statistical programming environment R with the metafor package was used for all statistical analysis. Random-effects meta-analyses were performed using the restricted maximum likelihood estimator. For ordinal categorical outcomes (NYHA), Agresti's generalized odds ratios (ORs) were first calculated for each trial before meta-analysis. 7,8 Interactions between groups were assessed using a mixed effects meta-analytical model with the variable in question as a moderator. We used the I 2 statistic to assess heterogeneity. Included RCTs were assessed using the Cochrane risk of bias tool.

Risk of bias
Trial quality for the four RCTs, assessed by the Cochrane risk of bias assessment tool, is shown in Table 3. Three were rated as low risk of bias due to double blinding but the fourth was single blinded. Lack of randomization and lack of blinding put all 31 observational studies at risk of bias.

Left ventricular outflow tract obstruction
Thirty-two studies reported mean change in LVOTg (DLVOTg) from baseline for at least one follow-up duration. The results are shown in Figure 2 for observational studies, Figure 3 for RCTs, and summarized in Table 4

2013
Observational single-arm (retrospective) 'longest interval associated with complete ventricular capture, at rest and during exercise. Radiofrequency modification of the AV junction was performed in patients whose short spontaneous PR interval precluded the complete capture of the ventricles. In patients whose P wave duration was > _120 ms, a third lead was placed in the coronary sinus and connected to a biatrial DDD pacemaker to resynchronize the atria'  Values are represented as mean ± standard deviation unless otherwise stated. Study type refers to arms of trials fulfilling inclusion criteria. Single-arm studies refer to studies where there is either one arm or where other arms are not control groups with conventional/medical/non-interventional therapy (instead they are alternative therapies). Optimal AV delay selection is the description of optimal AV delay selection in the wording of the source text with changes to wording made only to paraphrase and abbreviate. Optimal AV delay methods refer to the techniques used in determining the optimal AV delay. AV delay refers to the AV delay used in the trial; where the AV delay is specifically stated to be sensed AV delay, this is acknowledged by (s), and where paced (p).

ECG
a Unclear from publication whether invasive or echocardiographic but wording suggests invasive measurement. b Precise duration of follow-up not stated but longer than 1 year.
c Non-responders in initial acute study did not undergo implantation of pacemaker. d This study also included an arm where pacing was optimised to peak aortic flow but LVOTg response to this was not reported for all patients (only for responders and non-responders).
e Long-term follow-up data.
f Long-term follow-up of Fananapazir et al. 13 g Crossover periods were for 3 months but after the crossover study 6 months of DDD pacing occurred. In the four RCTs (follow-up ranged from 1 months to 3 months), pacing reduced LVOTg by 35%. The observational studies reported slightly larger LVOTg reductions ( Table 4).
We, therefore, compared gradient effect size of RCTs with observational studies at similar follow-up times (denoted short-term). The observational studies reported a gradient reduction, which was 18.6% greater than RCTs (CI 1.3-36, P = 0.0351).

New York Heart Association status
As pre-specified, because there were too few studies reporting NYHA, we did not sub-divide them by follow-up duration. The results are displayed in Figures 4 and 5 and summarized in Table 4.
In RCTs, there was a trend towards improved NYHA class with pacing (OR 1.82, CI 0.96- 3.44; P = 0.066) but with considerable heterogeneity (I 2 = 81.7%) as a result of one trial showing a particularly prominent favourable effect.
In contrast, almost all of the observational studies (8/9) reported marked improvement in NYHA class giving a combined OR of , P < 0.0001). This OR is over 450% of the OR for RCTs: ratio of ORs 4.54 (CI 1.61- 12.82, P = 0.0042).

Exercise and systolic function
For each of the other variables (ejection fraction, exercise duration, and peak oxygen uptake) fewer than five studies reported the data require, and therefore, meta-analysis was not conducted. Reported changes in ejection fraction ranged from þ3% to -11%, and in exercise duration from þ0.3 min to þ3.1 min. There was a single RCT report of peak oxygen uptake change, -0.1 mL/kg/min.

Discussion
This is the first meta-analysis assessing the role of right ventricular pacing as a treatment for LVOTO in HOCM. We found that, in blinded RCTs, pacing reduces LVOTg and shows a non-significant trend to reduce NYHA class. Unblinded observational studies report very much larger symptomatic effects, suggesting an unintentional bias much larger than any genuine effect of the pacing. Echocardiographic gradient assessments, which appear less vulnerable to this bias, suggest a progressive enhancement of the therapeutic effect with the passage of time.

Left ventricular outflow tract gradient reduction
The meta-analysed RCT data strongly supports the concept that pacing reduces measured LVOTg. There are broadly four mechanisms by which pacing can have an influence. Right ventricular pacing causes incoordination of ventricular activation which may reduce the driving force of ventricular ejection. The altered ventricular activation sequence attenuates the tendency of the left ventricular outflow tract (LVOT) lumen to become very small during systole. Atrioventricular (AV) sequential pacing alters ventricular filling through a change in AV delay, which impacts on ventricular ejection. Finally, knowledge that the patient is receiving a treatment believed to decrease LVOTg may cause an unintended bias of the echocardiographer to try less hard to find a high gradient.
Randomization with allocation concealment ('blinding') is the most effective approach to reduce unintended bias when evaluating therapies. Under blinded conditions each of the trials individually showed that pacing reduced gradient. The measured effect was much larger in the (unblinded) observational studies, -55% rather than -35% (P = 0.0351 for difference in study designs). While it is not certain that the much larger effect sizes reported by unblinded studies is due to unintended bias, it is difficult to imagine that the patient groups or procedural characteristics were so markedly different between the two study designs.

Symptoms without blinding
The blinded RCTs showed an encouraging trend towards a statistically significant reduction in NYHA class. A very much larger reduction in NYHA class was reported by the unblinded observational studies. The pooled estimates were so far apart their CIs did not overlap. Indeed, the point estimates for eight out of nine observational studies were for a greater NYHA reduction than even the highest upper limit of the CI of any individual RCT.
We conclude from this that there may be a symptomatic benefit, but that unblinded study design provides no useful information on it. We cannot even use the unblinded symptomatic relief data to compare different studies in order to select pacing approaches for future blinded trials. This is because the great majority of the unblinded symptomatic relief appears to be bias. Therefore, the differences between the unblinded symptom-relief effects reported by different studies will be dominated by differences in the amount of bias rather than clinically meaningful differences in protocol.
The origin of bias in unblinded observational studies investigating subjective outcomes, such as NYHA, may be due to the anticipation

Clinical implications
Knowing whether AV sequential right ventricular pacing is beneficial in HOCM is important because many patients with HOCM require a defibrillator and the modern era offers two changes from the traditional dual-chamber, transvenous defibrillator. First, omitting the atrial lead can reduce complication rates 43 but it prevents AV sequential pacing. Second, subcutaneous defibrillators are now available, which are easier to remove than trans-venous defibrillators, but do not have any pacing function.
If AV sequential right ventricular pacing genuinely helps patients, then dual chamber, transvenous defibrillators would be preferable over both single chamber transvenous and subcutaneous defibrillators. Although the blinded RCTs show only a non-significant trend to reduction in NYHA class, the point estimate of the pooled effect size is an OR of 1.82. Roughly speaking, this means patients are twice as likely to feel better with pacing switched on rather than off. This is a potentially meaningful clinical benefit and merits further investigation through further blinded RCTs. Such RCTs need not be expensive or resource-intensive, recruiting patients undergoing de novo implantation. They could be conducted in patients who already have a defibrillator, with two randomized periods (pacing on vs. off) and appropriately blinded evaluation. A 2012 Cochrane review 2 of observational and randomized studies concluded, as we do, that the existing observational studies are of low quality due to inadequate blinding but they did not quantify this effect as we have done. They also call for high quality trials to investigate the potential for a true symptomatic benefit.  The clear evidence of gradient reduction by pacing should not be assumed to prove that symptoms also improve. This is because pacing (i) reduces the force of contraction by inducing incoordination, (ii) reduces the impingement of the LVOT during systole, and (iii) necessarily alters filling since the paced activation must begin before the native activation would have otherwise occurred. The symptomatic effect will therefore be a result of not only changes in LVOT calibre but also ventricular filling and ejection. The choice of AV delay may be important. Applying a very short AV delay almost always reduces stroke volume, even if systolic LVOT calibre is increased. It is unclear how to programme the AV delay of a pacemaker in a patient with HOCM, to provide a net advantage. Each of the studies has taken a different approach, which may have contributed to the lack of translation of gradient reduction to symptomatic benefit, and more work on this is required. The TRICHAMPION trial is underway, which is examining the role of AV nodal ablation, for complete control of AV delay, with biventricular pacing. Importantly, not all symptoms contributing to NYHA status in HCM are due to LVOTO, some maybe due to the cardiomyopathic process itself or comorbidities, and thus unaffected by pacing.
The meta-regression of gradient with respect to follow-up duration suggests a progressive enlargement of the therapeutic effect with longer follow-up across all published data, as has been noted before within single studies. 34,40 It is not yet clear what the mechanism for such a progression might be. One possibility is progressive changes in the structure of the ventricle. However, it should be remembered that long-term right ventricular pacing is known to cause left ventricular function to deteriorate, and therefore, a longterm reduction in gradient should not automatically be assumed to be beneficial.

Limitations
There were only four RCTs, although all were blinded. Studies mostly did not report lead position or baseline QRS morphology, preventing separate analysis of the apical and septal positions or the presence of pre-existing bundle branch block or meta-regression for these variables. The apical lead position may be expected to result in more dyssynchronous activation and thus greater 'beneficial' LVOTg reduction, but this may be offset, or even superceded, by dyssynchrony-related reduction in myocardial performance. This has importance in procedure technique as achieving an apical lead position in severe hypertrophy can be technically challenging.
Many studies reported the mean baseline (unpaced) and the mean follow-up (paced) LVOTg and NYHA class with a categorical description of the P-value (e.g. 'P < 0.05') for the statistical test for differences between them. The most useful value, however, would be the mean intra-patient change in LVOTg along with its CI, standard error, standard deviation, or precise P-value. This has limited the precision with which we can calculate the CI for the pooled estimate for the change in LVOTg and NYHA, but has not affected the point-estimate.

Conclusions
The blinded RCTs show that AV sequential right ventricular pacing reduces LVOTg in HOCM, and shows a trend to benefit in symptoms that is not statistically significant but of a potentially clinically meaningful size. This may have implications for the choice of defibrillator to implant in HOCM. More research is needed into appropriate selection of AV delay. More blinded RCTs on symptom relief are required and they need not be very demanding or expensive.
Unblinded observational studies report substantially larger gradient reductions and very much larger symptom reductions. The difference between the effect sizes reported by the two study designs (RCT vs. observational) is so large that observational data are of uncertain value in progressing the field. This is particularly so for symptoms, where the unblinded effect sizes appear to be scaled up by 450%.

Supplementary material
Supplementary material is available at European Heart Journal -Quality of Care and Clinical Outcomes online.