Efficacy and safety of baroreflex activation therapy for heart failure with reduced ejection fraction: systematic review

Abstract

Baroreflex activation therapy (BAT) is a possible adjuvant treatment for patients with heart failure with reduced ejection fraction (HFrEF) who remain symptomatic despite optimal medical therapy and may be an alternative therapy in patients with contraindications or drug intolerance. Our aim was to evaluate the efficacy and safety of BAT in patients with HFrEF. The protocol for this study was registered with PROSPERO (CRD42022349175). Searches were conducted using MEDLINE, preMedLine (via PubMed), EMBASE, Cochrane Library, Web of Science, Trip Medical Database, WHO International Clinical Trials Registry, and ClinicalTrials.gov. We included randomized controlled trials that compared the effects of BAT with pharmacological treatment. We assessed the risk of bias of each study using the Cochrane RoB2 tool and the certainty of the results using the GRADE approach. We performed a meta‐analysis of treatment effects using a fixed‐effects or random‐effects model, depending on the heterogeneity observed. Two studies were included in the meta‐analysis (HOPE4HF and BeAT‐HF). The results showed that BAT led to statistically significant improvements in New York Heart Association functional class (relative risk 2.13; 95% confidence interval [CI, 1.65 to 2.76]), quality of life (difference in means −16.97; 95% CI [−21.87 to −12.07]), 6 min walk test (difference in means 56.54; 95% CI [55.67 to 57.41]) and N‐terminal probrain natriuretic peptide (difference in means −120.02; 95% CI [−193.58 to −46.45]). The system‐ and procedure‐related complication event‐free rate varied from 85.9% to 97%. The results show that BAT is safe and improves functional class, quality of life and congestion in selected patients with HFrEF. Further studies and long‐term follow‐up are needed to assess efficacy in reducing cardiovascular events and mortality.

Introduction

Heart failure (HF) is a chronic and progressive condition caused by functional or structural cardiac changes leading to an increase in intramyocardial pressure or insufficient cardiac output. There are three basic phenotypes, based on left ventricular ejection fraction (LVEF): (a) reduced LVEF (HFrEF), ≤40%; (b) mid‐range LVEF, 41–49%; and (c) preserved LVEF, ≥50%. ¹ Patients with HFrEF account for around half of all HF cases, and there are a number of potential pathological processes at play, including coronary artery disease, cardiomyopathy, endocrine disorders and some infections. ² Differentiating HF patients based on LVEF is important due to their relationship with different underlying aetiologies, co‐morbidities, and, above all, treatment response. ³ , ⁴ , ⁵

Knowing the prognosis for HFrEF in terms of morbidity, disability, and mortality is important for deciding the appropriate type and timing of treatment. ³

Together with recent advances in pharmacological treatments, which continue being the cornerstone in the treatment of HFrEF, there has been huge interest in device therapies as a new focus for HF treatment. Baroreflex activation therapy (BAT) works by inhibiting the sympathetic system and enhancing parasympathetic activity in order to modulate the harmful activation of the autonomic system observed with chronic HF. It is a therapeutic approach currently under investigation as a treatment for refractory hypertension and HFrEF. In August 2019, the Barostim Neo™ BAT device received premarket approval (PMA P180050) ⁶ and is indicated for the improvement of various intermediate endpoints [quality of life, 6 min walk test (6MWT), and functional status], for HF patients with LVEF ≤35%, who remain in New York Heart AssociationA (NYHA) class III or class II (with a recent history of class III) despite optimal medical treatment, and a N‐terminal pro‐brain natriuretic peptide (NT‐proBNP) <1600 pg/mL. This excludes patients indicated for cardiac resynchronization therapy (CRT) according to AHA/ACC/ESC guidelines. ⁶

The aim of this study was to evaluate the efficacy and safety of BAT for HFrEF through a systematic review of clinical trials.

Methods

This study followed the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses, which sets forth the main requirements for clear reporting of information in systematic reviews and meta‐analyses. ⁷

We designed a protocol which was registered in the PROSPERO database with number CRD42022349175; it described the aim of the review, the search strategy, the electronic databases used for the search, the study selection criteria, and the data summary procedures.

Information sources and search strategy

This study was based on a literature search in the MEDLINE (via OVID), PubMed (Ahead of Print/First online), EMBASE, Cochrane Library, Trip Medical Database, Cumulative Index to Nursing and Allied Health Literature, and Web of Science (SCI) databases. We also consulted the International HTA Database hosted by International Network of Agencies for Health Technology Assessment, Agency for Healthcare Research and Quality, Canadian Agency for Drugs and Technologies in Health, and the National Institute for Health and Care Excellence. Studies still in progress were identified using the WHO International Clinical Trials Registry Platform and the ClinicalTrials.gov website. The cut‐off date for the search was February 2023. Data S1 describes the strategies used for each of the databases.

The reference management tool Endnote X9 was used to select and classify the documents retrieved by the literature search in order to eliminate duplicate references. We also performed a hand‐search of the references of the documents and articles found.

Study selection

Two reviewers (J. M. M. L. and E. B. A.) worked in parallel and independently to assess the titles, abstracts, and keywords of every study identified from the literature search as being potentially relevant.

The full text was obtained of any study that appeared to meet the selection criteria or if there was insufficient information to take a clear decision.

Both reviewers then independently read the full text of the articles. Once this was done, they shared their results in order to determine which studies would ultimately be included in the systematic review. Any doubt and/or disagreement between the reviewers was settled by consensus.

Eligibility criteria and features of included studies

The eligibility criteria for the studies located by the literature search were established using the Population, Intervention, Comparison, Outcomes, and Design method.

Population

The study population included patients with HF who are in NYHA functional class III or class II (if they have recently been in class III) and have an LVEF of 40% or less, even with the best possible medical treatment according to HF guidelines.

Intervention

Devices designed to electrically stimulate the baroreceptors located in the carotid sinus for therapeutic purposes were used.

Comparators

The optimal pharmacological treatment with neuromodulators was applied for HF with reduced LVEF, including medications such as angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers and beta‐blockers.

Outcomes

We considered the following types of outcome:

• Efficacy and effectiveness:

  • Primary endpoints: change in LVEF, change in NYHA functional class, patient‐reported health‐related quality of life using validated generic or specific instruments, changes in 6MWT, number of hospitalizations, duration of hospital stay, and mortality.
  • Secondary endpoints: change in biomarkers (e.g. NT‐proBNP, estimated glomerular filtration rate, and cystatin C), change in arterial blood pressure, change in pulse, and technique learning curve.

• Safety: device‐ or procedure‐related adverse events, surgical revisions, and complications.

• Economic aspects.

Study design

We selected studies that evaluated medical devices, prior systematic reviews and meta‐analyses, and randomized controlled trials (RCT). Any studies containing an economic analysis would be described as a narrative. The following designs were excluded: narrative reviews, observational studies, conference abstracts, editorials, letters to the editor, and opinions. If the researchers had published more than one article on the same study population or various articles on overlapping populations, only the initial article was included.

Language of publication

Studies without language restrictions were included.

Critical appraisal of bias risk

Two reviewers (J. M. M. L. and E. B. A.) worked independently and in duplicate to evaluate the methodological limitations of the chosen studies. Any disagreement was resolved through discussion, or if no consensus could be reached, they consulted a third reviewer.

A number of tools were used to evaluate the methodological quality of the studies and the bias risk, based on study design. For meta‐analyses and systematic reviews, they used AMSTAR 2 (A MeaSurement Tool to Assess systematic Reviews). ⁸ Clinical trials were evaluated using the recommendations in the Cochrane Handbook for Systematic Reviews of Interventions together with the Risk of Bias in Randomised Trials tool (RoB 2). ⁹ Drummond's Checklist was used to assess the quality of economic studies. ¹⁰

Data extraction

Data were extracted from the chosen studies by one reviewer and verified by a second reviewer. Any disagreement was resolved through discussion, or if no consensus could be reached, they consulted a third reviewer.

If more than one publication was suspected to relate to the same sample of patients, the data were extracted from the first study only to avoid any possible duplication of results.

Data summary and analysis

We used the Nordic Cochrane Centre Review Manager (RevMan) ¹¹ for the statistical analysis. Study heterogeneity and analysis model (random effect or fixed effect) were assessed using the I ² index and χ ² tests. If the I ² index was greater than 20% and the P value of the χ ² test was <0.1, heterogeneity was rated high and the random effect model used. If the I ² index was <20% and the P value of the χ ² test was greater than 0.1, heterogeneity was rated acceptable and the fixed effect model used. ¹² Continuous variables were expressed as standard deviation ± mean and were analysed using difference of means (DM). Categorical variables were expressed as percentages and analysed using relative risk. If the text or images of the article did not provide concrete data, the figures were reconstructed digitally from the graphs using the WebPlotDigitizer software package.

Quality of evidence

Quality of evidence was assessed for any outcomes rated critical or important by the system proposed by the Grading of Recommendation Assessment, Development and Evaluation (GRADE) Working Group, ¹³ and evidence was given one of the following certainty ratings: high ⊕ ⊕ ⊕ ⊕ (further research is very unlikely to change our confidence in the estimate of effect), moderate ⊕ ⊕ ⊕ ⊖ (further research is likely to have an impact on our confidence in the estimate of effect), low ⊕ ⊕ ⊖ ⊖ (further research is very likely to have an impact on our confidence in the estimate of effect and is likely to change the estimate) or very low ⊕ ⊖ ⊖ ⊖ (any estimate of effect is very uncertain). We analysed bias risk, imprecision, inconsistency, indirectness, and reporting bias using the Guideline Development Tool (https://gradepro.org/). ¹⁴ , ¹⁵

Results

Search results

The systematic search identified a total of 1020 references. After eliminating duplicates and reading the titles and abstracts, we selected 130 articles that met the inclusion criteria for full‐text screening. After applying the inclusion criteria again, we finally selected six systematic reviews, ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ two randomized clinical trials (HOPE4HF and BeAT‐HF) and two economic studies ²² , ²³ for our systematic review (Fig. 1 ). After the full‐text screening, some documents were excluded because they were not clinical trials, because the population or intervention was not appropriate or because they included the same population as other studies. Detailed reasons for each of the exclusions are given in Data S2.

Figure 1.

Reference selection flow chart.

The HOPE4HF study comprised one published study ²⁴ and one ongoing study. ²⁵ The BeAT‐HF study comprised one published study ²⁶ and one ongoing study. ²⁷

Study features

Systematic reviews

The results of the systematic reviews ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ indicated that BAT is safe when performed by a well‐trained multidisciplinary team. There were improvements in LVEF, NYHA class, quality of life, 6MWT, NT‐proBNP levels, hospitalization rates and length of stay, and muscle sympathetic nerve activity.

However, although in most of the reviews there is no overlapping of studies ¹⁸ , ²⁰ , ²¹ or it is expressly indicated, ¹⁶ , ²¹ there is one review ¹⁹ and one meta‐analysis ¹⁷ that include data from different articles with the same patient population from the Hope for Heart Failure study (HOPE4HF). ²⁴ , ²⁸ , ²⁹ , ³⁰ This was not taken into account by the authors of the meta‐analysis, ¹⁷ so the population overlap probably overestimated the true effect of BAT.

Primary studies

The study by Abraham et al. ²⁴ was a multi‐centre, randomized, and controlled study designed to evaluate the efficacy and safety of BAT in patients with symptoms of NYHA class III, an ejection fraction ≤35% and on optimal medical management.

To date, the best study to have evaluated BAT plus optimal medical management versus optimal medical management alone is the BeAT‐HF study. ²⁶ This was a randomized multicentre trial of 408 patients with NYHA class II/III, LVEF ≤ 35%, on optimal medical management for at least 4 weeks, and not candidates for CRT. The study was part of the Breakthrough Devices Program (a scheme for speeding up the market release of promising devices) in collaboration with the Food and Drug Administration, which comprised one pre‐marketing phase and another post‐marketing phase. The pre‐marketing phase examined three primary efficacy endpoints (6 min hall walk distance, quality of life, and NT‐proBNP levels); the safety endpoint included the major adverse neurological or cardiovascular system or procedure‐related event rate, with an innovative design for identifying the patient cohort that experienced the greatest benefit. This cohort, the D Cohort, containing patients with NT‐proBNP ≤1600 and representing the intended use population in accordance with Food and Drug Administration‐approved instructions for use, consisted of 245 patients followed up for 6 months (120 in the BAT group and 125 in the control group).

Table 1 describes the main characteristics of these publications.

Table 1.

Description of primary studies

Study, year	Population	Study arms	Number of patients	Age, mean ± SD	Women, n/N (%)	Hypertension, n/N (%)	LVEF, mean ± SD (N)	Endpoints	Follow‐up (months)
Abraham et al., 20 (2015): HOPE4HF study	Chronic HF with LVEF ≤ 35%	BAT ‐ BAROSTIM NEO System plus guideline‐directed medical therapy	71	64 ± 11	9/71 (12.7)	19/33 (57.6)	24 ± 7 (70)	System‐ and procedure‐related major adverse neurological and cardiovascular events, NYHA functional class, quality‐of‐life score, 6 min walk test, cardiac structure and function assessed by echocardiography, serum biomarkers including N‐terminal pro–brain natriuretic peptide (NT‐proBNP), and an accounting of HF medications.	3 and 6
		Control: guideline‐directed medical therapy	69	66 ± 12	11/69 (15.9)	21/37 (56.8)	25 ± 7 (67)
Zile et al., 22 (2020): BeAT‐HF Study ‡	Chronic HF with LVEF ≤ 35%, NYHA functional class III or functional class II (patients who had a recent history of NYHA functional class III); stable medical management for 4 weeks; no Class I indication for CRT and 6‐min hall walk distance of 150–400 m.	BAT ‐ BAROSTIM NEO system plus optimal medical management	130	62 ± 11	23/130 (18)	ND	27 ± 7 (130)	System‐ and procedure‐related major adverse neurological and cardiovascular events, quality‐of‐life score, 6 min walk test, cardiovascular structure and N‐terminal pro–brain natriuretic peptide (NT‐proBNP) levels.	6
		Control: optimal medical management alone	134	63 ± 10	30/134 (22)	ND	28 ± 6 (134)

BAT, baroreflex activation therapy; CRT, cardiac resynchronization therapy; EQ‐5D, EuroQol 5 Dimension Long; HF, heart failure; LVEF, left ventricular ejection fraction; n/N, patients with event/total patients; ND, not described; NT‐proBNP, N terminal pro brain natriuretic peptide; NYHA, New York Heart Association; RCT, randomized controlled trial.

^‡ Results are based on the intended use population.

Appraisal of methodological quality and bias risk

Using the AMSTAR 2 system (Data S3), the systematic reviews were classed as low quality, ¹⁶ critically low quality ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ or moderate. ²¹ According to RoB2 (Data S4), the clinical trials had an uncertain ²⁶ or high ²⁴ risk of bias.

Summary of results

Efficacy

Primary endpoints

Table 2 describes the primary endpoints and outcomes. The meta‐analysis of each primary endpoint together with recommendations using the GRADE methodology can be found in Table 3 and Data S5.

Table 2.

Main efficacy results

Study, Year	Study arms, N	LVEF	NYHA class	Quality of life	6 min walk test (m)	Hospitalizations	Mortality
Abraham et al., 20 (2015): HOPE4HF study	BAT 71 Control 69	6 months vs. baseline Improved LVEF in the BAT group and a slight reduction in the control group, with a between‐group difference of 2.5 ± 1.7% (P < 0.15)	6 months vs. baseline BAT (n = 64) Improvement: 55% No change: 42% Worse: 3% Control (n = 54) Improvement: 24% No change: 67% Worse: 9%. P = 0.002	6 months vs. baseline using the Minnesota Living with Heart Failure Questionnaire BAT (n = 64): −17.4 ± 2.8 Control (n = 54): 2.1 ± 3.1 P < 0.001	6 months vs. baseline BAT (n = 56): 59.6 ± 14.1 Control (n = 43): 1.5 ± 13.2 P = 0.004	6 months vs. baseline Hospitalizations for HF (n) BAT (n = 57): −0.49 ± 0.2 Control (n = 50): −0.05 ± 0.2 P < 0.05 Hospitalizations for HF (days) BAT (n = 57): −6.28 ± 2.7 Control (n = 50): 0.08 ± 1.7 P < 0.05	ND
Zile et al., 22 (2020): BeAT‐HF Study ‡	BAT 130 Control 134	ND	6 months vs. baseline BAT (n = 130) Improvement: 65% (13% improved by two NYHA functional classes; 53% improved by one NYHA functional class) No change: 35% Worse: 0% Control (n = 134) Improvement: 31% (2% improved by two NYHA functional classes; 29% improved by one NYHA functional class) No change: 67% Worse: 2%. P < 0.001	6 months vs. baseline using the Minnesota Living with Heart Failure Questionnaire BAT (n = 120): −21 ± 1.53 Control (n = 125): −6 ± 1.2 P < 0.001	6 months vs. baseline BAT (n = 118): 49 ± 4.08 Control (n = 120): −8 ± 2.72 P < 0.001	ND	ND

BAT, baroreflex activation therapy; HF, heart failure; ND, not described; NYHA, New York Heart Association.

^‡ Results are based on the intended use population.

Table 3.

Meta‐analysis and GRADE‐based evidence quality for each primary endpoint

Left ventricular ejection fraction

Echocardiographic analysis indicated a nonsignificant trend toward improved LVEF in the BAT group and a slight reduction in the control group, with a between‐group difference of 2.5 ± 1.7% (P < 0.15). ²⁴

The BeAT‐HF study did not report this endpoint ²⁶ and so a meta‐analysis was not possible.

New York Heart Association functional class

The meta‐analysis found that BAT affected the possibility of changes in NYHA functional class. The probability of improved NYHA functional class at 6 months was double in the BAT group than in the control group, with a relative risk of 2.13 (95% confidence interval [CI, 1.65 to 2.76]).

Health‐related quality of life

Compared to the control group, the meta‐analysis found a significant improvement in quality of life for the BAT group. Due to significant heterogeneity between the studies when using a fixed effect model (DM −14.97; 95% CI [−1530 to −14.64]; [χ ² = 75.42 (P = 0.00001); I ² = 99%]), we conducted the meta‐analysis using the random effects model by way of precaution. The results did not vary significantly between the two models. The random effects model showed an improvement of −16.97 points, noting that the lower the score the higher the quality of life (95% CI [−21.87 to −12.07]), at 6 months versus baseline for the BAT group compared to the control group.

6 min walk test (6MWT)

The meta‐analysis revealed a significant improvement in favour of the BAT group for distance covered in the 6MWT (DM 56.54; 95% CI [55.67 to 57.41]) at 6 months compared with baseline.

Number of hospitalizations

Only the HOPE4HF study contained data about hospitalizations. ²⁴ It reported a lower hospitalization rate following BAT.

Due to the lack of information about hospitalizations for the BeAT‐HF ²⁶ study, it was not possible to conduct a quantitative evaluation using meta‐analysis.

Duration of hospital stay

The study by Abraham et al. ²⁴ showed that patients in the BAT arm tended to spend fewer days in hospital (P = 0.08).

The BeAT‐HF study did not report this endpoint, ²⁶ so a meta‐analysis was not possible.

Mortality

The articles about the HOPE4F ²⁴ and BeAT‐HF ²⁶ studies did not contain mortality data, so no meta‐analysis was possible.

Secondary endpoints

The meta‐analysis of each secondary endpoint together with recommendations using the GRADE methodology can be found in Table 4.

Table 4.

Meta‐analysis and GRADE‐based evidence quality for each secondary endpoint

Biomarkers

The meta‐analysis found a significant reduction in B‐type natriuretic peptide in favour of the BAT group at 6 months (DM −120.02; 95% CI [−193.58 to −46.45]). No significant changes were observed in other biomarkers (creatinine, estimated glomerular filtration rate, or cystatin C). ²⁴

Arterial blood pressure

At 6 months, the meta‐analysis found a significant difference in post‐treatment measurements in favour of the control group, with lower values. However, due to the high heterogeneity observed when using the fixed effects model (DM 2.87; 95% CI [0.28 to 5.46]; [χ ² = 4.60 (P = 0.03); I ² = 78%]), we instead used a random effects model, which found no between‐group differences (DM 2.17; 95% CI [−3.65 to 7.99]).

Neither was there a significant difference in diastolic blood pressure (DM −0.48; 95% CI [−2.07 to −1.10]).

Pulse

At 6 months, the meta‐analysis found a significant difference in post‐treatment measurements in favour of the control group, with lower values. However, due to the high heterogeneity observed when using the fixed effects model ([χ ² = 3.93 (P = 0.05); I ² = 75%]), we instead used a random effects model which found no between‐group differences (DM 2.28; 95% CI [−1.52 to 6.08]).

Learning curve

The studies did not report on this endpoint. ²⁴ , ²⁶

Safety

In the HOPE4HF study, ²⁴ the major adverse neurological and cardiovascular event (MANCE) free rate was 97.2% (lower 95% confidence bound 91.4%). Two MANCEs occurred during the course of the study, consisting of two haematomas adjudicated as related to the procedure. The system‐ and procedure‐related complication event‐free rate was 85.9% (lower 95% confidence bound 77.3%). All but one event occurred within 7 days of implantation and resolved without residual side effects. Complications included urinary retention, urinary tract infection, haematoma (n = 2), bradycardia, atrial arrhythmia (n = 2), hypotension, worsening HF, pneumothorax, and cervical neuralgia. The majority of BAT patients (93%) had pre‐existing cardiac rhythm management devices. Testing at the time of BAT system implantation revealed no device‐device interactions impeding the performance of either system.

For the BeAT‐HF study, Zile et al. ²⁶ reported four MANCEs (event‐free rate: 97%; 95% one‐sided CI: 93% to 100.0%; P < 0.001): one cerebrovascular accident, one participant with acute decompensated HF, and two participants acquired an infection requiring explant of the device. Nine system‐ or procedure‐related serious adverse events occurred in seven out of the 125 patients who underwent BAT in the 30 days post‐implantation (event‐free rate: 94%; 95% one‐sided CI: 90% to 100.0%). Complications included nerve damage/simulation, respiratory failure, hoarseness, pneumonia, cerebrovascular accident, and thromboembolism. There were no additional system‐ or procedure‐related serious adverse events between 30 and 180 days post BAT implantation.

Economic studies

A German team conducted an early analysis of the cost‐utility of BAT. Borisenko et al. ²² reported that BAT carried an additional cost of €33 185 (95% CI [€24 561–€38 637]), along with an incremental benefit of 1.78 life‐years (95% CI [0.45–2.71]) and 1.19 quality‐adjusted life years (QALY) (95% CI [0.30–1.81]). At a willingness‐to‐pay threshold of €35 000 per quality‐adjusted life year (QALY), BAT had a 59% probability of being cost‐effective but reached an 84% probability of being cost‐effective at a threshold of €52 000/QALY. The authors concluded that BAT can be cost‐effective in European settings at the commonly accepted willingness‐to‐pay threshold of €35 000/QALY in patients with advanced HF who are not eligible for CRT.

One study of a HF population in the United States ²³ reported that at baseline, the expected cost of BAT plus guideline‐directed therapy was $29 526 per patient more than guideline‐directed therapy alone. This difference is mostly attributable to the BAT device and implantation costs. At 3 years, the predicted cumulative cost per patient for BAT + guideline‐directed therapy was $80 565, while the cost of guideline‐directed therapy alone was $90 086, representing a saving of $9521 using BAT + guideline‐directed therapy. Within guideline‐directed therapy, medication utilization and the cost of left ventricular assist devices were the largest contributors to costs at the 3 year mark, consisting of 63% and 20% of predicted costs, respectively. The authors conclude that their cost impact model suggests that BAT is a promising treatment option for patients with HF and a reduced LVEF who remain categorized in NYHA class III or II (with a recent history of class III) and have been on optimal standard‐of‐care pharmacological therapies.

The quality of the studies according to Drummond's 10‐item checklist of elements is presented in the Data S6. The two economic evaluations obtained the highest quality score.

Ongoing randomized clinical trials

We identified two randomized clinical trials, HOPE4HF ²⁵ (NCT01720160) and BeAT‐HF ²⁷ (NCT02627196), which will provide new efficacy and safety information for BAT. Both are ongoing, although recruitment has ended. Data S7 contains full details of these ongoing trials.

Discussion

Patients with cardiovascular disease often suffer deregulated autonomic function. ³¹ , ³² , ³³ , ³⁴ There are numerous pharmacological treatments for correcting autonomic dysfunction and improving quality of life. ³⁵ However, even with optimal medical therapy, the prognosis for HF patients with reduced ejection fraction continues to be poor. Furthermore, we are approaching a ‘ceiling’ whereby pharmacological treatments are becoming harder to implement due to the cost, complexity of care, greater risk of pharmacological interactions, and side effects. ³⁶ Device‐based therapy has many advantages, such as mitigating the problems of treatment compliance and polypharmacy in certain patient populations. This is especially important for younger patients at a risk of non‐compliance and older patients with multiple co‐morbidities requiring numerous medications. BAT devices have the potential of treating resistant hypertension and preventing damage to target organs. ³⁷ BAT can also be used to inhibit the harmful sympathetic outflow observed in HF patients with reduced LVEF.

The results of our primary endpoint meta‐analysis and systematic review suggest that, compared with standard interventions for HF, BAT has a significantly positive impact on NYHA functional class by improving patient ranking, on quality of life and on the 6MWT. There was also a reduction in length of hospital stay following BAT implantation in one of the RCTs. ²⁴ However, for a hard endpoint such as mortality, we must wait for the results of the long‐term follow‐up of ongoing trials to have data in this regard. In addition, it has been observed that the learning curve has a short time that allows a rapid decrease in intraprocedural mapping time and total procedure time.

The selected studies did not evaluate mortality or changes in cardiovascular structure or function, and only the study by Abraham et al. ²⁴ evaluated hospital morbidity. Previous studies have suggested that a 25% BAT‐induced reduction in NT‐proBNP made it very likely that morbidity and mortality also reduced and that structural and functional remodelling occurred with BAT. ³⁸ , ³⁹ , ⁴⁰ However, all of these specific endpoints require additional studies and long‐term follow‐up, such as the ongoing post‐marketing phase of BeAT‐HF.

Although sympathoinhibition is often associated with a reduction in arterial blood pressure, a parallel change in these two outcomes is not always observed. Similar to clinical trials, non‐randomized comparative studies with very small sample sizes have shown no significant change in resting arterial blood pressure despite an improvement in spontaneous sympathetic vascular baroreflex sensitivity after an average of 43 months of BAT. ⁴¹ , ⁴² However, clinical improvements in LVEF, distance during the 6MWT, and quality of life were observed, while NYHA functional class, number of HF medications, and hospitalization rate decreased. ⁴¹ , ⁴² BAT also reduced NT‐proBNP levels. ⁴³ These improvements in clinical markers should be interpreted with caution as some changes were not marked or depended on the investigator's judgement (NYHA class and quality of life scale score).

Current studies and guidelines favour the use of angiotensin receptor neprilysin inhibitor (ARNI) as the renin angiotensin antagonist of choice. ⁴⁴ However, the role of ARNI on BAT is unclear. The study by Guckel et al. ⁴³ is the first study to directly assess the effects of ARNI on the BAT response. The authors note that the effects of BAT and ARNI appear to be comparable and that BAT + ARNIs leads to more pronounced effects than ARNIs alone.

Post‐implantation safety data show that the total rate of major adverse neurological and cardiovascular events was 2.8%. ²⁴ This is very similar to the rate in patients with hypertension. ⁴⁵ In addition, BAT devices were not found to interfere with implantable automatic defibrillators ⁴⁶ , ⁴⁷ , ⁴⁸ or with CRT. ²⁸

As for economic aspects, like other HF therapeutic devices such as CRT, the benefits of BAT for the treatment of HFrEF all come with financial implications and the cost‐effectiveness of the therapy should be further explored. However, it can be noted that the results of the models analysed indicate that, although the initial costs associated with BAT plus guideline directed therapy are higher, the economic benefits become evident in the long term. ²² , ²³

There is a long way to go before BAT can be established as a sound option the treatment of HF. In fact, the European Society of Cardiology (ESC) ¹ believes that the body of evidence for BAT is insufficient ²⁴ , ²⁶ to support its recommendation. There are also other options for device‐based therapy such as cardiac contractility modulation. This review has shown that BAT gives promising results for patients with symptomatic HF. Nevertheless, if the BeAT‐HF ²⁷ study reports positive results for the hard endpoints, it may change the indications in the main clinical practice guidelines, offering a solution for patients not eligible for resynchronization who remain symptomatic despite optimal treatment.

Limitations

This meta‐analysis has a number of limitations. Despite being aware of various sub‐analyses ²⁸ , ²⁹ , ³⁰ , ⁴⁹ for the primary publications, ²⁴ , ²⁶ we did not include them in the meta‐analysis because it would involve overlapping populations. We only included two RCTs due to a lack of RCTs with a large sample size and sufficient follow‐up to confirm the positive results. These limitations could be partly overcome by the ongoing trial NCT02627196, ²⁷ which aims to recruit a large number of subjects over a longer follow‐up period than the studies available to date. In addition, unpublished data could have created a certain bias in the pooled results. We attempted to remove this bias by extending the search to multiple databases and languages. The selection criteria were defined in advance in order to avoid any possible bias in their application.

Conclusions

Neuromodulation using BAT is a potential treatment option for patients with HF with reduced ejection fraction who remain symptomatic despite receiving optimal medical management, or who are drug intolerant. This systematic review shows that BAT improves the NYHA functional class, quality of life, 6MWT, and NT‐proBNP levels in patients not candidates for CRT with New York Heart Association functional class III (or class II with a recent history of class III), LVEF ≤ 35%, and NT‐proBNP levels <1600 pg/mL. Further studies and long‐term follow‐up are needed to evaluate efficacy in reducing cardiovascular events and mortality.

Conflict of interest

The authors declare that they have no conflict of interest.

Funding

Funding was received in the framework of collaboration with the Spanish Ministry of Health.

Supporting information

Data S1 to S7. Supporting information.

Molina‐Linde, J. M. , Cordero‐Pereda, D. , Baños‐Álvarez, E. , Rosario‐Lozano, M. P. , and Blasco‐Amaro, J. A. (2023) Efficacy and safety of baroreflex activation therapy for heart failure with reduced ejection fraction: systematic review. ESC Heart Failure, 10: 2760–2772. 10.1002/ehf2.14473.