Endovascular renal sympathetic denervation to improve heart failure with reduced ejection fraction: the IMPROVE-HF-I study

Chronic heart failure is a major public health problem, with a prevalence of 1–2% in the adult population [1]. While pharmacological treatment for heart failure with reduced ejection fraction (HFrEF) has shown to prevent hospitalisation and improve quality of life and survival, its long-term prognosis remains poor justifying a persistent need for novel therapeutic strategies that improve both morbidity and mortality [2‐6].

Increased sympathetic tone has been directly linked to severity of heart failure and adverse outcome [7, 8]. In response to a chronic low-output state in HFrEF, neurohormonal adaptations occur such as the activation of the renin-angiotensin-aldosterone-system (RAAS) and the sympathetic nervous system (SNS) in order to maintain vital organ perfusion [9, 10].

In the past decade, renal sympathetic denervation (RDN) emerged as a novel minimally invasive treatment option to reduce sympathetic tone and proved to significantly reduce blood pressure in hypertensive patients [11‐14]. Promising findings were subsequently reported on the effects of RDN in HFrEF animal models [15, 16]. Up to now, the clinical evidence for RDN in the treatment of HFrEF is limited and restricted to several small non-randomised studies [17, 18]. In contrast to several studies with pharmaceutical agents, data correlating the effect of RDN on cardiac sympathetic tone as measured using iodine-123 labelled meta-iodobenzylguanidine (¹²³I‑MIBG) is lacking. The present study aimed to assess the safety and efficacy of RDN in patients with HFrEF as measured using ¹²³I‑MIBG at 6 months.

This present study is a single centre open label prospective randomised controlled trial designed to allocate 70 patients to treatment with RDN and optimal medical therapy (OMT) or OMT alone (1:1).

Patients were eligible for enrolment when the following inclusion criteria were met: left ventricular ejection fraction (LVEF) ≤ 35% (as assessed by echocardiography), New York Heart Association (NYHA) functional class ≥ II, age between 18 and 75 years, renal arteries suitable for RDN (i.e. baseline diameter stenosis < 30%, main renal artery diameter of ≥ 3.5 mm and ≤ 7.0 mm for each kidney), a glomerular filtration rate (eGFR) of > 30 ml/min/1.73 m². Exclusion criteria included: systolic office blood pressure < 110 mm Hg, recent (< 3 months) stroke or myocardial infarction, acute heart failure (HF), presence of other medical diseases or conditions associated with a life expectancy of less than one year.

Work-up at baseline included laboratory analyses, 24 h ambulatory blood pressure measurement (24 h ABPM), echocardiography, ¹²³I‑MIBG, as well as a computed tomography (CT) scan to confirm renal artery eligibility. Clinical follow-up occurred at 1, 3 and 6 months and will be continued yearly up to 5 years. Follow-up renal artery imaging using CT was performed at 6 months in patients who underwent RDN.

The study was designed to compare the primary efficacy endpoint, late HMR and washout rate (WR) derived from ¹²³I‑MIBG, in the treatment group versus control group (supplementary material for sample size calculation). Baseline categorical variables were expressed as counts and percentages. Differences in baseline categorical variables between randomly allocated treatment groups were compared using the chi-squared test, Fisher’s exact test or chi squared test for trend (NYHA class) when appropriate. Baseline continuous variables were described as mean ± standard deviation (SD) when normally distributed. In case of non-normal data distributions, data were presented as median [interquartile range, IQR]. Continuous variables (such as HMR and WR, normally distributed) were compared between groups using independent samples t-test or paired samples t-test. To examine within-group changes, paired samples t-tests were used. Non-parametric tests (Wilcoxon signed-rank or Mann Whitney U test, when appropriate) were used to analyse differences in case of non-normal distributions. All statistical tests are two-tailed. A p-value < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS statistical analysis (version 24.0).

RDN in patients with HFrEF did not result in a significant change in cardiac sympathetic nerve activity as measured using ¹²³I‑MIBG late HMR and WR at 6 months. The therapy appeared safe. A significant difference was observed in LVEDD in the RDN group, and 26% of patients in the treatment group were in NYHA class I versus none in the control-group.

Percutaneous RDN was introduced about 10 years ago as a minimal invasive treatment option for patients with resistant hypertension, a condition linked to sympathetic overactivity [24]. Sympathetic overactivity proved to contribute to the progression of myocardial cell injury and left ventricular dysfunction in patients with HF and a significant correlation was found between the severity of overactivity and NYHA class [25]. As a result to the chronic low-output state in HFrEF, elevated sympathetic tone stimulates renin release by the kidneys, leading to sodium retention, volume expansion and renal vasoconstriction in order to maintain vital organ perfusion. However, due to a subsequent increase in peripheral resistance, myocardial contractility and increase in heart rate prognosis worsens. An inverse association was found between norepinephrine release and survival [26]. Human data from the REACH pilot study showed that RDN in seven patients with congestive HF was safe and associated with a significant increase in 6MWT [18]. A randomised study presented by Taborsky et al., showed that RDN in patients with more advanced heart failure (mean LVEF was 25%, N = 51) resulted in significant improvements in LVEF, left ventricular end-systolic diameter (LVESD) and LVEDD as well as in NT-pro-BNP while no change was seen in patients with OMT alone [17]. For reasons unknown, the study was never published.

To the best of our knowledge, our study is the first to assess the effect of RDN in patients with HFrEF using ¹²³I‑MIBG imaging to assess cardiac sympathetic nerve activity. HMRs remained unchanged at 6 months in both arms. While we aimed to enrol patients with symptomatic HFrEF, the vast majority of the patients in our study were in NYHA class II with relatively low NT-pro-BNP values, suggesting a less severe HF phenotype. The latter could have explained part of the lack of effect in the present study and should be put into perspective due to the fact that our study was one of the first dedicated studies on the safety and efficacy of RDN in heart failure. Whether a more pronounced effect can be observed in patients with more advanced or unstable heart failure should be assessed in future dedicated studies.

The relatively low risk profile of our patients could also explain the higher than expected baseline HMRs in our study as compared to previous studies on the topic with baseline late HMRs in the range of 1.2 to 1.6 in patients with more pronounced heart failure and late HMRs of 2.5 ± 0.3 in healthy control [27]. The same applies for the WR found in our study which, being around 12%, were significantly below the threshold of 27% associated with poor prognosis [28]. This might suggest that the stable HF population studied in the present study was on relatively well controlled heart failure therapy in which the additional treatment with RDN did not add substantially on top of pre-existent OMT to improve cardiac sympathetic nerve activity.

Although our study was not powered to detect a difference in clinical endpoints, the overall rate of HF-related events at 6 months in the present study was low which might illustrate the relatively low risk profile of the patients included.

Based on the data available at the time of our study design, a significant blood pressure lowering effect of RDN was anticipated in patients with hypertension. This raised concerns about a potential blood pressure lowering effect of RDN in HF patients which might have forced down-titration of HF drugs. The latter made that we refrained from including patients with a baseline systolic blood pressure < 110 mm Hg and might have resulted in the HFrEF population at relatively low risk. Finally, in contrast to previous studies suggesting a significant change in LVEF following RDN, no change in LVEF was found in the present study. Conversely, we did observe a small, albeit significant, decrease in LVEDD following RDN. The latter results, however, should be interpreted with caution given the known variability in measurements derived from transthoracic echocardiography. However, we did observe a significant decrease in the peak late diastolic filling velocity in the treatment arm, which could implicate an improvement in left ventricular relaxation [29].

The current study has a number of limitations. First, we enrolled a smaller number of patients than first intended due to slow inclusion rates. Therefore, the study was underpowered to reach its primary efficacy endpoint. Second, we cannot exclude the fact that we might have used a less efficacious RDN system. Whether the use of a different technology in the present study would have altered our findings remains unknown. Third, we included a lower risk HF phenotype (80% with NYHA II). Finally, the present trial was an open label trial and not sham-controlled.

RDN with the Vessix system in patients with HFrEF was safe, but did not result in significant changes in cardiac sympathetic nerve activity at 6 months as measured using ¹²³I‑MIBG.