Author + information
- Received January 10, 2013
- Revision received March 18, 2013
- Accepted April 7, 2013
- Published online July 16, 2013.
- Mark I. Davis, MD∗,
- Kristian B. Filion, PhD†,
- David Zhang∗,
- Mark J. Eisenberg, MD, MPH∗,‡,
- Jonathan Afilalo, MD, MSc∗,‡,
- Ernesto L. Schiffrin, MD, PhD∗,§ and
- Dominique Joyal, MD∗,‡∗ ()
- ∗Department of Internal Medicine, Jewish General Hospital, McGill University, Montreal, Canada
- †Division of Clinical Epidemiology, Jewish General Hospital, McGill University, Montreal, Canada
- ‡Division of Cardiology, Jewish General Hospital, McGill University, Montreal, Canada
- §Lady Davis Institute for Medical Research, Jewish General Hospital, McGill University, Montreal, Canada
Reprint requests and correspondence:
Dr. Dominique Joyal, Division of Cardiology, Jewish General Hospital, McGill University, 3755 Cote Ste-Catherine Road, Montreal, Quebec H3T 1W3, Canada.
Objectives This study sought to determine the current effectiveness and safety of sympathetic renal denervation (RDN) for resistant hypertension.
Background RDN is a novel approach that has been evaluated in multiple small studies.
Methods We performed a systematic review and meta-analysis of published studies evaluating the effect of RDN in patients with resistant hypertension. Studies were stratified according to controlled versus uncontrolled design and analyzed using random-effects meta-analysis models.
Results We identified 2 randomized controlled trials, 1 observational study with a control group, and 9 observational studies without a control group. In controlled studies, there was a reduction in mean systolic and diastolic blood pressure (BP) at 6 months of –28.9 mm Hg (95% confidence interval [CI]: –37.2 to –20.6 mm Hg) and –11.0 mm Hg (95% CI: –16.4 to –5.7 mm Hg), respectively, compared with medically treated patients (for both, p < 0.0001). In uncontrolled studies, there was a reduction in mean systolic and diastolic BP at 6 months of –25.0 mm Hg (95% CI: –29.9 to –20.1 mm Hg) and –10.0 mm Hg (95% CI: –12.5 to –7.5 mm Hg), respectively, compared with pre-RDN values (for both, p < 0.00001). There was no difference in the effect of RDN according to the 5 catheters employed. Reported procedural complications included 1 renal artery dissection and 4 femoral pseudoaneurysms.
Conclusions RDN resulted in a substantial reduction in mean BP at 6 months in patients with resistant hypertension. The decrease in BP was similar irrespective of study design and type of catheter employed. Large randomized controlled trials with long-term follow-up are needed to confirm the sustained efficacy and safety of RDN.
Resistant hypertension (RH) is defined as uncontrolled systolic blood pressures (BP) despite therapy with ≥3 antihypertensive agents from at least 3 different classes including a diuretic. In most studies, 10% to 15% of hypertensive subjects (1,2), but up to 20% of the hypertensive population in some publications (3), have RH, particularly those with advanced age, obesity, diabetes mellitus, sleep apnea, and chronic kidney disease (4–6). In patients with RH, pharmacological options are limited. Historically, a surgical option, namely sympathectomy, led to a significant reduction in BP but was associated with high surgical morbidity (7–9). Although surgical sympathectomy was largely abandoned in clinical practice, there has been a renewed interest in the concept as animal models (10,11) have shown that renal sympathectomy leads to significant reduction in BP and improvement in end organ function (12–15).
Percutaneous renal sympathectomy has emerged as a safer, although invasive approach using radiofrequency probes to ablate the sympathetic fibers along the renal artery. The proof of concept study (16) demonstrated surprisingly good results and was subsequently followed by a series of studies using different catheters. These studies have generated great enthusiasm such that percutaneous renal denervation therapy (RDN) has been adopted at a rate rarely seen in the hypertension field. RDN for RH is currently approved in Europe and Canada and is pending approval in the United States. One RDN catheter system (Medtronic Ardian Inc., Palo Alto, California) has been used to treat over 4,000 patients worldwide thus far (17). Despite the enthusiasm and rapid uptake, there has yet to be a comprehensive review of the available evidence to support the practice of RDN.
We have systematically reviewed the current body of evidence for RDN and quantified its BP-lowering effect in patients with RH using a random effects meta-analysis model.
Data sources and search strategy
We performed a systematic review and meta-analysis in accordance with the standards set forth by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement (18,19). We searched PubMed, EMBASE, and the Cochrane Collaboration database using the key words “renal denervation,” “blood pressure,” and “hypertension.” The search was limited to English language articles published in the last 5 years (this technology was only developed in that time frame). In addition, we hand-searched references of retrieved articles and used PubMed's related articles feature to identify studies not captured by our primary search strategy. The final search was run on December 1, 2012.
We included randomized and observational studies comparing BP response in patients treated with RDN versus patients treated with standard medical therapy (controlled studies) and observational studies comparing BP in a single group of patients before and after RDN (uncontrolled studies). Inclusion criteria were: 1) RDN performed using contemporary percutaneous catheters and radiofrequency probes; 2) patient population with RH (not meeting BP target despite therapy with 3 or more antihypertensive agents from at least 3 different classes); 3) at least 10 study participants; and 4) at least 3 months of follow-up for BP response. BP measurements could include manual, automated, or invasive BP recordings, as long as the same method was used before and after RDN. Reviews, editorials, letters, animal studies, case reports, and conference abstracts were excluded. Once full articles were retrieved, studies were further excluded if there was an overlap in patients with another study within the same analysis (in which case, the larger sample size of the 2 studies was selected). Thus, whereas some patients could possibly have been included in both the controlled and uncontrolled study analyses, they were only included once in any given analysis. Consequently, there was no overlap in patients included in our meta-analyses.
Data was extracted in duplicate by 2 independent reviewers (M.D., D.Z.). Disagreements were resolved by consensus. We extracted data pertaining to baseline characteristics of study subjects (number of subjects, age, sex, comorbidities, antihypertensive agents), trial inclusion and exclusion criteria, method of BP measurement, type of catheter used, BP response to RDN (including BP before and after RDN), nonresponder rate, procedural complications, maximal length of follow-up, and mortality.
The primary outcome measure was mean systolic and diastolic BP reduction following RDN between 3 and 6 months of follow-up. Secondary outcome measures included: 1) nonresponder rate, defined as an achieved decrease in systolic BP of <10 mm Hg; 2) mean BP reduction stratified by catheter type; and 3) reported procedural complications and averse outcomes including death from any cause.
To determine the quality of the included studies, we used the Cochrane Collaboration Risk of Bias Tool (Online Appendix 1) for the 2 randomized control trials and the Newcastle-Ottawa scale for the observational studies. We set a follow-up rate of >70% at 6 months as a limit to determine high risks of bias at follow-up for studies evaluated with the Newcastle-Ottawa scale in the outcome section of this scale (Online Appendix 2).
Data synthesis and statistical analysis
For controlled studies, the difference in BP change with RDN versus medical therapy was pooled across studies and analyzed using random-effects meta-analysis models with inverse variance weighting. Separate models were constructed for 3 and 6 months of follow-up. For uncontrolled studies, the BP change before versus after RDN was pooled and analyzed using the same meta-analysis models. The magnitude of heterogeneity present was estimated using the I2 statistic, an estimate of the proportion of the total observed variance that is attributed to between-study variance. To compare the magnitude of BP reduction based on the type of RDN catheter used, we constructed a separate meta-analysis stratified by catheter type using a random-effects generic inverse variance-weighting model to compare heterogeneity using the I2 statistic.
Certain studies reported measures of variability other than SD. In these cases, 95% confidence intervals or standard error of the mean were converted to SD to maintain consistency of the reported results. In a study by Witkowski et al. (20), the only measure of variability reported was interquartile range. By including this study in the meta-analysis models, we are assuming a normal distribution of change in BP. In the study by Prochnau et al. (21), no estimate of variance was reported, thus we assumed a SD equal to the mean of other reported SD. Sensitivity analyses were performed excluding these 2 studies. We considered p < 0.05 significant. Throughout, values are presented as mean ± SD unless otherwise stated. Analyses were performed using the Cochrane Collaboration Review Manager (version 5.1.7, Cochrane Collaboration, Copenhagen, Denmark) and the GraphPad Prism (version 6.0, GraphPad Software, La Jolla, California) software packages.
Study selection and characteristics
Our literature search identified 294 potentially relevant studies as shown in the flow diagram (Fig. 1). Of these, 18 studies met the inclusion criteria. Six additional studies were excluded due to overlap of patients (22–27). All six of these studies were controlled studies with overlap of patients with the Symplicity HTN-2 (Renal Sympathetic Denervation in Patients With Treatment-Resistant Hypertension) study and could not be included in the controlled study meta-analysis. On the other hand, the Ukena et al. study (28), which is an uncontrolled study, is included in the uncontrolled study meta-analysis even though 18 (of 136) patients overlap with Symplicity HTN-2, because these 2 studies are included in separate meta-analyses. Thus, 12 studies were included in our systematic review, encompassing 561 patients treated between 2008 and 2012. The 12-month follow-up data for a study was published in a separate article and the follow-up data was extracted (29,30). The follow-up duration varied between 1 and 24 months with a median follow-up of 6 months.
Table 1 summarizes the design and methods of the included studies. There were 2 randomized controlled trials (n = 133) and 1 observational study with a control group (n = 50) (i.e., controlled studies), and 9 observational studies without a control group (n = 396) (i.e., uncontrolled studies). The inclusion and exclusion criteria were in large part similar, although 1 study included patients at the lower end of the RH spectrum (31), and another study included only patients with moderate to severe chronic kidney disease (32). Ambulatory BP monitors were employed to measure the primary outcome of BP response (and used for the meta-analysis) in 2 studies (21,33), with these measurements being relatively lower than office measurements, which were used as the primary outcome measure (and used for the meta-analysis) in the other 10 studies (16,20,28,30–32,34–37). The risk of bias was low in the majority of studies, with a detailed assessment available in Online Appendixes 1 and 2.
Tables 2 and 3 summarize the patient characteristics and concomitant therapies, respectively. Sixty percent of patients were male and the average age was 60 years. Thirty-five percent of patients had type 2 diabetes mellitus and 18% had coronary artery disease. Patients with RH were receiving an average of 5 different antihypertensive medications.
Effectiveness of RDN
In controlled studies, there was a reduction in mean systolic and diastolic BP at 6 months of –28.9 mm Hg (95% confidence interval [CI]: –37.2 to –20.6 mm Hg) and –11.0 mm Hg (95% CI: –16.4 to –5.7 mm Hg), respectively, with RDN versus medical therapy (for both, p < 0.0001) (Fig. 2). At 3 months, there were reductions of –20.8 mm Hg systolic (95% CI: –26.4 to –15.2 mm Hg) and –7.6 mm Hg diastolic BP (95% CI: –11.0 to –4.2 mm Hg). At 12 months, there were reductions of –25.4 mm Hg systolic (95% CI: –27.8 to –23.0 mm Hg) and –10.0 mm Hg diastolic (95% CI: –11.0 to –9.0 mm Hg). There was a modest amount of heterogeneity (I2 = 0 at 3 months and 50% to 60% at 6 months) between these studies.
In uncontrolled studies, there was a reduction in mean systolic and diastolic BP at 6 months of –25.0 mm Hg (95% CI: –29.9 to –20.1 mm Hg) and –10.0 mm Hg (95% CI: –12.5 to –7.5 mm Hg), respectively, before compared with after RDN (for both, p < 0.00001) (Fig. 3). At 3 months, there were reductions of –22.8 mm Hg systolic blood pressure (SBP) (95% CI: –26.3 to –18.5 mm Hg) and –9.1 mm Hg diastolic blood pressure (DBP) (95% CI: –12.1 to –6.1 mm Hg). At 12 months, there were reductions of –22.8 mm Hg SBP (95% CI: –29.6 to –16.0) and –10.6 mm Hg DBP (95% CI: –15.0 to –6.0 mm Hg). There was a significant amount of heterogeneity at the 3-month (I2 = 66% and 80% for SBP and DBP, respectively) and 6-month (I2 = 70% and 50% for SBP and DBP, respectively) time points. However, sensitivity analysis excluding the Kaltenbach study (which included patients with lower baseline BP) indicated that the heterogeneity was no longer apparent (I2 = 0), whereas the BP change was maintained (–24/–10 and –27/–11 mm Hg at 3 and 6 months, respectively). Only 6 studies (21,30–33,37) had adequate 6-month follow-up of ambulatory BP following RDN. The overall BP response was smaller than that seen in studies solely evaluating office BP. For these 6 studies, there was a reduction in mean SBP and DBP at 6 months of –13.2 mm Hg (95% CI: –19.4 to –7.0 mm Hg) and –7.3 mm Hg (95% CI: –10.2 to –4.5 mm Hg), respectively, before compared with after RDN (p < 0.0001 for SBP and p < 0.00001 for DBP). Statistical heterogeneity was present for the SBP response only (I2 = 76% for SBP and 0% for DBP).
A total of 5 different catheters were employed among the 12 studies: the Symplicity/Flex catheter (Medtronic, Minneapolis, Minnesota) is a radiofrequency ablation catheter designed for RDN, the Celsius ThermoCool and Navistar ThermoCool (Biosense Webster Diamond Bar, California) are irrigated radiofrequency ablation catheters, the Marinr (Medtronic, Minneapolis, Minnesota) is a standard steerable radiofrequency ablation catheter, and the PARADISE (ReCor Medical, Ronkonkoma, New York) is an ultrasound ablation catheter. Given the overlap of 18 patients between 2 studies with separate study designs, we excluded the Symplicity HTN-2 study (30) from the comparison of catheter meta-analysis and included the Ukena et al. study (28), given the larger amount of included patients treated with RDN. With all 95% CI overlapping, there is no evidence of difference in the achieved BP response after RDN among the different catheters used (Fig. 4).
The pooled nonresponder rate was 13.3%. The Kaltenbach et al. (31) study had the highest nonresponder rate, and excluding this study from the pooled estimate decreased the nonresponder rate to 11.7%. No deaths were reported during the stipulated follow-up periods. A total of 5 procedural complications were reported (<1%). These included 1 renal artery dissection (remote from where RDN was performed) (16) and 4 pseudoaneurysms at the site of arterial puncture (16,30,34).
We have conducted the first systematic review and meta-analysis of the published body of literature pertaining to RDN in patients with RH. Our study was designed to evaluate the effect of RDN on BP reduction in a RH population. We found that there was a substantial reduction in BP after RDN at 6 months, which was apparent as early as 3 months and sustained up to 12 months. Importantly, the rate of procedural complications was quite low.
We chose to pool the results of the studies based on the study design. Hence, the 2 randomized controlled trials were pooled with the only controlled cohort study, and the uncontrolled observational studies were pooled together. Observational studies tend to overestimate treatment effects by confounding by indication. In our analysis, however, all studies have demonstrated a consistent BP reduction regardless of study design.
The current study population is composed of only 561 patients, which is a small number for a meta-analysis. It represents, however, most of the published experience of RDN in patients with RH. In a meta-analysis, more important than the number of patients is the number of included studies, especially when the outcome is continuous. As RDN is now used for clinical care in Europe, Canada, and parts of Asia, knowledge of the current extent of effectiveness of the procedure, and the source of the evidence, is of utmost importance. Before large studies are completed and reported, the quality of the evidence and effectiveness and safety of the procedure must be based on current data.
The short-to-intermediate term data suggests that RDN is safe and well-tolerated, with the most common periprocedural complaint being abdominal pain that responds to sedatives and/or narcotics (16,38). Serial vascular imaging follow-up up to 6 months in the Symplicity HTN-1 (catheter-based renal sympathetic denervation for resistant hypertension: durability of blood pressure reduction out to 24 months) trial (34) did not reveal renal artery abnormalities secondary to RDN (based on renal duplex, magnetic resonance imaging or computed tomography angiography). Serial biochemical follow-up did not demonstrate deterioration in renal function (30,34). The only complications identified in our review were pseudoaneurysms at the vascular access site in 4 subjects, and 1 case of renal artery dissection on initial placement of the catheter prior to delivery of the radiofrequency signal. Hence, the renal artery dissection was the only non-access-related complication reported. Among the 12 studies included in the meta-analysis, 8 studies presented data on follow-up imaging of the renal arteries via ultrasound duplex, computed tomography angiography or magnetic resonance imaging angiography (16,20,21,30,31,33,34,36). Of the 191 patients with follow-up imaging, there were no documented cases of renal artery stenosis, and only 2 patients had progression of previously visualized renal artery atherosclerosis. To date, 2 case reports (39,40) have been published of individual patients whose blood pressure initially responded to RDN but subsequently had an increase in blood pressure on follow-up visits. In both instances, renal Doppler and angiography demonstrated renal artery stenoses, which were treated by renal artery stenting. It is still unclear what proportion of patients develops renovascular abnormalities following RDN. Further studies are needed to evaluate long-term changes in renal artery anatomy after RDN as well as to determine the appropriate imaging follow-up.
The average nonresponder rate, defined as a reduction in BP of <10 mm Hg after RDN, was 13%. Careful examination of the studies shows that patients who had a lower baseline sBP (<150 mm Hg) were less likely to respond to RDN. The Kaltenbach et al. (31) study included patients with sBP between 140 and 160 mm Hg and had the highest nonresponder rate of nearly 50%. Therefore, it appears that patients with severe elevations of BP may derive the greatest relative benefit from RDN. It remains unclear if the response persists in this patient population at long-term follow-up or if a rebound phenomenon may occur.
In addition to the primary BP-lowering effect, secondary effects of RDN have been documented. Brant et al. (22) showed that RDN led to a decrease in left ventricular hypertrophy, a decrease in end-systolic volume, and an increase in left ventricular ejection fraction. Mahfoud et al. (24) showed that RDN led to a significant reduction in fasting blood glucose as well as a reduction in insulin and C-peptide levels. Additionally, Mahfoud et al. (23) showed that RDN reduced albuminuria. A recent double-blind study (36) showed that RDN prevented a recurrence of atrial fibrillation after pulmonary vein isolation. Another study (28) showed RDN led to a reduction in heart rate and prolongation of the PR interval.
We did not identify any studies directly comparing the effectiveness of different catheters. Based on the studies included in this meta-analysis, the magnitude of the achieved BP reduction was consistent with all of the different catheters employed. Although RDN-specific catheters with multiple electrodes are being evaluated, it remains to be seen whether they are more effective at lowering BP.
A number of RDN trials are currently underway. The Symplicity-HTN 3 (Renal Denervation in Patients With Uncontrolled Hypertension) trial (41) (NCT01418261) is a single-blind clinical trial that is randomizing 530 patients in a 2:1 fashion to RDN or control with a follow-up period of 6 months. The INSPiRED (Renal Denervation for Management of Drug-Resistant Hypertension) study (NCT01505010) is a randomized clinical trial with a longer follow-up period of 36 months. The DEPART (Study of Catheter Based Renal Denervation Therapy in Hypertension) (NCT01522430) is a double-blind clinical trial that is randomizing patients to RDN or sham procedure with a follow-up period of 6 months. The SymplicityHF (Renal Denervation in Patients With Chronic Heart Failure and Renal Impairment Clinical Trial) (NCT01392196) is evaluating the effect of RDN in patients with heart failure, and contrary to the aforementioned studies that will be focusing on BP response, this study will focus on procedural safety and renal and cardiac function. The DREAMS (Denervation of the Renal Artery in Metabolic Syndrome) study (NCT01465724) is evaluating the effect of RDN in patients with metabolic syndrome, with the primary outcome being a change in insulin-resistance parameters after 12 months of follow-up.
First, most of the included studies were observational in nature and thus may be affected by confounding by indication and/or selection bias. Whereas the BP reduction was modestly greater in the observational controlled study than in either of the randomized controlled studies, large clinically important treatment effects were reported in all controlled studies. Second, with the use of published aggregate data, we were unable to examine the effect of RDN in patient subgroups. Third, inclusion was restricted to published studies and may therefore be affected by publication bias. Fourth, the follow-up rate was quite limited in many of the included studies, resulting in less than 70% 6-month follow-up. Fifth, although we conducted secondary analyses that were stratified by catheter type, there were 5 different catheters used in the 12 included studies and studies of catheters other than the Symplicity catheter were of modest size. Consequently, there were insufficient data to draw meaningful conclusions regarding the comparative efficacy of the different catheters. Lastly, given the recent evolution of RDN, the pooled person-time remains modest.
The current available data suggest that RDN results in a substantial BP reduction at 6-month follow-up in patients with RH. With few adverse events reported, available data also suggest that RDN has a favorable safety profile. Nonetheless, large randomized controlled trials with long-term follow-up are needed to confirm the sustained efficacy and safety of RDN in this patient population.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- blood pressure
- confidence interval
- diastolic blood pressure
- renal sympathetic denervation
- resistant hypertension
- systolic blood pressure
- Received January 10, 2013.
- Revision received March 18, 2013.
- Accepted April 7, 2013.
- American College of Cardiology Foundation
- Sarafidis P.A.,
- Bakris G.L.
- Evelyn K.A.,
- Singh M.M.,
- Chapman W.P.,
- Perera G.A.,
- Thaler H.
- Hoobler S.W.,
- Manning J.T.,
- Paine W.G.,
- et al.
- DiBona G.F.,
- Kopp U.C.
- ↵(2012) Medtronic I. Symplicity™ RDN System Clinical Trial Data (Medtronic, Inc, Mountain View, CA).
- Brandt M.C.,
- Mahfoud F.,
- Reda S.,
- et al.
- Mahfoud F.,
- Schlaich M.,
- Kindermann I.,
- et al.
- Ukena C.,
- Mahfoud F.,
- Kindermann I.,
- et al.
- Brandt M.C.,
- Reda S.,
- Mahfoud F.,
- Lenski M.,
- Böhm M.,
- Hoppe U.C.
- Ukena C.,
- Mahfoud F.,
- Spies A.,
- et al.
- Esler M.D.,
- Krum H.,
- Schlaich M.,
- Schmieder R.E.,
- Bohm M.,
- Sobotka P.A.,
- Symplicity HTN-2 Investigators
- Hering D.,
- Mahfoud F.,
- Walton A.S.,
- et al.
- Ahmed H.,
- Neuzil P.,
- Skoda J.,
- et al.
- Pokushalov E.,
- Romanov A.,
- Corbucci G.,
- et al.
- Kaltenbach B.,
- Id D.,
- Franke J.C.,
- et al.