In this piece, I will argue that device-based therapies for resistant hypertension (RH), although interesting and meritorious of further study, do not yet deserve a prominent role in the management of patients with RH. This argument is on the basis of the limited availability of controlled studies using devices, the cost and uncertain long-term effectiveness of device modalities, and the successful track record of medical therapy.
First, let me define the demand for advanced therapies in RH. RH is defined as the BP of a hypertensive patient that remains above goal despite the concurrent use of three antihypertensive agents of different classes, preferably including a blocker of the renin-angiotensin system, a calcium channel blocker, and a thiazide-type diuretic (1). The prevalence of RH in the general hypertensive population is 12%–18%, on the basis of office BP measurements >140/90 mm Hg (1), an estimate that is expected to increase by approximately 2% when the 130/80 mm Hg threshold is applied to high-risk patients (2). Approximately one third of these patients have normal 24-hour BP (1), and a sizable portion is nonadherent to prescribed medications; therefore, approximately 5%–10% of hypertensive patients would ultimately qualify as having true RH. The current approach to the treatment of RH calls for adequate dosing of complementary medications, preferential use of long-acting thiazide diuretics (e.g., chlorthalidone), and the addition of a mineralocorticoid antagonist (e.g., spironolactone) (1). When such a strategy is used, BP goal is achieved in up to 90% of patients (3). Hence, about 0.5%–1% of hypertensive patients (approximately 750,000–1.5 million adults in the United States) would benefit from alternative approaches to treatment, such as referral to hypertension specialists for more advanced drug combinations or the use of device therapies.
I acknowledge the importance of treatment nonadherence and medication intolerance in RH. Nonadherence is present in 7%–60% of patients with RH (1). Intolerance to antihypertensive medications leading to treatment discontinuation is present in 7%–17% of unselected patients participating in clinical trials (4), and intolerance to three or more drug classes has been documented in approximately 3% of patients attending a hypertension clinic (5). Unfortunately, limited data are available on device therapy in this subgroup of patients. An uncontrolled, multinational study of renal denervation (RDN) in 53 patients not receiving antihypertensive drugs, of which 30 were due to intolerance to medications, observed a 5.7/4.0 mm Hg reduction in 24-hour BP 6 months after RDN, with high variability in responses (6). Therefore, although this is an area of opportunity, available data do not yet allow recommendations for use.
As noted in the Table 1, several device therapies have been studied in RH. All have been consistently effective in uncontrolled studies, but controlled trials have been negative or not yet completed. Some of the technologies, such as externally delivered ultrasound for RDN or central arteriovenous fistula creation, have been recently removed from further development because of lack of efficacy or complications. Baroreflex activation therapy is a technology that involves use of a device designed to produce chronic stimulation of the carotid sinus, resulting in prolonged decrease in sympathetic tone (7). Baroreflex activation therapy holds some promise on the basis of long-term results (up to 6 years of follow-up) of a previously used device (CVRx Rheos; CVRx, Minneapolis, MN), which was associated with excessive local complications at the time of implantation and is no longer available (8,9). The device currently undergoing testing (Barostim Neo, also produced by CVRx) provides advancements that minimize procedural complications (smaller lead, unilateral implantation) and battery longevity (3 years), and the US Food and Drug Administration recently approved it for use in advanced heart failure. Uncontrolled results in RH are encouraging (10) (Table 1); however, given the substantial changes in hardware design, it cannot be accepted as equivalent to the older device until controlled studies are available. Differently from other technologies, percutaneous (endovascular) RDN has been adequately studied, both in clinical trials and large registries. Therefore, the current state of affairs in device therapy impedes extension of clinical use beyond RDN, so I will restrict further discussion solely to RDN.
Summary of device therapies used in resistant hypertension
Efficacy of RDN in Untreated and Nonresistant Hypertension
RDN resulted in a 24-hour BP reduction of 5.5–7.0/4.4–4.8 mm Hg (4.1–4.6/1.8–4.3 mm Hg when sham-adjusted) in two sham-controlled trials of untreated hypertensive patients, serving as biologic proof-of-concept that the intervention lowers BP (11,12). Likewise, RDN produced a 9.0/6.0 mm Hg reduction in 24-hour BP (7.4/4.1 mm Hg when sham-adjusted) in a sham-controlled study of patients receiving one to three antihypertensive medications (13). However, the observed effect size observed in these trials is relatively small when compared with the effects of antihypertensive agents in similar populations, as shown by a meta-analysis of 15,289 patients receiving monotherapy with angiotensin receptor blockers documenting an average 24-hour BP reduction of 13.0/8.3 mm Hg (14). Therefore, it appears that the overall effect of RDN in uncomplicated hypertension is less than that of standard antihypertensive medications.
Efficacy of RDN in RH
Several recent meta-analyses evaluated the effect of RDN on uncontrolled or resistant hypertension, with variable results because of different methodology of study selection (15⇓⇓–18). Three of the meta-analyses did not find a significant effect of RDN on BP (15,17,18). However, I believe that it is the recently published meta-analysis by Cheng et al. that is most relevant to catheter-based RDN (16). The authors included 12 randomized, controlled trials with more than 40 hypertensive patients (six trials with sham procedure control) that reported ambulatory BP results during follow-up (2–6 months, mean 5.4 months) (16). Their meta-analysis was not restricted to RH; three of the 12 studies were not in RH patients, and one was of RH in patients with sleep apnea. They found a significant decrease in average 24-hour BP of 4.0/2.1 mm Hg (95% confidence interval, −2.6 to −5.5/−1.1 to −3.1 mm Hg) after RDN, with most studies finding numeric reductions in BP, but they were statistically significant in only six out of 12 studies (16). Therefore, the overall effect of RDN in RH is modest in the best-case scenario, and insignificant in other analyses.
Comparative Effects of RDN against Guideline-Directed Management of RH
On the basis of current consensus by leading hypertension specialists (1), aldosterone antagonists are the cornerstone of the treatment of RH, thus it follows that any proposed comparison of treatment efficacy in RH must be against aldosterone antagonists. A meta-analysis of the add-on effect of spironolactone in RH showed a BP reduction of 8.7/4.1 mm Hg (95% confidence interval, −8.6 to −8.8/−3.8 to −4.5 mm Hg) (19). In addition, the tolerability of spironolactone in RH clinical trials is quite good. In the PATHWAY-2 trial, overall adverse events (19%) and discontinuation due to side effects (1%) were similar to placebo (15% and 1%, respectively) (20). In the Anglo-Scandinavian Cardiac Outcomes Trial–BP Lowering Arm (ASCOT-BPLA), spironolactone was added as fourth drug in 1411 patients who remained uncontrolled on standard therapy (21). After a median exposure of 1.3 years, spironolactone was discontinued due to side effects in only 6% of patients. In contrast, data from a randomized trial of RDN versus optimized medical therapy including preferential use of spironolactone showed a high intolerance rate to spironolactone (39%) (22).
It is estimated that only about 30% of patients with RH would not qualify for aldosterone antagonist therapy on the basis of baseline low renal function (eGFR<50 ml/min) or serum potassium >4.5 mmol/L (23). Despite these estimates, guideline-concordant therapy using aldosterone antagonists is low in clinical trials and registries of RDN in RH (23%–26%) (22,24⇓–26). Understanding the impact of use of aldosterone antagonists on the effect of RDN is RH is essential to define the value of RDN. Data from three clinical trials help fill this gap. The Czech study PRAGUE-15 randomized 106 patients with RH to either RDN or medical therapy that included spironolactone if tolerated (22). At baseline, 25% of patients in the RDN group were taking an aldosterone antagonist and remained on it. In the medical therapy group, 24% were on one at baseline, and 61% were receiving spironolactone at 6 months (39% did not tolerate it due to hyperkalemia [11%] or other side effects). Both interventions resulted in similar reductions in average 24-hour BP at 6 months (8.6/5.7 mm Hg for RDN, 8.1/4.5 mm Hg for medical therapy; P=0.87/0.48) (22). The PRAGUE-15 investigators have published updated results after 12 and 24 months of follow-up showing persistence of effects and numerically better BP control in patients who continue to tolerate spironolactone (although not statistically significant) (27,28); findings also extended to patients who crossed over to the opposite group after 1 year (28).
The much smaller Spanish trial Denervacion en Hipertension Arterial randomized 24 patients with RH to RDN or spironolactone (25 mg daily, force-titrated to 50 mg daily after 1 month) (29). After 6 months, spironolactone resulted in a much larger 24-hour BP reduction than RDN (17.9/6.6 mm Hg in favor of spironolactone, P=0.01/0.04).
The third study was the French trial Renal Denervation in Hypertension (DENER-HTN), which randomized 106 patients with RH to either stepped medical therapy starting with spironolactone, followed by the addition of bisoprolol, prazosin, and rilmenidine, or to the same stepped therapy plus RDN (30). At 6 months, the 24-hour BP was 5.9/3.1 mm Hg lower in the RDN group (P=0.02/0.05), and BP control rates were higher in the RDN group (40% versus 19% on the basis of 24-hour BP; P=0.03). Most patients had similar progressive escalation of medications, and at the end of the study 79% of patients in both groups were receiving spironolactone (30).
My interpretation of these three relevant studies is that RDN is equivalent at best, and inferior at worst, than spironolactone-based treatment of RH. However, DENER-HTN suggests that RDN may provide additional BP control when added to a stepped medical therapy that includes spironolactone (30). A corollary to this would be that there might be value of RDN in patients who remain uncontrolled despite adequate doses of an aldosterone antagonist.
Safety and Cost Considerations
Bilateral RDN is performed under 60 minutes with the administration of approximately 129±78 ml of contrast (25). Among 998 procedures in the Global SYMPLICITY Registry, there were six peri-procedural complications (two renal artery dissections, three femoral artery pseudoaneurysms, and one groin hematoma) (31). The long-term safety of RDN has also been evaluated through clinical registries. One-year follow-up of 2112 patients identified three cases of de novo renal artery stenosis (0.1%), 19 cases of 50% increase in serum creatinine (0.9%), and nine cases of progression to ESKD (0.4%) (25). Extended follow-up to 3 years (N=1345) revealed a total of four cases of renal artery stenosis (0.3%), 24 cases of increased serum creatinine (2%), and 23 cases of ESKD (2%) (25). Fifty nine (4%) patients died after 3 years, 29 of which were due to cardiovascular causes. Unfortunately, these are not controlled data, so we cannot contextually interpret the magnitude of these findings. However, given the high complexity of this population, I suspect these numbers to do not reflect any increase in complication rates. Overall, my sense is that, in experienced hands, RDN is a safe procedure. However, one key area that requires further study is the adequacy of BP response to acute hypotensive illnesses (hemorrhage, sepsis) in patients who have undergone RDN. Hypertensive sheep treated with RDN exhibit larger BP reduction and blunted heart rate and plasma renin activity responses during hemorrhage (32). Clarification of this important question in patients who have undergone RDN is an essential element of the safety evaluation of the procedure.
A recent Australian study addressed the issue of cost-effectiveness of RDN (33). Using a model assumption of 24-hour systolic BP reduction of 5.7 mm Hg and procedural costs of AU$9531 (approximately US$6484), RDN would result in acceptable costs per quality-adjusted life years when the baseline 10-year cardiovascular risk was >13% (33). The major problem with the applicability of this analysis to the United States is the cost estimate, which is likely to be much higher than the authors’ imputed value on the basis of costs in Australia. In addition, they did not include the likelihood of reintervention. Despite the known regrowth of renal nerves within months of surgical denervation in humans and experimental animals (7), 3-year registry data show that only 10 patients (0.6%) were subjected to reintervention (25). Whether this will change over longer-term follow-up remains unknown. Because patients who undergo RDN require similar clinical follow-up as patients on medical therapy (office visits, laboratory testing for renal function and electrolytes) and may also require periodic imaging of the renal arteries, clinical costs at follow-up are likely to be higher than for medically treated patients.
Conclusions
RDN is the only device modality with enough data for detailed discussion at this time. Available data indicate a modest BP reduction that is sustained over at least 3 years with a good safety profile and infrequent need for reintervention. However, the observed efficacy and safety of aldosterone antagonists make them the first choice in RH. RDN may be considered in patients who remain resistant despite aldosterone antagonists. Perhaps the greatest value of device therapies will be in patients who are intolerant to medications, although this use has not yet been adequately studied. Therefore, I must conclude that device therapies do not yet deserve a prominent role in the management of hypertension, whether resistant or not.
Disclosures
A. Peixoto reports personal fees from Ablative Solutions, grants from Bayer, personal fees from Diamedica, grants from Lundbeck, personal fees from Relypsa, and grants from Vascular Dynamics, outside the submitted work.
Author Contributions
A. Peixoto conceptualized the study, wrote the original draft, and reviewed and edited the manuscript.
Footnotes
- Copyright © 2020 by the American Society of Nephrology
References
Jump to section
More in this TOC Section
Cited By...
- No citing articles found.