Renal Nerve Denervation—A Hypertension Bubble?

Tulip Mania of 16371 is generally considered to have been the first “economic bubble” in which investors paid many times their annual salary to purchase tulip bulbs with the expectation of becoming rich. Fast forward 4 centuries and we may have witnessed the first “hypertension bubble.” In just a few years, several hundred centers have been established to treat resistant hypertension by ablation of the renal sympathetic nerves using radiofrequency energy at the tip of an endovascular catheter. What led to the rapid adoption of this new technique for treating hypertension? How much was based on science and how much was due to aggressive promotion and the possibility of financial gain?

In 2009, Krum and colleagues2 reported the initial results from Symplicity HTN‐1, a nonrandomized series of 45 patients with resistant hypertension despite receiving a mean of 4.7 antihypertensive drugs who were treated with renal nerve denervation (RND). Mean blood pressure (BP) at baseline for all patients was 177/101 mm Hg. At 3 months post‐RND, the decrease in office BP averaged 21/10 mm Hg in the 39 patients for whom follow‐up was available. At 6 months postprocedure, the reduction in office BP was 22/11 mm Hg in the 26 patients with complete follow‐up. In a subset of 12 patients who had 24‐hour ambulatory BP monitoring (ABPM) performed, mean 24‐hour systolic BP was decreased by only 5.8 mm Hg. The RND procedure was relatively free of serious complications.

Few details were provided in the methods about the precise technique used to record BP. Measurements were performed according to the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7) guidelines in triplicate and then averaged. The type of sphygmomanometer used, the person recording the BP, the conditions under which the readings were taken, and the location of the visits were not mentioned.

This initial “proof of concept” study was followed by the Symplicity HTN‐2 trial published in 20103 in which 106 patients with resistant hypertension were randomized to either RND therapy plus current antihypertensive medication or continued drug therapy alone. The primary outcome was the average of 3 office BP readings taken using an automated Omron HEM‐705 recorder (Omron Healthcare, Inc, Lake Forest, IL) in accordance with JNC 7 and European guidelines. The person performing the office BP appears to have been aware of the patient's RND/control status and the “data analysers were not masked to treatment group assignment.” Home BP and 24‐hour ABPM were also performed in a subset of patients, although the reason why less than a half of the patients had ABPM was not stated.

At 6 months, mean office BP decreased by −32/−12 mmHg in 49 patients in the RND group vs a change in mean office BP of 1/0 mm Hg in the 51 control patients. However, mean 24‐hour ambulatory BP decreased by only −11/−7 mmHg in a subset of 20 RND patients compared with a fall of −3/−1 mm Hg in 25 controls, thus giving a “placebo‐corrected” (difference between differences) net fall in 24‐hour ambulatory BP potentially attributed to RND of −8/−6 mm Hg.

On the basis of these reports and other observational studies, regulatory agencies in a number of countries, including the European Union and Canada, gave their approval for the use of RND as a treatment for resistant hypertension. Thereafter, numerous centers embarked on clinical programs to treat resistant hypertension using RND. Was this rapid adoption of RND into clinical practice justified on the basis of the available scientific evidence? If RND had been a drug, would it have been approved for use on the basis of the two Symplicity HTN trials?

The open‐label study design used in Simplicity‐HTN 1 is subject to observer bias, which cannot be quantitated or excluded. Symplicity HTN‐2 did attempt to examine RND using a randomized intervention vs control study design. However, the primary outcome of office BP and data analyses were performed by nonblinded study personnel. Had this trial involved a drug for treating hypertension, these features alone would have precluded most regulatory agencies from approving it for clinical use. It is not clear why an invasive therapy was subject to less stringent criteria for regulatory approval.

Moreover, the failure of RND to significantly decrease ambulatory BP should have alerted the hypertension community to the possibility that the extraordinary decreases in office BP in the two Symplicity HTN studies may have been caused by factors unrelated to RND itself. From a regulatory perspective, the Food and Drug Administration requires the use of ABPM to evaluate potential antihypertensive therapy and also noncardiac drugs, which may have hypertension as an unwanted adverse effect. Furthermore, a thorough, evidence‐based assessment of office BP vs ABPM performed by the National Institute for Health and Clinical Excellence (NICE) group in the United Kingdom4 concluded that a diagnosis of hypertension should be made using ABPM and not office BP.

Several randomized controlled trials undertaken to evaluate RND using more rigorous study designs have been in progress during the past few years. Symplicity HTN‐3 was the first of these to be reported. This trial randomized 535 patients with resistant hypertension to either RND or a sham procedure that consisted of renal angiography alone. The patients were blinded to their intervention. The first indication of the results of Symplicity HTN‐3 came in January 2014 when the sponsor abruptly halted all ongoing research on RND under their sponsorship because of the study's nonsignificant findings. The actual data from this study have now been published.5

In Symplicity HTN‐3, 364 patients were randomized to receive RND and 171 patients to the sham procedure using a 2:1 randomization procedure. At 6 months, the decrease in office systolic BP in the RND group was a mean of 14.13 mm Hg compared with a fall of 11.74 mm Hg in the sham‐operated control group, giving a net decrease due to RND of 2.39 mm Hg. Similarly, RND was associated with a reduction in 24‐hour ambulatory systolic BP of 6.75 mm Hg compared with a fall of 4.79 mm Hg in the control group. Neither of these differences in BP met the prespecified criteria for statistically significant superiority (5 mm Hg difference in office and 2 mm Hg difference in ambulatory BP). It is noteworthy that the 6.75 mm Hg decrease in ambulatory systolic BP after RND was similar to the 5.8 mm Hg decrease in 24‐hour ambulatory systolic BP in the 12 patients in the Symplicity HTN‐1 study who had ABPM performed.

The observed decrease in both office and ambulatory BP in the sham‐operated control group is of particular interest and highlights the importance of utilizing a rigorous design when evaluating procedural interventions, similar to designs mandated for pharmacologic studies. Otherwise, changes in the behavior of either patients or study personnel can influence the results, a phenomenon known as the “Hawthorne Effect.”6 With Symplicity HTN‐3, the most likely factor contributing to a fall in BP in both groups was enrollment of a treated hypertensive population that included patients who were not taking all of their medication prerandomization but who then became more compliant once in the study.

The Symplicity HTN‐3 investigators concluded that RND could be performed safely, with a low complication rate, but that the procedure itself did not significantly decrease systolic BP in patients with resistant hypertension. They also speculated that further studies may “validate alternative methods of RND” or “confirm previously reported benefits of RND.”

Symplicity HTN‐3 may not be the last word in RND but it likely represents the climax of the RND story. The “hypertension bubble” has burst. The enthusiasm that has greeted RND will never quite be the same. After Symplicity HTN‐3, several aspects of resistant hypertension should now receive more serious consideration.

A diagnosis of resistant hypertension must be based on 24‐hour ABPM and not office BP, if pseudoresistant hypertension is to be identified.7 Otherwise, patients who are actually normotensive on multiple drug therapy may undergo an unnecessary invasive procedure.

A thorough assessment of a patient's BP status should be performed by a hypertension specialist before making a diagnosis of resistant hypertension. This assessment should include confirmation of a diagnosis of resistant hypertension and optimization of drug therapy to achieve better control of hypertension. In one series of 731 patients referred to specialty centers for possible RND, this approach decreased the number of patients eligible for RND by about 60%.8

Therapeutic drug monitoring should be performed to evaluate the patient's adherence to antihypertensive therapy. It is now possible to assess compliance by measuring the amount of drug or metabolites in the urine or plasma for most antihypertensive medications.9, 10, 11 Using this approach, nonadherence to antihypertensive drug therapy was found to be a contributing factor in a high proportion of patients with a diagnosis of resistant hypertension.9, 10, 11 Therapeutic drug monitoring can be performed at a reasonable cost when a procedure such as RND is being considered.11 In Symplicity HTN‐3, compliance was only assessed using patient diaries. As the authors noted, improved compliance is one explanation for the decrease in BP observed in both RND and control groups after randomization.

The final lesson to be learned from the RND story is that all therapies for hypertension must be evidence‐based. The same rigorous standards that currently apply to drug therapy must also be required for other treatment modalities, even if the intervention appears to be relatively safe. Otherwise, enthusiasm will prevail and another therapeutic bubble may develop.