Raising the Bar in Renal Sympathetic Denervation Research and Reporting

High blood pressure (BP) is frequently cited as one of the world's largest contributors to the global burden of disease.1, 2, 3, 4 Hypertension affects roughly 1 billion individuals worldwide5 and continues to be among the most prevalent conditions seen in adult primary care practice. As perhaps the most critical cardiovascular and cerebrovascular disease risk factor, it is ultimately a leading cause of death, stroke, myocardial infarction, and congestive heart failure worldwide.5 Considerable attention has accordingly focused for decades on improving the prevention and treatment of hypertension and cardiovascular disease by lifestyle modification, pharmaceutical therapy, and, more recently, medical device therapy.6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17 In this Commentary we focus on a device‐based approach to hypertension that is at once potentially transformational for various patient groups (eg, patients with various degrees of hypertension and other diseases resulting from sympathetic hyperactivity and patients with chronic lack of adherence to appropriate pharmacologic regimens) and, at the same time, intensely controversial among leading clinicians as they weigh accelerating treatment to patients in need against possibly delaying treatment to achieve acceptable scientific rigor during the device development process.

Resistant Hypertension and Renal Sympathetic Denervation

Resistant hypertension (RH), an important clinical problem in many global regions including the United States,18 can be defined as BP remaining above target levels despite the use of three or more antihypertensive medications at maximally tolerated doses (ideally, one being a diuretic) or BP requiring four or more drugs to achieve control.19 While apparent RH19 can result from suboptimal drug regimens prescribed by nonspecialist physicians20, 21, 22 or lack of adherence to appropriate regimens,23, 24, 25, 26, 27 true hypertension remains a significant clinical problem.19 According to the National Center for Health Statistics, more than 70 million adults in the United States alone (roughly one in three adults) have hypertension, which accounted for $46 billion in associated healthcare costs in 2011.28, 29

Renal sympathetic denervation (RDN) is an interventional procedure for treating RH, and potentially multiple other diseases, characterized by chronic pathological sympathetic nervous system input. Some systems employ a radiofrequency catheter to ablate renal nerves anatomically associated with renal arteries, while others use alternate methods, such as ablation via ultrasound energy or injection of pharmaceutical agents into the perivascular space around the renal nerves. The underlying logic to this medical device therapy is that the renal nerves play a key role in systemic hypertension via their effects in the kidney that enhance renal sodium retention and renin secretion while also exerting hypertensive effects in the central nervous system that lead to increased systemic sympathetic activity30 (see additional references for further discussion31, 32, 33, 34, 35, 36).

In the past decade, RDN has gained tremendous support as a potentially transformational new therapy for patients with RH who have not achieved appropriate BP control with standard‐of‐care oral medications. Perhaps early interest in RDN was generated in part from its potential to be another great example of breakthrough innovation leading to device‐based therapy for treatment of cardiovascular disease, similar to drug‐eluting stents and cardiac resynchronization therapy in its potential impact. The early adoption of RDN, especially in some European countries, has been driven most prominently by data generated from two studies of RDN in individuals with RH. These studies, Symplicity HTN‐137, 38, 39 and Symplicity HTN‐2,40, 41, 42 both initially appeared to demonstrate remarkable efficacy and a reassuring overall safety profile in patients with RH. Despite the investigators' assertion that Symplicity HTN‐1 was only intended to be a proof‐of‐principle study in the target population,37 and the fact that Symplicity HTN‐2 randomized only 106 participants to the two treatment arms, the overall promising data led to enormous interest. The results were widely regarded as compelling evidence that the procedure was both safe and effective. However, despite the proclamations of conclusiveness of these results in a large number of publications (the majority of which were review‐type papers, not empirical study reports), these studies were not methodologically designed to support such claims, ie, Symplicity HTN‐1 did not include a control group37 and Symplicity HTN‐2 did not utilize a sham‐control group40 (see also additional references43, 44, 45, 46, 47, 48, 49, 50). Subsequent negative results from Symplicity HTN‐3,51, 52, 53 the most rigorously designed study to date, as was appropriate for its stage of development (phase 3), not only failed to provide supportive evidence of the intervention's efficacy but also demonstrated why scientific rigor in study design is critical for informing decisional benefit‐risk assessments even in the presence of vocal and broad‐based expert scientific endorsement.

Pausing for Breath

Potential reasons for apparent discrepancies between results of Symplicity HTN‐3 and the earlier Symplicity HTN‐1 and Symplicity HTN‐2 trials have been discussed by various authors.54, 55, 56 Common themes include the appropriate use of a blinded control group in Symplicity HTN‐3 and the use in the early studies of an inappropriate assessment method––office BP––which may have allowed measurement bias, especially white‐coat hypertension, to influence primary endpoint results. Overall, many researchers in this field have taken a pause for introspection and deeper consideration of trial methodologies, with a view to determining how future research in this domain should be conducted. There now appears to be broad consensus that more rigorous study designs employed even earlier during development programs are imperative in future clinical trials of newer RDN devices and procedures to enable accelerated demonstration of RDN's compelling efficacy (perhaps in a subset of prospectively identifiable individuals) should such efficacy truly exist, and to allow for earlier no‐go decision‐making in the absence of sufficient efficacy. The efficiency gains and ultimate cost‐savings derived from speed‐to‐decision point development should garner praise for development plans well executed rather than retrospective criticism and claims of inadequate strategic planning in the face of late‐stage failure.

Recent Noteworthy Publications

Several recent publications have discussed options for accelerated pathways for future RDN development in the post‐Symplicity HTN‐3 era, with a continued focus on unmet needs in hypertensive patients. Lobo and colleagues57 presented the Joint United Kingdom (UK) Societies' 2014 consensus statement on RDN for RH; White and colleagues58 presented the American Society of Hypertension's (ASH's) scientific statement on pathways for moving forward the development of RDN for hypertension; Weber and a group of colleagues actively involved in the next generation of RDN clinical trials30 presented their thoughts in an editorial; and Mahfoud and fellow members of a multidisciplinary European Expert Group59 presented their recommendations in a European clinical consensus conference proceedings paper.

The Joint UK Societies' commentary is well summarized by their statement that they do not recommend “the use of renal denervation for treatment of resistant hypertension in routine clinical practice,” but they remain “committed to supporting research activity in this field.”57 The ramifications of their observation that “It is remarkable that to date there is no agreement on the precise anatomy of human sympathetic nerves, and in particular the variation in their trajectories relative to the renal arteries”57are also addressed by White and colleagues.58 Importantly, all expert opinions advocated the employment of ambulatory BP monitoring (ABPM)60, 61, 62, 63, 64 as an informative methodology likely to mitigate the impact of white‐coat hypertension. Interestingly, three of the papers advocated the use of a sham‐control treatment arm,30, 57, 58 while Mahfoud and colleagues59 did not, as discussed below.

Additional areas of general agreement include the following. First, while acknowledging differences between drug and device development, device trials may benefit from incorporation of trial‐design elements that have been successfully employed in antihypertensive drug programs. Second, to more fully understand the potential efficacy of newer RDN approaches, it may be important to start with rigorous assessment of RDN as the sole antihypertensive approach in an appropriate patient population: doing so may better position RDN as an adjunctive therapy for patients not well controlled with oral antihypertensive medications. Third, assessment of longer‐term safety should be required.

Regulatory Disharmonization?

The major difference between three of the four papers discussed30, 57, 58 and the paper by Mahfoud and colleagues,59 ie, their respective views on the inclusion of sham controls, reflects a potential ongoing regulatory discrepancy between US and European health authorities. Various European regulators have already given marketing approval to RDN for RH based on clinical studies that did not employ a sham‐control group. The US Food and Drug Administration (FDA), on the other hand, did not provide marketing approval for RDN prior to completion of Symplicity HTN‐3, indicating the agency's requirement for more rigorous tests of efficacy and safety even if it delays bringing potentially beneficial treatment to patients in need.

In the case of the Symplicity RDN system, the sponsor's decision to now pursue the SPYRAL HTN Global Clinical Trial Program,65 a program employing the SPYRAL catheter and modified study designs very much in line with those discussed in three of the four expert opinion papers just reviewed,30, 57, 58 leaves open the question of whether RDN by another method may be proven to be effective (perhaps in a more precisely defined patient population) or whether early studies simply benefitted from unrecognized confounders. In addition, one wonders whether more rigorous methods applied in earlier studies, such as sham‐control groups and/or ABPM, would have enabled an earlier “no‐go” decision, saving hundreds of participants in clinical trials from risk of invasive procedures and millions of dollars spent on a doomed phase 3 study.

Given the apparent regulatory disharmonization between US and European regulators, sponsors of future RDN development programs may find themselves in the awkward position of facing quite divergent probability models of technical and regulatory success and development costs, while at the same time realizing that patients' needs for better hypertension treatments are immediate, unremitting, and completely agnostic of political boundaries and regulatory precedents. Similar disharmonization among leading investigators only adds another layer of complexity to RDN development planning. Mahfoud and colleagues59 “expressed serious concerns,” including ethical considerations, regarding the employment of an invasive sham control in clinical trial participants with mild to moderate hypertension, while US‐based investigators supported use of invasive sham controls. Until a broader consensus is reached, perhaps via another RDN program, overcoming these regulatory and operational hurdles will remain a major source of uncertainty that will likely extend to research addressing other potential indications for RDN and potentially to other device development programs. It is conceivable that, in the future, multiple applications of RDN may be approved in Europe and not approved in the United States, and that other medical devices may realize a similar fate.

Lessons Learned Having Raised the Bar

McArdle and colleagues66 recently observed that “Despite the recent failure of Symplicity HTN‐3 to reach its primary end point, burgeoning evidence in many alternative areas suggests the potential to overcome current therapeutic hurdles.” They also stated that preliminary data continue to support the use of RDN to regulate sympathetic nervous system–derived pathology, “with suggestions for benefit outside of strictly antihypertensive effects.” These include heart failure (vascular stiffness, systolic heart failure, heart failure with a preserved ejection fraction), arrhythmias (atrial fibrillation, ventricular tachyarrhythmias), chronic kidney disease, metabolic disease, insulin resistance, obstructive sleep apnea, myocardial ischemia, and stroke (see also67, 68, 69, 70, 71, 72, 73, 74). The authors also noted that, as for initial results in the field of RH, “lack of adequate sample size and controls limits interpretation and application of these results,” and that there are currently more than 100 registered randomized clinical trials designed to further inquire into such indications.66

Given the recent directives for more rigorous nonclinical and clinical investigation of RDN for hypertension, what suggestions might we make for research addressing RDN as a potentially useful therapy for other conditions of clinical concern? Clearly, a high degree of methodological rigor is required. Also, appropriately toned reporting of clinical trials is needed. If a rigorous study design is employed, study data providing compelling evidence for RDN as a beneficial clinical therapy will communicate a sufficiently powerful message that does not need additional hype. Alternatively, if such evidence were not to ensue from an appropriately designed study, calm and collected clinical discourse would be needed to accept the results as conclusive and glean the important learning points for future research. We would like to share the following suggestions concerning the use of the term safe and the conveyance of emotions such as disappointment when publishing and commenting on study results.

A Tale of Terminology: Safe vs Favorable Benefit‐Risk Balance

Recent frequent comments favorably describing RDN have seemingly made the categorical statement that the procedure is “safe,” using the term in an apparently absolute manner. However, safety is a very difficult concept in medicine. In the field of pharmaceutical therapy, it is an unfortunate but immutable fact that no biologically active drug is free from the possibility of causing adverse reactions in certain individuals who are genetically and/or environmentally susceptible. Indeed, it is not actually safety that is assessed, it is harm. However, as Turner75 observed, “From the perspective of the media's treatment of drug‐related issues, the use of the term harm (while clinically meaningful) may not be an optimal choice since it would bring added emotional fuel to reporting that already often contains way too much. The term safety falls lower on the emotional thermometer.”

Since absolute statements of safety are not possible, safety is meaningfully operationalized in terms of a favorable benefit‐risk balance. The FDA's Sentinel Initiative76 commented as follows:

“Although marketed medical products are required by federal law to be safe for their intended use, safety does not mean zero risk. A safe product is one that has acceptable risks, given the magnitude of benefit expected in a specific population and within the context of alternatives available.”

At the time of a marketing application, regulators have to consider whether a medical product's overall benefit‐risk profile is favorable at the public health level. Physicians then have to make a benefit‐risk evaluation every time they consider prescribing an approved drug or recommending an approved interventional procedure to an individual patient. The benefit‐risk of exactly the same proposed treatment for one patient may be quite different from that for a different patient. An example would be a scenario where the same degree of therapeutic benefit is expected for two patients, but one has a condition that increases the risk of a serious side effect while the other does not. A benefit‐risk balance therefore requires consideration of both likely therapeutic benefit and the probability and degree to which the treatment is likely to cause harm. If the probability and degree of therapeutic benefit is considered to outweigh to a sufficient extent the probability and degree of potential harm, the drug or intervention may be considered acceptably safe for a given individual.

Given these considerations, no biologically active drug or interventional procedure with zero therapeutic benefit can meaningfully be regarded as acceptably safe, since any degree of risk whatsoever leads to an unfavorable benefit‐risk balance. Some reporting of the results of Symplicity HTN‐3 has therefore contained incongruous statements: that the intervention was safe (at a general level, implying for everyone), even though not effective (at the general level of mean responses). However, further investigation at the individual level is meaningful. When the earlier Symplicity studies suggested BP reductions in the order of 30 mm Hg, there was less incentive to look further than the mean response. However, results from Symplicity HTN‐3 suggested a more thorough examination of individual variation by looking at the BP changes of “good responders” and “less good responders” that were initially hidden behind the overall mean response. This is a completely sensible strategy. Learning again from the field of pharmacotherapy, it is now a fundamental tenet that not everyone responds in the same manner to the same drug,77 with individual differences determined by both genetic constitution and environmental assaults. The domain of oncology, where targeted therapies are perhaps more advanced than in other fields, provides examples relevant to current discussions.78 Crizotinib and vemurafenib were approved by the FDA in 2011 in combination with FDA‐approved companion diagnostic tests. Crizotinib is indicated for the treatment of locally advanced or metastatic non–small cell lung cancer that is anaplastic lymphoma kinase–positive as detected by the associated FDA‐approved test. Vemurafenib, indicated for melanoma, can only be prescribed for patients with a certain abnormal variant of the BRAF gene, BRAF^V600E, as identified by the associated FDA‐approved test. When there is no likelihood of benefit for a given patient, the benefit‐risk balance is immediately unfavorable.

An analogous challenge in RDN (for hypertension and any other condition) is to develop methods to prospectively and reliably identify individuals for whom the intervention is likely to have a favorable benefit‐risk balance. This first step necessitates the prospective identification of those likely to experience therapeutic benefit via translational research, which includes animal models, observational and mechanistic clinical research, and biomarker evaluation. As tantalizing as this precision medicine approach is, a second required step is to fully investigate those parameters that apply to all patients, ie, the potential that the device or procedure has for causing harm. In the case of RDN, a catheter must be inserted into the body, guided to the microenvironment of the renal arteries where therapy is delivered, and then extracted from the body. Expert and experienced interventional cardiologists and radiologists have the skills to conduct these aspects of the procedure extremely well, but, there is always the small chance that an unwanted event will happen, eg, damage to the artery through which the catheter reaches the renal arteries and failure to stop bleeding at the entry site of the femoral artery as quickly as is desirable. In addition, harm may be caused by application of the radiofrequency energy. These and other considerations may potentially prove through rigorous investigation that a device or invasive procedure is not suitable for use in humans regardless of their degree of unmet medical need.

What, therefore, may be more appropriate language when reporting future clinical trials providing compelling evidence of RDN's efficacy for certain individuals with a given clinical condition? If future longer‐term safety investigations indicate that the process is indeed associated with minimal risk, perhaps something like this: RDN is a technique associated with minimal risk, and for patients likely to experience therapeutic benefit of a certain magnitude, and those who do also not have a specific contraindication, it has a favorable benefit‐risk balance.

A Second Tale of Terminology: Disappointing vs Moving Closer to the Truth

Various publications have expressed the view that the results from Symplicity HTN‐3 were disappointing. While the following observation may initially sound harsh to compassionate clinicians who would very much like to offer the potential of therapeutic benefit to patients not attaining it from other modalities, such language can be reasonably argued to have no place in research endeavors: from a pure scientific perspective, our only hope as researchers should be to move iteratively closer to the truth, whatever that may be. Philosopher of science Karl Popper expressed this sentiment more forcefully: we should even rejoice in the falsification of a hypothesis that we have cherished as our brainchild, knowing that “the wrong view of science betrays itself in the craving to be right.”79 We readily admit that, in the real world, many forces (not the least of which are financial) come into play, but we are well advised to heed the spirit of Popper's admonition.

Concluding Comments

Turner and O'Brien44 noted that, by writing in the voice of his beloved character Sherlock Holmes, Sir Arthur Conan Doyle was able to wryly observe that it is a huge mistake to theorize before one has acquired the data, since to do so inevitably leads one to twist facts to suit theories rather than developing theories to suit facts. With the recent guidance from leading US and European investigators,30, 57, 58, 59 it now appears that we have good road maps to guide the acquisition of appropriately informative data and more authoritatively determine whether RDN is likely to be a useful clinical intervention for certain prospectively identifiable individuals with RH. Since RDN is also being explored in a multitude of other indications, researchers in those fields are encouraged to read the guidance documents pertaining to hypertension and transfer suggestions therein to their respective areas of investigation. We also encourage all authors to adopt appropriately conservative language when presenting and discussing study results in all venues, such as empirical publications and reviews, presentations at conferences, and media interviews––methodological rigor and calm, collected clinical discourse are abandoned at patients' peril.

Disclosure

No specific funding was used for the preparation of this manuscript. No editorial support was used.