Learning From FAME: The Need for Sham Controls in Trials of Stable Coronary Disease

Introduction

The negative impact of stable ischemic disease on quality of life is clear, and some patients have experienced remarkably reduced anginal burden after invasive interventions. In 1 case, a woman with previously intractable angina was able to climb “28 steps to her second floor apartment” (1). In another, a man “reported a 100 per cent improvement at 6 months and 75 per cent improvement after a year” (2). Both patients were enrolled in studies of IMA ligation in the 1950s, but the second patient had undergone a sham procedure. These famous and revolutionary sham-controlled studies more than 50 years ago demonstrated both the lack of efficacy of IMA ligation and the power of placebo in treating angina (3).

These powerful lessons unfortunately have not been widely applied. Few sham-controlled studies of invasive interventions to treat stable coronary disease have been performed since then, with notable exceptions including studies of transmyocardial laser revascularization, another procedure that proved ineffective (4). Although medications to treat coronary disease are routinely tested in double-blind, placebo-controlled trials, treatment assignment in most randomized interventional trials has been unblinded. This has important implications for clinical knowledge and practice. Physicians and patients must make decisions on the basis of incomplete information, patients may be exposed to risks not justified by clinical benefit, and the widespread use of ineffective or marginally effective therapies means that health care resources may be consumed inefficiently. Awareness of the role of sham-controlled trials in addressing these concerns has increased in the wake of recent revelations regarding the SYMPLICITY HTN-3 trial of renal sympathetic denervation (5). We discuss here their critical role in trials for stable coronary disease.

FAME 2

The FAME 2 trial highlights the need for sham controls in the context of interventions for stable coronary disease (6). Designed to evaluate the role of fractional flow reserve (FFR) assessment in revascularization decisions for patients with stable angina or documented ischemia and ≥50% stenosis in at least 1 major epicardial vessel, this trial randomized patients with flow-limiting stenosis (FFR < 0.8) to percutaneous coronary intervention (PCI) plus optimal medical therapy (OMT) or OMT alone. FAME 2 clearly addressed a critical question in the post–Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation era (7). It was well positioned to produce results with great social value.

FAME 2 was stopped early by its data safety and monitoring board, principally because of a significant increase in urgent revascularization, 1 of the components of the composite primary endpoint. Only 1.6% of PCI-treated patients underwent urgent revascularization, compared with 11.1% in the OMT arm (hazard ratio: 0.13; p < 0.001). Importantly, most urgent revascularization was driven by presentation of unstable angina not marked by objective elevations in cardiac enzymes or electrocardiographic changes. Despite rigorous adjudication procedures, this is an inevitably “soft” endpoint, and critics expressed frustration that the trial’s design failed to ensure clarification of the effect of an FFR-guided strategy on mortality or recurrent myocardial infarction (8). However, as defenders of the trial have also pointed out, reducing unstable angina is an important goal. What has received less attention is the vulnerability of this endpoint to placebo (or nocebo) effects and the easy avoidance of this problem through the use of a sham control.

By FFR, all FAME 2 participants had hemodynamically significant stenosis that would conventionally be treated with PCI. The experimental group was in essence the OMT group. These patients and their physicians were thus aware that they had “untreated” stenoses, knowledge that could profoundly affect their likelihood of hospital presentation and diagnosis with unstable angina. First, angina is a placebo-responsive condition, as illustrated by the IMA trials. Second, independent of the placebo effect, angina often persists in both medically treated and revascularized patients. However, the experience of angina for patients who know that they have been “fixed” is likely very different compared with the experience of those who are “untreated.” This difference may drive both greater anxiety and more likely presentation to a hospital when angina occurs. Finally, a physician encountering a patient with angina in the context of a known hemodynamically significant stenosis is surely more inclined to diagnose unstable angina, perform catheterization, and intervene than if the lesion is known to have been “fixed” (8). In summary, because of the lack of blinding, there is every reason to believe that an appreciable expectation-induced nocebo effect, operating on both patients and physicians, was operative in this trial and may have contributed to worse outcomes in the OMT group.

The Solution

These challenges could have been substantially mitigated by blinded treatment assignment using a sham control. Because patients were randomized in the catheterization laboratory after FFR was assessed, the sham could have been as simple as a few extra minutes on the table and the use of earphones or other modalities to prevent the patient from overhearing conversations about treatment. Perhaps the most complicated issue would have been the post-procedural use of dual-antiplatelet therapy (DAPT). The simplest solution would be to treat all patients with DAPT, although it would involve some excess risk for bleeding and could affect ischemic outcomes. A more conservative approach more reflective of current management would be to give placebo DAPT to the OMT group. Either choice would have been defensible and would substantially address concerns about placebo and nocebo effects in this trial.

The benefits of a sham design from a societal perspective are clear. Given the high volume of PCI performed yearly for stable coronary disease, we have a substantial collective interest in targeting therapy and reducing unnecessary intervention, a goal of the FFR strategy itself. Moreover, cost-effectiveness assessments of PCI and FFR, which are particularly important in the absence of a clear reduction in myocardial infarction or mortality, are directly affected by these design issues. Independent of cost, patients have a strong interest in avoiding potential complications of PCI and the risks of DAPT. Perhaps most important, data from a sham-controlled study would offer the most complete information possible to help physicians and patients arrive at shared decisions regarding the best approach to management, an important goal given that the principal benefit of PCI for stable disease is symptom reduction.

Ethical Implications of Sham Controls

Sham-controlled trials do raise ethical concerns, most of which relate to the risks of the sham procedure itself. However, a sham design in this trial would raise few such concerns. First, the potential value of the scientific information is a significant ethical advantage. Second, the sham procedure would involve little more than simple machinations to make patients believe they had undergone PCI after FFR. There would be essentially no additional risks imposed during the procedure. Some sham-associated risks may be necessary and justifiable in some cases (as with IMA ligation); however, sham controls are particularly important when those risks are minimal. Third, some may argue that depriving patients or clinicians of treatment assignment on presentation to the hospital would negatively affect management, but this argument seems unfounded. Appropriate unblinding mechanisms could be put in place to manage bleeding complications, for example, or need for surgical treatment (so that antiplatelet medication discontinuation could be evaluated). These concerns require planning but are not significant barriers.

Sham controls and other methods of blinding play a critical, and ethical, role in interventional trials in which endpoints are subjective or are prone to being influenced by expectations on the part of physicians or patients. The inclusion of a sham arm in the SYMPLICITY HTN-3 trial distinguished it from multiple prior and overwhelmingly positive studies that led to the approval and clinical use of this therapy in many countries (9). Although the full results have not yet been presented, the failure of this trial to meet its primary efficacy endpoint has shocked the cardiology community and illustrates the potentially pivotal role sham controls can play (5). Blinded lead deactivation has similarly been critical in rigorous trials of cardiac resynchronization therapy (10,11).

There are strong ethical reasons to favor the broader use of sham controls to study interventions for stable coronary disease. They have provided critical knowledge in the past and could be instrumental in answering 1 of the most important and lingering clinical questions in medicine: who benefits from interventions for stable coronary disease?