Presumed safe no more: lessons from the Wingspan saga on regulation of devicesBMJ 2014; 348 doi: https://doi.org/10.1136/bmj.g93 (Published 22 January 2014) Cite this as: BMJ 2014;348:g93
- Ari J Gartenberg, medical student1,
- Ariel Peleg, medical student1,
- Sanket S Dhruva, cardiology fellow2,
- Rita F Redberg, professor of medicine3
- 1Albert Einstein College of Medicine, Bronx, NY 10461, USA
- 2Division of Cardiovascular Medicine, UC Davis Medical Center, Sacramento, CA 95817, USA
- 3Division of Cardiology, School of Medicine, University of California, San Francisco Suite M-1180, 505 Parnassus Avenue, San Francisco, CA 94143, USA
- Correspondence to: R F Redberg
Use of stents throughout the vasculature has been increasing since the first coronary artery stent was introduced in 1986 to treat restenosis after balloon angioplasty. Coronary artery stenting has been shown to significantly benefit patients with acute myocardial infarction, and intracranial stents were developed in the hope that they would prevent strokes in patients with substantial stenoses who had already had strokes. In the brain, however, the procedure is far more challenging because it involves navigating a specialised, flexible device through much smaller and more tortuous vessels than in the heart. Intracranial stenting has serious possible complications, such as the formation of thrombi, which can embolize and cause acute ischemic strokes.
Two intracranial stenting devices have been marketed in the US. The first was the Neurolink system (Guidant Corporation, Menlo Park, California), which was approved in 2002 but then discontinued in 2006 in favor of the more technologically advanced Wingspan stent system (Stryker Neurovascular, Kalamazoo, Michigan).1 Treatment was once thought to be both feasible and safe, but new evidence has raised compelling questions about the safety of intracranial stenting and, more broadly, on the process by which these stents gained approval.
How the Wingspan got to market
Although cerebrovascular disease is common, intracranial atherosclerosis is considered rare. Intracranial stents are therefore able to obtain market approval in the US through the humanitarian device exemption, which is intended to offer incentives to develop devices for disease populations of fewer than 4000 individuals a year when a manufacturer’s research and development costs otherwise could exceed market returns.2 To be considered as a humanitarian use device, comparable devices must not be available through any other FDA approval pathway, the device must not pose an unreasonable risk of illness or injury, and its probable benefit must outweigh its risks. Devices approved under a humanitarian exemption do not have to have proof of effectiveness, as is required under the usual premarket approval process for high risk devices.3 The US regulatory system does not require new devices to be superior to existing therapies. Since 1997, the FDA has approved 58 devices through a humanitarian device exemption.4
The FDA approved the Wingspan intracranial stent system in 2005 on the basis of a company sponsored, single arm study of 45 patients enrolled in 2004 at 12 international centers (none in the US).5 All the patients had recurrent strokes thought to be due to ≥50% intracranial stenosis and were unresponsive to medical therapies such as anticoagulant or antiplatelet drugs. Participants were aged 18-80, at least seven days after stroke, and had a modified Rankin scale score (mRS) of ≤3, indicating a lack of severe disability due to stroke. The study’s primary end point was a composite of ipsilateral stroke or death at 30 days. Of the 44 participants treated with the Wingspan (one was not treated because of problems placing the stent), two (5%) had a stroke within 30 days of the procedure. At six months, a further patient had had a stroke and one of the three who had had a stroke died, giving a risk of 7% (3/44). Comparison with historical controls suggested that the risk in similar patients who did not receive intervention was 20%.
The FDA accepted this study as sufficient evidence for approval, stating that the reason there was no active control group was “because no alternative standard therapy was readily available for this disease state.”6 However, a subsequent study of the device sponsored by the National Institutes of Health managed to define a medically managed control group. The use of historical controls in device trials is common (13% of primary end points in studies of high risk cardiovascular devices approved by premarket approval pathway).7 Such studies meet the FDA’s requirement for comparison with controls, although the design is inferior to a randomized trial and open to bias.
Although the study had very specific inclusion criteria, as noted above, the FDA approval placed no limitations on the use of the Wingspan with regard to age, severity of previous stroke, or location of stenosis and did not specify a time interval between the stroke or transient ischemic attack and implantation of the device. The indications for use were given as “improving cerebral artery lumen diameter in patients with intracranial atherosclerotic disease, refractory to medical therapy, in intracranial vessels with ≥50% stenosis that are accessible to the system.”6 There was no evidence to support use of the device in this larger group. Furthermore, the agency did not require any specific postmarketing studies for the Wingspan except for the standard annual postapproval reports required for humanitarian device exemptions. These detail the number of devices shipped or sold, information describing clinical experience with the device, and a summary of any updated information or changes made to the device, as well as reporting of adverse reactions and device defects. These data are not publicly available.
The absence of clinical data comparing stenting with medical therapy left uncertainty about the safety and effectiveness of intracranial stenting.8 To establish the device’s efficacy definitively before it became the standard of care, the National Institutes of Health funded the Stenting and Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) trial.9 Enrollment at 50 US sites began in 2008, three years after FDA approval.9 The trial randomized patients who had experienced a recent stroke or transient ischemic attack that was attributed to stenosis in a major intracranial artery to either aggressive medical management plus stenting or aggressive medical management alone (table⇓). The primary outcomes were stroke or death within 30 days after a revascularization procedure or stroke in the territory of the qualifying artery beyond 30 days.
Enrollment was facilitated by Medicare’s decision to reimburse for this device only if patients were enrolled in the clinical trial, a structure known as coverage with evidence development. However, it was prematurely stopped after 451 patients had been enrolled (just 59% of the target sample size of 764 patients was reached) because the primary outcome was significantly higher in the Wingspan group (15%, n=33) than in the medical therapy arm (6%, n=13), P=0.002. This means that for every 11 patients treated with the Wingspan, one additional patient had a stroke or died.10
FDA reaction to fresh evidence
In December 2011, in light of the signal of harm, the consumer advocacy organization Public Citizen petitioned the FDA to withdraw approval of the Wingspan.11 The FDA convened a public meeting of its advisory neurological devices panel in March 2012 to review the SAMMPRIS results. Although there was no formal vote, an informal vote showed that panel members were unanimous that current data on the Wingspan “do not support its safety and efficacy as a treatment for ischemic stroke in adults.”12 13
However, rather than withdraw the device, the FDA tightened the indications for use, stating in a safety communication, “while Wingspan is not beneficial for the broad population of stroke patients studied in SAMMPRIS, there is evidence from the original HDE study to show there are probable benefits of using Wingspan to treat the specific population of patients outlined in the new indications for use.”14 The new indications limited the device to patients younger than 80 years who had not had a stroke or transient ischemic attack within seven days. The FDA said that it was keeping Wingspan on the market despite its increased mortality and stroke rate in SAMMPRIS because the patient population in SAMMPRIS was different from that in the original small approval study (table⇑).15 Although the differences have questionable clinical importance, the FDA argued that the narrowed indications limited the potential harm indicated in the SAMMPRIS trial’s disturbing results by reflecting the population in the original study. The FDA also required Stryker to conduct a postmarketing study. Seventeen months later, the only information on the FDA site says “study pending.”16
In the six years since the initial Wingspan trial, the company has shipped 11 000 Wingspan devices in the US alone.17 No further data are available on efficacy or safety in patients matching the original study criteria. No subset analysis has been done of the SAMMPRIS data to determine whether there is any benefit in the narrowed indication population defined by the FDA, and there is no evidence that the Wingspan outperforms medical therapy over longer periods either. Recently published data with longer follow-up (median 32 months) from the SAMMPRIS trial show that aggressive medical management maintains its advantage over intracranial stenting. The absolute risk reduction for the primary end point (stroke or death within 30 days or after a revascularization procedure or stroke in the territory of the qualifying artery beyond 30 days) from medical treatment was 8.9% at 30 days and stayed at 9.0% after three years.18
The FDA safety communication is vague in its terminology—for example, it uses but does not define “aggressive medical management.” Clinicians may remain unaware that there is a safety warning.19 The product page for the Wingspan on Stryker’s website does not mention the FDA’s revised guidelines20 despite the company’s stated commitment to work with physicians to “help them better understand the narrowed indications for product use reflected in the new product labeling.”21 The process is further hindered by broken links on the Wingspan’s product information webpage, checked on multiple occasions over six months. This public, outdated information presents a significant obstacle to clinicians interested in learning more about the Wingspan.
In a press release accompanying the FDA’s safety communication, Jeffrey Shuren, director of the FDA’s Center for Devices and Radiological Health, stated: “The FDA believes this device should remain available for this specific subgroup of patients who have exhausted other options.” However, restrictions on the approved use of medical devices will not necessarily protect patients because off-label use occurs and is not routinely tracked. Although institutional review boards generally have to approve use of humanitarian exempt devices, the Wingspan’s sponsor has stated that 30 off-label uses occurred in 2009, 36 in 2010, and at least 45 in 2011.11 The Wingspan has been used for treatment of intracranial dissections, even though it is approved only for use in intracranial stenosis and has been considered for use in acute ischemic stroke.22
The Wingspan also remains in use in Europe, where the approval process gives the manufacturer sole responsibility for defining the “intended purpose” of a device. In response to the SAMMPRIS study, European branches of Stryker, including in the United Kingdom and Germany, narrowed indications for Wingspan’s use to patients refractory to medical therapy—a stipulation that had already been required in the US under the initial approval. No regulatory authority in Europe has intervened to change the indications for use.23
Devices approved under a humanitarian device exemption must have proof of safety as well as probable benefit.24 The FDA’s decisions to approve Wingspan and allow the device to remain on the market after a clinical trial showed evidence of “unreasonable risk of illness or injury” are not consistent with these stated principles. Wingspan’s original approval was based on relatively poor quality evidence given the lack of randomization and blinding, use of historical controls, multiple strict inclusion criteria, and small study population. The study sponsor also analyzed, interpreted, and presented the data, a source of possible bias. The small number of patients enrolled per center (three to four) raises additional questions, as these characteristics are associated with “seeding trials,” which can be more for marketing than research purposes.25
Although avoidance of recurrent stroke is desirable, it does not justify use of a device that has been shown to increase stroke and mortality. When a high quality, multicenter randomized controlled trial has shown that a device is associated with serious harms, it is inadequate to posit benefit in a subpopulation based on a low quality small study. Additional randomized, unbiased data must be required to demonstrate safety—even in patients meeting the criteria of the original study.
The Wingspan saga shows how high risk devices can come to the market based on low quality clinical data. Furthermore, the Wingspan device has not been removed from the market, even though a high quality randomized trial was prematurely terminated because of harms and no benefit. When devices are approved under the humanitarian exemption, restriction of their use to within randomized controlled trials could prevent potentially dangerous devices from obtaining seemingly permanent market approval. A similar system would also be beneficial in Europe.
Such problems might be avoided by requiring a randomized controlled trial before a humanitarian use device is approved. In addition, data from postmarketing surveillance should be gathered and be publicly available. The responsibility to protect the public’s health means ensuring safety and effectiveness by clinical studies and includes removing dangerous devices in an effective manner and making information that shows the benefits outweigh harms publicly accessible.
FDA’s humanitarian device exemption allows devices for rare conditions to be marketed without a requirement to show effectiveness
The Wingspan device was approved through this route on the basis of a small, industry sponsored uncontrolled study
A subsequent randomized controlled trial (SAMMPRIS) was prematurely terminated because results showed that for every 11 patients treated with the Wingspan, one additional patient had a stroke or died
The device remains available despite additional FDA review, although the indications were tightened
When a device is shown to be harmful, there should be strong evidence to support its continued use, even in subpopulations
All high risk devices, including those for rare conditions, should be shown to be safe and effective
Cite this as: BMJ 2014;348:g93
We thank William Vodra for his comments
Competing interests: We have read and understood the BMJ policy on declaration of interests and declare that RFR is a member of the FDA circulatory system devices panel.
Provenance and peer review: Not commissioned; externally peer reviewed.