To RCT or Not to RCT? Impella Debate Pivots on How Best to Study Patients on the Brink of Death

A patient presents to the emergency department with acute MI and quickly takes a nosedive, just as the cath lab is gearing up for urgent PCI. Cardiogenic shock can’t be ignored: the heart races, blood pressure drops, breathing is a struggle, and the organs are starved of oxygen. Without some sort of intervention, the woman or man on the table may lose consciousness. Without action, they will almost certainly die.

It’s a rapidly changing scenario that unfolds minute by minute, sometimes even within seconds. Clinicians continue to grapple with how to make choices in that life-or-death moment, when the patient is fading in front of them and there are a range of tools at hand. Yet, as many told TCTMD, the path forward is far from clear.

Impella (Abiomed) is the one form of mechanical circulatory support (MCS) that currently has premarket approval (PMA) from the US Food and Drug Administration (FDA) for use in cardiogenic shock, and despite thousands of people having received the percutaneous left ventricular assist device (LVAD) for that condition, its efficacy has yet to be tested—much less demonstrated—in a large randomized controlled trial. Nor have any of the other LVADs or extracorporeal membrane oxygenation (ECMO) undergone this degree of study in cardiogenic shock. Even the intra-aortic balloon pump, once the gold standard for MCS, did not show any short-term survival benefit in the IABP-SHOCK II trial.

For some, the wealth of real-world evidence, as well as convincing results at bedside, are enough to justify continued use of Impella. But others are publicly calling for a randomized controlled trial that will settle the cardiogenic shock question in a definitive way: how do you pull patients from the precipice when their hearts begin to fail?

How Much Evidence?

Abiomed says that an Impella trial—currently known as DanShock—is still in the works 6 years after it was launched and is now likely to expand to additional sites.

As physicians await answers from DanShock and whatever trials may follow, Impella’s backstory has kicked off a storm of debate, much of it on Twitter, over how much evidence is required when treating vulnerable patients in acute need, and with a device that has such a high price tag.

From a clinical standpoint at the bedside, these are extremely sick patients who are in cardiogenic shock and our tendency is to want to do something to try and preserve their chances for recovery. Sunil Rao

“People are used to saying every question can be answered by a randomized trial, . . . but it also shouldn’t be something that impedes progress,” said Duane Pinto, MD (Beth Israel Deaconess Medical Center, Boston, MA), who estimated that Impella costs around $25,000 per disposable catheter.

“There are the purists of, ‘Show me the data.’ There are the realists of, ‘I need to try to save my patient.’ And then the two camps are actually quite similar,” Pinto told TCTMD. “They just want to know what’s best for the patients. . . . We’d all love to have a randomized trial that gave us perfect data and told us exactly who’s going to survive and who’s not. But that’s not practical in this population.”

In the absence of perfect data, “we’ve got to be continuously reflective” about whether devices are being used the right way, Pinto continued. However, “when it comes to something that’s a life-threatening disorder where we don’t have options, I feel freer about trying to make an attempt of best guess based on physiology while we accumulate data that inform us.”

Still, Prashant Kaul, MD (Piedmont Heart Institute, Atlanta, GA), said there’s a need for medicine as a field to aim higher.

“Although there is a definite improvement in hemodynamic parameters with the use of the device, contemporary practice is held accountable to the higher standard of demonstrable hard clinical endpoints. So there is a clear need for multicenter randomized control trial data in our sickest patients (ie, those presenting with cardiogenic shock, acute ST-elevation MI, as well as those undergoing complex PCI in the setting of high-risk features),” he noted to TCTMD in an email. Clinicians, he added, should help support research efforts.

For now, though, Kaul said he feels it’s “reasonable to use the device in carefully selected patients while we wait for more definitive data—this is a group of patients with a very high mortality rate and any attempt to improve outcomes should be carefully considered and not entirely dismissed due to a lack of large-scale RCT data.” Intra-aortic balloon pumps are still routinely used “in the absence of robust mortality benefit, albeit without the same cost or potential vascular risk,” he pointed out.

Pathway to Proof

Speaking with TCTMD, Sunil Rao, MD (Duke University Medical Center, Durham, NC), observed that some of the uncertainty over whether an RCT is a necessary element may relate to the “vagaries of the approval process for devices versus drugs.

“The pathway for [FDA] approval is quite a bit different than it is for devices than it is for drugs,” he explained, “and some would argue that the bar is quite a bit lower. For example, if a device is shown to be ‘safe’ by whatever definition or there is a high-risk clinical situation that this device addresses, then the device will get approved.”

It’s reasonable to ask for the best-possible data, Rao said. “On the other hand, from a clinical standpoint at the bedside, these are extremely sick patients who are in cardiogenic shock and our tendency is to want to do something to try and preserve their chances for recovery. So I certainly don’t fault any clinician for using Impella in that situation, because I think they’re trying to do the right thing.”

I just want proof so that I’m comfortable knowing that the thing that I’m using is of benefit to that particular patient that I’m treating. William Suh

William Suh, MD (David Geffen School of Medicine at UCLA, Los Angeles, CA), who has been outspoken in calling for a randomized controlled trial, told TCTMD that even he uses Impella “quite a bit” in cardiogenic shock.

“At my center, I’m actually the number one implanter of Impella, so I do have good experience with it,” he said. “Even though I’ve been pretty vocal about the lack of data, I do use it because I have seen it work.”

But faith alone isn’t enough, Suh suggested. “I just want proof so that I’m comfortable knowing that the thing that I’m using is of benefit to that particular patient that I’m treating.” Some randomized trials have debunked long-held strategies in cardiology, he pointed out, citing supplemental oxygen in MI and aspiration thrombectomy as examples.

“We need to have hope in what we do, but also a healthy dose of cynicism to really be able to provide the best care,” Suh stressed.

‘Real-world Evidence’

Impella originally gained clearance through the FDA’s 510(k) pathway. In 2015, Impella 2.5 garnered PMA for use in high-risk PCI, and then in 2016, this approval was extended to acute MI and postcardiotomy cardiogenic shock for not only the Impella 2.5 but also its successors, the Impella CP, 5.0, and LD devices. Earlier this year, the devices were cleared for the treatment of cardiogenic shock in the setting of cardiomyopathy—including peripartum and postpartum cardiomyopathy—or myocarditis stemming from isolated LV failure that does not respond to optimal medical therapy or conventional treatment. Impella RP, meanwhile, was approved for use in right heart failure in 2017, and the high-risk PCI indication has also continued to broaden.

“To date, there have been seven attempts at randomized controlled trials of Impella in cardiogenic shock,” Abiomed commented to TCTMD via a written statement. “All but one was stopped with inadequate enrollment to evaluate a survival benefit or had methodological flaws.”

The company was quick to point out, though, that the cache of observational data isn’t slim: to receive PMA in cardiogenic shock, Abiomed submitted an analysis of 415 patients from the RECOVER 1 study and cVAD registry, as well as a literature review including 692 patients treated with Impella from 17 clinical studies. As of 2016, the Impella Quality (IQ) database included more than 15,000 patients treated for acute MI cardiogenic shock over a span of 8 years, a number that has surely grown since. More than 60,000 patients have received the device for various indications in the United States.

All of this experience can provide guidance on how best to use Impella, proponents of the device say. A recent analysis of the IQ data, for example, showed that higher hospital Impella volume, use of invasive hemodynamic monitoring, and first-line versus salvage Impella use all are linked to better survival. Other practical information is coming from the National Cardiogenic Shock Initiative, which has 50 participating hospitals. Data has also been gleaned from the Nationwide Inpatient Sample.

We’d all love to have a randomized trial that gave us perfect data and told us exactly who’s going to survive and who’s not. But that’s not practical in this population. Duane Pinto

Seth Bilazarian, MD, chief medical officer of Abiomed, told TCTMD that “real-world evidence is in many ways preferred and more powerful to have us understand really what’s happening with the use of the device and outcomes, and how we can best improve training and technology to improve outcomes in cardiogenic shock.”

With the strictest interpretation, the findings of any randomized controlled trial can only be applied to patients who meet the same inclusion and exclusion criteria. For example, a product that hasn’t been tested in women couldn’t be used in women under this logic, Bilazarian said. “But of course we don’t do that. We say we have to think beyond what the data shows and make reasonable suppositions.”

Cardiogenic shock hasn’t had any therapies show effectiveness in a randomized controlled trial, he stressed, adding, “We know that the intra-aortic balloon pump had a very well-conducted randomized controlled trial that failed. It showed no difference compared to no therapy.” So here, according to Bilazarian, aiming only for the goal of a randomized trial becomes an intellectual exercise.

“If you really care for a patient who is dying in front of you, who has very low blood pressure, who is deteriorating, your options are to use therapies that either have been proven to be ineffective, like intra-aortic balloon pump, or to use therapies that don’t have randomized controlled trial data but make good sense from a biological standpoint, like Impella, [which] unloads the ventricle and supports the end organs,” he commented. “Or you can use therapies like vasopressors, which don’t have any randomized controlled trial [and] we think are toxic to the heart, or you can use other therapies like ECMO, which do support end organs but we know load the heart and cause the heart to likely fail and not be able to recover.”

What’s Standing in the Way of a Trial

So what are the obstacles to doing a randomized trial? According to everyone who spoke with TCTMD for this story, there are many justifiable reasons why this hasn’t been resolved yet for Impella.

In the United States, at least, the biggest challenge may be that Impella is already on the market, they said. Physicians who are accustomed to having the devices on the shelf might be loath to give them up, making them reluctant to enroll patients in a 1:1 trial where Impella use isn’t an option half of the time. “It’s going to be really hard when you have an FDA-approved device for cardiogenic shock to withhold that type of treatment because you’re in a trial setting,” Suh observed.

Several sources told TCTMD that, for this reason, a randomized controlled trial might best be done somewhere outside the United States, such as in the United Kingdom.

Also formidable is the level of informed consent that is required by institutional review boards (IRBs) in this country. “This is not only true in the shock space, it’s also true in just the regular STEMI space,” Ajay Kirtane, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY), commented. In doing US-based STEMI trials, “we started getting pushback from IRBs saying if you’ve got a patient coming in with a STEMI, that patient, even if they’re conversant, may be incapable of research consent.” In some cases, he noted, IRBs have required calling a family member in order to obtain consent. Add the time-sensitive nature of shock into the equation and informed consent becomes even more difficult.

If you don’t allow crossover and you have a sick patient that’s dying, you’re basically saying, ‘OK, let them die.’ Ajay Kirtane

Even if enrollment goes steadily, crossover could be rampant. “What happens if a patient starts failing? Do you cross them over or do you just not do anything? And the reality is, if you allow crossover, then that contaminates a treatment effect,” Kirtane said. “It’s going to be virtually impossible to show superiority of one device versus nothing, or another, when you’re allowing crossover. But if you don’t allow crossover and you have a sick patient that’s dying, you’re basically saying, ‘OK, let them die.’”

Bilazarian said another issue is that cardiogenic shock involves a “system of care.”

When investigating a pharmaceutical therapy, “comparators are very easy,” he explained. “You give a pill or you give a placebo and [you look at] a large enough number of patients to assess efficacy and safety.” And even with devices, he added, usually the products are used in a standard way and thus can be compared directly, such as in trials evaluating two different stent platforms.

Treatment of cardiogenic shock, however, is “really a care pathway that begins with identification [and] the use of the device appropriately,” Bilazarian commented. “But most importantly in my mind what’s really assessed in the survival of patients is the care after the cath lab, in the intensive care unit where doctors and nurses have to care for these patients 24/7” over the course of up to 6 days, depending on Impella type.

Information gleaned from real-world use of Impella can now be used to design trials, such that strategies are standardized across centers, he said. These trials would not be possible without the real-world experience that has come before, Bilazarian asserted.

Cardiogenic Shock a Heterogeneous Group

Importantly, “cardiogenic shock” as a category is very heterogenous, many said, not only as it relates to the root cause of the condition but also with regard to when patients present and how they deteriorate.

As Pinto noted, “there are many people that fit the definition of shock who are relatively well and are probably only harmed by the complications of a device, because they were going to survive no matter what. And then there are those people who are not salvageable, because their scenario and their illness is so severe that no matter what we do, it’s futile.

“Clinicians struggle with trying to find that middle ground at time zero with the patient and not being able to know is this a person salvageable or not,” he said, adding that prediction tools for gauging which shock patients will survive are “very rudimentary” at this point.

What this means is that “trying to design and implement a randomized trial is very difficult,” Pinto observed.

Daniel Burkhoff, MD, PhD (Cardiovascular Research Foundation, New York, NY), took a similar view. “Shock is a huge spectrum,” he told TCTMD. “The pathophysiology of cardiogenic shock is complicated. It’s more complicated than just hemodynamics, and so just addressing the hemodynamic aspects may or may not be sufficient. But for sure, it’s necessary.”

Clinicians can benefit from a better understanding of this pathophysiology and how the various MCS devices work, he said, adding that CRF’s TEACH program “has contributed significantly to providing such education to a large number of clinicians practicing throughout the world.” To improve survival, what’s needed is “more rapid deployment of devices, more rapid reductions in toxic pharmacologic agents [like vasopressors], and appropriate escalation of therapy when the first therapy is not reaching certain targets,” Burkhoff advised, noting that these targets are “still ill defined.” Other factors like inflammation and microvascular dysfunction may also be at play.

Rao, who said he doesn’t have a “standard approach” to MCS in cardiogenic shock, described how this heterogeneity can complicate decision-making: “If you’re dealing with acute MI and cardiogenic shock, how much of [the shock] is due to the artery still being closed? Would opening the artery allow you to overcome that shock? And how much of it is, this is a late presentation and opening the artery probably doesn’t have as much benefit as providing some kind of support upfront?”

All of these moving parts translate into cardiogenic shock being difficult to define and make it hard to predict who will benefit from different strategies, much less pin down a trial design that will embody these nuances. Rao said that the Cardiac Safety Research Consortium, a public-private partnership between the FDA and academic institutions including Duke, where he’s based, is going to hold a think tank devoted to these issues in December.

DanShock Remains in the Game

DanShock, too, has been “struggling” much like trials before it, “with a limited number of patients being enrolled annually,” Bilazarian acknowledged. “I think it is in jeopardy of [similar] issues related to these other trials in terms of trial fatigue, but [the investigators] are committed to trying to push it on and are interested in expanding the number of sites, with . . . the hopes that that will bring the trial to completion with an adequate sample size to scientifically assess the Impella effect in cardiogenic shock.”

All of the other randomized controlled trials in this area “were stopped prematurely and were reported as if they were complete trials, and that’s problematic because sometimes only the title or the abstract is read,” he cautioned. “And when very small trials are done and compared that way, it’s really a disadvantage to the patients and the scientific community that a full scientific evaluation hasn’t been completed to allow a fair assessment of a new technology.”

Currently, the DanShock trialists are attempting to expand into Germany, Bilazarian reported. “At the most recent German interventional congress, the DGK, which was about a month ago, the Danish PI met with German sites that were interested. So there’s general enthusiasm and commitment for funding from Abiomed, so we know that it’s something that will happen. We can’t predict the timing. Obviously sometimes with ethical review and contracting it can take a little bit of time, but we expect that this will happen and I think that this has bolstered the Danish investigators to press on.”

The investigator-initiated trial was originally sponsored by the Danish government, he explained, but because it was taking longer than expected to complete, investigators reached out for assistance and additional funding. “And we have committed to that, so that’s the current status,” Bilazarian said. “The funding of the trial infrastructure and trial support going forward will come partly from Abiomed.”

According to the company, the study will randomly assign patients to receive “conventional circulatory support” or to the Impella CP device, with inotropic support if needed. DanShock has a planned enrollment of 360 patients and a primary endpoint of death. Another randomized trial, STEMI DTU, is currently evaluating the best duration of Impella CP use prior to PCI in approximately 50 patients with anterior STEMI but without shock.

The question is whether DanShock will settle anything. As Rao put it, “If the trial ends up showing no difference between the treatment strategy and the control strategy, there is always the question of whether it was adequately powered. That’s the biggest risk—if a 300-patient trial shows no difference, does it add to our database, or does it simply fuel the controversy more?” Supporters of Impella will say the trial was just too small, while critics will point to it as yet another negative trial, he explained. And until DanShock reports results, the field won’t know whether the trial is sufficiently powered.

Hash(tag)ing It Out

Doubts over DanShock’s future, as well as back-and-forth over the randomized data gap, erupted on Twitter not long ago and seemed to unsettle many on both sides of the debate raging under the hashtag #RCT4IMPELLA.

What’s clear, according to those who spoke with TCTMD, is that online discussions are valuable but can get out of hand, even among professionals identifying themselves by their real names.

“As medical professionals, people are listening to what we’re saying, people are watching what we’re saying, and maintaining professionalism and civility is important,” noted Pinto, who said he likes to hear opposing views. “And if that erodes, I get very disappointed in myself and my colleagues as a representation of our profession, and that’s what I don’t like to see on Twitter.”

For Rao, the discussion has been “really fascinating.”

“I think that it’s good people are questioning the data,” he said. “I hope we come to some resolution on it.”

Suh, who stoked much of the debate, said that “Twitter has been a really eye-opening experience, because it allows for open conversation pretty quickly.” Not only can there be misunderstandings among those taking part, there are also bystanders—patients, too, can see what’s posted online. This public record could have implications for malpractice claims, Suh suggested. “It’s never happened before, but it’s definitely in the realm of feasibility.”

Kirtane, for his part, also expressed concerns over tweets being in a public forum, where nuance is easy to miss.

“While I understand that it’s important for transparency, so that patients and anybody out there can see how physicians interact with one another, I also think that with 280 characters you can’t appreciate the depth of the conversation and the things that often go unsaid,” he said, adding, “The other challenge is one sees sarcasm and jokes interspersed with serious conversations about patients dying and to be honest with you, I don’t think physicians ought to conduct themselves that way.”