A Look at the IMPRESS Trial: Impella CP vs IABP in Acute MI Complicated by Cardiogenic Shock

Why is the IMPRESS trial important? 

For such a critical area as acute MI with cardiogenic shock, we have very little data to guide the use of support devices to improve clinical outcomes. Over the last 15 years, there have only been a few small trials comparing a balloon pump with more powerful devices such as Impella (Abiomed) and TandemHeart (CardiacAssist). There has been a single, large trial looking at routine balloon pump use versus no routine balloon pump use. None of those trials have shown any clear direction of improved outcomes.Essentially, the issue is that mortality and outcomes for cardiogenic shock are unchanged over the last two decades. 

How frequently is acute MI with cardiogenic shock seen in cath labs today?

Caring for patients with ST-elevation MI, cardiac arrest, and cardiogenic shock is now a routine part of contemporary cath lab work. In many areas, STEMI and cardiac arrest care is delivered within regional systems organized in concert with pre-hospital emergency services. So, the expectation that shock is emergently managed and fixed in the cath lab is an increasing reality. Most busy centers are seeing this type of patient on almost a weekly basis.

Can you describe the IMPRESS trial design?

IMPRESS is unique in that it was an investigator-initiated, randomized comparison of Impella CP versus balloon pump in patients suffering acute MI complicated by cardiogenic shock. There are many challenges with conducting research in this patient population, and, of course, the treatments could not be blinded. The investigators endeavored to enroll extremely sick patients and so mechanical ventilation was a required inclusion criteria. For that reason, nearly all of the 48 enrolled patients had suffered a cardiac arrest prior to percutaneous coronary intervention. A few things are important to note about this trial. One is that the timing of device placement, meaning whether to place the balloon pump or Impella before PCI, was left to the discretion of the clinician. Also important is that investigators chose mortality at 30 days as the primary endpoint rather than a less meaningful surrogate endpoint. 

Results were presented at TCT in late 2016. Can you share some highlights?

Important to the results is that the trial was constructed with an anticipated 90% mortality rate in the balloon pump arm, and so for that reason, the trial was designed with a relatively small enrollment. As far as the trial endpoints, very importantly, the majority of patients, over 80%, had the support device placed after PCI. At 30 days, there was no difference in the primary endpoint. Furthermore, survival was identical at 6 months when mortality was 50% in both arms. 

How should clinicians interpret the trial results to apply to their everyday practice?

The IMPRESS trial is small and there are limitations on how one should interpret the data. I think the one message that clinicians should take away is that two or three small studies have now shown that routine use of novel support devices may not improve outcomes. The conclusion for clinicians might be that some measure of restraint in adopting new support devices is appropriate. The size and quality of the IMPRESS data are not enough to drive any major changes in guidelines or clinical practice, but I believe that it certainly puts devices like Impella in perspective. Increased bleeding risk is clear. Without definite benefit, the importance of bleeding events on device choice becomes more crucial. The risk profile of IABP is lower and has been very consistent across contemporary trials. Clinicians probably already appreciate the difference in bleeding risk, ease of use and experience between devices. The message from IMPRESS may be that the actual clinical benefits of the newer devices are likely far more similar to IABP than clinicians had expected.  

Outcomes in cardiogenic shock trials haven’t changed much in decades. For some segment of this population, there may be issues of medical futility though that is harder to determine for individual patients. Some patients do poorly no matter what we do. Regardless of hemodynamic support, neurologic injury may be the terminal problem following cardiac arrest. The same may be true for those with progressive cardiac failure, and we still don’t have data to support one device over another.  

Can you address the timing of device initiation?

An issue common to all these trials is that timing of device placement may have something to do with improved outcomes. If you look at the IABP-SHOCK II trial, the majority (more than 80%) of the patients that were supported with a balloon pump had the device placed after PCI. The same is true in the IMPRESS trial, where 80% of patients had the device, whether Impella or balloon pump, placed after PCI. There are good reasons why earlier placement might be better.

Why might clinicians make the choice to initiate device support after the procedure? 

The fact that device support was initiated after PCI probably reflects two things. One is the desire of clinicians to achieve reperfusion as quickly as possible in STEMI. The second may reflect the difficulty in enrolling a cardiogenic shock trial where the patients are very heterogeneous and much clinical decision-making is left up to the physician. There are some trial and registry data with various devices including balloon pump and Impella that suggest outcomes may be better with earlier placement of the device. In the IMPRESS trial, regardless of whether Impella or balloon pump was used, those patients who had the device placed prior to PCI had numerically superior survival.

Will clinicians respond to the idea that patients might benefit from taking the extra time to insert a support device prior to PCI?

Nationally, we have gotten to the point  where life-saving primary PCI is delivered well within target times of 75 or 90 minutes. Data shows that treatment and reperfusion times are better than ever even for shock and post-arrest patients. So, I do believe that physicians are getting comfortable with slowing down 5-10 minutes to insert and initiate a balloon pump or Impella. There isn’t a clear signal that a few extra minutes has a significant deleterious effect on outcomes though that might not be the case for individual patients. Impella is easy to insert, initiate and only requires an additional 10 minutes. A balloon pump can be placed in about half the time. Neither device creates unacceptable delays. So we’ve sped up treatment times to the point that I do believe clinicians can be comfortable with the concept of slowing down to stabilize before reperfusion.  

Could you tell us more about the patient population in the trial? 

The study included relatively young patients experiencing truly catastrophic illness. The average age was only 58 or 59 years, and while typical risk factors and co-morbidities were present, these were, in general, previously healthy patients. IMPRESS is too small to perform a valid subgroup analysis. But that reflects another challenge in designing and completing a cardiogenic shock trial. The investigators really should be applauded for their efforts here.  

How were the patients consented?

Because of rapid triage and very short door-to-balloon times, it is difficult to enroll and manage the consent process in STEMI trials, and still more difficult with cardiogenic shock or cardiac arrest. Informed consent was required as a part of this study, and as you can imagine, it was a significant challenge to investigators. They obtained consent from a patient’s family or legal representative, and tried to do so in a way that did not significantly delay care.

IMPRESS does seem to indicate that if physicians want to continue using an intra-aortic balloon pump, that they can feel more comfortable in doing so.

Correct. There have been dramatic changes in clinical practice and the use of support devices without any substantial evidence to guide clinicians. To that end, the IMPRESS trial is further support that physicians can still rely on a balloon pump in appropriate cases. It is very clear that novel devices such as Impella bring with them dramatic increases in cost and resource utilization, while IMPRESS and the other small studies have not shown a signal for improved outcomes. Furthermore, there are risks with any device. Bleeding and hemolysis are more common with Impella. Hemolysis didn’t show up as an issue during the largest study of Impella, but that may have to do with the very limited duration of support during high-risk PCI as opposed to the longer duration of support in cardiogenic shock. In IMPRESS, patients in both the balloon pump and the Impella arms received support for about 48 hours. 

Investigators expected mortality to be 90% in the balloon pump arm, but it was much better, at 50%. 

Yes, 40 and 50% 30-day mortality is the same as we saw in the late 1990s with the original SHOCK trial. It is the same mortality we saw in the more recent IABP-SHOCK II trial, and so the investigators probably erred in powering their study to a 90% risk of mortality in the balloon pump arm. They also probably erred in expecting a 30 or 40% improvement in survival with the use of Impella CP.

It seems like the next step might be looking at timing of insertion.

I would argue that because cardiogenic shock is such a critical challenge for our field, there are a few “next steps.” First, we can’t continue to evolve shock management without large, adequately-powered clinical trials to guide physicians and help patients. Then, issues like device timing clearly need to be part of trial design. Finally, we need to know whether we must rely on investigator-initiated studies or whether the device industry will step forward and make a stronger effort to help advance the science. Timing of placement is just one part of the puzzle. 

Any final thoughts?

Caring for STEMI, shock, and post-arrest patients is one of the most rewarding parts of interventional practice. It’s wonderful that we have so many tools at our disposal, and there is a role for all of them. Despite the randomized trial results, I do believe we are getting better at helping patients. In the future, I would expect it to be more apparent that we’ve turned the corner on outcomes. Until then, we need to approach these complex conditions with the same thoughtfulness and rigor that has brought about every advance within the field of interventional cardiology. 

Disclosure: Dr. Joseph Thomas has received speaker’s honorarium from Getinge/Maquet.

Dr. Joseph Thomas can be contacted at jthomas@labiomed.org.