Impella-Pump Bleeding, Mortality 'Signals' in MI-Shock and High-Risk PCI

November 26, 2019

PHILADELPHIA — Interventional cardiology has taken quite a shine to Impella ventricular assist pumps (Abiomed) for cardiogenic shock (CS) and to support high-risk procedures, but two new observational studies have independently called into question whether all of that luster is deserved.

Together, the analyses show a steady climb in Impella use in the United States since its 2008 regulatory approval, and a parallel decline in intra-aortic balloon pump (IABP) use for supported percutaneous coronary intervention (PCI) and in acute myocardial infarction (MI) complicated by shock.

Also, both studies point to significantly greater in-hospital risks of major bleeding, death, and other adverse events for patients supported by Impella devices, compared with those managed with an IABP.

The findings are hardly conclusive; neither study was randomized, and although they used propensity matching to smooth out some differences between the Impella and IABP populations and procedural characteristics, plenty of potential confounders remained.

But there are few randomized data available to support Impella use in either interventional setting. And the similar cautionary conclusions based on separate cohorts raised some eyebrows when the analyses were presented back-to-back here at the American Heart Association (AHA) Scientific Sessions 2019.

At the very least, researchers and observers said, the findings call for robust randomized trials to provide more definitive understanding of when to prefer or perhaps avoid Impella pumps compared to the IABP.

"Overall Consistent Findings"

One of the studies, a propensity-matched comparison of patients receiving mechanical circulatory support (MCS) for MI-related shock, saw a nearly one-third excess in mortality and almost a doubling in risk of major bleeding, both in-hospital endpoints, with use of Impella compared to IABP (45.0% vs 34.1%, respectively, for mortality and 31.3% and 16.0%, respectively, for major bleeding).

The analysis was based on 2015-2017 cohorts that underwent PCI for MI-CS, combined from two National Cardiovascular Data Registry (NCDR) data bases, the CathPCI Registry and Chest Pain-MI Registry.

In the other presentation, a comparison of Impella and IABP use based on nearly 50,000 cases of MCS-supported PCI from 2004 to 2016 in the Premier Healthcare Database, about 10% of which used Impella, that device was associated with significantly more deaths, bleeding, and stroke. The risk increases were by 24% (P < .0001), 10% (P < .05), and 34% (P < .0001), respectively, compared to IABP.

Deaths and strokes in the data base overall shot up after the Impella gained regulatory approval in 2008, compared to earlier years; mortality went up 17% and strokes more than tripled (P < .001 for both findings). There was also wide variation nationally in both Impella use and its risks.

"Impella is increasingly being used instead of the balloon pump to support PCI in the United States," Amit P. Amin, MD MSc, Washington University School of Medicine, St. Louis, MO, said while formally presenting the study, which was published in Circulation at about the same time.


But there is wide variation "not only the use of Impella across hospitals, but also its associated clinical outcomes," he said. For patients treated with Impella, "We saw a 2.5-fold variation in bleeding across hospitals, and a 1.5-fold variation in acute kidney injury, stroke and death across hospitals."

Those outcomes "did not show any substantial improvements" during the study period.

The findings are not a call to pull back on such Impella use, Amin told | Medscape Cardiology. "I think what the data support is a signal at best, and the signal indicates that there is no benefit." What they do show, he said, is a need for more definitive studies.

"That we have multiple studies that are showing overall consistent findings, it means that there's a need for additional evidence. Certainly, there are limitations to observational studies," Sanket Dhruva, MD, MHS, University of California San Francisco, told | Medscape Cardiology.

"We really need additional data on the use of Impella devices, especially given their increasing utilization in clinical practice, both for patients with acute MI and cardiogenic shock as well as in other settings," said Dhruva, who earlier had presented the NCDR data.

It's "concerning" that the two analyses didn't even show parity between the two MCS devices, and moreover, that the evidence "went the wrong way" for Impella, Deepak Bhatt, MD, Brigham and Women's Hospital, Boston, told | Medscape Cardiology.

"In general, I'd be skeptical of nonrandomized data showing that. But there's no randomized data to help us here, or what randomized data exists is from relatively small studies," said Bhatt, who isn't associated with either new study.

"I think they are actually well done analyses, as best as one can do with nonrandomized data," he said. "Nonetheless, a one-two punch, right? Now there are two different analyses that are basically coming to the same conclusion."

The findings, "really do question whether we're using too much Impella."

A Tale of Two Studies

The Premier Healthcare data analysis started with 48,306 patients who underwent PCI with MCS at 432 hospitals from the outset of 2004 to the end of 2016. They included 4,782 who received Impella after its 2008 approval. That overall use rate of 9.9% doesn't reflect the steady growth in Impella use; the device accounted for 31.9% of procedures in 2016, Amin reported.

Propensity-Adjusted Odds Ratio (OR) for Outcomes, Impella vs IABP
In-Hospital End Point OR (95% CI) P Value
Death 1.24 (1.13–1.36) <.0001
Bleeding 1.10 (1.00–1.21) .0445
Acute kidney injury 1.08 (1.00–1.17) .0521
Stroke 1.34 (1.18–1.53) <.0001

Overall outcomes in the population, regardless of MCS device, were significantly worse for patients after the 2008 approval of Impella. And counterintuitively, among hospitals using Impella, those using it the most had significantly worse outcomes with Impella than those using it the least.

Adjusted Odds Ratio (OR) by Time Period (Overall Outcomes) and by Hospital-Level Impella Use Level (Impella Outcomes Only)
In-Hospital End Point Overall, Impella Era vs pre-Impella Era,*
OR (95% CI) P Value
Impella Only, Highest vs Lowest Quartile Hospitals for Impella Use
OR (95% CI) P Value
Death 1.17 (1.10–1.24) <.001 1.48 (1.32–1.67) <.0001
Bleeding 0.99 (0.94–1.05) .843 1.17 (1.03–1.33) .015
Acute kidney injury 1.91 (1.81–2.01) <.001 1.29 (1.17–1.43) <.0001
Stroke 3.34 (2.94–3.79) <.001 1.26 (1.06–1.50) .0094
*After vs before US approval of Impella in 2008

The NCDR-based analysis presented by Dhruva was based on 1,768 patients managed with Impella and 8,471 who received IABP support among almost 270,000 patients undergoing PCI in the two subregistries from the last quarter of 2015 through 2017.

Although Impella use was about 6.2% overall in the total combined cohort, it increased from 3.5% to 8.7% (P < .001) throughout the study period, while IABP use declined from 32.1% to 27.3%.

The group derived 1,680 propensity-matched pairs of patients with acute MI and shock undergoing PCI supported by Impella and IABP, respectively; the pairs encompassed 95% of Impella-managed patients, Dhruva pointed out.

Propensity scores used for matching were based on 75 variables specifying demographics, clinical history, clinical presentation, MI, coronary anatomy, and laboratory parameters.

In-Hospital Outcomes Rates, IABP vs Impella, in 1680 Matched Pairs
End Point IABP, % Impella, % Difference in absolute % points (95% CI)
Death 34.1 45.0 10.9 (7.6 to 14.2)
Major bleeding 16.0 31.3 15.3 (12.5 to 18.2)

The group on Impella fared worse for in-hospital mortality and major bleeding than those managed with IABP, regardless of when MCS was initiated in relation to the PCI procedure.

In-Hospital Outcomes Rates, IABP vs Impella, by Timing of MCS Placement
End Point IABP, % Impella, % Difference in absolute % points (95% CI)
Placement pre- or during PCI, n = 573 pairs
Death 36.8 45.6 8.8 (3.1–14.4)
Major bleeding 16.6 27.4 10.8 (6.1–15.6)
Post-PCI placement, n = 662 pairs
Death 32.2 44.0 11.8 (6.6–17.0)
Major bleeding 15.7 34.4 18.7 (14.2–23.3)

There is plenty of potential for confounding in both studies, and the two populations were very different, agreed Ranya Sweis, MD, Northwestern University, Chicago, Illinois, speaking to | Medscape Cardiology.

But, "it was very interesting to me that these two databases that were completely different had very similar results," agreed Sweis, who was the invited discussant for the Dhruva and Amin presentations at the AHA sessions. "I think it basically forces us to do a randomized study of using Impella, particularly, in our patients."

Sweis didn't have the opportunity to publicly comment on the two studies after their formal presentations. Her presentation was to be the last of that morning session, but it was cut from the line-up ostensibly because earlier presentations went longer than their scheduled time slots.

Lacking good randomized data for now, she said in an interview, there are still lessons in the new observational findings.

"One of the probably more important take homes today is that even with all of the adjustments, there's variability in [MCS] use in outcomes at different hospitals. We know bleeding is associated with mortality. We also know that bleeding avoidance strategies have decreased risk of bleeding and of mortality related to bleeding. So that means we need to do better there."

The Premier Healthcare Database analysis, Bhatt observed, "documented a greater than five-fold variation in Impella use. That just tells me that a randomized trial could easily be done. When there's that much variation, it's basically just quite arbitrary who's getting the device and who isn't. And I think if we don't do a trial at this juncture, we may never get one done."

Amin discloses receiving an unrestricted grant from MedAxiom Synergistic Healthcare Solutions Austin, TX and is a consultant to Terumo and GE Healthcare. Dhruva and Sweis report they have no disclosures.

Bhatt discloses sitting on advisory boards for Cardax, Cereno Scientific, Medscape Cardiology, PhaseBio, and Regado Biosciences; conducting unfunded research in association with FlowCo, Fractyl, Merck, Novo Nordisk, PLx Pharma, and Takeda; and receiving research funding from Abbott, Afimmune, Amarin, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Chiesi, CSL Behring, Eisai, Ethicon, Ferring Pharmaceuticals, and Forest Laboratories.

Circulation. Published online November 17, 2019. Report.

American Heart Association (AHA) Scientific Sessions 2019: Abstract LBS.04. Presented November 17, 2019.

Follow Steve Stiles on Twitter: @SteveStiles2. For more from | Medscape Cardiology, follow us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.