Impella Saves Lives in CS, but Patient Selection Key

Let's start with the same comment I've heard from two of my interventional-cardiology partners when discussing the lack of positive randomized controlled trials with Impella, the microaxial flow pump (mAFP).

"John, I know what the trials say, but this patient wasn't surviving without mechanical support."

I believe them. And this idea is key to understanding the results of the DanGer Shock trial of the percutaneous ventricular assist device vs standard care in patients with cardiogenic shock due to acute myocardial infarction.

Primary investigator Jacob Eifer Møller, from Odense University Hospital in Denmark, presented the findings at the 2024 American College of Cardiology Scientific Sessions in Atlanta. He reported a statistically significant 26% reduction in death at 180 days for the device vs standard care. The hazard ratio (HR) was 0.74 (95% CI, 0.55 - 0.99) with a P value of .04.

As the authors predicted in their pretrial calculations, nearly 60% of patients in the standard care arm died vs just over 45% in the mAFP arm. The absolute risk reduction of nearly 13% and the number needed to treat (NNT) of 8 is clinically important. The Kaplan-Meier survival curves separate early with the standard arm curve continually increasing over the 180 days while the mAFP curve begins to flatten after 30 days.

The results for two subgroups — sex and mean arterial pressure — are of interest. Men had a 34% reduction in death with the mAFP vs standard of care; women treated with the device had a 1% increase in death. Patients with a mean arterial pressure ≤ 63 mmHg had a 39% reduction of death with mAFP vs 12% for those with a mean pressure above 63 mm Hg.

Patients in the mAFP arm had substantially more adverse effects. Overall, the composite safety endpoint occurred in 24% of patients in the mAFP vs 6% in the standard care arm. In the mAFP arm, the number needed to harm (NNH) was 6.

And the harms were serious, including a twofold higher rate of serious bleeding and renal replacement therapy, a fivefold higher rate of limb ischemia and a nearly threefold higher rate of sepsis with a positive blood culture.

Fragile Results of DanGer Shock

A P value of .04 for the main results suggests that the statistical significance is not robust. A fragility index calculation shows that it would require two more patients in the mAFP arm dying, and two fewer in the standard care arm living, for the results to lose statistical significance.

When interpreting the statistical fragility of a trial, it is important to consider what condition is being studied. DanGer Shock tested a treatment for a condition wherein 60% of patients die with standard care. In my view, such trials deserve more leeway than large cardiovascular outcomes trials where the mortality in the control arm is in the range of 2%–5%.

Another factor lending confidence to the result is a review of causes of death. Most of the early and late deaths were due to either cardiac causes or multiorgan failure. Relatively few deaths were unrelated to the acute MI and cardiogenic shock.

Higher Rates of Serious Adverse Events With Impella

Patients in the device arm incurred more treatment-related harm. In fact, the harm signal exceeded the benefit signal (NNH = 6 vs NNT = 8). The harm signal was more statistically robust than the benefit signal.

I've heard arguments that the harm outweighed the benefits in DanGer Shock. I don't agree with that thinking.

The main reason is the investigators choice of all-cause death as the primary endpoint. Serious adverse effects, such as major bleeding, limb ischemia, sepsis, and renal replacement therapy ought to be reflected in a 6-month endpoint of all-cause death.

Lessons From Subgroups and Patient Selection

The most important aspect in translating this trial to practice requires thinking about trial procedures and patient selection.

DanGer Shock took 11 years to randomize 360 patients in three countries. The researchers screened more than 1200 patients with acute MI and cardiogenic shock and excluded 70% of them. The most common reason for exclusion was comatose out of hospital cardiac arrest (OHCA). Others included shock duration greater than 24 hours, frailty, right heart failure, and poor vascular access.

DanGer Shock therefore enrolled a very select group of patients. Patients had to be sick with a median arterial lactate level of 4.5 mmol/l, impaired left ventricular ejection fraction (median LVEF 26%) and not present late (median time to randomization was 4 hours). But they could not be too sick.

The mean arterial pressure subgroup supports this selection criteria. Those who had lower mean arterial pressures, a marker of lower cardiac output, and greater need for support, had a larger benefit from the mAFP.

It is not nefarious to enroll patients most likely to benefit from the therapy in question. In fact, doing so allows for a more precise estimate of treatment effect. The challenge of highly selective trials comes in the translation of the results to practice.

Translation is the job of clinicians. And in this case, it is hard because there is a strong signal of harm from the mAFP. If we expect to save lives with this device, we must first choose patients similar to those enrolled in DanGer Shock and then employ the same skill in its use as those of the experienced centers in the trial. That is no small thing.

It would be wrong to use DanGer Shock to expand use of the mAFP for other indications. These other indications should be tested in trials similar to DanGer Shock.

Conclusion

These results strongly suggest that the intuition of my two partners was correct. Indeed, this device does save lives in carefully selected patients. Thing is, though, a doctor's intuition is not always correct.

Møller and his colleagues deserve congratulations for both the courage they showed in accepting the need for such a trial, and for their persistence in completing it.

Their trial should teach us far more than how best to treat cardiogenic shock in acute MI.

Here's what I hope the cardiology community takes from DanGer Shock: the need for proper testing of innervations before, not after, widespread acceptance.

Imagine if there had been no benefit from an expensive and harmful device that has been used in many tens of thousands of patients.

John Mandrola practices cardiac electrophysiology in Louisville, Kentucky and is a writer and podcaster for Medscape. He espouses a conservative approach to medical practice. He participates in clinical research and writes often about the state of medical evidence.

Comments

Commenting is limited to medical professionals. To comment please Log-in.

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.