Impella

What it is

A microaxial rotary blood pump mounted on a catheter. It is passed retrograde across the aortic valve so the pump body straddles the valve: inlet inside the left ventricle, outlet in the ascending aorta. An impeller spinning at tens of thousands of revolutions per minute pulls blood continuously out of the ventricle and throws it into the aorta.

That sentence contains everything that matters. The device removes volume from the left ventricle and delivers it to the systemic circulation. It does not wait for systole. It does not care whether the aortic valve opens. It works in a heart that is barely ejecting at all.

The physiology follows. End-diastolic volume and pressure fall, wall stress falls, oxygen demand falls. Wedge pressure falls, which is why the pulmonary edema clears faster than you expect. Coronary perfusion improves not by raising diastolic pressure the way a balloon does, but by lowering the end-diastolic pressure the coronaries have to push against. Because the aorta is fed continuously, the arterial trace becomes narrow and blunted. Do not panic about a low pulse pressure on a well-supported Impella. It means the device is doing the work.

One corollary people forget: the Impella unloads the left ventricle only. Every liter it moves forward had to come through the lungs from the right ventricle. If the right heart fails, the pump has nothing to pump, and flow falls with suction alarms no matter how high you set the support. Falling flow in a previously stable patient is a right heart problem until proven otherwise, which is one reason these patients usually need a PA catheter.

The platforms

Same idea at different sizes. The number is roughly the flow in liters per minute, and the hardware you have to put through an artery scales with it. Know which device your patient has before you open your mouth on rounds.

The right sided device is worth a second look because residents consistently misdescribe it. The RP is a venous device. Up the femoral vein, through the right atrium, across the tricuspid valve, through the right ventricle, across the pulmonic valve, outlet parked in the pulmonary artery. It draws deoxygenated blood from the IVC and delivers it to the lungs, bypassing the failing right ventricle. It does nothing for the left ventricle. In biventricular failure you may see both devices, sometimes called BiPella, and then you have two consoles and two purge systems.

Licensed duration differs too, on the order of days for the percutaneous devices and around two weeks for the surgical ones. Nobody should be vague about how many days the device has been in, because that number drives escalation, conversion and explant. The instructions for use govern.

Indications and evidence

Cardiogenic shock

The main reason you will meet one. A patient whose cardiac index will not come up, with a rising lactate, cold extremities, oliguria and escalating vasopressors, gets a device because the drugs are not working and because every extra microgram of norepinephrine buys blood pressure at the cost of myocardial oxygen demand. The Impella substitutes real flow for some of that pressor requirement, and does it while lowering the ventricle's workload rather than raising it. That is a different trade from what an inotrope offers.

Be honest about the evidence. For years the enthusiasm rested on registries, single-arm series and physiologic plausibility. Observational comparisons repeatedly showed worse outcomes with Impella than with balloon pump, which is close to uninterpretable given how much sicker the Impella patients were. Guideline bodies noticed the absent trial data and have been cautious, giving percutaneous support in shock a weak recommendation in selected patients rather than a routine one.

DANGER-SHOCK changed the conversation. It randomized patients in cardiogenic shock complicating ST-elevation myocardial infarction to routine early Impella CP plus standard care, or standard care alone, and reported a reduction in death over follow-up in the device arm. It also reported a clear excess of adverse events: more renal replacement therapy, more bleeding, more limb ischemia, more hemolysis, more device complications. Both halves are the finding. This is not a free intervention and the trial says so.

Read the population before generalizing. STEMI shock, selected patients, entry criteria requiring genuine hypoperfusion, and exclusion of patients who arrested out of hospital and had not woken up. It does not tell you what to do in decompensated chronic heart failure, myocarditis, post-cardiotomy shock or a comatose post-arrest patient.

High-risk PCI

Protected PCI means placing a device before intervening on someone in whom losing the vessel for the length of a balloon inflation would be catastrophic: low ejection fraction, unprotected left main, a last remaining conduit, heavily calcified multivessel disease. The device supports through the ischemic moments and comes out in the lab. The data are less clean than the marketing: the pivotal randomized comparison against balloon pump did not meet its primary endpoint at 30 days, with a more favorable signal on later and non-primary analyses, and further randomized work is ongoing. The honest summary is that it makes hard procedures safer to attempt, not that it is proven to save lives.

The rest

• Bridge to decision. Buying 48 to 72 hours of perfused end organs so a transplant or durable device evaluation can happen, or so a neurologic prognosis can declare itself.
• Bridge to recovery in fulminant myocarditis, peripartum cardiomyopathy, or stunning after revascularization.
• Refractory ventricular arrhythmia, including support during ablation, where the device perfuses through non-perfusing rhythms.
• Venting a distended left ventricle on VA ECMO, covered in the last section.

Contraindications

Most are anatomic, and most come down to the fact that this catheter has to sit across the aortic valve and suck blood out of the ventricle.

Reading the console

This is the section that matters. Almost every Impella disaster on a ward round is a console somebody looked at, found confusing, and walked away from. There are four things on that screen and you should be able to explain all four.

1. The placement signal

A pressure waveform from a sensor near the outlet, which is to say from wherever the outlet currently sits. It is the best single answer to "where is this device."

• An aortic waveform, with a dicrotic notch and a diastolic pressure well above zero, means the outlet is in the aorta. That is what you want.
• A ventricular waveform, diastolic pressure falling toward zero and no notch, means the outlet has slid back into the ventricle. The whole device is now inside the ventricle, pumping blood from the ventricle into the ventricle, which achieves nothing.
• A dampened or flat signal may mean the sensor is against tissue, or thrombus, or a genuine sensor fault.

Newer consoles also display a derived left ventricular pressure and a calculated end-diastolic pressure, which is useful for weaning and for spotting a distending ventricle.

2. The motor current

Motor current is proportional to the work the motor is doing, and that work is set by the pressure difference between inlet and outlet. So it is a free position check requiring no imaging.

With the pump correctly across the valve the gradient changes through the cycle. In diastole the ventricle is at 10 or 15 and the aorta at 70, so the gradient is large and the motor works hard. In systole the ventricle pressurizes toward aortic pressure, the gradient collapses, and the motor works less. The result is a pulsatile motor current waveform, and that pulsatility is your evidence that the two ports are on opposite sides of the aortic valve.

3. P levels

The performance level is the impeller speed setting, running from P0 to P8 or P9 depending on platform. P0 is off, P1 is minimum speed, and each step up raises RPM and therefore flow. A P number means nothing by itself, only the flow it produces in this patient at this filling pressure, and the same P level gives a different flow in the same patient an hour later.

Two warnings about the ends of the range. P0 with the device still in situ is dangerous: with the impeller stopped, blood runs backward down the cannula from aorta to ventricle, functionally acute aortic regurgitation, and the stationary pump thromboses. If it is off it needs to come out or be restarted. Prolonged running at the bottom of the range also risks retrograde flow across the pump, which is why weaning trials at the lowest levels are time-limited rather than a place to park somebody overnight.

4. Suction versus malposition

The two alarms you will see most are "Impella position wrong" and "suction." They look similar at the bedside because both drop the flow and change the waveforms, and the correct response to each is the opposite of the correct response to the other.

The unifying principle: suction is a volume problem, malposition is a geography problem. If in doubt the echo answers it in 90 seconds. Parasternal long axis, find the inlet, measure how far below the annulus it sits.

The purge system

The purge is not an infusion for the patient. It keeps the motor alive. The impeller runs on a bearing, and blood must not get into that bearing or it will clot and the motor will seize. So the console continuously drives fluid outward through the motor housing, against the blood, to keep blood out.

That fluid is dextrose, typically between 5 and 20 percent, with heparin added, commonly on the order of 25 to 50 units per milliliter, hung in the purge cassette. It runs at a few to a few tens of milliliters per hour and generates a purge pressure the console displays and holds within a wide band. The dextrose is there for viscosity, not calories: a higher concentration is thicker fluid, which means a higher purge pressure at the same flow.

Daily management

Put this in the plan of your progress note as its own block rather than burying it in the cardiovascular paragraph.

Hemolysis

Some hemolysis is intrinsic. Blood is being accelerated through a narrow cannula by a fast impeller and red cells do not enjoy it, so a mildly raised LDH and plasma free hemoglobin with a suppressed haptoglobin is the expected background. What matters is the trend, and many units act on a plasma free hemoglobin above roughly 40 mg/dL and act urgently on a rising one.

The single most useful thing to know: new or worsening hemolysis is usually a position problem, not a blood problem. An inlet drifted into the septum, a device too deep or too shallow, or repeated suction events all shear cells. So the first response to a jump in plasma free hemoglobin is not a transfusion. It is an echo. Other contributors are high P levels, low volume with recurrent suction, pump thrombus and aortic regurgitation causing recirculation. The consequences are pigment nephropathy, hyperkalemia and a falling hemoglobin that is not bleeding. Fix position, drop the P level if the hemodynamics allow, optimize volume, protect urine output, review anticoagulation, transfuse by your usual thresholds rather than reflexively, and if it is severe and uncorrectable the device comes out or is exchanged.

Anticoagulation

Two sources of heparin, and this is where people get confused. There is heparin in the purge fluid, delivered continuously and absorbed systemically, and there may be a separate systemic infusion. Account for both. Stopping a systemic drip does not stop this patient's anticoagulation, and raising the purge concentration to fix a purge pressure problem changes their systemic anticoagulation as a side effect.

Typical practice is fixed-concentration heparin in the purge plus systemic heparin to a unit-specific target, often lower than you would use for venous thromboembolism precisely because the purge contributes. In active bleeding many units run purge-only and accept the thrombotic risk. In heparin-induced thrombocytopenia a direct thrombin inhibitor is substituted and a non-heparin purge solution used per protocol. Follow your local heparin protocol, and never write "continue heparin" without saying which heparin and to what target.

The limb, and the bed

The leg is what will hurt this patient after the heart recovers, and the patient is often sedated and cannot tell you. Doppler both feet hourly and document it, because a monophasic signal where there was a biphasic one is a change. A distal perfusion catheter, a small sheath placed antegrade into the superficial femoral artery below the access site and perfused from the arterial sheath sidearm, is the answer to a threatened but salvageable limb, and it should go in before the foot is mottled rather than after. Watch for compartment syndrome and rhabdomyolysis after reperfusion.

For a femoral device the leg stays straight and the head of bed stays under 30 degrees. Hip flexion is the commonest cause of a sudden position alarm, because bending the groin displaces and kinks the catheter. Use a knee immobilizer if the patient is restless and treat delirium aggressively, because a confused patient who sits up can dislodge the device. Any turn, log roll or trip to CT needs someone whose only job is the device, and a look at the console before and after. An axillary 5.5 changes all of this: the patient can sit in a chair and do physical therapy, which matters enormously if the plan is transplant listing or a durable device.

Weaning and removal

Before you start, the patient should look like someone who no longer needs a pump: off or nearly off inotropes and pressors, normalized lactate, good urine output, an acceptable mixed venous saturation, no ongoing ischemia or malignant arrhythmia, and adequate right ventricular function. If they are on norepinephrine at 0.2 mcg/kg/min you are not weaning, you are gambling.

1. Come down two P levels at a time, with observation at each step. Many units use 15 to 30 minutes per step. Follow your protocol.
2. At each step record mean arterial pressure, pulse pressure, heart rate, PA pressures and wedge, cardiac index, mixed venous saturation, right atrial pressure, urine output, and any new arrhythmia or ischemic change.
3. Watch the pulse pressure widen as the native ventricle takes the work back. That is the visible marker of a ventricle ejecting again.
4. Echo at low support: an acceptable outflow tract velocity time integral, an aortic valve opening on every beat, an ejection fraction improved from the nadir, no new mitral regurgitation. On newer consoles watch the calculated end-diastolic pressure, because a steadily rising number at low support means the ventricle is not tolerating it.

Stop and go back up if the cardiac index falls below roughly 2.2, the wedge climbs, the mixed venous falls, the lactate turns up, urine output drops, or you have to restart pressors. A failed wean is information. It means the ventricle is not ready, and the conversation becomes duration of support, escalation, or evaluation for transplant or a durable device.

Do not park the device at a low P level overnight because nobody wanted to make a decision. Low settings risk retrograde flow and thrombus, and a device delivering half a liter a minute contributes risk without contributing support. Either the wean succeeded and it comes out, or it failed and the support goes back up.

At removal: hold or reverse anticoagulation per protocol, come down and off with removal happening promptly, and close with prolonged manual pressure or a pre-placed suture-mediated closure device depending on how access was obtained. Surgical devices come out surgically. Afterward watch the site and the distal pulses for hours, because bleeding and limb ischemia both appear after removal as well as during support.

Complications

• Limb ischemia. The complication that defines femoral access. Higher risk with larger sheaths, smaller vessels, women, diabetes, peripheral arterial disease and high-dose vasopressors.
• Hemolysis. Common, sometimes severe, and usually the first sign that the inlet is not where you think it is. Causes acute kidney injury through pigment nephropathy.
• Device migration, into the ventricle or back into the aorta. Presents as a changed placement signal, a flattened motor current, falling flow, new hemolysis, or all four.
• Aortic valve injury. The catheter sits across the leaflets continuously for days. New or worsening aortic regurgitation after explant is described, and is more likely the longer the device is in and the more it has been repositioned. Look for it on the post-removal echo rather than assuming it is absent.
• Access site bleeding, hematoma and pseudoaneurysm, and retroperitoneal hemorrhage, which is the one that kills. High access above the inguinal ligament plus full anticoagulation plus a large sheath is the combination. Unexplained tachycardia, hypotension, back or flank pain, or a falling hemoglobin with no visible source is a retroperitoneal bleed until CT says otherwise.
• Purge system failure. Disconnection, air, occlusion or pump thrombosis. Rising pressure suggests obstruction, falling suggests a leak. Either threatens the motor, and a seized pump in the aorta is a surgical problem.
• Stroke, from device thrombus, embolized ventricular thrombus or atheroembolism during a large-bore retrograde crossing. Examine the neurology every shift, including a sedation hold where safe, or you will find the stroke on day 5.
• Ventricular arrhythmia from catheter irritation, sometimes the first hint the device has migrated deeper.
• Right ventricular failure unmasked by left sided support, seen as suction alarms and falling flow with a rising CVP.
• Infection at the access site and in the bloodstream, and pulmonary artery injury with the right sided device, which is uncommon and catastrophic.

Troubleshooting

Escalation and ECPELLA

Sometimes the Impella is not enough or is the wrong tool alone. Refractory hypoxemia, biventricular failure, or an output a left sided pump cannot support moves the conversation to VA ECMO. The comparison page covers the choice between the three, and it belongs to a shock team rather than to whoever happens to be awake.

Peripheral VA ECMO drains venous blood and returns it oxygenated and under pressure, retrograde up the descending aorta. Excellent systemic flow and gas exchange, and one large problem: it raises left ventricular afterload. The ventricle now has to eject against a column of blood being driven back toward it. A weak ventricle cannot, so it does not empty, while bronchial and Thebesian return and any residual pulmonary flow keep filling it. That matters because:

• End-diastolic pressure and wall stress rise, and the subendocardium becomes ischemic in exactly the ventricle you are trying to rest.
• Left atrial and pulmonary venous pressure rise, giving florid pulmonary edema and sometimes frank pulmonary hemorrhage.
• Blood stagnates in the non-ejecting ventricle and the aortic root and thrombus forms, which is a stroke waiting for the moment the ventricle starts moving again.
• The myocardium never truly rests, so the recovery you are running ECMO to permit does not happen.

How you detect it: a pulse pressure narrowing toward flat, an aortic valve that stops opening on echo, a dilating ventricle with spontaneous echo contrast, a rising PA diastolic or wedge on the PA catheter, and a chest film that whites out. Look for these every day on every VA ECMO patient, because the treatment is time-sensitive.

The fix is to vent the left ventricle, and one of the most effective ways is to add an Impella to a patient already on VA ECMO. The combination is usually called ECPELLA. ECMO does the systemic flow and the gas exchange. The Impella pulls blood out of the left ventricle and delivers it forward, lowering end-diastolic pressure and volume, restoring pulsatility, opening the aortic valve, clearing the pulmonary edema and reducing thrombus risk. Observational data favor venting over not venting, and it is standard practice at many high-volume centers. Other venting strategies, chosen on local expertise, include a balloon pump, a percutaneous atrial septostomy, a surgical vent, or a transseptal or pulmonary artery drainage cannula added to the venous limb.

Running an ECPELLA is a two-device, two-console, two-anticoagulation problem and it multiplies every hazard on this page: two access sites, two sources of hemolysis, two things that can migrate, and a great deal of hardware in one leg. Your job on the round is to know which device is doing what, to check position and purge on the Impella and circuit and oxygenator on the ECMO, and to be the person who notices the pulse pressure narrowing before the chest film does.

Related: balloon pump for the device this is most often compared with, ECMO for the next step up, choosing between the three, and arterial lines and CVP for the traces the console is competing with.