A microaxial pump, not a balloon
Impella is a catheter-mounted microaxial pump that pulls blood from the ventricle and ejects it into the great vessel. For the LV devices (CP, 5.5) that means LV → aorta across the aortic valve. For RP it means IVC/RA → PA. It unloads the chamber it sits in and forwards stroke volume. It is not IABP counterpulsation and it is not VA-ECMO.
- Impella CP — 14-Fr femoral percutaneous. Peak flow about 3.5–4 L/min in a filled ventricle.
- Impella 5.5 — 19-Fr, usually a surgical axillary graft. Peak about 5.5–6.2 L/min. Same SmartAssist rules, hungrier inlet.
- Impella RP — femoral venous, inlet in the IVC/RA, outlet in the PA. Forwards a failed RV.
- ECMELLA — VA-ECMO plus an LV Impella. ECMO is the afterload; Impella is the unload.
Abiomed / J&J IFU language is consistent across platforms: confirm position, then titrate. The motor does not invent blood.
What SmartAssist actually shows
The Automated Impella Controller (AIC) with SmartAssist is the screen you will live on. Three traces matter: a red Ao placement signal, a white LV estimate, and a green motor current. Displayed flow is derived from current and P-level. SmartAssist can also trend CO and CPO; it still does not print a diagnosis. How to read the traces is the next section — it is the skill that separates a titration from a guess.
| Readout | What it is | How you use it |
|---|---|---|
| P-level | The support setting you choose (P1–P9) | Not RPM. Titrate after position is confirmed. |
| Flow (Q) | Estimated L/min | Falls with suction, afterload, malposition. Not a promise. |
| Placement signal | Ao (red) + LV (white) | Morphology tells you which side of the valve the sensor is on. |
| Motor current | Electrical load, green | Pulsatile = two chambers. Flat = same chamber or no native beat. |
| Purge pressure | Cassette line pressure | Roughly 300–700 mmHg. Climbing is a cassette problem. |
| Purge flow | mL/h of D5W ± heparin | Dribble plus high pressure = fix the purge. |
Reading the AIC waveforms
The Placement screen is two stacked windows. The lab console is drawn to match the real AIC. Learn the pictures here, then open a case and name the tracing before you touch P-level. Waveforms are how you decide whether the next click is support, volume, a pull-back, a pressor wean — or leaving the catheter alone.
What the screen is for
SmartAssist is not a diagnosis engine. It is three synchronized pictures of the same pump: where the optical sensor sits, what the estimated LV is doing, and how hard the motor is working to move blood from one chamber to the other. You use them, in that order, to answer four questions:
- Is the catheter across the aortic valve?
- Is the inlet seeing blood, or sucking wall / apparatus / vacuum?
- Is the motor loaded (afterload, clot) or unloaded (no work, no spin)?
- Is there a native beat left, or is the pump the entire circulation?
Displayed flow is derived from motor current and P-level. It will lie if position is wrong, if the inlet is empty, or if the motor is no longer moving blood. The traces are the ground truth. The number is a guess.
The three traces
- Red Ao (placement signal) — optical sensor at the outlet, where the cannula meets the outflow. When the device is seated this is an aortic pressure waveform: upstroke, dicrotic notch, diastolic decay. It is not your arterial line. Manage blood pressure off a real A-line. The red trace answers “which side of the valve is the sensor on?”
- White LV (SmartAssist estimate) — not a second transducer. It is derived from the Ao signal plus motor current. When seated it should look ventricular: high systolic peak, deep diastolic valley. That valley is the visual signature that the inlet is in the LV. A white trace that never dives is either the wrong chamber or an LV that is still loaded (token ECMELLA, no native beat, afterload clamp).
- Green motor current — energy the motor is using to move blood, in milliamps. When inlet and outlet sit in two different pressure chambers, current is pulsatile: it peaks when the aortic valve is open (smaller LV–Ao gradient, more flow through the cannula) and falls when the valve is closed. Flat current means both ends of the pump see the same chamber — or the ventricle is not contracting. Mean current rising with flow stuck is afterload. Mean current falling with flow collapsing is a motor that has stopped doing work.
How the traces line up on one beat
Read them as one cardiac cycle, not three independent graphs. On a seated device:
- Systole — white LV peaks first. Red Ao upstroke follows. Green motor current rises (easier pump, more flow).
- Early diastole — red notch, then decay. White falls toward its valley. Green current falls.
- Late diastole — this is where suction declares. A healthy white valley stays a few mmHg above zero and recovers before the next beat. A suction valley crashes below zero. Continuous suction never recovers.
If the three traces no longer share a rhythm, believe arrhythmia or a placement signal the AIC has already flagged as unreliable — not a new P-level.
Does the LV waveform ever cross the Ao waveform?
This is the native-ejection test. On a seated device, look at whether the white LV peak climbs through the red Ao line each beat.
- White crosses red — the LV generated a higher pressure than the aorta. The aortic valve opened. There is leftover native stroke volume on top of pump flow. Red keeps an aortic pulse and a notch; green motor current is clearly pulsatile.
- White never crosses red — the LV never generates enough pressure to open the aortic valve. Native cardiac output contributes little (often ~0) to overall cardiac output. Displayed Impella flow is the cardiac output. Red collapses to a near-flat line (pulse pressure of a few mmHg). White stays ventricular — it still dives in diastole — but its systolic peak remains below the Ao line. Green loses most of its pulse because there is no AV opening to cycle the load; mean current stays in the working band if the pump is actually moving blood.
This is not malposition. Red is still aortic (flat, but aortic). White is still ventricular. The inlet is doing its job — sometimes so well, at P-8 / P-9 on a 5.5, that the ventricle is fully unloaded and the valve never opens. Echo will show a closed AV and an inlet mid-cavity. Do not pull that catheter.
Split it from token ECMELLA: same “AV closed” physics, opposite LV. Token VA afterload leaves white diastolic high (the cavity is still loaded). A working Impella that has taken over the circulation leaves white diastolic low or slightly negative, with high displayed flow and a CO that almost equals pump flow.
The numbers next to the traces
| Box | What it prints | What you actually use |
|---|---|---|
| Ao 88/72 (76) | Sys / dias (mean), mmHg | Morphology first. Mean is not an A-line MAP. |
| LV 90/12 | Estimated LV sys / dias | Diastole is the suction tell. Negative = empty inlet. |
| MC 822/738 (762) | Max / min (mean), mA | Pulse of current = pulsatility. Mean rising = load. Mean falling = no work. |
| Flow 3.8 / 3.0 → 3.3 | Max / min → mean L/min | Max–min shrinks in suction. Mean stuck low = afterload or position. |
Motor-current pulse pressure (max − min) is the fastest “is it across the valve?” check. Seated CP at working P is typically tens to a couple of hundred mA of pulse. Same-chamber or no-native-beat tracings sit under ~20 mA. Afterload keeps the pulse and lifts the mean. Motor failure drops the mean.
One algorithm, every time
- Is motor current pulsatile? Yes → the pump is crossing a valve (or you have a beating ventricle). No → either both ends are on the same side of the AV, or there is no native contraction. You cannot tell those two apart from the green trace alone. Echo.
- Do Ao and LV look like different chambers? Red aortic + white ventricular = seated. Both ventricular = too deep (sensor in the LV). Both aortic = too shallow (sensor in the aorta).
- Does white ever cross red? If the LV peak stays below the Ao line, the aortic valve is not opening. Native output is negligible. Displayed pump flow is the cardiac output. Echo before you call malposition.
- Look at white diastole. A deep negative valley that recovers by end-diastole is diastolic suction (empty LV). A valley that never recovers is continuous suction (empty, obstructed, or against the wall).
- Look at green mean, not just shape. High mean + low displayed flow = afterload or a working motor against a brick wall. Low mean + low flow at the same P = the motor is no longer doing work (clot, seized purge, inlet occluded).
- Then look at the patient. Suction plus a high CVP and a low PAPi is the RV, not a dry tank. High MAP, high motor current, low displayed flow is afterload, not under-P. Equalized filling pressures and a swinging septum is tamponade, not “needs volume.”
Atlas at a glance
Same three traces. The diagnosis is the combination, not any one line.
| Pattern | Red Ao | White LV | Green MC | First move |
|---|---|---|---|---|
| Seated | Aortic, notch | Ventricular valley | Pulsatile, mid mean | Titrate P |
| Too deep (in LV) | Ventricular | Ventricular | Flat | P-2, pull 1 cm |
| Too shallow (in Ao) | Aortic | Aortic | Flat | P-2, advance 1 cm |
| Hypovolemia | Aortic | Diastole < 0, recovers | Pulses, diastolic chatter | Ease P, fill, find the leak |
| Continuous suction | Overlaying white | Stays negative | Flat + chatter | Drop P now, echo |
| RV failure | Looks like hypovolemia | Empty LV valley | Same as suction | Ease P, treat the RV |
| Tamponade | Low pulse | Suction-like | Low pulse | Drain. Do not fill blindly. |
| High afterload | Tall aortic | Valley present | High mean, pulse kept | Wean pressor / vasodilate |
| No native ejection | Flat aortic | Ventricular, never crosses red | Small ripple, mid mean | Echo. Do not pull. |
| VT | Fast spikes | Fast spikes | Irregular | Treat the rhythm |
| Motor failing | Damped | Damped, diastole high | Mean collapsed | Echo, prepare exchange |
| Token ECMELLA | PP ~4 mmHg | Diastolic high | Barely pulses | Titrate unload, not P-2 |
| RP seated | PA pulse | CVP lower | Pulsatile | Titrate RP |
| RP failing | PA pulse gone | CVP high / overlay | Flat | Fix the RV forward |
What the screen should look like
Each card is the Placement window (red Ao + white LV, or PA + CVP on RP) over the motor-current window (green). Same colors and sweep as the AIC in the lab. Name the tracing, then read why.
- Red aortic
- White ventricular
- Green pulsatile
- Both ventricular
- Green flat
- Both aortic
- Green flat
- Position still correct
- White diastole < 0
- Green chatters in diastole
- White stays negative
- Traces overlay
- Green flattened + chatter
- Position still correct
- White diastole < 0
- Green chatters in diastole
- Position still correct
- White diastole < 0
- Green chatters in diastole
- Position correct
- Green mean high
- Flow stuck low
- White never crosses red
- AV not opening
- Native CO ≈ 0
- Fast irregular spikes
- Signal not reliable
- Green mean collapsed
- Opposite of afterload
- Pulse pressure ~4 mmHg
- White diastolic high
- AV not opening
- PA over CVP
- Green pulsatile
- PA pulse collapsed
- Green flat
- CVP not falling
Looks like X, is actually Y
Several catastrophes share an Impella picture. The AIC names the inlet. You name the reason.
| AIC picture | Could be | What splits them |
|---|---|---|
| Diastolic suction | Hypovolemia, bleed, over-diuresis | CVP low, small LV, dry exam, falling Hb |
| Diastolic suction | RV failure | CVP high, PCWP low, PAPi < ~1, D-sign |
| Diastolic suction | Tamponade | Equalized filling, blunted y, swinging heart |
| Diastolic suction | High PEEP / abdominal pressure | Vent / abdomen. Drop PEEP or decompress. |
| White never crosses red | Pump-dependent, AV closed | High pump flow, white diastolic low, echo AV shut |
| White never crosses red | Token ECMELLA, LV still loaded | White diastolic high, pulse pressure ~4, VA running |
| Flat green current | In LV or in Ao | Both pressure traces look like the same chamber |
| Flat green current | No native ejection, seated | Red and white still different; white stays below red |
| Flat green current | RP not forwarding | You are on an RP. PA ≈ CVP. |
| Low flow, high MAP | Afterload | Green mean high, pulse kept, white valley present |
| Low flow, falling MAP | Motor / clot / purge death | Green mean collapsed, white diastolic rising |
| Placement not reliable | VT, CPR, extreme suction | Look at the patient, not the next P click |
What waveforms will not tell you
- They will not distinguish hypovolemia from RV failure or tamponade. You need CVP, PCWP, PAPi, and echo.
- They will not give you a true arterial pressure. Use the A-line. AIC “Ao” is an optical sensor at the outlet.
- They will not diagnose HIT, a cold foot, or a soaking groin. Look at the patient.
- A flat motor current is not a diagnosis. It is a fork: malposition versus no native beat versus a dead motor.
- Displayed flow is not cardiac output. Add leftover native stroke volume, and only after position is honest.
- They will not tell you the inlet is 3.5 cm below the AV. Echo or fluoro does that.
P-levels are not RPM
You do not set revolutions. You set a P-level from P-1 (wean / insertion) to P-9 (max). On a filled CP, P-2 is about 1.6 L/min and P-8 is about 3.3 L/min — afterload and preload then take their cut. 5.5 numbers are higher at the same P; RP numbers live in the CP range but on the right side.
Titrate P only after the placement signal is aortic and echo (or fluoro) has the inlet mid-cavity. Climbing P into suction, malposition, or a brick-wall SVR shears red cells and does not raise effective flow.
| P-level | CP (filled) | Typical use |
|---|---|---|
| P-1 / P-2 | ~1.1–1.7 L | Insertion, wean, brief imaging |
| P-3 – P-5 | ~1.9–2.7 L | Modest support, recovering LV |
| P-6 – P-8 | ~2.7–3.3 L | SCAI C/D working range on CP |
| P-9 | ~3.5–4.0 L | Ceiling. If still shocked, escalate the platform. |
Inlet mid-cavity, outlet above the valve
For CP / 5.5 the inlet cage belongs in the mid-LV, about 3.5 cm below the aortic valve. The pigtail sits in the LV. The outlet sits in the aorta, above the AV. Too shallow: outlet in the LV, no forward work, odd signal. Too deep: inlet against the mitral apparatus or the wall, suction and hemolysis.
SmartAssist first, then echo. A ventricular placement signal plus a flat-high motor current is malposition until imaging says otherwise. Reposition under echo or fluoro. Do not climb P on a device sitting in the wrong chamber.
Suction means the inlet is empty
The axial pump can only move what the cavity contains. SUCTION on the AIC is the Impella equivalent of ECMO chatter: the well is dry. Causes, in the order you should think them:
- Hypovolemia / over-diuresis / bleed
- RV failure — the LV never fills (watch PAPi and a D-sign)
- Malposition — inlet against wall or MV
- Tamponade or high PEEP / abdominal pressure
- VT / no diastolic fill time
Do not climb P into suction. Ease P-level, restore fill (or treat the RV / rhythm / position), then come back up. Repeated suction is how these devices hemolyze.
High SVR is a clamp
Axial pumps are afterload-sensitive. A P-8 CP into an SVR of 2100 may deliver 1.8 L with a reassuring MAP — because MAP is flow × resistance. The ventricle is not unloaded. Motor current runs high. Wean the leftover norepinephrine or vasodilate. Climbing to P-9 into a brick wall just shears cells.
Purge keeps blood out of the motor
A dextrose purge (D5W ± unfractionated heparin, or a bicarbonate / heparin-free recipe) runs through the motor at a few to a few dozen mL/h. Working purge pressure is typically a few hundred mmHg (about 300–700). The point is to keep blood from entering the motor gap.
A climbing purge pressure with a falling purge flow is a cassette, line, or seal problem. Exchange / prime the purge. It is not a P-level problem. Ignore it and the motor seizes.
Hemolysis is shear
Rising plasma-free Hb, tea-colored urine, falling Hb, drifting motor current. The number is a symptom. The cause is almost always suction, malposition, or high P into a dry or tight ventricle. Fix the inlet. Transfusion alone does not stop the shear.
SCAI shock, CPO, and PAPi
SCAI stages A–E are how adult cardiogenic shock is now spoken. This lab lives in C–E. Two numbers from the cath-lab literature belong on the glass:
CPO (W) = MAP × CO / 451
Fincke et al. (JACC 2004): CPO below about 0.6 W is the shock cutoff associated with mortality. In this lab, CO is approximated as pump flow plus leftover native output (~1.2 L when the LV is wrecked). CPO still under 0.6 W on max CP means escalate the platform.
PAPi = (PASP − PADP) / RA
Korabathina et al. (CCI 2012): PAPi below about 1.0 unmasks an RV that will not feed the Impella. Suction + high CVP + low PCWP + a D-sign is this pattern. Support the RV. More P-level on an empty LV is how you write a hemolysis note.
The groin, the foot, and the cassette heparin
- Limb ischemia — the 14-Fr CP sheath occupies the femoral artery. Cool foot, weak Doppler: antegrade / distal perfusion sheath now, not after the CK peaks. 5.5 trades groin ischemia for an axillary cutdown.
- Access bleed — the most common Impella complication. Hold purge heparin, pressure / figure-of-eight, transfuse. Do not explant a working pump because the dressing is red.
- HIT — cassette heparin is systemic enough to cause and perpetuate HIT. Day-5 thrombocytopenia and mottling: stop every heparin source, switch to bivalirudin (or argatroban), run a heparin-free purge.
Step the P-level down
Wean by stepping P, not by pulling at P-8. Watch MAP, lactate, CPO, and whether the LV dilates on echo. P-2 / P-1 is a wean setting. P-0 is off — do not linger (stasis, thrombus). Have a re-escalate plan. One good gas is not a graduation.
VA-ECMO loads the LV. Impella unloads it.
Femoral VA return is retrograde aortic flow. That afterload can close the aortic valve, smoke the LV, and thrombose a still cavity. An Impella across the AV unloads that LV and lets the valve open — if you actually titrate it. Danial and the ECMELLA series are consistent: token P-2 next to 4–5 L of VA is not an unload.
Titrate Impella P until PCWP falls and the AV opens. The inlet still needs volume. If you climb into suction, VA is over-draining or the RV is out. Wean leftover vasopressor; more afterload is the opposite of the job.
CP, 5.5, and RP are not interchangeable
| Device | Access | Job | Ceiling |
|---|---|---|---|
| CP | 14-Fr femoral | LV → aorta | ~3.5–4 L. Escalate if still shocked. |
| 5.5 | Axillary graft | LV → aorta, higher flow / longer support | ~5.5–6 L. Still needs fill. |
| RP | Femoral vein | IVC/RA → PA (RV forward) | CVP should fall; LV should fill. |
| ECMELLA | VA + LV Impella | ECMO for flow, Impella for unload | Titrate both. Watch the inlet. |
How to work a case in this lab
Same order every time — it is the order that keeps people alive at 03:00.
- Look at the patient: MAP, lactate, urine, the foot, the urine color, the groin.
- Look at the AIC: P-level vs flow, placement signal, motor current, purge pressure. Compare signal to what an aortic waveform should look like before you touch P.
- Name the problem in one sentence (settings too low, suction, malposition, afterload, purge, RV, HIT, token ECMELLA…).
- Do the matching move. Harmful clicks treat the wrong organ.
- If you neglect shock, suction, or a misplaced inlet, the patient will declare — pre-arrest, then a fail. There is no ROSC mini-game. Restore MAP and flow.
Guided mode forces three decisions. Free play is the same physiology without the multiple choice. The debrief is the attending in the doorway.