Artificial heart

Artificial heart is the general term for a mechanical pump that takes over the circulatory work of a failing heart. The category divides into ventricular assist devices, which work alongside a heart left in place, and total artificial hearts, which replace the ventricles entirely after the native heart is excised. The first are a routine therapy implanted in thousands of patients a year; the second remain rare, used mostly to keep a patient alive until a donor organ appears. The engineering history of both is a record of trading one failure mode for another.

§How it works

An adult heart moves roughly five litres a minute against arterial pressure, adjusts that output over a fivefold range in response to demand, and does so for decades without maintenance in a fluid that clots on contact with foreign surfaces. Any replacement has to approximate all four properties.

Pulsatile designs copy the heart's method: a flexible chamber compressed by air or a pusher plate, with inflow and outflow valves. They generate physiological pulse pressure and can be volume-matched to the patient. They are also large, contain flexing membranes that fatigue, and require valves that wear.

Continuous-flow designs use a rotating impeller. They have one moving part, are far smaller, and have no flexing components to fail. Early axial-flow pumps rested the rotor on mechanical or hydrodynamic bearings; current designs suspend it magnetically, so nothing touches, wear is negligible, and the gaps through which blood passes can be widened to reduce shear damage. The cost is that the patient often has little or no pulse, which turns out to have physiological consequences of its own.

Control is the third problem. A native heart increases output automatically with venous return. A fixed-speed rotary pump does not, so devices either run at a set speed with limited adaptation or use pressure and flow sensors to autoregulate, an approach used in some total artificial hearts.

§Development history

1. 1957
   First animal implant
   Willem Kolff and Tetsuzo Akutsu implant a polyvinyl chloride total artificial heart in a dog at the Cleveland Clinic, which survives about ninety minutes.
2. 1969
   First clinical use
   Denton Cooley implants a device designed by Domingo Liotta in a patient at the Texas Heart Institute as a bridge to transplant, supporting him for roughly three days.
3. 1982
   Jarvik-7
   William DeVries implants the pneumatic Jarvik-7 in Barney Clark at the University of Utah as permanent therapy. Clark survives 112 days, tethered to a large pneumatic console, and the case triggers lasting debate about experimental surgery.
4. 2001
   REMATCH
   A randomised trial shows that a left ventricular assist device roughly halves one-year mortality compared with medical therapy in patients ineligible for transplant, converting assist devices from a bridge into a destination therapy.
5. 2004–2010
   The rotary turn
   Continuous-flow pumps displace pulsatile ones. A pneumatic total artificial heart is approved as a bridge to transplant, and remains the only one in sustained use for two decades.
6. 2024–2025
   Magnetically levitated total heart
   A titanium rotary total artificial heart with a single levitated rotor is implanted in its first patients, one of whom is supported for more than a hundred days before transplant.

§Ventricular assist devices

The success story is the left ventricular assist device. A pump the size of a small fruit is implanted at the apex of the left ventricle, draws blood from it, and returns it to the ascending aorta, unloading the failing ventricle while preserving whatever native function remains. Power and control run through a driveline that exits the abdominal wall to an external controller and batteries.

The randomised evidence dates to 2001, when a trial in patients ineligible for transplant found that an assist device substantially reduced one-year mortality compared with optimal medical management, at the cost of frequent adverse events.¹ Devices have improved considerably since. The current standard is a centrifugal pump with a fully magnetically levitated rotor and an intermittent speed modulation that produces an artificial pulse; in a large randomised comparison against the previous axial-flow generation, it eliminated pump thrombosis almost entirely and reduced stroke, and the survival advantage held at five years.² An earlier competing device was withdrawn from the market in 2021 after neurological events and pump-stop failures.

Assist devices are now implanted in patients who will never receive a transplant, and some remain supported for years. They do not restore normal physiology. Patients carry batteries, cannot swim, and take anticoagulants indefinitely.

§Total artificial hearts

Total replacement is a different problem, because both ventricles must be removed and both circulations driven, and because there is no residual native function to fall back on if the device stops.

The pneumatic total artificial heart approved in the United States in 2004 as a bridge to transplant is the descendant of the Jarvik-7 and has been implanted in well over a thousand patients; the pivotal study reported that most recipients survived to transplantation, against poor odds without it.³ It works, and it is bulky, noisy, and tethers the patient to a driver. A fully implantable pulsatile device tested in the early 2000s reached a handful of patients, one supported for well over a year, and was discontinued.

A French bioprosthetic design combines a hybrid membrane of bovine pericardium on the blood-contacting side with electrohydraulic actuation and pressure sensors that adjust output automatically, and received European marking; its manufacturer has struggled financially, and its long-term availability is uncertain as of 2026.

The most recent entrant abandons the anatomical metaphor entirely. It uses a single titanium rotor, magnetically levitated, with impeller surfaces on both faces so that one spinning part drives both the systemic and pulmonary circulations. There are no valves and no flexing membranes, and therefore nothing to fatigue. First implanted in a patient in 2024 as a bridge to transplant, it has since supported a patient for more than a hundred days, with early feasibility testing continuing. Whether a valveless rotary total heart is haemocompatible over years rather than months is the question the device exists to answer.

§The haemocompatibility problem

Every complication that limits mechanical circulatory support traces to blood meeting an engineered surface at non-physiological shear.

Thrombosis and stroke. Blood clots on foreign surfaces and where flow stagnates. Patients require anticoagulation, which shifts risk toward haemorrhage; stroke, both ischaemic and haemorrhagic, remains the most feared complication.

Acquired von Willebrand syndrome. High shear in a rotary pump unfolds and cleaves the large von Willebrand factor multimers that mediate platelet adhesion. Combined with reduced pulsatility, which promotes intestinal angiodysplasia, this produces gastrointestinal bleeding in a substantial fraction of patients — a bleeding disorder manufactured by the device in a patient who also needs anticoagulation.

Haemolysis. Shear damages red cells directly, releasing free haemoglobin, which scavenges nitric oxide and injures the kidney. The mechanism is the same one that defeated cell-free oxygen carriers described in Artificial blood.

Right heart failure. Unloading the left ventricle increases venous return to a right ventricle that is often also diseased, and right-sided failure after implantation is a leading cause of early death.

§Power, infection and the untethered goal

The percutaneous driveline is the single largest source of avoidable morbidity. It is a permanent breach of the skin, and infection along it is common with time. The problem is generic to devices that cross the body surface: bone-anchored limbs face the same skin-implant interface, discussed in Osseointegration, and it is the reason most Neuroprosthetics are fully implanted with an inductive link rather than wired through the skin. Transcutaneous energy transfer, in which power is coupled through intact skin by induction, has been demonstrated and has not become standard, because the coupling efficiency, heat generation and internal battery lifetime all impose their own costs. Implanted stimulators such as those used in Deep brain stimulation solved a version of this problem by drawing milliwatts; a circulatory pump draws several watts continuously, which is what makes wireless power hard. A fully implanted system also has to solve what happens when the internal battery fails.

Why mechanical support is not a bridge to nowhere

Roughly four and a half thousand heart transplants are performed annually in the United States against a far larger population with advanced heart failure, the imbalance described in Organ shortage. Devices are the only intervention whose supply is not capped by donation, which is why they persist despite complication rates that would be unacceptable in almost any other field. The alternative supply strategies are edited pig hearts, described in Xenotransplantation, and grown tissue, described in Lab-grown organs — neither of which has supported a human for as long as a pump has.

§Outlook

Two directions are being pursued. One is incremental: better surfaces, smarter speed control, fully implantable power, and reduced anticoagulation requirements, all aimed at making an assist device something a patient can live with for a decade rather than tolerate for a few years. The other is replacement of the pump concept altogether by biological means: a regenerated myocardium, an engineered cardiac patch built by Tissue engineering from cardiomyocytes differentiated from Induced pluripotent stem cells, a construct assembled by Organ bioprinting, or a heart rebuilt on the matrix of a donor organ as described in Decellularized scaffolds. On the scale set out in Technology readiness level, mechanical support sits near the top and every biological alternative near the bottom, which is unusual for a field where the biological option is usually the older one.

Cost is the quiet constraint. A durable assist device and its implantation run to hundreds of thousands of dollars, which restricts the therapy to health systems that can absorb it and makes it a standing example of the problems collected under Access and inequality.

The honest comparison is that mechanical support has a track record measured in decades and hundreds of thousands of patients, while every biological alternative has a track record measured in months and single figures. The question for the next decade is not whether pumps will be replaced but whether a total artificial heart can become a destination therapy rather than a waiting room, which requires a device that neither clots nor bleeds its recipient, and no design has yet demonstrated that over years.

§See also

• Artificial blood
• Xenotransplantation
• Organ shortage
• Lab-grown organs
• Organ bioprinting
• Neuroprosthetics
• Decellularized scaffolds
• Tissue engineering

§References