An MD&DI September 1997 Column
Complete pump systems may someday replace damaged or defective organs.
Without much fanfare, miniature implantable pumps are ushering in a new era of disease management. The latest developments have targeted two specific areas - cardiac support and drug delivery. Devices for both applications are undergoing clinical trials domestically and in some cases are already available in Europe.
Currently available only in Europe, this patient-controlled device eases pain in cancer patients using smaller dosages than needed orally for the same level of relief. Photo courtesy of Medtronic (Minneapolis)
One of the more attractive applications for implantable pumps is insulin delivery. According to the American Diabetes Association, diabetes mellitus affects approximately 16 million people in the United States. About 10% of these have type 1 diabetes, which can only be controlled by injections of insulin. This group of patients will benefit most directly from the new technology.
In the cardiac field, miniature pumps are saving lives by replacing and restoring heart function. Congestive heart failure afflicts 3 million to 4 million people in the United States and is responsible for 30,000 to 40,000 deaths annually. Drug therapy is effective in many cases but cannot address the needs of all patients. Nearly 300 new patients require heart transplants each year, but with a typical wait of more than 300 days for a donor organ, many never make it to the operating table. Most of these patients can benefit from left ventricular assist devices (LVADs), which take over the job of pumping blood for the heart.
In both cases, the underlying technology is not necessarily new, but optimization and commercialization have been slow. One of the biggest obstacles to an implantable insulin pump, for example, was not the pump but the insulin. Designers realized early that an implantable pump would have to retain a stock of insulin in a reservoir for extended periods; however, insulin must be stored at cool temperatures and becomes unstable in the warm environment of the body. Complicating matters was the need for a concentrated formula to minimize the need for refilling. It was not until Hoechst AG synthesized a new, more stable insulin compound that an implantable insulin pump seemed possible, and even then, device safety came under intense scrutiny. In fact, Eric Kentor, senior vice president and general counsel at Minimed Technologies (Sylmar, CA), says that when the company developed its implantable insulin pump, FDA demanded a joint submission of both the device and the drug that goes with it.
Minimed's pump is in the final stages of clinical trials, where it competes with devices manufactured by Strato/Infusaid (Norwood, MA) and Siemens-Elema (Solna, Sweden). Some of these devices are already available in Europe. Christopher Saudek, MD, of Johns Hopkins University (Baltimore), worked with Minimed to develop its MIP 2001 pump. He notes that the CE-marked device has more implantation in France than in any other country, primarily because of the French purchasing system. "National health insurance allows them to progress more quickly with trials," Saudek says. "The federal government approved purchase of these pumps in certain hospitals, and that helped."
In the United States, he says, manufacturers can't afford to give the pumps away for free. Minimed's device, for example, costs between $10,000 and $12,000, not including the cost of surgery and insulin refills. "It's going to be a lot more expensive than insulin injections," Saudek admits. "It's a very sophisticated device. It's got all the electronics and casings of a pacemaker, plus a complete fluid-delivery system that pacemakers don't have."
An implantable insulin delivery system by Minimed Technologies (Sylmar, CA) includes a refillable reservoir and a pump. A pencil is displayed for scale.
The telemetry systems that are used to program the devices are similar to those used to program pacemakers. As for the actual pumping mechanism, the three major versions differ slightly. Minimed's MIP 2001 and Siemens' Promedos ID 3 use electrically driven piston pumps: displacement of the piston draws the insulin from a reservoir into a piston chamber; when the piston returns to its original position, if forces the insulin through a free-floating catheter, which is usually inserted into the patient's peritoneal cavity. The Infusaid pump uses freon gas to produce positive pressure, which pushes insulin from a reservoir into a valve-type accumulator and into the catheter. The reservoirs in all three models have to be refilled every three months. A hypodermic needle, inserted directly through the patient's skin into the pump's reservoir, removes any unused insulin and replaces it with a fresh supply.
MONITORING INSULIN DELIVERY
Of course, patients will still have to draw blood frequently to monitor their glucose levels. A glucose monitor that can function without finger sticks--or one that can be permanently implanted--has not yet been marketed, despite significant work in the field. According to Saudek, reliability and repeatability remain elusive. "They're very hard technically to establish. To create a sensor that can continuously measure glucose in the blood . . . the technology is very difficult. That's the real point. It's not for lack of trying or not wanting to do it." Minimed's ultimate goal, says Kentor, is to marry an implantable pump with a glucose sensor in a fully closed-loop system, but the technology is still a long way off. Interim fixes include acute or short-term glucose sensors, currently in development, that would involve a soft-tipped cannula that would need to be changed every three-and-a-half days. Minimed hopes to introduce a cabled system that would be worn by high-risk patients for about three days, after which a physician could download recorded information for analysis.
A completely closed-loop system is also the goal of Michael Huff, an electrical engineer and associate professor at Case Western Reserve University, Cleveland. Huff has used his background in microelectromechanical systems (MEMS) and wafer-bonding technologies to create a microminiature pump about the size of a contact lens. The process entails taking a shape-memory nickel-titanium alloy and depositing it as a thin film on a silicon substrate, which is released to create an actuation site. In a typical valve design, the unit would be connected to a plunger mechanism. "We can cause the silicon to deform by straining it," Huff says, by heating the shape-memory material. "Once we heat it, it will regain its shape and become a diaphragm again." Another valve design involves a plunger pushed down over a sealing orifice; when heated, the film reverts to its original shape and compresses a syringe. "We use an integrated spring that can be made out of a polyamide," Huff says, "so it's a complete batch-fabrication process. We can do that same type of thing with a pump using that diaphragm and check valves on either side."
Huff has achieved even better performance with his latest design. "The one problem," he explains, "was with the spring--we had a linear load deflection. What we'd like is a spring that becomes more squishy" as it's compressed. A second recessed nickel-titanium spring solved the problem. By alternating the heating of each spring, top and bottom, Huff says, "we can achieve 100 to 500 times the pumping capacity" of other designs.
Huff has attained two important goals in his designs. By generating a lot of displacement, he says, he's reduced the flow resistance by three orders of magnitude. The high strain forces also enable control of fluids at high pressure. Both factors contribute to a high stroke capability, which is important in terms of power consumption. Minimizing power input per cycle, Huff says, translates to a reduction in power consumption overall. The high force levels involved also generate a large amount of pressure head, enabling the device to pump far more insulin than would actually be required.
Power consumption is proportional to the temperature differential, which can be readily tailored by scaling the thermal time constant. Devices can consume as little as 100 mW while maintaining a fast thermal response time. "Right now," says Huff, "we're operating at a tenth of a hertz." Accuracy, he says, will hardly be an issue. Units integrated with uncalibrated flow sensors operate at about ±1%, and units with calibrated sensors would get down to ±0.1%. The important thing, he says, would be the closed-loop infusion system--currently unheard of.
The technology is compatible with delicate insulin on several levels. For example, the phase-transformation temperature, which initiates the actuation, can easily be adjusted to 40°C, and a polyamide membrane separates the heated membrane from fluid. Huff adds that more testing is needed for concentrated insulin, which, he's been told, has a tendency to crystallize. Commercialization is not necessarily too far off, he says. "We've demonstrated that we can make these devices, and they're something that can be scaled up to large volume at reasonably low cost. We've demonstrated we can control the fatigue of the shape-memory material." As for future developments, Huff expects even greater miniaturization and intends to create a device "the size of a pinhead."
Huff's primary focus is on insulin delivery, but manufacturers of both insulin pumps and other drug pumps have expressed an interest in his work. The market for implantable pain-therapy systems, for example, would certainly benefit from the technology. Medtronic (Minneapolis) estimates that in the United States alone, as many as 70 million people suffer from chronic pain, and an additional 9 million suffer from cancer-related pain; most are effectively treated with oral medication, but some 510% are not. These patients, "many of whom were told that nothing more could be done for them," constitute the primary market for pain-therapy devices, says Jessica Stoltenberg, public relations manager for Medtronic.
Medtronic is actively pursuing this market and already has several drug pumps in its line, including the SynchroMed, an externally programmable implantable device, and the AlgoMed, a patient-activated device currently available only in Europe. The SynchroMed has been approved for the administration of morphine sulfate to treat chronic pain, baclofen to treat severe spasticity of both spinal and cerebral origin, clindamycin to treat osteomyelitis, and several drugs used in chemotherapy. The device sells for just over $7000, and the surgical implantation can run another $20,000. The AlgoMed, designed to treat intractable pain in cancer patients, includes a drug reservoir implanted just under the skin of the abdomen, a control pad implanted over the rib cage or breast bone, and a small catheter that delivers medicine to the spinal cord. The patient or caregiver activates the device by depressing two buttons on the control pad. A single activation delivers a maximum bolus of 1 ml, and a lockout mechanism prevents overdosing.
Studies indicate that patients require as little as 1% of the equivalent oral dosage to experience the same level of pain relief. Moreover, according to the company, many of the genetically engineered drugs currently under development would decompose too easily in the digestive tract. Stoltenberg predicts that the fusion of devices and drugs for site-specific drug delivery will become increasingly common. The next major hurdle, she says, will be overcoming the "brain-blood barrier," which could allow effective treatment of Alzheimer's, Lou Gehrig's disease, and similar conditions.
Medtronic is also active in the cardiac market, and recently released its Hemopump cardiac-assist system in Germany. The device, as described by Dick Reid, director of media relations for Medtronic, is a screw-type pump "about the size of a slightly longer ballpoint pen." The tubular pump uses an internal impeller to push the blood so that the heart can be slowed during minimally invasive coronary bypass surgery. Medtronic originally acquired the pump from a division of Johnson & Johnson, says Reid. "They had brought it to the point where it became quite obvious that it was going to require more clinicals, more design--and that's what we did. We redesigned it, worked with some physicians in upgrading various aspects, and then started the clinical and data-collection process."
Absolute reliability was naturally a primary goal in the refinement process. "The high-speed motor and control console are all outside the body," Reid explains, but the catheter-mounted pump is implanted directly in the ascending aorta, "which doesn't leave much room for error." Moreover, the axial-flow pump "required some rather sophisticated design work" to get the right combination of impeller type and speed.
Medtronic's device is intended as a surgical assist and not as a bridge-to-transplant. Baxter Novacor (Oakland, CA), on the other hand, has developed an LVAD to serve not only as a bridge-to-transplant, but as a permanent alternative to transplant. The device is currently available in Europe but is still in clinical trials in the United States. What separates the Novacor device from competitors, says Linda Strauss, Baxter Novacor vice president, is that it was "the only one designed from the beginning with the intention of providing an alternative [to transplant], which is why our system is the only one that has been electronic since its inception." Other systems, she says, are pneumatic, and although they are technically simpler and certainly satisfactory for the short term, their design limits their long-term feasibility. The Novacor system was designed as an implantable electrical system, with the blood pump and energy converter constructed as an integral unit. The system uses a pusher-plate pump implanted within the abdominal wall and connected by a conduit to the ascending aorta. The controller is external and fits in a pack like a camera bag that can be worn over the shoulder. "There still is a percutaneous lead," Strauss says, noting that currently "there are no systems that are completely implanted, with the power source implanted as well."
A fully internalized LVAD is still a long-term goal, but the need may not be as pressing as some might imagine. "Twenty-five years ago," Strauss says, "it was thought that if you had a percutaneous lead, it would be fine for a short time, but it would pose a limiting factor" because of infection. New developments have proved otherwise. Three recipients have had the Novacor device longer than three years, and a number have had it for more than two. "Infection still remains an issue," Strauss says, "but it turns out not to be as limiting as quickly as everyone thought."
Strauss lists three key areas requiring further development before widespread commercialization of long-term LVADs: device optimization, patient selection, and clinical management. Optimization is concerned ultimately with device reliability. "After all," she says, "if the device isn't reliable, it doesn't matter whether it's designed for long-term use." As for patient selection, experience should help physicians determine not only which patients to approve but when to begin using the device. Clinical management involves more immediate concerns such as prediction and prevention of thromboembolic events. Heart disease patients are by nature more prone to thromboembolic events than the general population, and some researchers have also suggested a link between thrombosis and infection in heart patients. "Certainly," Strauss adds, "when you implant anything into the blood stream, you also have a risk of embolic events." The challenge, she says, is in identifying all of the contributing factors and finding ways to deal with them, perhaps through anticoagulation therapies or through the design of the device itself.
Device redesign is certainly an option being investigated by engineers such as Steve Winowich and Jim Antaki at the University of Pittsburgh Medical Center (UPMC). Winowich and Antaki are looking to replace the current generation of pulsatile-flow pumps with axial pumps, which operate on a turbine principle--much like Medtronic's Hemopump. As Winowich explains, the first heart-pump designers understandably tried to mimic the heart's natural rhythm; early pumps therefore fit into the chest cavity and employed a positive-displacement action. Recently, however, scientists have begun to question whether the body really needs or desires a pulsatile flow--particularly as more information about continuous-flow pumps comes to light. "It was discovered that these high-speed impellers can produce the pressure and flow without completely destroying the blood cells," says Antaki. "That was a surprising discovery."
One of the principal advantages of axial pumps is their small size. Prototypes are about the size of a D-cell battery, but even smaller units should be possible. Smaller pumps would suit a greater variety of patients--not just large individuals. Moreover, the turbo pumps do not involve check valves or use flexing diaphragms that could suffer fatigue. Because they're more efficient, they require less power and can use smaller batteries. Antaki also believes they can be made more economically than pulsatile pumps. Current LVADs typically cost between $50,000 and $80,000; the new pumps could conceivably cost half or even a quarter as much.
UPMC's design team, relying heavily on computer-based simulation and optimization, remains focused on three interrelated factors: blood, material, and flow. "What we're doing," says Antaki, "is trying to design a flow path that's devoid of high shear stress--which would damage red cells and activate platelets--and eliminate stagnation and separation, which could cause blood to clot."
The same technologies used to extend performance life might also help to minimize blood damage. For example, the turbo pumps use magnetic levitation to support the internal impeller. The magnetic bearing would involve no contact and no friction and would obviate the shear forces that could cause hemolysis. Conventional pumps, on the other hand, are lubricated by blood; a rotating element is suspended by a ball and socket, and the space between is occupied by whatever blood seeps in. The design produces a combination of heat and shear force that can be highly damaging to human blood. As for biocompatibility, Antaki says, the new pumps use "tried and true biocompatible materials" such as titanium alloy, diamond-like coatings, and other types of ceramic coatings.
It may be some time before the UPMC device is ready for commercialization. In the meantime, new LVADs will be released from other sources. Robert Jarvik (of Jarvik-7 fame) is reportedly going to attempt some clinical trials in Europe, says Antaki, and Nimbus Medical (Rancho Cordova, CA), with which Antaki's team has been working, also has a device to debut. Both devices rely on blood-lubricated bearings. The Cleveland Clinic is also developing a continuous-flow centrifugal pump, and a few scattered universities have similar projects in the works. In a couple of years, they'll start seeing clinical trials.
The implantable pumps currently in clinical trials have sufficiently addressed --but not overcome--a number of critical design and performance obstacles. Accuracy has reached remarkable levels, but there's always room for improvement. Further refinements in biomaterials will help advance the technology, and greater miniaturization will increase the potential market. Device manufacturers need to coordinate their efforts with researchers in the fields of pharmacology and sensor technology to ensure that future developments contribute to the goal of a complete closed-loop implantable pump system.