Flexible Microsystems Deliver Drugs Through the Ear

A MEMS-based microfluidic implant could open up many difficult-to-treat diseases to drug therapy

13 min read
human body illustration by Bryan Christie Design
Illustration: Bryan Christie Design

img Illustration: Bryan Christie Design

The bedrock of modern medicine is an age-old technology: the pill. For the most part, it works. The drug inside the pill finds its way to whatever part of the body it’s meant to treat, and the patient recovers.

But sometimes, oral medications just aren’t effective. The medication may interact with other drugs or food. It may break down in the stomach before entering the bloodstream. Or, even more frightening, it may trigger other, bigger problems even as it aims to cure. Recall the painkiller Vioxx, which did wonders for arthritis patients but also raised the risk of heart attack and stroke. Or Baycol, a cholesterol-lowering medication that was linked to dozens of reports of fatal kidney failure. In one way or another, traditional oral medications fail hundreds of thousands of patients each year, according to the U.S. Food and Drug Administration.

For years, researchers have searched for better ways to deliver drugs. The ideal method would administer the right dose to the exact area being treated, whether it’s an arthritic knee or a tumor in the lungs. From the patient’s perspective, it would be both convenient and unobtrusive—as easy as, if not easier than, taking a pill [see sidebar, “The Push-Pull Method”].

My collaborators and I are pursuing that better way. The implantable drug-delivery device we are developing at the Charles Stark Draper Laboratory, in Cambridge, Mass., and at the Massachusetts Eye and Ear Infirmary (MEEI), in Boston, merges aspects of microelectromechanical systems, or MEMS, with microfluidics, which enables the precise control of fluids on very small scales. Unlike rigid implants, such as pacemakers or titanium hips, our device is a flexible, fluid-filled machine: Stretchy tubes expand and contract, and fluids flow in and out of channels according to a preset rhythm. A tiny pump acts as the machine’s heartbeat, and software lets the device adapt to new demands from its environment.

Microscale machines are nothing new, of course. For the last four decades, engineers have built them for a variety of purposes, harnessing the mechanical properties of silicon and the same assembly techniques used to manufacture microchips. The resulting MEMS products now appear in all kinds of consumer electronics, such as air-bag sensors, inkjet printers, and the tiny accelerometers inside Nintendo Wii controllers.

Microsystems may have a much greater impact, however, in biomedical engineering. Engineers and clinicians have dreamed up a whole world of assistive devices that could enhance, sustain, and prolong human life. But these health-care innovators are stymied by one key constraint: the difficulty of making the devices small enough to sit comfortably and unobtrusively inside the body. Microscale systems built not from silicon but from flexible, stretchy polymers could be the answer. And to take that one step further, our implantable device may be the first therapeutic MEMS machine to include the active control of fluids.

The polymer MEMS device my colleagues and I are developing, with funding from the United States’ National Institutes of Health, will treat the most common form of hearing loss, which now affects more than 250 million people worldwide. By delivering tiny amounts of a liquid drug to a very delicate region of the ear, the implant will allow sensory cells to regrow, ultimately restoring the patient’s hearing.

Our system, which is still under development, consists of a programmable micropump powered by a small battery and controlled by an electronic circuit. It pulses precise quantities of a drug from a small reservoir into the inner ear. A flow sensor meters the delivery and sends out an alert if anything goes wrong. What we have so far is about the size of a D-cell battery, but we’re working to get it down to the volume of a single AA battery, which ought to be small enough to suit most patients. The device’s reservoir would hold enough medication for about one year. We’ve already tested the system on guinea pigs, and our results show that it can successfully deliver medication to the inner ear without damaging hearing.

ear gear

Illustration: Bryan Christie Design
Ear Gear: Once it has been fully miniaturized, our implantable drug-delivery device may end up resembling a “UFO” in shape. It will hold enough medication to last a patient about one year, after which the patient would undergo a minor procedure to refill the reservoir. The final device will be about the volume of one AA battery.
Click on image for larger view.

At our current pace of progress, we could see clinical use of such drug-delivery systems in the next five years. Eventually we intend to extend the uses of our system to treat a number of other diseases. Keeping track of medications can become impossible for the elderly and those with neurological or psychiatric conditions, and their doctors never know how closely these patients are following their medication schedules. A tiny, programmable electronic system that delivers the proper drug dose at exactly the right time would clearly benefit patients and caregivers and would also ease the burden of patient monitoring on medical personnel and hospitals. Indeed, this work could revolutionize medical care.

Do we really need to replace the simple act of swallowing a pill with complicated, invasive electronics? Yes—for many diseases and conditions, pills are either ineffective or unsafe. The human body has a number of defense mechanisms that will attempt to outsmart invading foreign substances like drugs. A drug, when swallowed, travels through the bloodstream to circulate widely. Doctors must often prescribe a higher dose than would otherwise be necessary if the drug could be delivered directly to the spot where it’s most helpful. Sometimes that higher dose is also more toxic.

Another challenge is that not all parts of the anatomy are in direct contact with the bloodstream. As a result, many conditions, including hearing loss and several neurodegenerative diseases, can’t be treated effectively with oral or injected medications, The blood-brain barrier, for instance, is a thin layer of endothelial cells lining the blood vessels in the central nervous system that prevents contact between the bloodstream and the brain. A similar blood-cochlea barrier protects the ear. A device implanted directly into a hard-to-reach target, however, could circumvent these blockades.

dose on demand

Photo: Charles Stark Draper Laboratory

Dose on demand: The 2007 version of the device [above] uses a traditional pump and is about the size of a D-cell battery. A smaller, push-pull system is under development.
Click on image for larger view.

We aren’t the first to explore implantable drug delivery. Nonelectronic implantable systems are now used to treat brain tumors that can’t be surgically removed. In such cases, the tumor is typically treated with a drug called BCNU, which is effective but also highly toxic. In 1987, Robert Langer, a chemical engineering professor now at MIT and a pioneer in the field of drug delivery, and his colleagues Henry Brem of Johns Hopkins and W. Mark Saltzman, now at Yale, came up with the idea of encapsulating the drug in a biodegradable polymer wafer and surgically implanting it in the brain. As the polymer slowly dissolves over a period of weeks to months, the BCNU is released and delivered directly to the tumor. Studies show that, on average, the treatment extends patients’ lives by several months at least, and they avoid the severe side effects—such as anemia, liver problems, and lung damage—that can occur when the drug is taken by injection. Similar passive systems have also been developed into drug-delivery patches for treating nicotine addiction and motion sickness and for birth control.

The dissolvable wafer was a landmark development, but not all ailments respond to its passive functions. In the cochlea, for example, injected medication and drugs in controlled-release gels would not travel far enough without the assistance of a pump. The wafer also has a limited life span; an electronic delivery device, by contrast, could operate for years and be programmed to deliver multiple drugs of varying dosages at different times. Several teams of engineers are now developing implantable drug-delivery systems to treat pain, neurological diseases, cancer, and many other conditions. One leading system is the “pharmacy on a chip” being developed by MicroChips, in Bedford, Mass., also founded by Langer and his colleagues. This drug-delivery microsystem has spurred the development of other implantable delivery devices, including our own microfluidic approach.

We’ve focused our work on hearing loss because it is so widespread and because existing treatments don’t offer a cure. In sensorineural hearing loss, the most prevalent form, the tiny, sensitive hair cells inside the organ of Corti, a structure within the cochlea, can be damaged or lost, which means they would no longer function normally as receptors of sound. The disease has several causes, including exposure to excessive noise or toxic chemicals, genetic factors, or simply aging. In addition, roughly three in a thousand children in the United States are born with significant hearing impairment, making it one of the most common disorders at birth.

The only therapeutic interventions available to these individuals are hearing aids and cochlear implants. Hearing aids amplify sound, but they don’t always make the sound clearer. Cochlear implants bypass the damaged cochlea altogether to electrically stimulate auditory nerve fibers. Neither device can halt or slow the loss of hearing.

Once thought to be irreversible, sensorineural hearing loss may soon be treatable with drugs. Clinical researchers and biologists are rapidly accumulating insight into the molecular signals involved in generating new hair cells. One approach involves inserting genetic material into the nuclei of hair cells, with the hope of restoring their function.

Delivering therapeutic compounds to the ear, though, is no small task. Any drug-delivery system for the ear will require extremely high-precision engineering to navigate the intricate internal geometry of the cochlea. This structure contains a coiled, snail-like tube with a membrane that’s covered with sensory elements and stretched across the middle of the tube. The membrane moves in response to sound, and hair cells sense the shearing forces of that motion. Fluid, called perilymph, bathes many of the functionally important elements of the cochlea, including the hair cells.

Despite the cochlea’s peculiar and complex environment, the system we’ve built is quite simple. Our approach is to drill a single inlet hole in the scala tympani, one of the perilymph-filled cavities of the cochlea, and insert a small tube. A highly efficient, electronically controlled pump about the size of a pencil eraser will push the drug out of its reservoir through a network of valves and channels. The drug will travel through the tube, and the pump will help mix the drug with the inner ear fluid. Sensors in the device will detect the fluid flow and transmit that information to an external receiver, which will monitor the performance of the implant.

Photo: Charles Stark Draper Laboratory

Fluid Circuit: A piston compresses a chamber [left] to pump a drug reservoir [right], in this 2.5- by 1.3-centimeter element.
Click on image for larger view

The device’s defining feature is its fluid circuit. Just about every introductory electrical engineering course uses the analogy of water flowing through a pipe to explain the basics of circuit design. We’ve taken the reverse approach, using electrical circuit elements to model the fluid-dynamic behavior of the drug-delivery system and the ear. Critical parts of our model include the resistance to flow both in the delivery device and in the cochlea, the compliance (or stretchiness) of the device and the cochlear membranes, and the driving force that the micropump must deliver to the flow.

If a tube or vessel has some elasticity, energy can be stored in the expansion and contraction of the vessel’s walls, giving it a storage function comparable to electrical capacitance. So we model compliance using a capacitor. Modeling the resistance to flow in a pipe or tube is very straightforward: We use an electrical resistor analogue. The flow rate of a fluid in a pipe is analogous to the flow of current. The driving force for fluid flow—the micropump—acts like a voltage source, or power supply. The micropump generates a pressure difference between the drug reservoir and the inner ear by working against the fluidic resistance and using the storage capability of the tube’s elasticity to drive the drug into the cochlea. In our model, the resistance and capacitance of the overall system includes values for inner ear structures, which are fixed, and values for our device, which can be tuned.

All of this is important because the cochlear structures in the inner ear are extremely sensitive to forces and flows. We need to provide just enough flow to deliver sufficient quantities of a drug far enough down the spiral tube. Too much flow could damage the ear, and too little would have no therapeutic effect. The resistances and capacitances within the device determine how much flow is generated by the pump and how quickly the drug is delivered to the cochlea. In other words, we can choose a particular diameter and length of tube to achieve a specific resistance and a particular elasticity to impart a specific capacitance. To tailor the timing and magnitude of doses, we can vary the intensity of the flow and the frequency of the pumping by programming a microcontroller in the device.

When we started building our device, we initially worked with silicone rubber materials that are prevalent in microfluidics, namely Silastic tubing and PDMS (for polydimethylsiloxane) components. These materials are transparent, inexpensive, and simple to fabricate, and they are generally suitable for contact with biological fluids and tissues. They also stretch nicely and therefore are excellent capacitor structures for the device. But we soon discovered a problem: The materials are permeable to air and liquid, and tiny air bubbles, barely visible to the eye, were forming inside the narrow tube that carries fluid into the cochlea. The tube is narrower than a human hair, so even a very small bubble can completely stop up the system. The air bubbles ended up acting as capacitors in our circuit, and they became an uncontrolled and unstable element in the design—sort of like having a frayed wire in intermittent contact in a circuit.

To avoid the bubbles, we switched to a less familiar material, polyimide, for most of our polymer components. It’s much less permeable to air and water, and we coated the remaining permeable materials with impermeable surface layers. Of course, there was a trade-off. Most microfluidics research is done with PDMS, and engineers typically mold and cast PDMS by micromachining a silicon wafer, pouring the liquid polymer into it, and peeling out a replica molding. But micromolding polyimide is impractical, so we now use laser machining to build our micropump.

With the new materials in place, we immediately faced another challenge. The total volume of perilymph in the human cochlea is only about one-fifth of a milliliter, and so very small changes in fluid volume or pressure can damage the hair cells and other delicate structures. To avoid causing any damage, we implemented a “push-pull” system in which the drug is infused into the perilymph and then an equivalent volume of fluid is withdrawn from the cochlea, thereby maintaining a constant volume. To give the drug a chance to mix with the perilymph, we use a low-resistance, low-capacitance component to release the drug and a high-resistance, high-capacitance element to extract the fluid. That translates into a relatively rapid drug infusion pulse and a much slower and gentler withdrawal pulse. The dosage can be programmed to be constant over time, but more complex dosing profiles and combinations of multiple drugs are also possible.

The push-pull system has an added advantage that has nothing to do with the cochlea’s twisted passages but everything to do with why implantable medical devices often fail. Anything that gets inserted into the body undergoes a process known as biofouling, which occurs when a foreign object triggers an inflammatory response from the body. Our concern was that fouling could potentially slow or stop the pumping action. But our system isn’t static (unlike most implants), and we believe that minimal fouling will occur, because the pushing and pulling would break up any cellular or protein build-up before it becomes permanent.

Our clinical collaborators at MEEI have tested a version of our drug-delivery system on guinea pigs. Many compounds with known effects on hearing exist, and we are using these formulas to evaluate our system.

So far, the results have been very encouraging: We were able to circulate a liquid drug well into the depths of the cochlea, and we found that the fluid manipulations didn’t harm the guinea pigs’ ears. Even more compelling were our experiments in which we deliberately modulated their hearing. We started by infusing about 1/1000th of a milliliter of a test compound into a guinea pig’s ear through a catheter that had been surgically implanted. The catheter’s inner diameter was 100 micrometers, or about the width of a human hair, and it was connected to a pump and control system worn by the guinea pig. We then withdrew roughly the same quantity of fluid, now mixed with the perilymph, and repeated the infuse-withdraw cycle for several hours.

One of the beautiful things about working with the auditory system is that the exact location in the cochlea at which a sound is perceived depends on the sound’s frequency. High-pitched sounds register near the entrance of the cochlea, while low-pitched sounds are perceived much deeper inside. (One octave equates to about 3 millimeters of cochlear tube.) To test how effective our device is at delivering a drug, we can fill the device’s reservoir with a drug that has a temporary dampening effect on hearing and begin pumping it into the subject’s ear. When we observe that the drug is deadening only high-pitched sounds, we know that the drug has reached the entrance to the cochlea. Once the subject stops responding to low-frequency tones, we know that the drug has approached the apex of the coiled tube. And that’s what happened—we literally watched the guinea pigs lose their hearing in a cascade of frequencies, first as the high-frequency hearing diminished, then as the lower-frequency hearing was lost.

We succeeded in demonstrating for the first time that the parameters of hearing can be modulated in a controlled manner using an engineered, preprogrammed delivery device. We also established a margin of safety within which drugs can be effectively delivered without damaging sensitive hearing structures. The downside of the device is that any drug supply will eventually be depleted, but we are optimistic that by the time the device is ready for market, we’ll be able to implant at least a year’s worth of a highly concentrated drug.

As is true with all implantable devices, one of the key challenges we still face is how to power the device. We envision one small battery that’s implanted and one slightly larger battery that’s either on the surface of the skin behind the ear or tucked just underneath the ear. This larger battery would be easily rechargeable and replaceable and could, in turn, wirelessly recharge the embedded battery.

There are many hurdles to overcome in the development of this or any other electronically controlled implantable drug-delivery system. These obstacles run the gamut from microfluidic challenges to surgical and biological considerations. Once these hurdles have been cleared, implantable drug-delivery devices ought to see a healthy future. With electronics taking the bulk of the work away from them, patients can look forward to healthier, simpler, and most of all, more enjoyable lives.

This article originally appeared in print as “Medicine By Micromachine.”

IEEE Spectrum has agreed to include the following statement at the request of the authors. This article was updated 9 November 2009.

The project described was supported by Award Number R56DC006848 from the National Institute on Deafness and Other Communication Disorders.  The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

About the Author

Jeffrey T. Borenstein’s first experience with the Massachusetts Eye and Ear Infirmary was as a patient in the hearing department. Three years later he started collaborating with MEEI on microfluidics-based drug delivery for the inner ear, a project he describes in “Medicine by Micromachine”. A solid-state physicist at the Charles Stark Draper Laboratory, in Cambridge, Mass., Borenstein says he “always had a fascination with medicine and wanted to find opportunities to get involved in human health.”

To Probe Further

The microfluidic technology described in this article is the product of a decade of research at Draper Laboratory, in Cambridge, Mass. Initially, the research was targeted toward lab-on-a-chip systems, a field that has experienced explosive growth and now has its own journal, Lab on a Chip, and an international annual conference, MicroTAS. In one early application, Draper’s microfluidic system was developed as a front end for a biosensor that could be used in medical applications, such as clinical diagnostics, or in biodefense. Mark Mescher and Jason Fiering have led the development of microfluidics at Draper, with contributions from Erin Swan, Sarah Tao, Ernest Kim, and Maria Holmboe.

The project on drug delivery for the inner ear  got its start in 2002, when Draper and the Massachusetts Eye and Ear Infirmary,  or MEEI, began a seed project aimed at developing new therapeutic approaches to sensorineural hearing loss. The MEEI is home to the Eaton-Peabody Laboratories (EPL), a 50-year-old institute devoted to hearing and deafness research. Three MEEI researchers—Sharon Kujawa, William Sewell, and Michael McKenna—have played key roles in developing the technology and finding ways to apply it to the treatment of inner-ear diseases. MEEI postdoctoral fellow Zhiqiang Chen conducted the first demonstration of the system in an animal model, and this work was published in the Journal of Controlled Release in 2005. More recent publications have described the device and its adaptation for an animal model.  A review article on drug delivery in the inner ear and an overview of micro- and nanoscale technologies for drug delivery are also informative. 

For a broader introduction to drug delivery, Mark Saltzman wrote an excellent textbook. To learn more about the early pioneering work of Robert Langer and his colleagues, see this article in the online version of Hopkins Medicine. MicroChips’ technology is described in several articles, including this one, published by MIT’s Technology Review.

The intracochlear drug delivery project is funded by the National Institute on Deafness and Other Communication Disorders (NIDCD), an institute of the National Institutes of Health, in Bethesda, Md.

The Conversation (0)

How Can We Get​ Blockchains to Talk to Each Other?

The field is fragmented, but common protocols are on the way

4 min read
An artists impression of two blockchain blocks trying to communicate with each other.

This article is part of our exclusive IEEE Journal Watch series in partnership with IEEE Xplore.

Keep Reading ↓ Show less

Clever Compression of Some Neural Nets Improves Performance

MIT researchers find an efficient way to prune speech-recognition AIs while still boosting accuracy

3 min read
Icon based illustration showing an AI language processing cycle.

As neural networks grow larger, they become more powerful, but also more power-hungry, gobbling electricity, time, and computer memory. Researchers have explored ways to lighten the load, especially for deployment on mobile devices. One compression method is called pruning—deleting the weakest links. New research proposes a novel way to prune speech-recognition models, making the pruning process more efficient while also rendering the compressed model more accurate.

The researchers addressed speech recognition for relatively uncommon languages. To learn speech recognition using only supervised learning, software requires a lot of existing audio-text pairings, which are in short supply for some languages. A popular method called self-supervised learning gets around the problem. In self-supervised learning, a model finds patterns in data without any labels—such as “dog” on a dog image. Artificial intelligence can then build on these patterns and learn more focused tasks using supervised learning on minimal data, a process called fine-tuning.

Keep Reading ↓ Show less

COMSOL News Special Edition Biomedical

How simulation and apps have enabled engineers and scientists to develop biomedical design

1 min read
COMSOL News Special Edition Biomedical features 12 stories of how simulation and apps have enabled design engineers, researchers, and scientists to develop biomedical designs.
Keep Reading ↓ Show less