At the Department of Microsystems Engineering in Freiburg, Germany, in a lab run by the biomedical engineer Thomas Stieglitz, a cuff the width of a shoelace gets wound around a nerve under a surgical microscope. The nerve belongs to a pig, and it is the cervical vagus, chosen because its diameter and internal architecture track closely with the human version. The cuff itself is not metal tubing. It is a thin sheet of polyimide, a flexible polymer familiar from flex circuits in consumer electronics, patterned with a dozen or more hair-fine platinum contacts. Once wrapped, it looks less like a surgical implant than like a tiny ribbon cable that has learned to hug a wire.
That ribbon cable is the point. For nearly three decades, implantable vagus nerve stimulators have used a design that has barely changed: a cylindrical silicone cuff with two or three electrode contacts, wrapped around the whole nerve, delivering one undifferentiated pulse to everything inside. LivaNova’s VNS Therapy System, first approved by the FDA for epilepsy in 1997 and for treatment-resistant depression in 2005, still uses a variant of that approach. It works by brute force. It also, by design, cannot distinguish between the thousands of individual nerve fibers running through the cuff, which is why patients on these devices commonly report hoarseness, coughing, or a dropping heart rate when the stimulation amplitude climbs high enough to be clinically useful.
The flexible, multi-contact cuff is an attempt to fix that at the hardware level, not the software level. Instead of asking whether a drug or a stimulation protocol can be made more selective, it asks whether the electrode itself can be redesigned to talk to only part of the nerve.
The Vagus Is a Bundle of Different Conversations, Not One Wire
The cervical vagus nerve in an adult human carries somewhere in the neighborhood of 100,000 individual nerve fibers, and the overwhelming majority of them are afferent, meaning they carry information from the heart, lungs, gut, and immune organs up to the brainstem rather than commands down. Efferent motor fibers controlling the larynx run through the same nerve, which is why turning up a non-selective cuff too far changes a patient’s voice before it does much else. All of this traffic is compressed into a nerve roughly two to three millimeters across.
Nicole Pelot and Warren Grill at Duke University have spent years building detailed anatomical models of the human, pig, and rat vagus using microCT imaging, mapping exactly how fibers are grouped into fascicles, the internal bundles within the nerve’s outer sheath. Their work, along with functional mapping research out of the Feinstein Institutes for Medical Research in Manhasset, New York, where Stavros Zanos has studied how the vagus’s internal organization relates to the organs it serves, established something important for electrode designers: fascicles are not randomly arranged. Fibers headed to the heart tend to cluster separately from fibers headed to the lungs or the gut, though the degree of that separation varies by species, by the level of the neck where you cut, and by individual anatomy.
The vagus is not one wire. It is a cable with dozens of circuits bundled inside.
That partial, imperfect organization is both the opportunity and the catch. If fascicles were perfectly and consistently segregated by target organ, selective stimulation would be a straightforward engineering problem: put a contact next to the cardiac fascicle, another next to the gastric one, and stimulate independently. Because the organization is messier than that, and shifts from one animal to the next, electrode designers have had to build hardware that can adapt to anatomy it cannot fully predict in advance.
Why a Rigid Cuff Cannot Do This Job
Standard VNS cuffs are built from silicone with embedded metal contacts, and they are mechanically stiff relative to the nerve they encircle. A stiff cuff tends to compress the nerve into a round cross section and sit loosely enough that contacts do not maintain consistent proximity to any one fascicle. Over months of a beating heart and a moving neck, that loose, rigid geometry drifts further.
Dominique Durand’s group at Case Western Reserve University addressed part of this problem with the Flat Interface Nerve Electrode, or FINE, which gently reshapes a round nerve into a flattened oval, pulling fascicles that normally sit deep in the nerve’s core closer to the surface where electrode contacts can reach them. It was a clever mechanical solution, and it has been tested on peripheral nerves beyond the vagus. But reshaping a nerve introduces its own mechanical stress, and a FINE electrode is still a relatively bulky, rigid scaffold.
The Freiburg approach, and similar thin-film work pursued by companies like CorTec, a Freiburg spinout, takes a different route. Instead of reshaping the nerve to fit the electrode, the electrode is made soft enough to fit the nerve. Polyimide and parylene, both flexible polymer substrates borrowed from the flexible-circuit industry, can be patterned with multiple independently addressable platinum or iridium oxide contacts and then wrapped into a spiral that conforms to the nerve’s actual, native shape. Because the material flexes with the tissue rather than against it, chronic studies have found less mechanical irritation and less of the scar tissue, or fibrosis, that gradually raises electrical impedance and degrades signal quality around stiffer implants.
A cylindrical cuff asks the whole nerve the same question. A multi-contact cuff asks each fascicle separately.
Current Steering, Borrowed From a Different Organ
Having twelve or sixteen contacts wrapped around a nerve only matters if there is a way to decide, electrically, which ones to use and how. Here the field has borrowed a technique that already works commercially in a different context. Spinal cord stimulators from companies like Boston Scientific and Abbott use “current steering,” splitting a single pulse across multiple contacts in programmable ratios to sculpt where the resulting electric field actually peaks, sparing some fiber populations while driving others harder. Cochlear implants do something structurally similar, directing current toward specific regions of the cochlea to approximate specific pitches.
Applied to a vagus nerve cuff, current steering means a clinician, or eventually an automated closed-loop controller, can fractionalize current across the ring of contacts to push the effective stimulation field toward one fascicle and away from its neighbors, without needing to know in advance exactly which physical contact sits closest to which target. In bench and acute animal testing, this has let researchers selectively recruit fibers associated with heart rate changes while leaving fibers associated with laryngeal muscle activation largely unaffected, and vice versa, by adjusting which contacts fire and in what proportion rather than by physically repositioning the implant.
The chronic studies are the harder test. Selectivity measured once, immediately after implantation, says relatively little about whether a device will still work as intended six months later, after the nerve has settled, scar tissue has formed, and the animal has been moving, eating, and breathing around the implant the entire time. Multi-month implantations in pigs and other large animal models, whose cervical vagus anatomy approximates the human nerve far better than a rat’s does, are where flexible multi-contact cuffs have had to prove that the selectivity holds up, that impedance stays within a usable range, and that the softer substrate does not shed contacts or crack under repeated flexing.
Selectivity measured once on a bench is a different claim than selectivity that survives six months inside a living animal.
Programs funded through the NIH’s SPARC initiative, Stimulating Peripheral Activity to Relieve Conditions, launched in 2014 specifically to fund this kind of anatomical mapping and electrode engineering, have pushed several academic groups toward exactly this chronic testing stage rather than stopping at proof of concept.
Where This Sits Against What Is Already Implanted in Patients
None of this flexible, fascicle-selective hardware is implanted in a human patient today. What is already in patients is the older, non-selective generation, and a newer but still non-fascicle-selective generation built around dosing rather than anatomy.
| Approach | Developer | Selectivity mechanism | Status |
|---|---|---|---|
| Bipolar helical cuff | LivaNova | None, whole-nerve activation | FDA approved, epilepsy (1997) and depression (2005) |
| Flat Interface Nerve Electrode (FINE) | Case Western Reserve University (Durand lab) | Mechanical reshaping to expose fascicles | Preclinical and early human testing, other peripheral nerves |
| Thin-film multi-contact spiral cuff | University of Freiburg (Stieglitz lab), CorTec | Flexible conformal contact plus current steering | Chronic large-animal studies |
| Miniaturized bipolar microregulator | SetPoint Medical | Low-dose, intermittent duty cycling rather than anatomical targeting | Clinical trials, rheumatoid arthritis and inflammatory disease |
SetPoint Medical’s device, which commercializes Kevin Tracey’s original inflammatory reflex findings from the Feinstein Institutes, sidesteps the selectivity problem a different way. Rather than targeting a specific fascicle, it uses very low current and short, infrequent stimulation bursts, enough to engage the anti-inflammatory pathway through the spleen without driving the nerve hard enough to trigger the off-target effects that come with higher amplitude. It is a dosing solution to a targeting problem. Galvani Bioelectronics, the bioelectronics joint venture that grew out of GlaxoSmithKline’s electroceuticals program, has pursued a related strategy of precise anatomical placement on smaller branch nerves rather than the main cervical trunk, which reduces the selectivity burden by choosing a nerve that is already narrower in scope.
The flexible multi-contact cuff is making a different bet: that the main cervical vagus trunk, with all its mixed traffic, is still the right place to intervene, provided the hardware is finally good enough to listen to, and talk to, one fascicle at a time.
The Distance Between a Pig Study and a Patient
Chronic animal data is not clinical data, and the gap between the two has swallowed plenty of promising peripheral nerve interfaces before. Human nerves scar differently than pig nerves. Human patients move, cough, and live for decades with an implant, not months. Manufacturing a flexible, sixteen-contact thin-film cuff at the reliability and cost structure required for an FDA submission is a separate, unglamorous engineering problem from getting one to work in a university lab. Connector technology, the part of the system that routes sixteen or more independent channels from a soft cuff to an implanted pulse generator without a wire breaking after the ten-thousandth flex cycle, has historically been a weaker link than the electrode itself.
Still, the direction of travel in the field is unambiguous. Vagus nerve stimulation has spent twenty-five years as a blunt instrument, effective for some patients and intolerable for others largely because the hardware could not tell one fiber from another. Flexible, multi-contact cuffs do not change the biology of the vagus nerve. They change what the electrode is physically capable of asking it. Whether that translates into a device a patient can live with for ten years, rather than a pig for ten months, is the question every lab working on this is now racing to answer.
Credit: Robina Weermeijer on Unsplash