Brain implants: What’s standing in the way of pivotal trials, FDA approval

STAT News · Feb 26, 2026 · Collected from RSS

Full Article

Brain-computer interfaces don’t cure disease, so it’s tricky to define and measure therapeutic value O. Rose Broderick reports on the health policies and technologies that govern people with disabilities’ lives. Before coming to STAT, she worked at WNYC’s Radiolab and Scientific American, and her story debunking a bogus theory about transgender kids was nominated for a 2024 GLAAD Media Award. You can reach Rose on Signal at rosebroderick.11. Mike Willis just wants to go to the pub. He wants to stroll down the street, walk into a bar, and shoot the breeze with his friends. But the Cambridge, England, resident can’t do that anymore. Over the last six years, Willis, 71, has lost the ability to speak and socialize because of a neurodegenerative disorder. “If I go to a bar or restaurant to meet a friend, I take my wife to give them someone to talk to,” Willis told STAT in an email. “Otherwise I’d be sitting in near silence, incapable of making myself heard.” Scientists have developed devices that can help people with disabilities, like Willis, communicate. But in the United States they’re hitting an obstacle: the Food and Drug Administration. Thirty years after scientists demonstrated how brain implants can help rhesus monkeys move robotic limbs using only their mind, the clinical pathway for the technology is still out of focus. Startups have built devices that, when implanted, help people with ALS and paralysis move, speak, and expand their overall functioning. But to take these devices out of the lab and into people’s homes, the field must vault its most significant hurdle yet — convincing federal regulators that their devices are safe and effective. The first step, experts say, is demonstrating that the medical benefit outweighs the risks, especially as implanting most brain-computer interfaces require open brain surgery. And that’s the riddle the field has been wrestling with — how to design clinical trials that measure the benefit that regulators, startups, scientists, and patients can all agree would outweigh the considerable risks. “If you’ve been with patients who have used a BCI, you really sense how impactful it is,” said Mariska Vansteensel, a former president of the International BCI Society and a neuroscientist at University Medical Centre Utrecht in the Netherlands. “But to put this in numbers, it’s quite challenging.” To tackle that challenge, the industry and regulators joined forces in 2024 to form a “collaborative community.” Two years later, consensus is still elusive. Industry insiders say that conflicting messaging about non-medical uses of brain implants from Elon Musk’s Neuralink, a well-funded startup, hasn’t helped. Even Neuralink and Synchron, the companies closest to potential FDA approval, have yet to publicly announce the primary endpoints for their pivotal trials. The journey to the market for most medical devices starts with establishing relatively simple criteria with the FDA: therapeutic value. Want to alleviate stroke? Remove blood clots. Treat schizophrenia? Reduce hallucinations. Then conduct clinical trials to show a device does that better than alternatives on the market, or more safely. That’s easier said than done for brain-computer interfaces, which don’t cure or change the underlying medical condition. “How do we verify the device is doing the thing that the patient wants it to do? How do we ensure that whatever is being said by the machine is true?” said Doug Weber, a professor of mechanical engineering and neuroscience at Carnegie Mellon whose lab is currently working on developing neural interfaces for controlling prosthetic limbs. “If you look at medical device history, particularly neurotechnology, there are many examples of so-called breakthrough technologies … that ultimately failed.” For a field deeply rooted in its engineering legacy, reorienting around patients’ wants and needs is almost like learning a new language. If the field can solve this technological and regulatory riddle, the reward could be immense. The FDA’s decision with the early startups will pave the path for how other brain implants could improve the lives of millions of patients with paralysis and other disabilities. Prioritizing patient-centered outcomes Sporting glasses and a jaunty, multi-colored tie, Leigh Hochberg stepped behind a lectern to deliver an unequivocal message to federal regulators and the implanted brain-computer interface field gathered for a workshop in September 2024. Without robust clinical outcome assessments, Hochberg warned, “we can’t really have a meaningful pivotal clinical trial. Those companies need to exist or our field stops.” A neurologist at Massachusetts General Hospital, Hochberg is as close to a godfather as the field has, having been a key member of the early 2000s trials that demonstrated that brain implants could work in humans and not just monkeys. So when Hochberg talks, people listen. And on that September morning in the FDA’s cavernous “Great Room,” he brandished the field’s burgeoning optimism: “We’re all going to figure out exactly what those outcome assessments should be, hopefully in the next few days.” Hochberg might have been too optimistic. Seventeen months later, the field has stuffed a spreadsheet with more than 100 possible metrics to measure in clinical trials, based on input from federal regulators, industry leaders, clinicians, academics, and patients. The interest from FDA officials is real, too, as the agency has given all of the top startups breakthrough device designations or like Paradromics, accepted them into the Total Product Life Cycle Advisory Program (TAP). And yet, startups are still stumped on which benchmarks to choose. Part of the delay stems from the field struggling to shed its engineering origins, said Cristin Welle, who helped evaluate the viability of brain implants at the FDA in the early 2010s. The researchers Welle interacted with were obsessed with a device’s technical performance, hawking bit rates and electrode counts as proof of the technology’s promise while mailing in “thousand-page thick” applications that were “horribly naive and maybe even dangerous.” Agency officials balked at green-lighting trials. “There was some concern about, well — even if these things work, is there really going to be any benefit to anyone? Is this even worth doing at all?” said Welle, now a neurosurgery professor at the University of Colorado Anschutz. After years of robust communication and multiple workshops to align priorities, industry leaders say they are in lockstep with FDA officials: Patient outcomes, not technological specifications, should dictate clinical trial benchmarks. “Until you bring [the device] into the clinic and show that those technical superlatives translate to clinical outcomes, it’s still — you know, people are skeptical,” said Matt Angle, CEO of Paradromics, which is trying to decode people’s speech from brain activity. Stuttgart, Germany – 01-23-2025: Person holding smartphone with webpage of US neurotechnlogy company Neuralink Corp. on screen in front of logo. Focus on center of phone display.Adobe The technological problems of BCIs Why, then, is the field still struggling to pick the right metrics to measure? It’s complicated. If you can prove a medical device’s safety and efficacy, you’ve got a good shot at FDA approval. Early human trials from the top brain implant startups have met initial safety standards, showing few adverse outcomes. Now, to prove efficacy, they must demonstrate their device improves how a patient feels, functions, or survives. Experts see promise in assessing function, in particular. Most medical device makers can repurpose existing outcome measures used in clinical trials for similar devices. A continuous glucose monitor can measure endpoints like “time in range” that ensure a user’s blood sugar levels stay within a specific band. But the newness of the brain-computer interface field means that startups don’t have the luxury of previously approved devices. Implanted brain-computer interfaces are also a novel technology, acting on the brain in ways that the FDA has not regulated before. Stents prop up weak blood vessels. Deep-brain stimulators reduce seizures. But brain-computer interfaces don’t inherently restore function — they enable patients to use external devices to perform an action. Like a wheelchair for the mind, these devices parrot sight, speech, movement, and more using a computer and hundreds of tiny electrodes implanted in the brain. Each burst of neuronal activity is a chance to reopen a person’s world. For patients with spinal cord injuries and vision loss, this parroting is not necessarily an impediment to FDA approval. Blackrock Neurotech and Neuralink — whose devices help users manipulate robotic limbs with just their mind — might be able to quantify the therapeutic benefit using existing metrics such as the Action Research Arm Test, which assesses hand and arm function by measuring grip, pinch, grasp, and overall movement. Science Corporation’s device — which boosted visual acuity and reading ability in dozens of users with age-related macular degeneration — is primed to enter the European market this year after the device’s performance convinced regulators that the benefit outweighs the risk of surgery. Founder and CEO Max Hodak says the company is talking with FDA officials about potentially using data from their European trial to secure approval in the U.S. as well. Measuring benefit is less straightforward in the case of patients with ALS and other neurodegenerative diseases who want to communicate with friends and family. Synchron, Blackrock, Neuralink, Paradromics, and Precision Neuroscience have all conducted trials or made public overtures about restoring a person’s speech. Existing devices like eye-gaze trackers offer tediously slow conversations that average 10 to 20 words per minute. Synchron’s early trials only permitted 14 characters typed per minute