By Deborah Borfitz

April 16, 2024 | Study sponsors making choices about if, when, and how to include brain monitoring in their clinical studies do not often ground those choices in science or optimize for subject safety and trial de-risking, according to clinical neurophysiologist Dona Murphey, M.D., Ph.D., chief scientific officer for Lifelines Neuro. The key problem is a lack of electroencephalogram (EEG) expertise with no guidance from the Food and Drug Administration (FDA) to mitigate the situation. 

In the absence of standardized guidelines from the agency, companies too often get the timing, duration, and spatial extent of EEG data collection wrong—and, potentially, inappropriately forego the expense altogether, she says. What they should be doing is, “incorporating brain monitoring into any trial where there’s either a known preclinical risk or a strong theoretical suspicion that an agent could cause any kind of damage to the brain... measurable by EEG.” 

EEG is especially good at capturing “epileptiform discharges,” which demonstrate risk for seizure, and seizures, which can sometimes be subtle and are often short-lived, says Murphey. The list of potentially seizure-causing medications includes anti-cancer drugs, antimicrobials, hypoglycemic agents, immunosuppressants, pulmonary drugs, stimulants, and sympathomimetics. 

The worst-case scenario is a dangerous condition known as “convulsive status epilepticus,” which would be clinically evident and not necessarily require an EEG, she says. But also concerning would be nonconvulsive seizures where patients may blank out mentally, perhaps with elusive involuntary movements. Without an EEG, it’s easy to miss.  

Brain monitoring didn’t become commonplace in critical care settings until the 1990s after EEG data were digitized, allowing waveforms to be presented in a more accessible format, says Murphey. Even then, it took quite a while to overcome technical hurdles in interpreting the deluge of information, especially when video data were being streamed alongside brain electrical behavior. 

It soon became apparent that between 15% and 20% of patients in hospital intensive care units (ICUs) were having non-convulsive seizures that were not clinically apparent without an EEG. They’re in an altered state of consciousness and seem confused or are perhaps comatose—not in a position to self-report their condition. 

Expertise Shortage

Unlike EEG monitoring, cardiac monitoring with an electrocardiogram (ECG) test is routine in both critical care and clinical research, notes Murphey. For decades now, ECG technicians have been monitoring ICU patients around the clock with cardiologists overreading the tracings. Likewise, thanks to FDA guidance on QT prolongation first issued in 2005, ECG is also routinely used in clinical trials, particularly phase 1 studies.  

As first became clear to Murphey a few years ago while working at another EEG company and subsequently as an independent consultant, clinical trial sponsors and contract research organizations (CROs) both lack the required personnel to acquire, read, and interpret EEGs—and understandably so, she says. It’s a complex craft that takes years to learn and has yet to be well automated. 

EEG expertise is nonetheless vital to evaluating the safety and efficacy of candidate drugs in much the same way as ECGs have long been used, says Murphey. This is particularly true given the number of drug developers that have recently re-committed themselves to the central nervous system (CNS) space. 

When it comes to epilepsy and pediatric syndromes and developmental disorders where seizures are a manifestation of the disease, epileptiform activity is generally monitored as an endpoint to establish if a treatment is reducing hyperactivity in the brain. But EEG has not been systematically implemented for CNS drug development, or in general for any other class of drugs where there may be seizure safety concerns, and this may not happen without formal FDA guidance, Murphey says.  

The fact remains that EEG monitoring can help de-risk trials, she adds. And to their credit, sponsors seem motivated to incorporate the test when animals exposed to a drug experience seizure. “The problem is that they maximize for cost, which is not always the safest thing that they can do for the subject, nor does it serve their long-term commercial interest in approval and staying power of a drug.” 

Calculating Risk

It is sometimes sufficient to incorporate EEG testing only in a drug’s phase 1 trials, says Murphey. But the risks worth evaluating include longer term repercussions as well as whether the experimental agent is acutely causing seizures. 

Even if an individual doesn’t appear to be having a full seizure-like event, an EEG can pick up abnormal “interictal discharges” that are thought to be a sign of a hyperexcitable brain and can cause injury to the brain over time, she says. In patients with epilepsy, this paroxysmal firing is often corrected by increasing their medication.    

Healthy subjects in phase 1 studies of psychedelic drugs intended to treat post-traumatic stress disorder and depression should be getting EEG monitoring to watch for these interictal discharges, Murphey adds, who doesn’t consider this an acceptable risk. There is mixed literature on whether hallucinogens induce seizures. “In this setting, we should absolutely be monitoring the brain.  

“An argument can be made that we need to be exceptionally cautious when phase 1 populations are disproportionately Black and Latino and often without routine care,” she continues, meaning they are underdiagnosed for conditions like traumatic brain injury (TBI), stroke, and epilepsy that make them more susceptible to seizures. Exposing this population to compounds that can be epileptogenic without informed consent and monitoring becomes a serious issue of ethics and potential perpetuation of racialized harms. The FDA and pharmaceutical industry have been largely silent on this issue.”  

The reality is that “everyone in the industry knows phase 1 subjects tend to be people who are lower income and are disproportionately Black and brown,” says Murphey. Since they may never have been evaluated for conditions that make them vulnerable to seizures, if they’re so afflicted, they may not know it and it won’t be in their medical history. 

In these populations, she and her colleagues have anecdotally observed a “higher-than-expected number of abnormalities” identified in their EEG screenings done prior to drug exposure, she says. “That worries me... [when] we don’t have any [regulatory] guidelines around monitoring for brain dysfunction.”  

Medical racism—meaning, historically entrenched practices whereby risks being disproportionately shouldered by certain populations get overlooked—is real, Murphey says. Global solutions are needed to tackle these structural problems. “We have to ask ourselves these questions, especially if we are making public statements about our commitments to things like diversity, equity, and inclusion [DEI].” 

Opposing The Norm

Murphey says it is not uncommon for single experts to be used as EEG readers, despite significant interrater variability and her advice to the contrary. “Sometimes, when that single reader identifies abnormalities that may be inconvenient to drug development, the sponsor may reflex to adjudication. But consensus and adjudication are not utilized routinely, even though the scientific standard for reading EEGs to publish peer-reviewed papers is to do just this.”  

Similarly, protocols will sometimes dictate collecting a different quantity of EEG data before and after an intervention, says Murphey, likely to reduce costs. They might call for a 30-minute screening study in the month prior to drug dosing and equally brief pre-exposure data acquisition but eight hours of data collection post-exposure for purposes of pharmacodynamic analysis. Such “apples to oranges” comparisons are never appropriate. “The longer you study something to identify an abnormality, the likelier it is you are going to find it.” It does not de-risk a drug development program to capture more data after drug exposure than prior to drug exposure.  

And importantly, “there is more variability to epileptiform discharges around sleep and sleep-wake transitions,” as well as circadian variation, she adds. Her standard advice is to capture 24 hours of EEG data before and after dosing in phase 1 trials when participants are on-site for multiple days anyway. This is of course expensive, so Lifelines Neuro does offer alternatives such as episodic (every few hours) rather than continuous real-time monitoring. This both acknowledges the cost limitations of sponsors while still lowering the risk shouldered by subjects. 

Guidance Needed

Recommendations of Lifelines Neuro reflect clinical practice guidelines issued by professional societies such as the American Epilepsy Society and American Clinical Neurophysiology Society as well as technical guidelines regarding quantitative EEG data collection offered by the International Pharmaco-EEG Society.  In the clinical trials space, Murphey says she has found only a single position paper specific to epilepsy that talks about the need to formulate guidelines around the use of EEG parameters as potential endpoints (Epilepsy Research, DOI: 10.1016/j.eplepsyres.2022.107028). 

Murphey says she first raised issues related to the lack of understanding of EEG technology and the need to acquire and analyze the data in standardized ways six years ago—first with colleagues in academia who were EEG readers for studies and later with study sponsors and CROs. With industry, which has much to gain by de-risking trials in this way, it has been the same conversation again and again with little real movement toward substantive change.  

“It would be great to have leadership... from the FDA” in the form of guidance around the development of CNS drugs or any other agents that could potentially cause seizures, says Murphey. In the meantime, she’ll continue her “grassroots” efforts at the company level, which could involve collaboration with CROs in a position to propose recommendations with their sponsor-clients. 

At the end of the day, FDA guidance is going to be required, she adds. “The reason we have incorporated cardiac monitoring and DEI in trials the way that we have... [is because] leadership top-down says this is important.”  

For patients taking the latest FDA-approved Alzheimer’s medication, lecanemab, appropriate use recommendations mention the drug should not be prescribed in patients with seizures since phase 3 trials excluded patients who had seizures within the 12 months prior to screening, she notes. “But EEG was not performed routinely in trials nor is it recommended in post-prescription monitoring for a drug demonstrating up to 20% risk of brain bleeding and swelling, which can put one at risk for seizures.” Another anti-amyloid antibody currently under investigation carries a risk of imaging abnormalities that can lead to seizures, adds Murphy. 

EEG monitoring would similarly be wise, in clinical trials as well as clinical care, among older individuals as well as people with high blood sugar levels, prior TBI, chronic drug exposure, and those regularly ingesting antidepressants—all of which have been linked to a higher incidence of seizures, she says. In many cases, the risks may not be immediate but longer term due to chronic exposure to a compound in vulnerable populations. 

“I think it would be impractical for a pharmaceutical study with EEG to last 30 or 40 years, but we need to be mindful of these things up front,” says Murphey. “The brain is who we are as human beings.”