Streamlining Seizure Detection with Kaha Telemetry

By
Blaise Cahill-Lane

Scientific Writer

The team at the Anticonvulsant Drug Development (ADD) Program, a lab within the University of Utah, is hard at work testing new compounds to treat epilepsy. It’s an intensive job, with blinded trials lasting for months at a time to amass the data needed to push a compound on to the next stage of drug development or end its journey altogether.

ADD uses spontaneous seizure models in rats and mice to replicate epilepsy. “Traditionally, the way drugs were screened was usually to induce a seizure either electrically or chemically. You would give the drug beforehand, then induce the seizure and ask: ‘Did my drug block that seizure?’,” ADD Associate Director, Dr. Cameron Metcalf says, “But as we’ve expanded our ability to identify new targets and better validate new drugs, we are also looking at seizures as they occur spontaneously in various contexts… it’s a more clinically relevant seizure type, and it better models what happens in people.”

To accommodate this more realistic model of epilepsy, however, the animals have to be recorded at all times. EEG biopotentials are continuously recorded from the animal’s brain, synced up with video recordings that show the animal’s behavior and any potential seizures. ‘Potential’ is the operative word here, because all kinds of signal noise can look like a seizure in an EEG readout. In order to confirm that the signal was a seizure, and not just a scratch of the ear or a bite of food, the researchers have to look through the video recordings.

The EEG signal was originally recorded through a tether, a long cable directly connecting the sensor in the animal to a recording device. Over the past few years, however, the ADD Program has gone wireless for their rats. Using implantable Kaha biopotential telemeters the team is able to record EEG signals without the extra signal noise that comes with an external tether.

“We have an algorithm for seizure detection, but missing a seizure is really bad, so you have to train your algorithm so that it misses nothing,” says Dr. Kyle Thomson, senior staff scientist in the ADD program, “but that means it’s going to pick up a lot of false positives… The tether is just too noisy [for the algorithm]. You end up with the same amount of data to review as just doing it by eye.”

“The reason why we use, and continue to use, the Kaha system is because the cleanliness of the data has made it possible to trust our seizure detection algorithms to pick potential signals up and then reliably discard false positives (~95% of the data), look at the remaining 5% of it and be done quickly,” he says, “we're talking the difference between 4 hours a day and 20 minutes for the same amount of data.”

That signal clarity, however, would be useless if the system wasn’t flexible enough to handle the realities of lab life. “Until [recently] every other system has been on batteries,” Kyle says, “batteries need to be replaced. And that limits your studies. A lot of our animals sit around for six months enrolled in drug studies. A three-month battery would just be a no-go.”

"Things in the pipeline always get delayed,” he says, “there's always something going on that might force a study to go back a week. And if you're hitting a three-month battery, even if you plan your strike perfectly to fit in those three months, all of a sudden you're a week out from the battery dying potentially.”

“The flexibility that we have because of the rechargeable system and being able to explant and re-implant has been so huge for what we need to do,” Cameron says.

“We have, I think, two telemeters on their eighth implant now. So we’ve been using them pretty much continuously for 3 or 4 years,” Kyle adds.

The more that the process is streamlined, the more effectively the work can get done. And that is key in the drug development process.

“About a third of patients with epilepsy are still not experiencing full seizure control,” Cameron says, “We are fortunate to have that two-thirds number where most people can have their seizures primarily stopped and taking medicines with a minimal amount of side effects. But we still, at least over the last two decades, have not gotten any better with that number,” he stresses, “that's one of our biggest hurdles in the field, is finding new drugs, even new combinations of drugs or new modes of therapy, in general, that help this group of people that are currently suffering from seizures without adequate seizure control.”

“We've seen some compounds that have come through our program that just blow our minds in these spontaneous seizure models. They look amazing. But it could be 20 years between when we start doing this work and the actual end result hits the patient just because there is so much that goes into drug development.”

You want to be able to greenlight or redlight a compound quickly, and without concern that the process may have been compromised by the technology used or excess strain on researchers.

“The difference between a drug completely eliminating all seizures and an animal having one seizure could be huge or even worse, you missed a seizure on [the baseline]. And that brings the average down for that [baseline] treatment. And all of a sudden, the drug doesn't look as good.” Kyle says, “All of these things are so bad, but you're looking for this little bit of video, 10 to 30 seconds in 24 hours of data.”

As the National Institutes of Health moves further toward mouse models, the cost and energy needed to assess those models becomes increasingly steep. “We've just kind of kept going with seizure review in mice by eye and we have so much staff time dedicated to looking at EEG this way because we're doing so many mouse experiments,” Kyle says, “And at the end of the day, when I go to do my review, those reviewing mouse EEG hate me. I've had like five, eight-minute files and it's just like boom, boom, boom. Oh, I'm done for the day. Oh, I just did four days,” he laughs, “and the mouse review staff are sitting there, their eyes are bleeding, mainlining coffee, because it's just an arduous process. You have to be very focused.”

With the introduction of the implantable Kaha mouse biopotential telemeter, all of that is about to change. “We just got the 40 pads [for the mice], and we’re building that suite,” Kyle says, “I imagine once the mouse telemetry is going we’ll have 12 rats and 40 mouse telemeters running pretty much continuously.”