Lara Jehi, MD, professor of neurology and an epilepsy specialist at the Cleveland Clinic Lerner College of Medicine, spoke to her work on individualized risk assessments for epilepsy surgery candidates and the impact it can have for both physicians and patients alike.
Lara Jehi, MD
Evaluating patients for epilepsy surgery has long involved a mix of neuro and psychiatric evaluations, imaging, and inferences from previous cases. While generally effective, those assessments often still fall short of providing physicians and patients alike a clear, personalized expectation of whether epilepsy surgery will be successful in reducing or eliminating seizures.
At the 73rd annual meeting of the American Epilepsy Society (AES),December 6-10, 2019, in Baltimore, Maryland, Lara Jehi, MD, professor of neurology and an epilepsy specialist at the Cleveland Clinic Lerner College of Medicine, detailed her work on the seizure freedom score and eventually the individualized risk prediction model nomogram, a tool used to more accurately identify patients who might have success following epilepsy surgery.
In an interview with NeurologyLive, Jehi described the novel nomogram and how the results can lead to more personalized clinical care decisions.
Lara Jehi, MD: That's an excellent question, and that's actually one that patients ask most often after they go through a very detailed comprehensive surgical evaluation. They want to know at the end, “How do we know if surgery is going to work or not?” Up until about 2015, that decision was made based on clinical judgment. We have looked, as epilepsy specialists, into data that has been accumulated over years with small series and different hospitals, giving us some signals as to what patient characteristics could highlight people that could benefit from surgery versus not. For example, we knew that people who had an abnormal spot in their brain on an MRI would be more likely to benefit from surgery. We also knew that patients who haven't had epilepsy for a very long time are more likely to benefit from surgery. We were putting patients into these buckets where if you have this one category, you'll get a green mark. If you have this opposite one, you would get a red checkbox. But the challenge that we had as physicians is that patients don't really fall into neat boxes. As an individual patient, you could have one part of your epilepsy that puts you in a good outcome category and another that doesn’t. We would have to just guess, essentially, combining all of that information and making a prediction. So that's why there was a need to develop something that is more systematic, that looks at large data sets, that incorporates all of these different characteristics that patients can have into a comprehensive measure. One that would account for all different characteristics that patients have and come up with a final assessment. Putting in the good with the bad, how does it add up at the end? That's why there was an interest in developing outcome prediction scores.
The seizure freedom score was a score that we developed a couple of years back. It was our initial attempt at trying to address this challenge of having multiple factors that could affect outcomes and figure out a way to account for them. We identified, based on our prior experience, the top 8 clinical characteristics that matter. We worked on them some more to simplify them further and we ended up with 5 that seemed to matter the most. We put them in a scoring system, where if you take the example I just gave about the MRI abnormality, if it's there, you get a point. If it's not there, you get zero. Same thing for how long they had seizures. Same thing for the other 3 characteristics that we looked at which were all fairly straightforward. Then we added it up. We looked to see how good the performance of that edition would do in categorizing patients into these outcome buckets. Either very good outcomes, very poor outcomes, and then people in the middle. We ended up finding that the score does predict very well. If you are a patient with the best-case scenario or the worst-case scenario, then that seizure freedom score performs very well. Its challenge was “How can we improve it to help us better sort out the people who are in the middle?” That’s why after the seizure freedom score, we pivoted towards an individualized risk prediction model that was a bit more sophisticated, called the nomogram, which is a risk calculator. It's essentially a website that you go to and you enter the individual characteristics of your patients. You don't want to have to split them into long or short epilepsy duration, you don't have to make those assessments ahead of time. You just input how long they've had seizures for. It gives them an outcome prediction that is for that specific one patient that you're interacting with. There are no more buckets that we are forcing people to fall under. The seizure freedom score gave us a big cookie cutter, a medium cookie cutter and a small cookie cutter, but it was still forcing people to fit. That nomogram just took every cookie cutter out of the equation. You're then designing that outcome assessment individually for each patient.
It helps the patients get a better expectation of what's to come, should they decide to have surgery. The whole idea of starting projects related to this individualized outcome prediction came to me from a patient. It was a patient who was seeing me in the clinic and was considering temporal lobectomy, which is the most common type of surgery that we do. She was asking me about her odds of being helped with that surgery. I was giving her the numbers that we had at the time and I said, “Well, half the patients who get it are seizure free 10 to 15 years after surgery,” and she looked at me and said, “Dr. Jehi, I'm asking about me. What are my odds of this helping me? I don't want you giving me the numbers of everybody else who had it. I want to know if it's going to work for me or not.” It was that challenge that pushed me in that direction. That's when we developed the seizure freedom score. From it, we developed that online risk calculator, which we are still continuing to improve and refine to make it more accurate. But patients want to know if whatever we're offering them is going to work for them. We try to compensate for that by telling them how well it works in general, but we're talking in this context about brain surgery for epilepsy. It's a huge commitment that a patient is making. They deserve an individualized opinion and not a general statement.