New Approach May Revolutionize Epilepsy Surgery

Pauline Anderson

July 11, 2016

Researchers have created what seems to be a ground-breaking method of identifying regions of the brain that trigger seizures in patients with epilepsy, a finding that could revolutionize surgery for the disorder.

The mathematical model measures the relative contribution of different brain regions to the development of seizures, thereby helping to pinpoint areas to resect for optimal surgical outcome in individual patients with epilepsy.

This model-based approach to quantifying the ictogenicity of brain networks could function as "a decision support tool" for neurosurgeons, said John R. Terry, PhD, professor of biomedical modelling, and director, EPSRC Centre for Predictive Modelling in Healthcare, College of Engineering, Mathematics and Physical Sciences, University of Exeter, United Kingdom.

"It's not going to replace surgeons; it's going to provide additional information that will hopefully be valuable to them when it comes to planning the surgery," Dr Terry told Medscape Medical News.

The paper outlining the research behind the new approach was published online July 7 in Scientific Reports.

Low Surgical Success Rates

Surgery is a valuable option for patients with epilepsy in whom pharmaceutical therapy has failed. However, long-term postsurgery success rates are only about 50%, and they can be as low as 15% with extratemporal surgeries.

Many cases that are evaluated don't even proceed to surgery because of the inability to identify an optimal resection zone.

There are "myriad" potential reasons why this brain surgery fails, but a likely one is the way in which it's currently performed, said Dr Terry.

From patterns of electrical activity in the brain, surgeons typically look for "markers" that suggest that a particular region — or epileptogenic zone — represents the origin of the seizures or is involved in some way in creating them.

"We see these patterns in a particular region of the brain and therefore we assume that that's the region of the brain that the seizure is originating from, whereas in actual fact it could be, and our research suggests, that…there are other regions involved in generating the seizures," said Dr Terry.

The patterns surgeons see may merely be a consequence of the seizure mechanism itself rather than the actual cause, he added.

In epilepsy, aspects of the brain that might correlate with the epileptogenic zone might include brain regions that are first to produce pathologic electrographic activity when seizures occur and regions that display evidence of abnormal structure that may contribute to ictogenicity, the investigators note.

For this new predictive approach, researchers start with the electrical recordings that are routinely collected as part of the presurgical planning process.

"We take periods of data when patients are not having seizures, and we use these recordings to build up a representation of how the different regions of brain are communicating with each other," said Dr Terry.

"We have essentially created an in silico representation of the person's brain."

Patient-Specific Approach

A "network of interactions" extracted from these recordings is put into the computer model. "Then we run the computer simulation and we examine the regions of the brain where seizures appear to be starting from."

The model identifies a set of brain regions that, if removed, "would result in the network becoming seizure free, the brain becoming seizure free, and this would be the recommendation to the surgeon."

In principle, the computer model can always identify an optimal set of regions that should be removed to stop seizures, but that might not always be recommended.

"There may be a number of patients for whom surgery isn't an option because the regions of the brain involved in generating the seizures are too important," said Dr. Terry. "You may remove their seizures but leave them paralyzed."

The researchers have validated this approach retrospectively in 16 adults with epilepsy from Switzerland ranging in age from 19 to 59 years (median age, 31 years) who had focal epilepsies. All had undergone standard surgical resection as determined by current clinical methods up to 5 years previously, with a median postsurgical follow-up of 3.0 years, said Dr Terry.

Investigators were given access to the presurgical electrographic recordings for these patients, from which they created a patient-specific representation of the brain and used this model to identify the relative contribution of each brain region in generating seizures.

They used this information to determine the optimal set of regions to resect in order to reduce the seizure likelihood, and they compared these model predictions with the brain regions the surgeons actually removed.

"What was really interesting was scenarios where patients had a good outcome. For example, their seizures were dramatically reduced, the regions of the brain the surgeon actually removed correlated very nicely with regions of the brain that the model suggested should be removed," said Dr Terry.

On the other hand, there was poor agreement between the model prediction and surgical sites in cases where there was a bad outcome. 

Dr Terry stressed that every patient has subtly different brain connectivity, with brain regions communicating in slightly different ways. When seizures occur, they are a consequence of these subtle differences.

"The whole point is to make this patient specific."

Decision-Making Tool

The researchers have funding to develop a working prototype of the model, which will serve as a tool to support decision-making.

"It would be a computer that would sit alongside the other tools that the surgeon uses in presurgical evaluation to make the decision about which regions of the brain they should remove."

The plan is to meet over the next few months with neurosurgeons from across the United Kingdom and Europe to get their input into the prototype.

"We want to know some of the issues they have around the use of a computer model in terms of surgery planning; what are their concerns; what do they believe are the opportunities; how would they want to see the information presented, et cetera."

Once the prototype is "up and running," which Dr Terry predicts will be soon, it may not be long before it will be put into practice.

"It could be sooner than we think that this prototype device is being trialed," said Dr Terry, adding that many surgeons "are keen" to work with the researchers to test it.

But it could be much longer before it reaches clinical practice because a prospective study and validation must be done first, followed by regulatory approval. "For regular use, we're talking a number of years," said Dr Terry.

He noted that the mathematical model could be tested prospectively in an animal model but that validating it in an animal model may be limited given the concern that the mammalian phenotype may not adequately represent human epilepsy.

Alternatively, it could be tested directly in people with epilepsy.

Mathematical Twist

Commenting on the findings for Medscape Medical News, Joseph I. Sirven, MD, professor, neurology, Mayo Clinic, Phoenix, Arizona, said the study is "interesting," but not in the way one might think a new test or technique might be interesting.

"Rather, this study shows that if you crunch up existing data from EEG [electroencephalography] and apply a mathematical twist, it may provide a better way of identifying the best place to resect for successful epilepsy surgery."

But while the paper discussed the model and theory behind it, "what is missing for me is the actual proof from patients a priori undergoing this testing model and then surgeons operating based on the results and getting great results."

Despite that, Dr Sirven said the approach is "novel" and may eventually prove useful to neurosurgeons.

Better targeting of brain regions for optimal results "is considered a holy grail of epilepsy surgery," said Dr Sirven.

Michael D. Privitera, MD, professor, neurology, University of Cincinnati, Ohio, and president, American Epilepsy Society, noted that the validation study was retrospective and the sample size small, and that many variables affect outcome in epilepsy surgery.

"Without seeing the supplemental information on these patients, it's hard to say how much heterogeneity is in their sample."

He also called the model-based approach an "interesting idea" and agreed that "it needs to be replicated in a much larger sample, then tested prospectively."

The authors and Dr Sirven and Dr Privitera have disclosed no relevant financial relationships.

Sci Rep. Published online July 7, 2016. Full text

For more Medscape Neurology news, join us on Facebook and Twitter.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: