Skip to Content

New Study Published in Nature Neuroscience Shows Neural Fragility EEG Marker and Computational Network Modeling Is Predictive of Surgical Success for Drug-Resistant Focal Epilepsy

March 14, 2022

Drug-resistant focal epilepsy remains a challenging clinical disorder to treat. Complete surgical resection of the epileptogenic zone (EZ) is the only means currently available that in some patients can achieve a complete cessation of seizures, either alone or in combination with drug therapy or neurostimulation.

Precisely identifying the EZ and the critically important SOZ is paramount to achieve successful surgical outcomes and a complete cessation of seizures; however, this can be exceedingly difficult to accomplish in many cases because current testing and imaging modalities are unable to provide the level of precision necessary to completely localize the SOZ in all cases due to the heterogeneous nature of focal epilepsy and the complexity of the disease.

Sensitive and predictive biomarkers for confirming the SOZ localization would be transformational for the field of epilepsy surgery and significantly improve surgical outcomes for many patients. However, these clinical biomarkers have remained elusive, with no potential candidates yet translated for use in routine clinical assessment. 

However, in a study1 published in October 2021 in the journal Nature Neuroscience, a multicenter research team, including Jorge González-Martínez, MD, PhD, FAANS, director of the Epilepsy and Movement Disorders Program in the Department of Neurological Surgery at the University of Pittsburgh School of Medicine, and co-director of the UPMC Comprehensive Epilepsy Center, describes the development of a new computational neuroscience modeling technique, and a sensitive electroencephalograph/computational marker called neural fragility, to aid in confirming the localization of the SOZ in patients with focal epilepsy. Confirmation of the precise locality of the SOZ using the concept of neural fragility as a marker and computational neuroscience modeling of the complex nature of an epileptogenic network was found to be highly sensitive in predicting surgical outcome success in a retrospective study of 91 patients with varying etiologies of focal epilepsy and surgical outcomes. 

The research team included members from the University of Pittsburgh, Johns Hopkins University, Cleveland Clinic, University of Miami Miller School of Medicine, University of Maryland Medical Center, and the National Institutes of Health.

“While surgical resection for some patients with drug-resistant focal epilepsy can be curative, there remains significant room for improvement in outcomes and success rates, but also in the number of potential patients that could be helped through surgery if we can improve upon our ability to precisely map the seizure onset zone in patients with focal epilepsy,” says Dr. González-Martínez. “The new computational neuroscience approach that our team has developed using neural fragility as a computational-based biomarker could radically change and improve how we evaluate and plan for surgical resections or ablations in cases of focal epilepsy, and it could lead to further discoveries as we advance our understanding.”

The Concept of Neural Fragility

The concept of neural fragility in epilepsy is built upon the notion that epilepsy is fundamentally a network-based disorder involving a complex, interconnected neural network. Within this network exist what the team describes as nodes – regions within the brain identified by intracranial EEG (iEEG) signal recording data that are both highly interconnected with other areas and electrically unstable. The unstable nature of the nodes' electrical signaling capabilities, coupled with its dense neural connections to other areas of the brain, allow for the propagation of dysregulated electrical activity and hence the manifestation of seizures. Within this framework, the nodes are representative of the seizure onset zone.

"Accurately and completely identify and remove the seizure onset zone, and you disrupt the aberrant network's starting point and thereby eliminate the seizures. The problem thus far has been our imprecise ability to visualize and therefore identify the SOZ with the accuracy needed such that every patient can achieve an optimal surgical outcome. The location of the SOZ can provide the medical team better guidance to localize the areas in the brain responsible for the organization of seizure activity. This area is also known as the epileptogenic zone. Our computational modeling of the SOZ using neural fragility as a marker for confirming our hypotheses has shown to be powerful when comparing it to other measures,” says Dr. González-Martínez.

Study Highlights and Key Findings

Using iEEG recording data from a heterogeneous group of 91 individuals with drug-resistant focal epilepsy, the team used their computational approach of assessing neural fragility, and hence the SOZ, and then compared those findings against the surgical outcomes from the 91 cases to determine the ability of neural fragility to predict the outcomes in each case.

In the analysis, neural fragility predicted 43 of the 47 cases in which surgical resection failed to eliminate the patient's seizures. In essence, fragile areas of the brain making up the SOZ were not resected in these cases, hence the patient's continuing seizures. 

Additionally, the study’s analysis found that neural fragility had an overall predictive accuracy among the cases of 76%. This was far greater than the 48% accuracy rate of the clinicians’ predictions for surgical success.

The research team also compared the predictive ability of neural fragility against 20 other EEG data characteristics that could be used to predict surgical outcome success. Neural fragility was able to outperform all 20 measures in predictive ability.

Clinical Importance of Neural Fragility as a Biomarker and Future Validation Studies

The potential clinical significance of neural fragility as a biomarker for confirming the localization of the SOZ is profound and twofold.

First, this marker and computational approach developed to evaluate it could improve surgical outcomes because it could predict in advance from the iEEG data whether or not the area targeted for resection is sufficient to remove the node or nodes necessary to achieve a cessation of seizure activity. The data can be collected quickly, in a relatively noninvasive manner, by examining iEEG data from the interictal period. The analysis and interpretation of that data can also be achieved expediently – in minutes to hours.

“Neural fragility and the computational approach to assessing it becomes a new, powerful tool in the diagnostic armamentarium for neurologists and neurosurgeons to surgically treat focal epilepsy with better outcomes,” says Dr. González-Martínez.

In addition, Dr. González-Martínez explains that he and his colleagues are also beginning to investigate the approach's feasibility using scalp-recorded EEG data, which in theory would simplify the diagnostic for some patients by avoiding the use of intracranial electrode placement to record seizure activity.

Furthermore, this new diagnostic tool could help clinicians better understand which patients may not be amenable to surgical resection or ablation, thus avoiding the need for and potential risks and morbidities of a surgical resection that is unlikely to be curative. 

"In this case, the negative predictive capabilities of the modeling approach are just as important as the positive ones. If we can exclude patients from having a surgical resection who, in all likelihood, will not benefit from it clinically or whose seizure may only be marginally improved, we avoid subjecting these individuals to the risks and perhaps false hope that the procedure may work,” says Dr. González-Martínez.

In the future, Dr. González-Martínez and his research colleagues will conduct a larger-scale randomized prospective clinical trial to further validate the efficacy and predictive power of this computational network analysis tool and the role of neural fragility in localizing the complex neural networks involved in patients with medically intractable focal epilepsy.

Learn more about the study at the reference below. Learn more about Dr. González-Martínez, the University of Pittsburgh Department of Neurological Surgery Epilepsy and Movement Disorders Program, and the UPMC Comprehensive Epilepsy Center.


1. Li A, Huynh C, Fitzgerald Z, Cajigas I, Brusko D, Jagid J, Claudio AO, Kanner AM, Hopp J, Chen S, Haagensen J, Johnson E, Anderson W, Crone N, Inati S, Zaghloul KA, Bulacio J, González-Martínez J, Sarma SV. Neural Fragility as an EEG Marker of the Seizure Onset Zone. Nat Neurosci. 2021 Oct; 24(10): 1465-1474.