A Statistical Tool to Predict Outcomes with Epilepsy Surgery
ARTICLE IN BRIEF
Nomograms that incorporate a patient's complex clinical characteristics can be used to predict outcomes for patients who have been judged eligible for epilepsy surgery.
Doctors advising patients on whether to get epilepsy surgery are often asked a seemingly simple question: “What are my chances for being seizure-free after surgery?”
The response is often based on statistics from research studies that show a range of possible outcomes — useful data, for sure, but not necessarily enough for patients who want to know whether surgery will put an end to their seizures.
In the Jan. 28 online edition of The Lancet Neurology, researchers at the Cleveland Clinic describe a statistical modeling tool that they say may make it easier to predict an individual's chance for a favorable outcome post-surgery.
Referred to as a nomogram, the tool — found here at http://bit.ly/nomogram-Lancet — calculates the odds for seizure freedom using different variables.
“We aren't really doing a good job as physicians in guiding patients through a decision as major as having brain surgery for epilepsy,” said Lara Jehi, MD, the study's lead author and director of research at the Cleveland Clinic Epilepsy Center.
Dr. Jehi said “the main goal of the nomogram is to supplement rather than replace the critical individual aspects of physician decision-making.” Nomograms have been shown to be useful prediction tools in other fields of medicine, including cardiology and cancer treatment, Dr. Jehi noted, adding, “In neurology, we have been lagging behind.”
Nomograms provide “another tool to help with clinical decision making,” Dr. Jehi said, but she cautioned that the nomograms proposed in the current study need to be further validated.
STUDY METHODOLOGY, RESULTS
The Cleveland Clinic researchers developed the nomogram to predict a patient's odds of being seizure-free at two years and five years and another to predict the odds of having an Engel score of 1 — that is, they might have some initial seizures in the first weeks after surgery or seizures occurring only with physiological stress such as drug withdrawal — at two years and five years after surgery.
They based the nomograms on a review of 846 cases involving patients who had undergone epilepsy surgery at the Cleveland Clinic Epilepsy Center between January 1995 and December 2012. Cases involving hemispherectomy, multilobar resection, and re-operations were excluded. They looked at nine characteristics — sex, seizure frequency, secondary seizure generalization, type of surgery, pathological cause of seizures, age at seizure onset, age at time of surgery, epilepsy duration at time of surgery, and surgical side — to determine how the various factors influenced a positive outcomes. Six were ultimately relevant to the nomogram, including sex, seizure frequency, secondary seizure generalization, type of surgery, pathological cause of seizures, and epilepsy duration at time of surgery.
The review found that 57 percent of patients were completely seizure-free at two years after surgery and 40 percent were seizure-free at five years. Sixty-nine percent of the patients had an Engel score of 1 at two years and 62 percent at five years.
The two nomograms developed from that analysis were validated using a cohort of 604 patients from four epilepsy centers — the Mayo Clinic in the US, University of Campinas in Brazil, INSERM in France, and Ospedale Niguarda in Italy — and they were found to be highly predictive of outcome. The researchers reported that the “proportion of patients in the developmental cohort who were completely seizure-free was 0.57 at two years and 0.40 at five years; the proportion who had an Engel score of 1 was 0.69 at two years and 0.62 at five years. In the validation cohort, the models had a concordance statistic of 0.60 for complete freedom from seizures and 0.61 for Engel score of 1.”
“Long epilepsy duration, high baseline seizure-frequency, and a history of secondary generalization were associated with worse seizure outcomes,” Dr. Jehi said. “Outcomes were better in women, and in patients with tumors and mesial temporal sclerosis compared to malformations of cortical development. The extent to which these different variables influenced outcomes varied in different situations, particularly in relation to the type of surgery being proposed. For example, the relation between epilepsy duration and outcomes was most relevant in frontal lobe resections.”
“In an era when health care costs are increasing for brain surgery, both patients and physicians would benefit from a prediction of the probability of success on a case by case basis,” the researchers wrote, though they also noted that the nomograms had some shortcomings.
The statistical models do not take into consideration such factors as quality of life, mood, and psychosocial function, nor do they include information gleaned from EEG testing, MRI scans, or other sophisticated neuroimaging. Dr. Jehi said it is difficult to capture in a statistical model “all the nuances a case might have,” though her team is continuing to refine the nomograms to include EEG and imaging results.
In an accompanying editorial in The Lancet Neurology, Jerome Engel Jr., MD, FAAN, the Jonathan Sinay professor of neurology, neurobiology, psychiatry and behavioral science at the University of California, Los Angeles, cautioned against overreliance on nomograms, stressing that they should not replace the expertise of doctors or outweigh what's learned from test results in a presurgical workup.
A nomogram “could be useful when patients are referred to an epilepsy center and are ambivalent about whether they want to go through the surgical workup,” said Dr. Engel, who directs the UCLA Seizure Disorder Center. But he worries that if neurologists or general practitioners use the tool earlier in the deliberation process it could have the unintended result of keeping patients from being referred to an epilepsy center for further evaluation, not only for surgery but also for non-surgical treatments and services such as psychiatric, psychological, and social management.
“Ideally all patients with refractory epilepsy deserve a consultation at a full service epilepsy center,” he said, but currently only about 1 percent of patients who meet the definition of medically intractable epilepsy — their seizures persist despite trying two antiseizure medications — are referred. He said the nomograms devised by the Cleveland Clinic researchers fail to take into consideration added information that can be gleaned from an evaluation by a multidisciplinary team at an epilepsy center.
Jacqueline French, MD, FAAN, director of the Clinical Trials Consortium at New York University's Comprehensive Epilepsy Center and a member of the Neurology Today editorial advisory board, said she applauds the Cleveland Clinic team for their work in developing a nomogram for epilepsy surgery, but she questions the use of only patients who had already undergone surgery to develop their prediction model. She said the nomogram would look different if the researchers had included as a starting point in their calculations all patients who had come into the epilepsy center for a surgical evaluation, whether they ended up having surgery or not.
“The odds for a positive outcome change depending on which group of patients you look at,” she said.
EXPERTS: ON NOMOGRAMS TO PREDICT EPILEPSY SURGERY OUTCOMES