BRECKENRIDGE, COLO. — Gamma knife surgery shows some promise for drug-resistant epilepsy, but the definitive proof of its effectiveness remains a few years away, C. Akos Szabo, M.D., said at a conference on epilepsy syndromes sponsored by the University of Texas at San Antonio.
Radiation-induced changes, both good and bad, require time to evolve. The durability of the improvement seen to date as well as the possibility of future late side effects remain open questions, explained Dr. Szabo, a neurologist who is director of the epilepsy surgery program at the university's South Texas Comprehensive Epilepsy Center.
Gamma knife surgery—also known as stereotactic radiotherapy—has been around for a long time. Worldwide, 200,000 procedures have been performed, chiefly for treatment of benign and malignant brain tumors, trigeminal neuralgia, and vascular malformations. Only relatively recently have investigators pursued in earnest the gamma knife's potential as a less invasive therapy for epilepsy that spares patients the risks associated with craniotomy and cold-steel resective surgery.
Proposed candidates for gamma knife surgery have a single well-defined temporal lobe seizure focus. The procedure utilizes a collimator to focus a hemispheric array of 201 beams of gamma radiation from cobalt-60 sources on a 5,500- to 9,000-mm3 targeted volume of brain tissue. The dose at the target margin is 24 Gy. The operation is done in a single sitting, with separate anterior and posterior exposures.
The sole reported prospective study of gamma knife surgery to date was conducted in Marseille, Prague, and Graz, Austria. It involved 21 patients with refractory mesial temporal lobe epilepsy deemed suitable for temporal lobectomy but who instead had stereotactic radiotherapy. One died of amyocardial infarction, leaving 20 with 2-year follow-up.
“Of course, this type of study would have to go on for 10 years before it would gain any type of acceptance within the epilepsy surgery community,” Dr. Szabo observed.
Targets of the focused radiation included the anterior parahippocampus, head and body of the anterior hippocampus, the rhinal sulcus, and the basal and lateral amygdala.
The tricky thing about gamma knife surgery is the seizures don't begin to improve until 9–12 months later. During the interim they actually get worse; for the first months following surgery, patients may begin to have complex partial seizures rather than their usual simple partial seizures.
Still, at 2 years the median seizure frequency was reduced to 0.33 per month, compared with 6.16 per month before treatment, and 65% of patients were seizure free. Quality of life measures improved significantly (Epilepsia 2004;45:504–15).
With the target tissue's close proximity to the optic nerve, it was unsurprising that 8 patients in the European study had quadrant visual field defects resulting from the gamma knife. Another patient experienced narrowing of the visual hemifield, a more disabling side effect.
Radiation-induced brain swelling, often accompanied by extremely severe headaches, was a common late side effect, peaking in occurrence at 1 year.
Of gamma knife-treated patients, 62% required corticosteroids for this condition, and 15% were hospitalized. However, the brain swelling resolved over time, and at 2 years none of the study participants had any neuropsychologic impairment.
The big problem for proponents of gamma knife surgery, Dr. Szabo noted, is that conventional open temporal lobectomy is a relatively mature operation whose long-term benefits and risks are well documented.
A 2003 evidence-based practice parameter issued by the American Academy of Neurology concluded that two-thirds of patients become essentially seizure free following anterior temporal lobectomy, while 10%–15% are unimproved. Three percent of patients develop permanent neurologic deficits, and an equal percentage experience permanent cognitive or behavioral deficits.
Thirty-day mortality was 0.4% in the patient series analyzed by the AAN panel, but none of the deaths was deemed surgery related (Neurology 2003;60:538–47).