Abstracts

Rationale: At Children's Hospital of Michigan (Detroit), PET scans of glucose metabolism are integral to the presurgical evaluation for hemispherectomy in order to evaluate the functional integrity of the contralateral hemisphere. Patients with significant contralateral PET and/or MRI abnormalities are more likely to be excluded from surgery, although some are offered a 'palliative' resection. In the present study, we hypothesized that our stringent criteria for surgical selection might be associated with better seizure outcomes than reported in meta-analyses of published hemispherectomy series (73% Class 1 outcome; Griessenauer et al., 2015; Hu et al., 2015, J Neurosurg) where PET was not routinely used in presurgical evaluation. Methods: 55 patients (ages 0.2-20 years; 31 females) operated between January 1993 and September 2015 with adequate follow-up were studied. At the Children's Hospital of Michigan in Detroit, most of the patients had anatomical hemispherectomies. 'Palliative' cases were excluded from this analysis. One-stage hemispherectomy was performed in 44 children; chronic intracranial monitoring with subdural electrodes was performed in 11 cases but the results led to a decision for hemispherectomy. Mean follow-up was 39.6 months (range 6 to 174 months). 52 were lesional cases and 3 were non-lesional based on MRI. Results: 50 of the 55 patients achieved an ILAE Class 1 outcome (91%). 25 patients were no longer taking AEDs. Surgical failures included 2 patients with epilepsy following previous meningoencephalitis, one with anti-GAD antibody encephalitis initially diagnosed with Rasmussen encephalitis, one with neonatal thalamic hemorrhage of undetermined etiology, and one with focal epilepsy associated with photomyoclonic epilepsy. 13 patients (23%) developed isolated hydrocephalus requiring shunting. Four other patients developed a post-operative complication. These included one child who suffered a hypoxic brain injury during surgery subsequently developing hydrocephalus requiring shunting but was seizure free, one child who developed a scalp granuloma requiring excision, one who developed an epidural hematoma postoperatively requiring evacuation, and one who developed an infection requiring cranioplasty and shunting. Conclusions: The surgical results of hemispherectomy are excellent in carefully selected cases (91% Class 1 in this series), and the post-operative complications are generally low, with the exception of hydrocephalus (23%). Although our surgical failures were few (9%), there is still opportunity to improve prediction of surgical failures (and thus exclude surgical candidacy) by appropriate diagnostic testing. Finally, 'palliative' surgery (excluded in this analysis) might be offered to some patients, but this must be determined a priori and analyzed separately from 'curative' surgical cases. Funding: None