Brain Waves Shift in Predictable Patterns After Pediatric Epilepsy Surgery

When surgeons disconnect an entire cerebral hemisphere to stop relentless seizures, the isolated half of the brain does not fall silent. Instead, it begins to broadcast a strange and evolving electrical signature, one that shifts predictably over the following two years, according to a new retrospective study of 16 children with drug-resistant epilepsy published in Acta Epileptologica. The findings offer one of the first systematic longitudinal portraits of what the electroencephalogram looks like after hemispheric disconnection surgery, a procedure that seizes control of otherwise untreatable epilepsy by cutting the communication lines inside a diseased hemisphere.

The study, conducted at the Epilepsy Center of Jinan Children’s Hospital, enrolled patients who underwent the surgery between November 2020 and May 2023 and followed each of them for at least two years. The clinical results were striking: 14 of the 16 children, or 87.5 percent, experienced no seizure recurrence within the two-year follow-up window, and at the last follow-up visit all 16 patients had reached Engel Class I, the classification’s highest tier, meaning complete seizure freedom. Two children did seize again during the first year, one at three months and one at one year after surgery, but in both cases the recurrences were traced to specific, correctable causes, postoperative hydrocephalus in one patient and incomplete disconnection in the other. Both were seizure-free after reoperation.

Hemispheric disconnection is reserved for a narrow and severe subset of epilepsy. It is indicated when drug-resistant seizures arise from extensive damage or malformation confined to one hemisphere, a situation that accounts for roughly 20 to 40 percent of epilepsy surgeries in the relevant literature. Rather than removing the hemisphere, the surgeon disconnects it: the operation described in the paper involved resecting opercular tissue, severing the corpus callosum, interrupting the connections between the fronto-parieto-occipital lobes and the basal ganglia, and isolating the insular cortex, temporal lobe, posterior hippocampus and amygdala. The goal is to wall off the epileptogenic network so its abnormal discharges can no longer propagate to the healthy hemisphere, which retains its motor, sensory and language functions.

To track what happens electrically after this radical intervention, the team performed video-EEG monitoring before surgery and at three months, one year and two years afterward, using a 32-channel system with electrodes placed according to the International 10-20 System. Preoperative recordings lasted at least 24 hours; postoperative sessions were shorter, at least three hours each, a methodological constraint the authors themselves flag as a limitation. Two experienced neurophysiologists reviewed all 64 recordings independently, and their agreement was nearly perfect, with Cohen’s kappa coefficients ranging from 0.868 to 1.000 across the various EEG classification categories.

The preoperative baseline was itself pathological. In 12 of the 16 children, the hemisphere destined for disconnection showed continuous slow-wave activity, theta and delta rhythms of low to moderate amplitude, while two children already displayed very low-voltage activity below 20 microvolts. Epileptiform discharges, the sharp spikes and sharp-and-slow-wave complexes that mark epileptogenic tissue, appeared in the affected hemisphere in every patient. Notably, more than half the cohort, 11 of 16 children, also showed discharges over the supposedly healthy contralateral hemisphere, a finding that in many epilepsy contexts would raise concern. The study’s data suggest that in this surgical population, such contralateral discharges are not a barrier to surgery or necessarily a bad omen.

After surgery, the disconnected hemisphere began a choreographed electrical evolution. At three to four months, half the children showed persistent slow waves on the operated side and half showed low-voltage activity. Epileptiform discharges there initially increased rather than decreased: 11 of 16 children, or 68.8 percent, showed more discharges with higher amplitude than before surgery, and nine children, 56.3 percent, exhibited a burst-suppression-like pattern, in which bursts of medium-to-very-high-amplitude slow waves mixed with spikes lasting half a second to two seconds alternated with near-flat suppression phases below 20 microvolts lasting one to three seconds.

By one year, the low-voltage pattern had become dominant, present in 11 of 16 children, while the three children whose healthy hemisphere had shown abnormal slow-wave background before surgery all returned to normal. The burst-suppression-like pattern became more prominent or newly appeared in six children during this period. By two years, the pendulum swung back: overall background amplitude on the disconnected side rose, slow waves again predominated in 11 children, discharges diminished and stabilized, and the burst-suppression-like pattern improved in six children and resolved completely in three. On the healthy side, the story was one of gradual quieting: among the 11 children with preoperative contralateral discharges, the discharges decreased in eight and vanished entirely in three, and by two years only three patients still showed any persistent contralateral discharges.

The authors propose a mechanistic account for these dynamics. The disconnected cortex loses the rhythmic drive and modulatory input it normally receives from the thalamus and can no longer maintain stable, desynchronized activity, falling instead into a highly synchronized, imbalanced state. Local network reorganization disturbs the excitation-inhibition balance, with reduced GABAergic inhibition and increased glutamatergic excitation lowering the burst threshold. Meanwhile, blood perfusion and glucose metabolism in the isolated hemisphere drop substantially, with blood flow reduced by an estimated 30 to 50 percent, and diminished ATP production struggles to sustain continuous neuronal firing. Early slow waves likely reflect acute disruption of thalamocortical fibers and transient metabolic suppression; the later low-voltage phase reflects permanent deafferentation; and the eventual return of slow waves may represent the isolated cortex settling into a new equilibrium, regaining weak autonomous activity from local intracortical circuits. Prior work by Wennberg and by Joshi, cited by the authors, described similar burst-suppression patterns in functionally isolated cortex, supporting the idea that this pattern marks physiological isolation rather than pathology.

Critically, none of these EEG features predicted outcome. Fisher’s exact tests found no significant association between seizure recurrence and low-voltage background, the presence of a burst-suppression-like pattern, or preoperative contralateral discharges, with all P values above 0.05 and phi coefficients between roughly minus 0.28 and plus 0.22. The co-occurrence rate of low-voltage activity and burst-suppression-like patterns reached 56.3 percent, and 11 of the 14 children tapering anti-seizure medications displayed the burst-suppression pattern, yet neither feature tracked prognosis. The authors are unusually candid about why these null results carry little weight: a post-hoc power analysis showed the study had only about 17.2 percent power to detect a clinically meaningful association, defined as an odds ratio of 3.0, with just two recurrence events in the cohort. Power estimates for the individual analyses ranged from 11.2 to 17.5 percent, meaning the findings cannot distinguish a true absence of association from a simple failure to detect one.

The practical message for clinicians is nonetheless concrete. A burst-suppression-like pattern over a disconnected hemisphere should be read as a marker of cortical isolation, not as an indicator of poor outcome, and decisions about withdrawing anti-seizure medication should not hinge on the persistence of abundant ipsilateral discharges, which routinely increase for the first year after this operation. The two recurrences in this cohort stemmed from hydrocephalus and incomplete disconnection, detected through combined EEG, MRI and clinical evidence, rather than from any EEG pattern per se. The study’s limitations are substantial and acknowledged: it is a single-center retrospective analysis of 16 young children, mean age at surgery 45.9 months, with short postoperative EEG recordings, no diffusion tensor imaging, no postoperative PET, and no standardized quantitative assessment of disconnection completeness. The authors call for multicenter prospective studies with at least 100 to 150 patients, extended 24-hour postoperative monitoring and quantitative neuroimaging to establish whether EEG features after hemispheric disconnection can ever serve as reliable prognostic tools. Until then, the evolving waves of the silent hemisphere remain a fascinating physiological phenomenon whose clinical meaning is still being mapped.

Subject of Research: Electroencephalographic changes after hemispheric disconnection surgery in children with drug-resistant epilepsy

Article Title: Characteristics of electroencephalographic changes after hemispheric disconnection surgery in patients with drug-resistant epilepsy

Article References: Cui, X., Gao, Z., Chen, S., Liu, L., Shi, J., Geng, G., & Zhan, X. (2026). Characteristics of electroencephalographic changes after hemispheric disconnection surgery in patients with drug-resistant epilepsy. Acta Epileptologica, 8(1), Article 28. https://doi.org/10.1186/s42494-026-00269-z

Image Credits: AI Generated

DOI: 10.1186/s42494-026-00269-z

Keywords: drug-resistant epilepsy, hemispheric disconnection, EEG, burst-suppression pattern, pediatric epilepsy surgery, seizure freedom, video-EEG, retrospective study, hemispherotomy, prognosis, Jinan Children's Hospital, anti-seizure medications