Epilepsy surgery in early infancy: A retrospective, multicenter study

Abstract Although epilepsy surgery is the only curative therapeutic approach for lesional drug‐resistant epilepsy (DRE), there is reluctance to operate on infants due to a fear of complications. A recent meta‐analysis showed that epilepsy surgery in the first 6 months of life can achieve seizure control in about two thirds of children. However, robust data on surgical complications and postoperative cognitive development are lacking. We performed a retrospective multicenter study of infants who underwent epilepsy surgery in the first 6 months of life. 15 infants underwent epilepsy surgery at a median age of 134 days (IQR: 58) at four centers. The most common cause was malformation of cortical development, and 13 patients underwent a hemispherotomy. Two thirds required intraoperative red blood transfusions. Severe intraoperative complications occurred in two patients including death in one infant due to cardiovascular insufficiency. At a median follow‐up of 1.5 years (IQR: 1.8), 57% of patients were seizure‐free. Three patients where reoperated at a later age, resulting in 79% seizure freedom. Anti‐seizure medication could be reduced in two thirds, and all patients improved in their development. Our findings suggest that early epilepsy surgery can result in good seizure control and developmental improvement. However, given the perioperative risks, it should be performed only in specialized centers.


| INTRODUCTION
Epilepsy is one of the most common neurologic disorders with the highest incidence in the first year of life in childhood. 1 One third of patients do not achieve seizure freedom through anti-seizure medication (ASM). 1,2 Drug-resistant epilepsy (DRE) in infancy is caused most frequently by vascular etiologies and can contribute to cognitive deterioration and low quality of life. [3][4][5] For this reason, rapid seizure control is crucial to minimize the damage to the developing brain. 6 While epilepsy surgery is an established treatment for MRI-positive (lesional) DRE, both families and physicians are reluctant to consider surgery in the first months of life. 7 Thus, surgery in early infancy is seldomly performed, resulting in a lack of data on the specifics of the procedures and their outcomes. 8 Especially comprehensive data on complication rates and cognitive development in this specific age group are lacking. Our retrospective multicenter study aims to show real-world data on complications and outcomes of epilepsy surgery in the first 6 months of life.

| METHODS
We performed a retrospective multicenter study of all infants who underwent epilepsy surgery in the first 6 months of life (until the 190th day of life) at four epilepsy centers in Germany and Switzerland from 2012 to 2022. Surgical techniques where either classified as hemispheric or focal (ie, [multi]-lobectomy and lesionectomy). Intraoperative complications and postoperative morbidity were defined as unexpected events. 9 In all patients with cognitive testing the BSID-II/III was used (Mental Development Index and Cognitive Scale). Statistical analysis was performed with R (4.2.1, psych, crosstable, dplyr, resphape2, ggplot2 and ggpubr). Wilcoxon test and Fisher's exact test were used as applicable. The study was approved by the local ethics committee (approval no. EA2/084/18).

| RESULTS
The study cohort comprised 15 patients (12 female: 3 male) who underwent epilepsy surgery in the first 6 months of life (Table 1). Median seizure onset was 13 days (IQR: 56, range 1-128), and first surgery was performed at an age of 134 days (IQR: 58, range 61-186), ( Figure 1A). The youngest patient was operated at a corrected age of 1 week. 10 Cranial MRIs revealed malformation of cortical development as the most common cause (n = 14).
Preoperatively all patients had multiple seizures a day, and seven patients had at least one status epilepticus until surgery. Most patients had focal motor seizures; six patients had epileptic spasms. The infants had a median life time number of five ASM (IQR: 2.5) and were taking at surgery a median number of three ASM (IQR: 1) ( Figure 1B). While long-term video-EEG was performed preoperatively in all patients, no patient received invasive recordings. Interictal epileptiform discharges were detected in all patients ipsilaterally and in four patients also contralaterally to the MRI lesion. EEG seizure patterns were recorded on the ipsilateral (surgical) side in ictal EEGs in every patient, while in three patients also contralateral EEG seizure patterns could be identified. Preoperative standardized cognitive testing showed belowaverage development quotient (DQ) values with a median of 55 (IQR: 24.5) in seven patients. Prior to surgery, a third The most common cause was malformation of cortical development, and 13 patients underwent a hemispherotomy. Two thirds required intraoperative red blood transfusions. Severe intraoperative complications occurred in two patients including death in one infant due to cardiovascular insufficiency. At a median follow-up of 1.5 years (IQR: 1.8), 57% of patients were seizure-free. Three patients where reoperated at a later age, resulting in 79% seizure freedom. Anti-seizure medication could be reduced in two thirds, and all patients improved in their development. Our findings suggest that early epilepsy surgery can result in good seizure control and developmental improvement. However, given the perioperative risks, it should be performed only in specialized centers.

K E Y W O R D S
anti-seizure medication, cognition, epilepsy, epilepsy surgery, infants, neurosurgery, outcome, pediatrics of all patients had a hemiparesis (n = 5), all of whom had a hemispheric lesion. 13 infants (87%) underwent hemispherotomy, which was extended in one patient with a unilateral frontal resection, and in another with an additional tumor resection in the lateral ventricle during the same surgery. Two patients underwent epilepsy surgery for a vital indication due to intractable epilepsy requiring mechanical ventilation. Infants had a median weight of 6200 g at surgery (IQR: 1733, range: 3830-9000). The median time from incision to suture was 228 min (IQR: 221, range: 80-540, n = 11). Hemispheric surgeries took longer than focal surgeries (237 vs 163 min, IQR: 255.0 vs 123.5, n = 7). More than two thirds of patients (n = 11) required intraoperative red blood cells (RBC). Patients received a median of 341 mL of RBC (IQR: 1252.5, range: 118-3000, n = 8), which translates to 57.67 mL/kg of body weight (IQR: 193.66, range: 17.05-425.81, n = 8). In one patient with hemimegalencephaly, blood loss during hemispherotomy was so severe that he required six RBC transfusions during surgery due to a strong tendency to bleed when touching the cortex.
One patient died intraoperatively during hemispherotomy at the age of 172 days. He also showed cortical bleeding which was well controlled during surgery. However, the patient required resuscitation during the surgery and died secondary to cardio-circulatory insufficiency.
Overall, hemoglobin levels remained stable from 1 day postoperatively (10.4-9.2 g/dL, P = 0.203, n = 9) and throughout the hospital course. Approximately one third of patients received RBCs postoperatively (n = 4). Patients received a median of 95 mL, corresponding to 18.4 mL/ kg body weight (IQR: 16.1, range: 25-330). After surgery, patients spent a median of 3 days on the intensive care unit. During hospitalization, five patients developed complications, with respiratory complications being the most common (Table 2). Postoperatively, five patients newly developed a hemiparesis, and seven infants showed an overall worsening of their motor function. Only patients who underwent hemispheric surgery developed hemiparesis (P = 0.0005) and showed worsening of motor function (P = 0.021). Four of the seven patients with worsened motor function improved in the postoperative course, while one did not show an improvement. In two patients it was not possible to evaluate a putative improvement of motor function due to their young age at surgery.
Outcome after epilepsy surgery in the first 6 months of life was evaluated at a median age of 1.9 years (IQR: 1.6), with a follow-up of 1.5 years (IQR: 1.8). Four patients showed early seizures, between 7 and 14 days after surgery. Eight patients where seizure-free (57%) and two patients only had up to three seizure days a year (ILAE Class 3) ( Figure 1C). Early seizures were significantly associated with persistent seizures (P = 0.003), resulting in reoperations beyond the first 6 months of life in three patients. In one patient, the hemispherotomy was extended to an anatomical hemispherectomy at 10 months of age. One patient received a stereotactic commissurotomy twice at 1.7 and 2.4 years of age. In the third patient, the parieto-occipital lobectomy was extended to a hemispherotomy at the age of 6 months and 20 days. 79% of patients were seizure-free at the time of the study at a median age of 2.6 years ( Figure 1C). Further, ASM could be reduced in nine patients (64%) from a median of three ASM preoperatively to a median of 1.5 ASM postoperatively (P = 0.013). In two of these nine patients, ASM treatment was discontinued ( Figure 1D). One patient developed a hydrocephalus after hemispherotomy which was treated with a ventriculoperitoneal shunt.
The attending physicians reported cognitive progress based on subjective assessment in all patients. In four patients, cognitive development could be assessed using the same test before and after surgery at a median postoperative follow-up time of 12 months (IQR: 7.25, range 6-35). Median DQ values declined from 79.5 preoperatively to 57.5 postoperatively (P = 0.427). Later retesting was associated with a greater increase in DQ points after surgery (r = 0.95, P = 0.051). Individually, two patients lost DQ points, one patient gained and one stagnated. However, patients who lost or stagnated all showed an increase in raw scores from a median of six (IQR: 8.5, range 3-20) to a median of 30 (IQR: 13.5, range . This indicates that these patients individually showed cognitive gains, but at a slower rate than their healthy peers.

| DISCUSSION
Here, we report the outcome of 15 infants with intractable epilepsy operated in the first 6 months of life. In children with DRE, epilepsy surgery is superior in achieving seizure freedom over further chronic ASM treatment, with postoperative seizure freedom in about two thirds of patients depending on etiology and surgical techniques. [11][12][13] We recently reported a seizure freedom rate of 66% after epilepsy surgery in the first 6 months of life in a metaanalysis including 158 patients matching the rates reported in older children and adults. 8 In this study, 57% of the children were seizure-free after a first epilepsy surgery. Most patients in the study had focal motor seizures, while few had epileptic spasms. Early surgical treatment of epileptic spasms is rarely performed but can result in a high rate of seizure freedom. 14 However, it is often difficult to narrow down a putative epileptogenic zone. However, a significant number of patients with epileptic spasms have an identifiable lesion on MRI, which is associated with a favorable outcome after epilepsy surgery. 14,15 In this study, 60% of patients with epileptic spasms were seizure-free after one surgery, adding to the growing body of evidence supporting the effectiveness of epilepsy surgery in the treatment of epileptic spasms. There is a general reluctance to operate infants in the first months of life out of fear of anesthesiologic and surgical complications due to low bodyweight, blood volume, and the often more extensive surgical interventions offered at that age when compared to older children with more localized/demarked lesions resulting in only a small number of reports. The major hypothesized benefit of early epilepsy surgery is that these patients will show improved cognitive development due to the short disease duration compared to later surgery, as longer duration of epilepsy is associated with poorer postoperative outcomes and results in cognitive deterioration and impaired quality of life. 16 In this line, it has been proposed that early seizure control and, thus, a short epilepsy duration and a reduction or discontinuation of ASM allow infants to develop to the best of their abilities postoperatively. [17][18][19] In line with previous research, all infants in the present study had belowaverage DQ values before surgery. 8 DQ values declined but raw scores increased, indicating that these patients individually develop, but at a slower rate than average. This further strengthens the evidence that raw scores must be evaluated in patients after epilepsy surgery to assess individual development. 20 In addition to short epilepsy duration, many modifiable and non-modifiable factors influence potential cognitive development after surgery. Epilepsy duration is the only preoperative modifiable parameter. 19,21,22 However, many factors of neurodevelopment are not modifiable, such as the epileptogenic pathology itself. 23 Children with malformation of cortical development have lower postoperative IQ/DQ. 24 Low-grade epilepsy-associated tumors show a positive correlation with the postoperative IQ and perinatal stroke/porencephaly, encephalitis are associated with a decrease as well as multilobar surgical interventions. 22 Contralateral MRI abnormalities result in lower postoperative ID/DQ and less gain. 25 As the preoperative IQ/DQ results from these (non)-modifiable factors, the IQ/DQ value itself is a strong predictor of postoperative development. 20,22 (Post)operatively, complete resection is associated with greater postoperative improvement as well as reduction in ASM. 22,26 However, genetic/environmental background such as parental education also plays an important role. Children whose parents had higher education show greater improvement, demonstrating that IQ change is also determined by environmental and genetic mechanisms. 27 In addition to cognitive development, functional (motor) outcome is important, especially since most children underwent hemispheric surgery. In our study, most patients showed improvement in motor function after developing hemiparesis. Overall, gross motor function remains stable or improves after pediatric epilepsy surgery. 28 However, fine motor function of the hand is particularly important for daily living. To date, clinical predictors of fine motor development are lacking and require the use of focal transcranial magnetic stimulation or MRI analysis of corticospinal tract asymmetry. 29 In general, it can be assumed that patients with post neonatal or progressive lesions and brainstem symmetry will lose grasping abilities. 29 Whether early hemispheric surgery leads to potentially better gross and fine motor outcomes due to a greater brain plasticity is unclear and cannot be extrapolated from the data of this study due to the high heterogeneity and small number of patients, which requires further studies prospectively looking at fine and gross motor development after early epilepsy surgery.
As mentioned, the most common causes of neonatal seizures are hypoxic-ischemic encephalopathy and intraventricular hemorrhage. 3 However, here we report that the most common etiology in the operated infants was malformation of cortical development. This seemingly paradoxical finding may be due to several factors, including the limited sample size of our study but most probably a selection bias of infants who underwent epilepsy surgery. In addition, a recent retrospective analysis of 64 infants reported cortical dysplasia and hemimegalencephaly as the most common etiologies. 30 This is consistent with a previous study that analyzed the histopathological findings of 2623 pediatric patients in which cortical malformations were the most common etiology overall. 31 It remains, however, unclear why infants with vascular etiologies were not operated on at a higher rate. Potentially, children with vascular etiology are evaluated for epilepsy surgery or referred to epilepsy surgery centers later. However, this paradoxical finding requires further investigation, especially analyzing selection of potential epilepsy surgery candidates in infancy and referral of potential patients to epilepsy centers.
The results of this study are promising; however, they cannot be generalized to all infants with structural DRE. The operated patients all had severe epilepsy with a high seizure burden despite many ASM, and many had status epilepticus. In these patients, the risks of the uncontrolled epilepsy were weighed against the potential benefits and risks of surgery. Two thirds of infants received intraoperative RBCs and 29% later. This rate for intraoperative RBCs is lower than the 98% calculated in our recent metaanalysis. 8 However, this pooled rate is mostly based on studies published more than one decade ago and infants undergoing epilepsy surgery in the first 3 months of life. In our cohort one patient died intraoperatively.
Overall, epilepsy surgery is a safe procedure, but this cannot be extrapolated to infants without caution. Infants and neonates have a much higher rate of critical cardiovascular and respiratory events compared to older pediatric patients undergoing anesthesia. 32 The fragile brain tissue, smaller situs, and low body weight combined with low blood volume increase the surgical and anesthetic risks. 8 At this age, infants also exhibit a decrease in hemoglobin levels, which independently increases the risk of cardiovascular complications. Further prematurity combined with lower body weight increases the risk of complications. 32 Despite these surgical risks and overall low body weight, the infants in this study underwent surgery because their epilepsy was severe and potentially lifethreatening. It remains unclear whether delaying surgery until the infant has reached a higher body weight may mitigate some of these surgical risks. In addition, the potential adverse effects of continued epileptic activity and the long-term effects of polypharmacy in these patients must be weighed against the surgical risks. Because of these risks, other therapeutic options, such as treatment with novel ASM, ketogenic diet, or vagus nerve stimulation, should be considered in parallel for infants with relatively low seizure frequency. [33][34][35] However, all these treatments are not curative. Therefore, infants with lesional DRE should always be evaluated at a specialized epilepsy center with experience in infantile epilepsy surgery.

| CONCLUSION
Here, we show that early epilepsy surgery in infancy results in good seizure control, intraoperative complications occurred in 13% patients including mortality in one patient. For this reason, patients undergoing epilepsy surgery in infancy should only be treated at epilepsy centers specialized in infants with DRE, where experienced neurosurgeons, anesthesiologists and a pediatric intensive care specialist team are available in addition to the pediatric epileptology team. Infants with structural DRE should not be denied the evaluation for surgery or epilepsy surgery based on age. However, the benefits must be weighed of against the potential risks due to early epilepsy surgery.

AUTHOR CONTRIBUTIONS
Konstantin L. Makridis, Ingo Borggraefe, and Angela M. Kaindl contributed to the conception and design of the study. Konstantin L. Makridis, Kerstin Alexandra Klotz, Georgia Ramantani, and Ingo Borggraefe contributed to acquisition of data. Konstantin L. Makridis organized the database, analyzed the data, wrote the first draft of the manuscript, and created tables and figures. All authors discussed the results, revised the first draft, and approved the final manuscript.