The Predictors of Seizures in Patients with Encephalocele: An 11-Year Experience from a Tertiary Hospital

Introduction: The aim of the study was to investigate and identify the predictors associated with the incidence of seizures in patients with encephalocele (EC). Methods: A retrospective analysis was undertaken of patients treated for EC at a tertiary medical center in Tehran between 2010 and 2021. Data including age at presentation, gender, location, size, and content of EC, ventriculomegaly, hydrocephalus, associated anomalies, and neurodevelopmental delay (NDD) were evaluated for their prognostic value. In addition, univariate and multivariate analyses were performed to identify the correlation between independent predictors and seizure incidence. Results: One hundred and two cases of EC were identified. Seventy-one ECs (69.6%) were posterior ECs, while 31 (30.4%) were anterior. Neural tissue was found in 43 (42.2%) of the ECs. Thirty-three patients (32.4%) had ventriculomegaly, of which 90.9% underwent shunt placement for progressive or symptomatic hydrocephalus. Seizure was found in 26 (25.5%) patients. On univariate analysis, presence of other anomalies, postoperative infections, and NDD were associated with seizures (p < 0.05). When the anomalies were categorized into intracranial and extracranial groups in univariate analysis, none was associated with statistically significant increase in seizure (p values of 0.09 and 0.61, respectively). Although according to multivariate analysis, only the association between other associated anomalies and seizure was near significant (OR: 2.0, 95% CI: 0.95–4.2, p = 0.049). Children with NDD and postoperative infection were, respectively, 3.04 and 1.3 times more at risk to experience seizures compared to other patients. Conclusion: We found a rate of 25.5% risk of seizure in patients with EC. This study could not find any significant predictors of seizure in children with EC. However, pediatric patients with postoperative infections including sepsis, wound infection, and NDD require more consideration to reduce the risk of seizure.


Introduction
Encephaloceles (ECs) are classified according to their location [1].The prevalence of EC is estimated to be 0.8-4 in every 10,000 live births [2,3].Although rare, the association between ECs and seizure is well established in some patients with seizure syndrome including focal and generalized tonic-clonic seizures [4,5].These ECs are diagnosed only by neuroimaging including either computerized tomography (CT) scan or magnetic resonance imaging (MRI) to confirm a bony defect on the CT scan or differentiate the EC content in MRI [6,7].Although EC has been discovered in 0.3% of the MRI performed for newly diagnosed or unprovoked seizures [8], neuroimaging, particularly MRI T2-weighted sequence, has high sensitivity for all types of ECs, even the small ones associated with temporal lobe seizures [9][10][11].Additionally, Campbell et al. [12] observed that application of 3T MRI increased the detection rate of temporal ECs in patients with seizures.
Literature has shown that ECs associated with seizure consisted mainly of case report and series of seizures associated with temporal ECs [5, 12, 13, 15-18, 21, 27].Temporal ECs include anterobasal, lateral, or medial variants.Anterior and lateral temporal ECs can be associated with cerebrospinal fluid (CSF) otorrhea but rarely seizures [5,8,9,14,18,23,32].Contrarily, medial temporal ECs commonly present with seizures/epilepsy that are often intractable to medications and mostly require surgical intervention [16,17,[33][34][35][36].Interestingly, ECs of several locations including the motor cortex [30], frontal [37], or occipital lobe [3,22,24,29] have also been reported to cause seizures.However, besides the vast knowledge about surgical repair for temporal EC associated with intractable seizures [6, 11, 13, 15-17, 20, 21], little is known about the proper management of seizures in children with EC before and after EC resection.Also, till date, there has not been any study that sought to find predictors of seizure in pediatric patients with EC, even though seizures have been reported in large studies [2,3,24,31].Prediction of seizure development in children with ECs could provide a more realistic perspective for their caregivers and help physicians tailor their follow-up sessions.In this study, we reviewed our series of patients with EC and analyzed the predictors associated with seizure incidence.The management, follow-up, and overall outcome are presented.

Methods
All patients with EC who underwent operations in the pediatric neurosurgery department of Children's Medical Center, Tehran, between the years 2010 and 2021 were enrolled in this study.The demographic information of 102 patients treated for ECs was retrospectively collected.Information about EC (location, size, neural tissue in the sac), ventriculomegaly/symptomatic hydrocephalus, associated anomalies, CSF shunting, seizure, meningitis, postoperative infection (sepsis, wound infections), neurodevelopmental status [3], and final status (alive or dead) was gathered.The diagnosis of seizures was made clinically when a child showed signs and symptoms of seizures.Ventriculomegaly was defined based on brain imaging.The diagnosis of hydrocephalus was also primarily based on imaging including ultrasound, brain CT scan, or MRI associated with clinical signs and symptoms of intracranial hypertension including unusual enlargement of head size, persistent vomiting, bulge and tens fontanel, or sunset eye.
The presence of neural tissue in the EC was confirmed by intraoperative findings, histopathological reports, or imaging reports.The EC sac was categorized into two parts: anterior and posterior.The anterior lesions included basal (sphenoorbital, sphenomaxillary, and trans-sphenoidal) and sincipital (nasofrontal, nasoethmoidal, nasoorbital), while the posterior lesions included temporal, parietal, occipitocervical, and occipital ECs.After surgery, all patients were followed up in the outpatient clinic.During clinical visits, all patients were examined for signs of hydrocephalus (abnormal increase of head circumference, sunset eye, bulge, or tense fontanel) and neurodevelopmental status.Also, a pediatric neurologist examined patients with a history of seizures and managed them with regular clinical visits, electroencephalography, and antiepileptic medications.Follow-up data were documented in patients' medical files.Besides the collection of data from the medical records of patients, patients were also interviewed for additional data about seizures and the NDD status of the children (according to the Center for Disease Control and Prevention/CDC developmental milestone scores [3]) through interviews in the clinic or phone calls to their parents.

Statistical Analyses
Statistical analyses were performed using SPSS statistical software (version 25.0 for Windows, IBM Corp.).Continuous variables were presented as median ± standard deviation, and dichotomous variables were expressed as percentages.Categorical variables were defined as frequency, constituent ratio, and crosstab.Cross-tabulation analyses and univariate analyses of various predictors of seizure incidence were accomplished.Variables of statistically significant associations with seizures were included in multivariate logistic regression.The exact logistic regression was used to estimate and test the association between independent variables and seizures.The means of continuous variables were compared between seizure and non-seizure groups using independent sample T test.All statistical tests were 2 sided, and data were deemed to be statistically significant when p < 0.05.

Demographic and Clinical Characteristics
Among 102 patients involved in this study, 52 were male.Patients aged from 1 day to 7.5 years at the time of surgery had a median age of 4 months.Seventy-one patients (69.6%) had posterior ECs, while 31 (30.4%) had anterior ECs.Seven children were born preterm (6.9%) (Table 1).

Demographic and Seizure Outcomes
Twenty-six (25.5%) patients had seizures, which occurred before surgery in 2 and after EC resection/ postoperative in 24 patients.Male children (61.5%) had more seizure episodes than females (38.5%).Sixteen (61.5%) of the seizure cases had posterior ECs, while 10 (38.5%) were associated with anterior ECs.Majority of seizures were observed between 2 and 6 months of age (26.9%) and between 6 and 12 months (26.9%), respectively.History of prematurity was found in 2 (7.7%) of 26 patients with seizures.Presence of prematurity did not have a statistically significant association with seizure occurrence (p = 0.22).Interestingly, most seizures occurred in the patients with ECs size below 5 cm.The difference between the presence and absence of other anomalies was marginal in terms of seizure development (p = 0.01).Categorizing the anomalies into extracranial and intracranial anomalies, a p value of 0.61 for extracranial and 0.09 for intracranial anomalies was calculated that was not statistically significant.For postoperative seizure, 15 (57.7%) patients had seizures after EC repair, 9 (34.6%) had seizures after ventriculoperitoneal shunting for progressive ventriculomegaly with symptomatic hydrocephalus, and 2 (7.7%) seizures occurred after an endoscopic procedure for anterior ECs.Among 6 (5.9%) patients with meningitis, 3 had seizures.However, 12 of the 25 patients with postoperative infection had seizures (Table 2).
Neurodevelopmental status was evaluated with CDC milestones in 94 patients, which revealed 17 patients (18.1%) had severe NDD, 14 (14.9%) had mild/moderate delay, and 63 (67%) had a normal neurodevelopmental outcome.Eight (30.8%) of 17 patients with severe NDD had seizures, while 4 (15.4%) of 14 with mild to moderate NDD found seizures, and 14 (53.4%) of 63 with normal neurodevelopment had seizures.This difference was statistically significant (p = 0.011).The median follow-up according to the last clinical visit was 9 months, while the median follow-up by phone calls was 96 months.At the last follow-up visit, seizures in 19 (70.4%) children were controlled (Engel class 1 and 2) with antiepileptic medication, and 1 (3.7%) patient was dead.Seven (25.9%) patients had lost to follow up in terms of seizure control.The overall survival rate of 102 ECs was 87.3%.Five (4.9%) patients in the total population were dead during followup, while 8 (7.8%) patients were lost to follow up (Table 3).

Cross-Tabulation Analyses of Various Predictors of Seizure Outcomes
Table 2 illustrates the cross-tabulation analyses between predictors and seizure incidence.The results of univariate analysis showed no correlation between age at surgery (p = 0.165), gender (p = 0.154), time of delivery (p = 0.662), EC location (p = 0.213), size of EC sac (p = 0.284), ventriculomegaly (p = 0.126), meningitis (p = 0.171), CSF leak (p = 0.446), type of surgery (p = 0.357), and seizure development.However, according to univariate analysis, there was a strong correlation between postoperative infection and seizures (p = 0.004).Of 27 patients who developed seizures, 14 (53.8%) had neural tissue inside the sac, while 5 (19.2%) did not.χ 2 test showed a marginal correlation between presence of neural tissue in the sac and seizure (p = 0.067).Although only 14 of 39 patients with other anomalies had seizures, this correlation was statistically significant (p = 0.01).Additionally, NDD correlated with seizure development (p = 0.011).

Logistic Regression Analysis of Predictors with Seizures
Considering seizure as a dependent outcome, all variables were reanalyzed using the univariate and multivariate models.The analyses were adjusted for age at surgery, gender, term delivery, posterior location, EC sac size, NDD, ventriculomegaly, associated anomalies, meningitis, postoperative infections, and CSF leak.Results of the univariate model indicate that the presence of other anomalies, postoperative infection, and NDD were associated with an increased risk of seizure.Again, the three variables were inputted into a multivariate model.The multivariate model showed that only the presence of other anomalies was statistically near significant associated with seizure (OR: 2.0, 95% CI: 0.95-4.2,p = 0.049).Categorizing the associated anomalies into intracranial and extracranial was performed to evaluate their importance for seizure occurrence in a multivariate model.Neither extracranial nor intracranial anomalies in patients with EC had a statistically significant association with an increased risk of seizure (p = 0.61 and p = 0.09, respectively).Although not statistically significant, children with postoperative infection were 3.04 times more likely to have seizures (95% CI: 1.07-8.58;p = 0.17), while children with NDD were 1.3 times (95% CI: 0.9-4.2;p = 0.49) more likely to develop seizures (Table 4).

Discussion
There are previous series that have documented seizures in children with ECs [2,3,31,38,39].However, this is the first retrospective large series that has sought to investigate risk factors for seizures in patients with ECs.According to univariate analysis, our results showed that postoperative infections, the presence of neural tissue inside the ECs, other associated anomalies, and NDD were significant or marginally significant predictors of seizure development.However, in a multivariate model, none of them was a significant predictor of seizure occurrence.
Other evidence suggests that hydrocephalus and its management could also provoke seizures.Almost half (42.3%) of the patients with seizures in this series had progressive ventriculomegaly/symptomatic hydrocephalus or underwent shunting for hydrocephalus treatment.Particularly, our data showed that 9 patients had tonicclonic seizures after shunt placement for progressive hydrocephalus.Additionally, Leelanukrom et al. [40] reported three EC cases in which hydrocephalus resulted in tonic-clonic seizures.However, we found no correlation between ventriculomegaly/hydrocephalus and seizure occurrence (p = 0.126).Contrarily, Da Silva et al. [31] and Lo et al. [2] in a univariate analysis showed strong correlations between hydrocephalus and seizures (p = 0.001).Seeburg et al. [39] reported two cases in which one premature hydrocephaly infant developed seizures 1 month after shunting.Bui et al. [38] argued that the presence of hydrocephalus and seizures/epilepsy can independently and significantly lower the overall prognosis of children with ECs.However, the constellation of associated brain anomalies in ECs, progressive hydrocephalus, resection of dysplastic brain inside the sac, shunting procedure, and infection can explain the pathophysiology of postoperative seizures.Notwithstanding, further studies are needed to provide concrete evidence to support this explanation.

Anomalies, Seizures, and NDD
As mentioned earlier, seizures may also occur as a presenting or preoperative symptom.In this cohort, only two seizures were found preoperatively.The first child was a term female infant who began having seizures a few weeks after birth.Subsequent evaluation showed progressive hydrocephalus with Dandy Walker malformations associated with an occipital EC.EC resection was performed, and VPS was inserted for hydrocephalus.Although her seizure was under control, a 72-month follow-up showed severe NDD.The second child was a male microcephaly neonate who had a huge occipital EC of 10.5 cm at birth.Neuroimaging confirmed neural tissue inside the sac, Chiari malformation, and corpus callosum agenesis.On the third day of birth, the patient developed a seizure and surgical team decided to perform shunting before EC resection.Again, 120 months of follow-up revealed severe NDD in this patient.
Our univariate analyses confirm a correlation between the presence of other anomalies, presence of neural tissue in the sac, postoperative infections, and NDD with seizure occurrence.However, on the multivariate model, only the association between the presence of other anomalies and seizures was statistically near significant (p = 0.049).Subcategorizing the associated anomalies as intracranial and extracranial showed no statistically significant association with seizure.We observed that children with NDD were 1.3 times more likely to have recurrent seizures even with antiepileptic medication.In turn, seizures can further deteriorate NDD outcomes and negatively impact the overall health of the affected child [38].Consistent with our finding, Lo et al. [2] showed that patients with seizures had a 4.09 more risk of developing NDD.Rehman et al. [29] also reported a patient with NDD who developed a seizure.According to them, although the seizure was well controlled, the patient's developmental milestones showed no improvement [29].Contrarily to the above findings, Da Silva et al. [31] found no association between seizures and NDD (p = 0.12).However, there are several studies that have reported patients with ECs and associated anomalies who developed seizures and NDD [22,41,44].

Association between Other Predictors and Seizures
The mean EC size in seizure (N = 26) and non-seizure patients (N = 76) in our study was 3.4 ± 2.5 and 4.1 ± 3.3, respectively.This difference was not statistically significant according to the independent sample T test (p = 0.286).Pettersson et al. [10] in their series characterized MRI features of middle cranial fossa ECs in seizure and non-seizure patients.Similar to our findings, they observed that the EC size and degree of parenchymal morphologic distortion may not be useful in predicting the likelihood of seizure occurrence [10].Our present results show a higher prevalence of seizure incidence in male children.However, this was not statistically significant (p = 0.154).Temporal EC is known to cause seizures [12, 13, 15-18, 21, 27].Also, 61.5% of patients with posterior lesions (with only 1 case of temporal EC) in this cohort had seizures.However, EC sac location did not significantly correlate with the occurrence of seizures in the affected patients.Our data also showed no correlation between time of delivery, age at surgery, and seizure development.Particularly, the mean age at surgery between patients with seizures (N = 26; mean age at surgery of 12.8 ± 14.7 months) and those without seizures (N = 76; mean age at surgery of 10.2 ± 16.6) were not different (p = 0.484).Moreover, we have shown that the type of surgery (open or endoscopic surgery) does not predispose the patients to develop seizures.
Management, Follow-Up, and Prognosis of Seizures In this study, seizures were managed mainly with antiepileptic medication.Seizure control was achieved in 70.4% of cases, with only 1 patient died from seizure related complications upon follow-up.Apart from Mahapatra and Agrawal who documented intractable recurrent seizures in four children in their series in spite of antiepileptic medication [23], the overall prognosis of seizures in patients with EC is favorable [24,29].Because postoperative seizures can develop in patient with no preoperative seizures some authors suggested administration of antiepileptic medications before surgery [40].Additionally, proper and adequate treatment of postoperative infections is mandated since this complication could trigger seizures development.The overall survival rate of children with EC has been shown to be high [2,24,31].However, adequate management for worse outcomes like seizures and hydrocephalus are needed to reduce NDD and improve the survival.

Limitation
The main limitation of this study was missing data.For instance, although 102 patients were included in this study, data of histopathological diagnosis of the presence or absence of neural tissue were available in 74 cases only.We attribute the incomplete patients' data to the retrospective design of the study.We admit that the final follow-up clinical evaluation through phone interviews was limited as the parents' responses might be overestimated or underestimated according to their expectations, social background, or level of education.In spite of these limitations, we believe the findings of this retrospective study will be relevant to the literature and provide pertinent information for the neurosurgical management of patients with ECs.

Conclusion
This is the first large retrospective cohort that has evaluated seizure outcomes and investigated the effective factors.Univariate analysis showed a statistically significant correlation between postoperative infections, associated anomalies, the presence of neural tissue in the EC sac, NDD, and seizure incidence.However, in the multivariate model and dividing the anomalies into intracranial and extracranial groups, none of them found significant association with seizures.Children with postoperative infections had a 3.04 times higher risk of seizure, while those who had NDD had a 1.3 times higher risk of developing seizures.

Statement of Ethics
This study was approved by the Institutional Ethics Committee of the Iran University of Medical Sciences (ethics code: IR.-IUMS.FMD.REC.1401.460).The study is a retrospective study, in which data were drawn from the medical records of encephalocele patients admitted to our hospital.In their medical record, we have taken a medical consent from their parents that their data (with respect to their privacy and confidentiality) is used for research.If it is needed, we can send you copy of the informed consent form signed by parents in Farsi language to the journal.

Table 3 .
Follow-up outcomes according to duration, management of hydrocephalus, seizure, and overall survival

Table 4 .
Comparison of seizure predictors in patients with EC according to univariate and multivariate analysis