Introduction

Awake craniotomy has been used for several decades for the resection of intraaxial lesions near eloquent brain regions [1, 2]. When epileptic tumors are localized in eloquent areas of the brain, a multimodal integrated diagnostic approach may help maximize the extent of resection while preserving cerebral function [3].

Seizure is one of the most commonly reported complications associated with awake craniotomy [1, 4, 5]. Brain manipulation and electrical cortical stimulation used for intraoperative mapping may cause seizures. Intraoperative seizures during awake craniotomy may complicate the procedure by affecting the ability to map and monitor patients. Furthermore, seizures following a craniotomy can lead to hypoxia, intracranial hemorrhage, neurological deterioration, and increased intracranial pressure [6, 7].

The etiology of brain tumor-associated seizures is still not fully understood, while pathophysiology of tumors and the recently identified tumor markers were known to be potential factors [8]. Before the surgeries, WHO grade II–III gliomas are more likely to induce epilepsy than grade IV glioblastomas [9], and one possible mechanistic link is that the former group of tumors usually have mutations in isocitrate dehydrogenase 1 (IDH1), whereas the latter usually do not [10]. However, whether IDH1 mutations play a prognostic role in postoperative seizure control is not determined yet.

Seizures significantly contribute to morbidity in patients with brain tumors. Herein we hypothesized awake craniotomies with comprehensive strategies benefit seizure control in these patients and improve functional outcomes, and we sought to figure out the prognostic factors for postoperative seizure recurrence.

Methods

The authors performed a retrospective cohort study of all consecutive patients having gliomas and with seizures as initial presentation. The patients underwent awake craniotomies for tumor resection at Chang Gung Memorial Hospital between August 2013 and March 2017. Clinical information was retrospectively obtained from the patient medical records, including radiology data and pathology specimens. This study was approved by the Chang Gung Medical Foundation Institutional Review Board. Informed patient consent for the study was waived due to the retrospective nature of the study. To ensure the analysis of de novo epileptic brain tumors in patients who underwent awake craniotomies, patients with confounding neurosurgical histories and those with tumor recurrence following previous treatment were excluded from the study.

Patients selected for awake craniotomies

Whether the patients’ tumors located within an eloquent area was determined by two neurosurgeons through preoperative MRI scan. Cooperative, consenting patients, with some potential for monitoring were selected to undergo an awake procedure after neuropsychologic assessment. Patients who had medical comorbidities prohibiting awake procedures, severe dysphasia prohibiting any cooperation, or significant preexisting anxiety were excluded from the candidates for awake craniotomies.

Medical management for seizure control

Seizures were classified as recommended by the latest International League Against Epilepsy (ILAE) [11]. All patients were initially treated with first line antiepileptic drug (AED) monotherapy. Levetiracetam and valproate acid were both started at a maintenance dose of 1000 mg. Patients who did not respond to either levetiracetam or valproate acid alone received an additional AED for dual therapy. Through the AEDs treatment, all patients’ seizures were under control before the surgeries. Any patient whose seizure was active and intractable would not be suggested to have awake craniotomies. Immediate postoperative AED therapy was identical to the preoperative regimen unless intractable seizures recurred in the early postoperative stage. In cases of ongoing early seizures, dual therapy was initiated rather than increasing the dose of the initial agent and was continued for a minimum of 6 months. AEDs serum levels were checked every 3 days during hospitalization and whenever patient returned to the hospital for follow-up. Dosing was adjusted to obtain a normal serum level.

Anesthetic management: asleep–awake–asleep protocol

Total intravenous anesthesia (TIVA) with propofol and fentanyl infusions was used for the first asleep phase in all patients to regulate the depth of anesthesia, along with a tube in the nasal airway facilitating ventilation. The scalp was injected with local anesthetic (20 ml 2% lidocaine with 0.1 ml 1% epinephrine) before the skin incision. After the craniotomy the dura mater was infiltrated with 2% lidocaine using a 30 gauge needle tip on a 1 ml syringe.

Once the dura mater was opened, TIVA was paused and the patient was allowed to wake up naturally. The awake phase of the procedure concluded once functional mapping and tumor resection was achieved. General anesthesia with TIVA was then resumed until the end of the operation.

Intraoperative stimulation mapping (ISM) and linguistic tests

This procedure was performed as previously described [12]. Briefly, when a patient awoke, number counting combined with continuous movement of the upper limbs was utilized to determine the optimal intensity of ISM. The intensity of electrical stimulation was increased gradually until oral twitching, speech arrest, dysarthria, or the motor response of a limb was observed, and electrical stimulation at this magnitude of current was used for subsequent stimulation. Specifically, a bipolar electrode with 5 mm spaced tips was applied to deliver biphasic current stimulation (pulse frequency, 50 Hz; single pulse phase duration, 1 ms; amplitude, 4–7 mA; Model OCS2 Ojemann Cortical Stimulator, Integra LifeSciences Corporation, Saint Priest, France).

Number counting was applied as a visual object naming test and Pyramids and Palm Trees Test (PPTT) was used as a semantic test of association for intraoperatively linguistic testing and functional monitoring [12]. When the patient revealed naming errors or incorrect responses in PPTT, the neurosurgeon was alerted and tumor excision was stopped to preserve the patient’s speech function.

Intraoperative seizure

When seizures were clinically observed or there was epileptiform activity on the electrocorticography (ECoG), cortical or subcortical stimulations was stopped immediately and the brain was irrigated with cold normal saline solution (0.9% NaCl) until the seizures ceased. AEDs or sedatives bolus giving was avoided where possible to allow for uninterrupted functional mapping and clinical evaluation. Once the patient was again able to perform cognitive tasks, resection was resumed.

Postoperative management and follow-up

All patients were sent to the intensive care unit for postoperative monitoring. The modified Rankin Scale [13] (mRS) is a neurological functional assessment, which uses the same tests as those administered preoperatively; it was performed at around the third day after the surgery. The length of hospitalization was defined as the period from the date of surgery to the date of discharge.

The term “post-operative early onset seizure” (early seizure) was applied to a seizure that occurred within 7 days after the craniotomy. “Post-operative late onset seizure” (late seizure) was defined as a documented seizure from the 8th day after the surgery to any later time during follow-up.

The ILAE classification system [14] was used to evaluate late seizure outcomes following surgery. The outcomes were considered Class 1 and 2 if the patient was seizure free or only had auras; Class 3 and Class 4 if at least one active seizure occurred during follow up but still 50% reduction of baseline seizure days. Class 5 and 6 if no response to surgery, which meant < 50% reduction of baseline seizures to > 100% increase in seizure days. Seizures were diagnosed via observation, clinical examination or electroencephalography (EEG).

Statistical analysis

Descriptive statistics were given as the mean ± standard deviation (SD) for continuous variables and the frequency distribution for categorical variables. Univariate analysis was performed to determine the association between each variable and the seizure group. The Pearson 2 test or Fisher’s exact test were used to compare the groups with respect to categorical variables. A two-sample t test or Mann–Whitney U-test was performed to compare two groups with respect to variables. Only significant variables in the univariate analysis were included in the multivariate regression analysis. All statistical analyses were performed using SPSS software (version 19; IBM Corp., Armonk, NY, USA). P < 0.05 was considered to indicate a statistically significant difference.

Results

Patients

Between August 2013 and March 2017, a total of 41 patients with epileptogenic gliomas at the eloquent brain area, received awake craniotomies. Demographic and clinical data are presented in Table 1. Thirteen patients (31.7%) had focal seizures, and 19 patients (46.3%) had generalized seizures preoperatively. Tumors involved solely frontal lobe were 56.1% of all patients, while others involved in frontotemporal (14.6%), frontoparietal (19.5%) or insula area (9.8%). The median preoperative tumor volume was 23.0 cm3 (range 1.1–78.2). The histopathology of all the tumors was glioma and they were stratified into WHO grades II (n = 19), III (n = 12) and IV (n = 10). The average proportion of removed tumor volume was 94.5 ± 12.5%. IDH1 status was obtained in 33 tumors of this cohort, and 20 were IDH1 mutation (60.6%). There was a significant difference between the mean mRS before (2.4) and after (2.1) the awake craniotomies (P = 0.032). The average follow-up time was 11.2 months (range 6–12 months). One of the patients was demonstrated in Fig. 1.

Table 1 Patient demographics
Fig. 1
figure 1

A 29-year old male patient had focal seizure with impaired awareness, followed by temporary right hemiparesis. a Preoperative T1 weighted with contrast medium MRI revealed an intra-axial tumor at left superior frontal gyrus invasion to left precentral gyrus (arrow). He had intraoperative seizure, but the awake craniotomy was undergone successfully. b Postoperative MRI showed grossly total removal of the tumor (arrow), and c pathology exam with hematoxylin and eosin stain showed uniformly rounded nuclei and clear haloes imparting a classical “fried egg” appearance, indicating the diagnosis of oligodendroglioma (grade II glioma). d Immunochemistry stain revealed the positive granular cyto-plasmic demonstration of tumor cells for mutant IDH1 protein. This patient had postoperative mRS 2 scores and no early nor late seizure recurrence postoperatively

Intraoperative seizures

In three patients (7.3%), subcortical stimulation mapping elicited an intraoperative seizure. The tumors of the three patients all located within frontal lobes. However, none of the seizures were unmanageable by immediate action. Accurate functional mapping and monitoring were not affected and the awake craniotomies were all completed successfully. One of the three patients had a postoperative early seizure, however no late seizures were observed in these patients. Intraoperative seizures were not significantly associated with the postoperative reccurrence of early (P = 0.657) or late seizures (P = 0.512).

Postoperative early seizure

A total of 14 (34.1%) patients had early seizures. There is no difference in tumor location or preoperative tumor volume between patients with and without early seizures (P = 0.513 and 0.374, respectively). The histopathology of tumors was associated with seizure recurrence, as grade III gliomas had the highest incidence rate (58.3%), followed by grade II (31.6%) and grade IV gliomas (10.0%; P = 0.046). Surgical resection to gliomas with IDH1 mutation contributed to better seizure reduction rate than to those with IDH wild type (75.0% vs. 38.5%, P = 0.041; Table 2). Notably, there was no difference in percentage of tumor removal volume postoperatively between these two IDH1 subgroups (95.8% vs. 92.9%, P = 0.564). Univariate regression analysis revealed grade III gliomas (odds ratio [OR] 4.40; 95% confidence interval [CI] 1.06–18.36; P = 0.043), tumor ≤ 23 cm3 (OR 3.64; 95% CI 1.11–14.60; P = 0.042), and IDH1 wild type (OR 4.80; 95% CI 1.06–21.68; P = 0.041) were significant factors correlated to early seizure recurrence postoperatively. Patient’s sex, age, location of tumor or surgical extent were not risk factors (Supplemental Table). However, all the factors were not statistically significant in multivariate analysis (Table 4).

Table 2 Early seizures after awake craniotomies

The most frequently prescribed initial AED was valproate acid, which was given to 21 patients, followed by levetiractam (11 patients) and phenytoin (4 patients). A total of five patients had dual therapy with valproate acid and levetiracetam as their seizures were intractable to the monotherapy. There was no significant difference in early seizure control between the four AED regimens (P = 0.138).

Analysis of the postoperative length of hospitalization revealed the average stay was 8.9 days for the non-seizure group and 14.4 days for the seizure group (P = 0.03; Table 2).

Postoperative late seizure control

At the 6-month follow-up for late seizures, 33 patients (80.5%) were considered ILAE class 1 and 2. Eight patients (19.5%) were classified as ILAE 3 and 4. There were no patients who qualified as ILAE 5 or 6. The 6-month late seizure free rate was 84.2%, 83.3% and 70.0% in the WHO grade II, III, and IV gliomas, respectively. Patients with active late seizures (2.9) had a higher average postoperative mRS compared with those without late seizures (1.9; P = 0.035). Early seizures occurrence was associated with higher rate of ILAE class 3 and 4 (P = 0.012, Table 3).

Table 3 Late seizures after awake craniotomies

There was no change in AED administered pre and post-operatively in 38 of the patients. The remaining three patients, all treated with valproate acid monotherapy, had intractable early seizures and were given levetiracetam adjuvant treatment. There was no significant correlation between the AED regimen and the incidence of late seizures (P = 0.446).

Multivariate regression analysis revealed the presence of early seizure (OR 30.75; 95% CI 1.36–80.30; P = 0.039) and a postoperative mRS ≥ 3 (OR 7.00; 95% CI 1.03–47.42; P = 0.047) were independently associated with the occurrence of late seizures (Table 4).

Table 4 Risk factors for postoperative seizures

Discussion

Glioma-related seizures are characterized by their pharmacological resistance [15, 16], and surgical resection is key to seizure freedom. The pathophysiological mechanisms of seizure recurrence after brain tumor resection appear to be multifactorial and can often be unclear. Early recurrence might reflect an incomplete resection of the epileptogenic zone or irritation of the normal cortex. Late recurrence might reflect the development of a new epileptogenic process, possibly due to an underlying epileptogenic tendency [17]. Therefore, we believe that early and late recurrence of seizures after surgery may be better considered as two distinct events. To the best of our knowledge, the present study is the first to evaluate the incidence of early and late seizures separately after awake craniotomies for Asian population. In the current patient cohort, 65.9% were early seizure free and 75.5% achieved late seizure freedom at least 6 months (IALE class 1–2). Additionally, the neurological function was improved following the awake craniotomies.

Intraoperative seizures

Previous studies reported a 4.9–20% rate of intraoperative seizures during awake craniotomies, with severe intraoperative seizures requiring intubation and induction of general anesthesia [1, 4, 18]. Nossek et al. demonstrated that the incidence of seizures during the procedure was 12.6% with 18% of these episodes resulting in conversion to general anesthesia, which results in a failed awake procedure [19]. Due to the small number of intraoperative seizures in this series, it is difficult to analyze statistically the risk factors. However, it is notable that all the three patients had tumors located solely within frontal lobe. The impact of tumor locations on intraoperative seizures occurrence need to be determined in further studies.

In our institution, the ISM followed the principle that threshold stimulation levels for mapping should be as low as possible to elicit a response without triggering seizures [20]. However, it was beyond conclusion in this study whether the current amplitude used varied significantly across patients for mapping because the maximal current was not documented routinely. On the other hand, the epileptiform activity on ECoG that led to avoid further escalation of the stimulation current was not specified, after discharge thresholds vary across the cortex though [21]. Besides, for certain cortical sites, mapping may be successful only at currents above the after discharge thresholds [22]. Therefore, ECoG does have some limitation during mapping, despite it is recommended by most to improve the reliability of mapping and to prevent intraoperative seizures [23].

Although patients who experienced any type of preoperative seizure seemed more likely to seize intraoperatively [18], our result of 7.3% intraoperative seizure rate and none awake procedure failure rate supported that the presumably higher rate of intraoperative seizure would be under control for patients with initial seizures.

Postoperative early seizures

The incidence rate of reported post-craniotomy early seizures have ranged from 1.1 to 29% [24, 25]; this is mainly due to different patient demographics and tumor characteristics across the studies. Our data revealed that patients who experienced early seizures had a significantly longer hospitalization. Following awake craniotomies, early seizures may be emotionally troubling for the patients and their families, as they raise concerns of a failed operation or surgical complications. Therefore, intensive control to early seizure recurrence would not only shorten the length of stay but also influence the quality of care during hospitalization.

Glioma grade

Histopathology has been shown to be a prognostic factor for seizure freedom postoperatively on epileptogenic brain tumors [25, 26]. In the present study, the 6-month late seizure freedom rate for low grade gliomas (grade II) was higher compared with high grade gliomas (grade III and IV). This result was similar to previously published studies [16, 27, 28]. Interestingly, grade III but not grade IV tumors were associated with the highest number of postoperative early seizures. One potential explanation for the predominance of grade III gliomas was initial recruiting criteria. As rapidly growing and invasive tumors, grade IV gliomas would cause more severe neurologic symptoms earlier than seizures, and these group of patients with grade IV gliomas might be excluded from this study consequently. The correlation between this specific histopathology and postoperative early seizures needs further investigation.

IDH1 status

Previous studies indicated that IDH1 mutant genotype is more epileptogenic than IDH1 wild-type genotype in both low grade and high grade gliomas preoperatively [10, 29,30,31]. The data from the current cohort was consistent with those literatures, since 60.6% patients with epileptogenic gliomas were IDH1 mutation. Besides, we demonstrated a better early seizure control rate after resection to IDH1 mutation gliomas than for wild type gliomas. Given similar percentage of removal volume to the two groups, it is reasonable that removal of the IDH1 mutation-related epileptic zone would control seizure better right after the surgeries. No significant difference was noted in late seizure control rate in IDH-1 mutation and wild type. The result also inferred to better seizure remission after resection to IDH1 mutation gliomas, since IDH1 mutation induce epilepticogenesis more initially.

The correlation between early and late seizures

In the current study, early seizure presentation was an independent risk factor for late seizure recurrence. At present, the definition of early and late seizure is arbitrary and the mechanism of epileptogenesis is not uniform [32]. However, early seizures have been shown to play a predictive role in the occurrence of late seizures in different structural etiologies. In traumatic brain injuries, significant risk factors for the development of seizures > 1 week after trauma, included seizures within the first post-trauma week [33]. In addition, early seizures are associated with a significant risk of late seizure occurrence in patients with acute stroke [34]. Zheng et al. reported that after epileptogenic meningioma resection, the majority of patients with early postoperative seizures exhibited late postoperative seizures [35]. Elucidating the connection between early and late seizures is important because it can help earlier establish a more accurate long-term prognosis for patients following surgery.

Functional outcome and long-term seizure control

Significant improvements in neurological function were noted in the present study when the mRS was compared before and after awake craniotomies. At least moderate disability (mRS ≥ 3) around the 3rd day postoperatively was an independent risk factor for late seizure recurrence. All patients included in the present study had seizure as an initial symptom, which causes a preoperative mRS score of at least 1. Decompression of the brain via tumor debulking and early seizure freedom decreased the mRS, while new neurological deficits after surgery and active seizure increased the mRS. In addition, the mRS score of patients did improve over time during follow up (data not shown). Delayed seizures were associated with worse long-term functional outcomes in patients with cerebrovascular insults [36], while our results suggested long-term seizure control may be correlated with improved immediate postoperative functional outcomes. The results of the present study provided evidence that an awake craniotomy could definitely benefit long-term seizure control.

Anticonvulsant therapy

Tumor-related seizures are difficult to control with AEDs before tumor resection [37]. There is varying evidence regarding the efficacy of postoperative AEDs [25, 38]. We demonstrated no significant differences in the incidence of postoperative seizures among different AEDs, and levetiracetam seems to have better seizure control as a monotherapy compared with valproate acid though. Previous studies showed that postoperative seizure prophylaxis with levetiracetam and valproate acid had equal efficacy [28, 39]. It should be noted as well that levetiracetam has been shown to have no effect on the metabolism of other drugs, including chemotherapy agents [40, 41].

The retrospective design of the present study makes it susceptible to random bias. The sample size was insufficiently powered to compare the relative impact of some variables in the patients. Data obtained from a single institute are vulnerable to selection bias. The electrophysiologic parameters during operation were not documented in this study, and this omission limited the investigation for seizure control during awake craniotomies. In addition, only clinically-diagnosed seizures were included in the study, which may underestimate the percentage of seizures in patients with epileptic brain tumors.

Conclusions

As a preoperative history of seizure is a significant predictor for seizures following tumor resection [25, 42], patients with epileptic tumors deserve increased attention. The present study suggests that resection to IDH1 mutation gliomas resulted in relatively better postoperative seizure control than to IDH1 wild type gliomas. Postoperative early seizures significantly prolonged the hospitalization period and contributed to late seizure recurrence. Better neurologic function following an awake craniotomy is beneficial to long-term seizure control.