Early postoperative seizures (EPS) in patients undergoing brain tumour surgery

Early postoperative seizures (EPS) are a common complication of brain tumour surgery. This paper investigates risk factors, management and clinical relevance of EPS. We retrospectively analysed the occurrence of EPS, clinical and laboratory parameters, imaging and histopathological findings in a cohort of 679 consecutive patients who underwent craniotomies for intracranial tumours between 2015 and 2017. EPS were observed in 34/679 cases (5.1%), with 14 suffering at least one generalized seizure. Patients with EPS had a worse postoperative Karnofsky performance index (KPI; with EPS, KPI < 70 vs. 70–100: 11/108, 10.2% vs. 23/571, 4.0%; p = 0.007). Preoperative seizure history was a predictor for EPS (none vs. 1 vs. ≥ 2 seizures: p = 0.037). Meningioma patients had the highest EPS incidence (10.1%, p < 0.001). Cranial imaging identified a plausible cause in most cases (78.8%). In 20.6%, EPS were associated with a persisting new neurological deficit that could not otherwise be explained. 34.6% of the EPS patients had recurrent seizures within one year. EPS require an emergency work-up. Multiple EPS and recurrent seizures are frequent, which indicates that EPS may also reflect a more chronic condition i.e. epilepsy. EPS are often associated with persisting neurological worsening.


Materials and methods
patients. We identified all 679 consecutive patients who underwent a craniotomy for an intracranial tumour between January 2015 and August 2017 in the Department of Neurosurgery, Evangelisches Klinikum Bethel, Bielefeld, Germany, by searching the departmental electronic database. Patients operated for medication-refractory epilepsy within the epilepsy surgery program were excluded from this analysis, i.e. all cases specifically referred to us following a presurgical work-up aiming at the establishment of a surgical concept for epilepsy rather than tumour control 10 . During the study period, we used the current (= 2009) ILAE (International League Against Epilepsy) definition of pharmacoresistant epilepsy, i.e. "… failure of adequate trials of two tolerated, appropriately chosen and used antiepileptic drug schedules (whether as monotherapies or in combination) to achieve sustained seizure freedom" 11 . clinical data. Pertinent clinical data and follow-up information were retrospectively retrieved through a chart review and entered in an electronic database. Clinical parameters included age at surgery, gender, histopathological diagnosis, surgery for tumour recurrence, extent of resection, pre-and postoperative (time of discharge) Karnofsky Performance Index (KPI), occurrence of seizures prior to operation and their frequency, type of the seizure (according to the current ILAE (International League Against Epilepsy) classification) 12 and preoperative anticonvulsive therapy. The preoperative MRI work-up was reviewed in each case and the respective tumour location was recorded including specifically the following items: supra-vs. infratentorial and intravs. extra-axial growth as well as involvement or compression of the frontal, temporal, central or insular lobe. Tumours affecting the latter cerebral structures are believed to be particularly prone to cause epilepsy 13 . EPS were defined as (generalized) involuntary movements, abnormal sensory phenomena or an altered mental status that could not otherwise be explained, occurring within 30 days post-surgery, and categorized according to the current ILAE classification 8,12,14 . Acute electroencephalograms were ordered in cases in which the diagnosis of an EPS was questionable 8 . We specifically reviewed all clinical, laboratory and in particular imaging data from all patients with EPS in order to identify their likely cause.
Maximum tumour diameter and perifocal oedema > 1 cm were assessed in all meningioma patients using axial contrast-enhanced T1-and axial FLAIR-weighted scans, respectively 4 . Oedema formation and large tumour size have been associated with preoperative seizures in meningioma patients by some authors 4,15,16 . We also reviewed all available early (< 24 h) postoperative imaging studies from cases undergoing supratentorial meningioma surgery. Specifically, we recorded extra-axial (epi-or subdural) bleeds and pneumocephalus with mass effect, cerebral contusions, and measurable (i.e. > 0.5 cm) hematomas or bleeds with mass effect in the resection cavity as well as new small/perforator and territorial infarcts. Such imaging findings may cause neurological deficits and/or have clinical consequences ranging from prolonged observation to revision surgery 17 . Of note, they would also be regarded as plausible causes for an acute symptomatic seizure 2 .
Only 1/137 (0.7%) patient with an infratentorial tumour, but 33/542 (6.1%, p = 0.007) patients with supratentorial growths suffered an EPS. EPS were seen significantly more often following surgery for tumours involving or compressing the frontal, central, temporal and/or insular lobes (32/452, 7.1% vs. 2/227, 0.9%, p < 0.001). At least some involvement of the primary sensorimotor cortex was seen in 83 patients, 4 of which experienced an EPS (4.8%). Extra-axial growths carried an increased risk for EPS (extra-axial vs. intra-axial: 23/273, 8.4% vs. 11/406, 2.7%, p = 0.001). More precisely, the risk of EPS varied significantly with tumour histology. The majority of the patients suffering an EPS underwent surgery for a meningioma (22/34, 64.7%). EPS were seen in 22/218 (10.1%) cases with meningioma, but only in 7/177 (4.0%) with glioblastoma, 2/64 (3.1%) with other gliomas (i.e. diffuse astrocytoma or oligodendroglioma WHO grades II and III, pilocytic astrocytoma, pleomorphic xanthoastrocytoma and glioneuronal tumours), 2/139 (1.4%) with metastasis, and 1/81 (1.2%) with other histologies (p = 0.001). Repeat surgery was not associated with an increased risk for EPS (Table 3). www.nature.com/scientificreports/ Next, we performed a multivariate binary logistic regression analysis. Using preoperative seizures (none, single, ≥ 2), tumour location (infra-vs. supratentorial growth and involvement/ compression of the frontal, central, temporal and/or insular lobes vs. not) and histology (meningioma, glioblastoma, other glioma vs. metastases and all "other" histologies combined) as covariates, the analysis revealed only histology (p = 0.007) as an independent predictor of EPS. Specifically, patients with meningiomas had a 4.79-fold (95% CI 1.37-16.78, p = 0.014) increased relative risk for suffering EPS using the combined metastases and "other" histologies subgroup as reference category. early postoperative seizures in meningioma patients. The majority of our EPS cases had surgery for a meningioma. Preoperative seizure history, supratentorial tumour location, growth in association with the frontal, central, temporal and/or insular lobes and a worse postoperative KPI were associated with EPS not only in the overall cohort but also in the meningioma patient subgroup. Likely due to the limited sample size, only the latter two correlations proved statistically significant (Table 4). We found no correlation between oedema and EPS, while patients with EPS had indeed somewhat larger tumours (largest tumour diameter, axial T1-weighted contrast-enhanced images, EPS vs. no EPS: 471.4 ± 243.3 vs. 391.8 ± 184.2 mm, p = ns; Table 4).
All meningioma cases with EPS had supratentorial surgery. Early postoperative imaging studies could be made available in 166/186 (89.2%) patients (151 cCT, 15 MRI) with supratentorial meningiomas. Neither extraaxial hematomas, nor brain contusions, postoperative infarcts, resection cavity hematomas or pneumocephalus were found to significantly predict EPS. However, there was a statistical trend for an association between EPS and resection cavity hematoma (p = 0.055).
Early postoperative seizures: work-up, clinical relevance and outcome. Following an EPS, all except one of the 34 patients received cranial imaging within 24 h. MR and/or CT scans revealed a plausible cause for the seizure in 26/33 (78.8%) of the cases ( Table 2). This includes 22 patients managed conservatively and four patients who required surgery for removal of epidural, subdural and/or bleeds into the resection cavity. Of note, one of these cases reported a preoperative seizure history, yet no postoperative or postictal deficit (Fig. 1). However, the size of the (epidural) bleed was felt to warrant surgery. Only two patients undergoing revision surgery recovered to a KPI at discharge of 80 or better ( Table 2). The one case (no. 34) without emergency CT or MR scanning presented initially with a generalized seizure. His postoperative MRI was unremarkable. He suffered a second seizure with identical semiology several days later, from which he recovered quickly and completely. A CT scan was obtained a few days later showing no acute pathology.  www.nature.com/scientificreports/ Laboratory testing diagnosed relevant serum electrolyte disturbances in two cases (1 hyponatremia: Na + = 126 mmol/l, 1 hypernatremia: Na + = 159 mmol/l; Table 2). Meningitis was believed to be responsible for seizures in two cases. These two cases made a good recovery following appropriate antibiotic treatment ( Table 2). Awake craniotomies and intraoperative cortical electrostimulation were thought to play a causative role in two other cases, and a temporary external ventricular drain in one patient.
Medical management of EPS patients relied heavily on levetiracetam. 20/22 (90.9%) of cases with de-novo EPS were treated with levetiracetam. In the remaining two cases, no anticonvulsive treatment was initiated. Ten EPS patients with preoperative seizures had their levetiracetam dose increased or adjusted based on serum level determinations. In two cases with preoperative epilepsy and multiple EPS, who were already on levetiracetam, concomitant medication with lacosamide was initiated.
EPS were associated with very relevant morbidities and even mortality. Thirteen patients (38.2%) displayed a new neurological deficit following EPS not explained by imaging or other findings, with seven of those persisting at discharge (20.6%). One patient developed pneumonia subsequent to multiple seizures and ultimately a fatal respiratory insufficiency.
Seizure recurrence after EPS was frequent. As pointed out above, 17/34 (50.0%) patients experienced multiple seizures, i.e. suffered already a recurrent seizure within 30 days of the index surgery. Patients were routinely managed by outside neurologists after discharge. Nevertheless, seizure follow-up for at least one year could be made available for 26 of these cases. Nine patients (34.6%) reported at least one other seizure within a year. Interestingly, five of these nine patients (55.6%) had a negative seizure history prior to their surgery.

Discussion
The rate of early postoperative (= perioperative or in-hospital) seizures is usually reported as < 5-10% in most studies with some authors detailing EPS rates only for seizure-naïve patients 4,8,9,15,18,19 . We observed a 5.1% overall and 6.1% rate in patients with supratentorial tumours. Posterior fossa operations were included in our analysis primarily in order to avoid selection bias. Seizures in patients with infratentorial tumours are generally rare 19 . However, some patients with posterior fossa tumours require (temporary) ventricular drains, i.e. an (albeit minor) supratentorial operation, and certain perioperative complications thought to underlie EPS such as meningitis occur after both infratentorial and supratentorial surgeries. Interestingly, our series includes a patient undergoing surgery for a large vestibular schwannoma who required temporary CSF drainage and suffered a generalized seizure on postoperative day 13. EPS are not benign. In a substantial number of cases, they are associated with and reflect major complications such as bleeds requiring operative revision (4/34, 11.8%) or meningitis (2/34, 5.9%). Also, early postoperative seizures per se can result in neurological worsening (13/34, 38.2%) which often persists (7/34, 20.6%). Seizures can result in pulmonary complications due to acute or silent aspirations in confused or stuporous patients. One of our cases ultimately succumbed to a pulmonary sepsis initially triggered by an early postoperative seizure.
Risk factors for EPS in our cohort included a positive preoperative seizure history, a supratentorial tumour location, tumour growth involving or compressing particularly epileptogenic brain tissues (i.e. the frontal, central, temporal and insular lobes) 13 and, importantly, tumour histology. Meningioma patients accounted for the majority of cases and the highest overall rate (10.1%). However, contradictory results, i.e. an association with glioma histology and infiltrating growth and relatively lower perioperative seizure rates in meningioma patients have also been reported 8 . Awake craniotomy (i.e. intraoperative cortical electrostimulation) figured prominently as a risk factor for perioperative seizures in a recent study by Oushy and co-workers and was therefore felt to contribute to seizure formation in two of our cases 8 .
In accordance with the literature, we found an (albeit not statistically significantly) increased risk for EPS in patients with WHO grade II/III meningiomas, a convexity/parasagittal tumour location, perifocal oedema > 1 cm and an association with a worse postoperative KPI 4,15 . Meningioma-associated epilepsy has recently attracted some attention in the neurosurgical community. A non-skull base location, tumour size, peritumoral oedema, malignancy, tumour progression and recurrence, age and sex (as a possible corollary of a higher WHO grade), seizure history and EEG findings, clinical symptoms and surgical complications have all been (variably) associated with the risk of preoperative and postoperative seizures in meningioma patients 1,4,5,15,16,20 . We also investigated if early postoperative neuroimaging following supratentorial meningioma surgery might predict EPS. We were unable to identify significant correlations, however, there was a statistical trend for an association between EPS and the presence of a hematoma in the resection cavity.
Are EPS simply acute symptomatic seizures? By definition, acute symptomatic seizures occur within 7 days of the underlying brain insult 2 . Indeed, the majority of our cases reported seizures within the first three days following surgery and 29/34 (85.3%) within the first 7 days. Some authors have reported statistical correlations between surgical complications (including new neurological deficits) and the occurrence of early postoperative seizures 1,4 . We identified surgical complications such as haemorrhages and increased oedema, meningitis, and electrolyte disorders as the most likely cause of the seizure in 27/34 (79.4%) of cases. Clinical worsening (a possible corollary of a structural postoperative or other e.g. infectious complication) also correlated with the EPS rate in our series.
On the other hand, EPS tend to recur. 17/34 (50.0%) of our cases with EPS had a recurrent seizure within 30 days of the surgery (i.e. multiple EPS). The one year-recurrence rate was 34.6%, i.e. in a sizable proportion of   4 . Also, the risk of EPS varies with the preoperative seizure history in this as well as in several other published cohorts 1,4 . These data suggest that not only acute perioperative brain insults but also more chronic factors such as the anatomic and metabolic alterations caused by the brain tumour contribute to early postoperative epileptogenesis. In our view, these figures together with the adverse clinical course seen in several cases with multiple perioperative seizures justifies institution of anticonvulsive treatment already following a single perioperative seizure. Of note, we have no longer term follow-up from our patients and while we use antiepileptics quite liberally during the early postoperative period, we strongly urge patients and their physicians to re-evaluate this medication at the latest after one year. Antiepileptic medication may occasionally carry significant adverse effects. Side effects of levetiracetam, which is the current drug of choice for most neuro-oncological patients requiring anticonvulsive treatment, include e.g. fatigue, insomnia, mood and behaviour changes, headaches and decreased white blood counts 21,22 . These effects of have to be weighed against the risk of recurrent seizures with its attendant socioeconomic sequelae (e.g. restriction of driving privileges).
Given the rather high rates of adverse outcomes and recurrence after EPS, it is tempting to speculate about a role for prophylactic anticonvulsive medication. This is a controversial topic. Most believe that routine use of prophylactic anticonvulsants does not lower the risk of postoperative seizures in patients without preoperative seizures 14,[23][24][25] . However, there are some data including a small randomized prospective trial suggesting a benefit from prophylactic levetiracetam 9 . Based on our experience and the data just outlined, we consider prophylactic levetiracetam in vulnerable patients with a high risk of EPS, e.g. an elderly patient with a large convexity meningioma.
Our study allows some conclusions with respect to the proper management of EPS. Establishing a specific diagnostic algorithm for such emergencies is important. Neuroimaging revealed the presumed cause of the seizure in the great majority (26/33, 78.8%) of our cases. In four patients, the imaging finding prompted an operative revision. We therefore feel that obtaining a CT or MRI scan after an EPS is mandatory. Meningitis and electrolyte disorders may also contribute to the formation of perioperative seizures, i.e. laboratory blood testing and a very low threshold for a lumbar puncture should be part of the diagnostic work-up.
The major limitation of our study is its retrospective design. Not all seizures reported by the patients or observed by the attending staff may have been properly documented. In addition, it is quite possible that subclinical seizures have simply escaped detection. Another shortcoming is that this is a single institutional series which somewhat limits the generalizability of our findings. Nevertheless, we present a sizable, unselected and recent experience with the diagnosis and management of EPS.

conclusion
Early postoperative seizures (EPS) following brain tumour surgery are common. EPS often reflect serious complications of brain tumour surgery and are associated with a relatively high rate of adverse neurological and medical sequelae. We found that structural causes not infrequently requiring surgical revision are a common cause, which suggests CT (or MR) imaging as a mandatory part of the work-up of an EPS. Multiple EPS and recurrent seizures during follow-up are frequent, which indicates that EPS are not just acute symptomatic seizures, but not uncommonly reflect a more chronic condition i.e. epilepsy. Finally, our data indicate that meningioma surgeries may carry a particularly high risk for EPS, which raises the question of an antiepileptogenic prophylaxis in selected cases.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.