Time course of neurological deficits after surgery for primary brain tumours

Background The postoperative course after surgery for primary brain tumours can be difficult to predict. We examined the time course of postoperative neurological deficits and analysed possible predisposing factors. Method Hundred adults with a radiological suspicion of low- or high-grade glioma were prospectively included and the postoperative course analysed. Possible predictors of postoperative neurological deterioration were evaluated. Results New postoperative neurologic deficits occurred in 37% of the patients, and in 4%, there were worsening of a preoperative deficit. In 78%, the deficits occurred directly after surgery. The probable cause of deterioration was EEG-verified seizures in 7, ischemic lesion in 5 and both in 1, resection of eloquent tissue in 6, resection close to eloquent tissue including SMA in 11 and postoperative haematoma in 1 patient. Seizures were the main cause of delayed neurological deterioration. Two-thirds of patients with postoperative deterioration showed complete regression of the deficits, and in 6% of all patients, there was a slight disturbance of the function after 3 months. Remaining deficits were found in 6% and only in patients with preoperative neurological deficits and high-grade tumours with mainly eloquent locations. Eloquent tumour location was a predictor of postoperative neurological deterioration and preoperative neurological deficits of remaining deficits. Conclusions Postoperative neurological deficits occurred in 41% and remained in 6% of patients. Remaining deficits were found in patients with preoperative neurological deficits and high-grade tumours with mainly eloquent locations. Eloquent tumour location was a predictor of neurological deterioration and preoperative neurological deficits of remaining deficits. Electronic supplementary material The online version of this article (10.1007/s00701-020-04425-3) contains supplementary material, which is available to authorized users.


Introduction
After surgery for primary brain tumours, it is not uncommon with a deterioration of the neurological function [6,7,21,15]. In some cases, postoperative neurological deterioration is expected due to either perioperative ischemic injury or surgery in eloquent areas with corresponding deficits or when a supplementary motor area (SMA) syndrome occurs after surgery in the premotor cortex [17]. However, the reason for the postoperative neurological decline is not always clear, and often, it is difficult to predict the course of the deteriorated function. In the preoperative information to the patient, it is desirable to give more precise information regarding postoperative outcome. The primary aim of this study was to analyse the occurrence and time course of postoperative neurological deficits and the secondary aim to find possible predisposing factors.

Methods and materials
Patients One hundred patients, with a presumed glioma (WHO grades II-IV) planned for surgery at the Department of This article is part of the Topical Collection on Tumor -Glioma Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00701-020-04425-3) contains supplementary material, which is available to authorized users. Neurosurgery, Uppsala University Hospital during the period 22 August 2016 to 7 December 2017, were prospectively included. There were 60 men and 40 women with a mean age of 53.5 ± 16.2 years.
Preoperative neurological deficits were evaluated with clinical examination by a specialist in neurosurgery and documented in the medical record the day before surgery. The motor deficit was scored according to our clinical scale used for pre-and postoperative evaluation: no motor deficitdiscrete motor deficit, pronounced motor deficit or complete motor deficit. Cognitive deficit was defined as the presence of confusion, disorientation, personality change or memory disturbances judged by clinical examination or medical reports.
Eloquent tumour location was assessed according to Chang et al. [5]. The presumed eloquent areas included sensorimotor strip (precentral and postcentral gyri), dominant hemisphere perisilvian language areas (superior temporal, inferior frontal and inferior parietal areas), basal ganglia/internal capsule, thalamus and calcarine visual cortex.

Surgery and postoperative care
The surgical procedure is described in detail earlier [13]. Briefly, tumour resection was done through craniotomy using microsurgery guided by neuronavigation and intraoperative ultrasound. Intraoperative neurophysiological monitoring of motor function was performed if the tumour was located in close connection to eloquent cortical or subcortical areas. In an awake surgery, speech function and visual fields were monitored. 5-Amnolevulinic acid (5-ALA) (Gliolan, Medac Pharma, Varberg, Sweden) was used in 21 patients with presumed high-grade (contrast enhancing) tumours if total resection was the goal of surgery. After surgery, the patient was awakened in the operating theatre and brought to the postoperative neurointermediate ward, and EEG-and video monitoring was initiated [13]. A neurologic examination was performed by the responsible surgeon. The level of consciousness according to Reaction Level Scale 85 (RLS85) [31] and the presence and grade of postoperative neurological deficits were then monitored according to our clinical postoperative protocol by specially trained nurses. According to his protocol, RLS85 is checked every 30 min and neurological status (motor deficits) every 60 min for the first 6 h, RLS85 and neurological status every 60 min for 6-12 h postoperatively and every 120 min 12-24 h after surgery. After 24 h, the monitoring is prolonged if indicated in selected cases. An acute CT scanning was performed in any case of postoperative deterioration or new neurological deficits. In uncomplicated cases, postoperative monitoring continued for 24 h in the neurointermediate ward, and the patient was then discharged to the general ward. If there was a complicated postoperative course, for example, with seizures or new neurological deficits, postoperative monitoring in the neurointermediate ward continued until the patient was considered stable.
Patients were followed up and the neurological status checked in the outpatient clinic 3 months postoperatively. In some patients with high-grade tumours undergoing oncological treatment, the neurological status was evaluated by the responsible doctor. A complete regression was defined as no visible deficits left, and the performances of the patient were unchanged compared with those of before surgery. An almost complete regression was defined as a there was a slight remnant of the deficits, but the patient was not impaired by it in daily life and it was not clearly visible for the examiner. Remaining deficits were defined as the deficits were still there (but might had improved) at the 3 months following up.

Radiology
Postoperative magnetic resonance imaging (MRI) was performed within 48 h after surgery. In contrast enhancing (high-grade) tumours, contrast enhancement on T1-weighted turbo spin echo sequences and in non-contrast enhancing tumours, high signal intensity on T2 fluid-attenuated inversion recovery (FLAIR) sequences was considered a tumour tissue.
Postoperative ischemic lesions were evaluated on DWI with B 1000 value and corresponding ADC map. The total volume of ischemic lesion (in cm 3 ) was calculated using the Vue PACs software (Picture Archiving Communication System, v11.1.4) and its semi-automated lesion management application (livewire algorithm) [20]. The software is supported by an algorithm that uses an active contour model in order to evolve and segment the lesions. In defining the volume of the surface voxels, a clear difference in pixel contrast (black/ white) assisted the operator, increasing the ability to better adapt or correct the ischemic contour line even where it was less defined. To investigate the correlation between ischemic lesions and white matter tracts, FLAIR and DWI sequences were normalized into MNI space using the built-in software of DSI studio (DSI Studio, http://dsi-studio.labsolver.org/ download-images). The ischemic areas were defined as new regions of interests (ROIs) on patient-specific sequences and reconstructed into the HCP-1021 template. A group average template was constructed from a total of 1021 subjects enrolled by the Human Connectome Project (the WU-Minn HCP consortium which is an institutional, review board-approved, NIH-funded project led by Washington University, University of Minnesota and Oxford University) [35]. A multishell diffusion scheme was used, and the b values were 990, 1985 and 2980 s/mm 2 . The number of diffusion sampling directions was 90, 90 and 90, respectively. The in-plane resolution was 1.25 mm. The slice thickness was 1.25 mm. The diffusion data were reconstructed into MNI space using qspace diffeomorphic reconstruction [41] to obtain the spin distribution function [42]. A diffusion sampling length ratio of 2.5 was used, and the output resolution was 1 mm. The restricted diffusion was quantified using restricted diffusion imaging [40]. Major projection, commissural and association white matter pathways were reconstructed within the HCP-1021 template following the anatomical criteria already published with the Brain Grid DTT reference atlas [19] and matched with ROIs defining the ischemic areas. The method's workflow is visually described in Supplementary Fig. 1. The white matter structures impinged by the ischemic lesions are displayed in Table 4.

Statistics
Comparisons between groups were made with Mann-Whitney U test for continuous and categorical variables and Fischer exact two-tailed test for proportions. Possible predictors of postoperative neurological deterioration were evaluated in a simple regression analysis. Factors with a p value < 0.1 were chosen to be tested in the multiple regression analysis. A p value < 0.05 was considered statistically significant. Statistica, version 13.2 (StatSoft, Inc. Tulsa, OK, USA), was used for statistical calculations.

Ethics
The study was approved by the institutional ethics review board (2016/112). Informed consent was obtained prior to participation.

Tumour locations, diagnosis and tumour volumes
Tumour locations and tumour diagnosis are presented in Table 1. The most common tumour location was frontal, n = 33 (33%) followed by temporal, n = 27 (27%). Forty patients had a right-sided tumour, 52 patients left-sided, 6 patients bilateral and 2 patients had midline tumours. Thirty-nine patients harboured tumours in presumed eloquent areas ( Table 1) and two patients in the SMA. Intraoperative neurophysiological monitoring of motor function was used in 25 patients which was combined with awake surgery and monitoring of speech functions in 10 of these patients.
In 21 of the patients with tumours, according to Chang [5], in eloquent areas, neurophysiological monitoring was used. However, 18 of the patients with eloquent tumour location went through surgery without neurophysiological monitoring. In 8 of these patients, a part of the tumour extended into the basal ganglia, and the intention was not to resect this portion. Ten of the patients showed preoperative neurological deficits, and radical surgery was not planned (n = 7) or the deficits were already maximal and considered not to be worsened by surgery (3 patients with hemianopsia).
In 5 patients with tumour location that was not considered eloquent according to Chang, neurophysiological monitoring of the motor functions was used due to close connection to subcortical motor tracts (3 patients with tumours in the parietal area) or cortical motor areas and subcortical motor tracts (2 patients with tumours in the SMA). Numbers in italics are the total numbers in every subgroup *Fronto-insular, n = 2; temp-insular, n = 1; fronto-temporal-insular, n = 8; fronto-temporal-insular + central, n = 1. The presumed eloquent areas were sensorimotor strip (precentral and postcentral gyri), dominant hemisphere perisilvian language areas (superior temporal, inferior frontal and inferior parietal areas), basal ganglia/internal capsule, thalamus and calcarine visual cortex The most common diagnosis was high-grade glioma, found in 69 patients (69 %) followed by low-grade glioma (WHO grade II) in 24 patients (24%). Median (IQR) preoperative tumour volume was 32.4 (11.2-74.5) cm 3 and resection grade was 96.5 (72-100) %.

Pre-and postoperative neurological deficits
Pre-and postoperative neurological deficits are shown in Fig. 1. Preoperative neurological deficits were present in 40 patients. Cognitive deficit was the most common (13%), followed by visual field (10%) and motor deficit (8%). New postoperative neurologic deficits were found in 37% of patients. In addition, 4% of patients exhibited worsening of a preoperative existing neurological deficit postoperatively. Most commonly, motor dysfunction occurred (21 patients) and 20 patients showed dysphasia.
Among the 25 patients who went through intraoperative neurophysiological monitoring, 18 patients developed new postoperative deficit. Six of these patients went through awake surgery, and in 12 patients, intraoperative monitoring of motor function was used. In 2 patients with neurologic deterioration after awake surgery, there were intraoperative fluctuations of speech functions which made intraoperative speech evaluation difficult and one patient showed dysarthria due to motor impairment of the tongue (with intact motor signals). In the remaining 3 patients, there were only stimulation-induced speech disturbances which is an expected finding at awake surgery. Among the 12 patients with intraoperative neurophysiological monitoring of motor functions and postoperative neurological deterioration, there was a change of the intraoperative motor signals in one patient who postoperatively showed an ischemic lesion in the right corona radiata. The intraoperative finding was an increase in the motor threshold for the left hand, but no change of motor signals for the face was noted. A left facial palsy with a quick improvement was noted postoperatively, but the left hand was intact. In the other 11 patients, there was no intraoperative change of motor signals. Table 2 describes the 18 patients with intraoperative monitoring and postoperative neurological deterioration.
The time course of the neurological deficits is shown in Fig. 2. In 32/41 patients (78%), the deficits occurred or worsened directly after surgery and in 9/41 patients (22%) after a delay, median (IQR) 12 (4-50) hours. In 27/41 patients (66%, 27% of the whole group of patients), there were complete deficits, and four patients exhibited worsening of a preoperative existing neurological deficits. In 13 of these patients, there were two deficits regression of the new postoperative deficits. Six patients (6% of the whole group of patients) showed almost complete regression with only slight deficits remaining. In 6 patients (6% of the whole group of patients), the deficits still remained 3 months postoperatively, and for two patients, there was no information. The six patients with remaining neurological deficits are described in Table 3. In summary, all patients showed preoperative neurological deficits and had high-grade glioma (grades III-IV), mostly located in eloquent areas. There was no ischemic lesion in any of these patients. The probable reason for the neurological deficits was resection of eloquent tissue. In addition, two patients showed very fast tumour growth which probably contributed to an impaired plasticity and remaining deficits.
In a subgroup analysis, we compared low-grade gliomas (WHO grade II), n = 24, with high-grade (WHO grades III-IV) gliomas, n = 69. We found that 12/24 (50%) patients with low-grade gliomas developed postoperative neurological deficits, 7 eloquent and 13 non-eloquent tumour locations, but there were complete or almost complete regression of neurological deficits in all these patients and no patients with low-grade gliomas showed remaining deficits. Among patients with high-grade gliomas, there were 27/69 (39%) patients with postoperative neurologic deterioration, 14 patients with eloquent tumour location and 13 patients with tumours in non-eloquent areas. Remaining deficits were found in 36% (5/14) of patients with eloquent tumour locations and in 8% (1/13) of patients with non-eloquent tumour locations. Thus, some trends were found with better recovery for patients with low-grade gliomas compared with high-grade gliomas (p = 0.15) and in the high-grade glioma group better recovery for patients with non-eloquent tumour locations compared with those with eloquent tumour locations (p = 0.16).
The two patients with a tumour in the supplementary motor area showed a postoperative hemiparesis with slightly slower movements (almost complete regression) after 3 months. The four patients with a decline of a preoperative neurological deficit are included in the numbers above. A complete regression in those patients was defined as that the postoperative decline of the function returned to preoperative level.
The improvement occurred within one day, n = 7 (17%); within 1 week n = 8 (24%); within 1 month, n = 11 (27%); within 3 months, n = 7 (17%). For two patients there was no information after discharge from hospital. Figure 3 shows the number of patients with neurological deficits at different time points after surgery.

Seizures
EEG-verified seizures were detected in nine patients after surgery (seven patients < 24 hours and two patients > 24 hours after surgery) and caused postoperative deterioration in eight of these patients. In one patient with subclinical seizure activity for totally 22 h, no postoperative neurological deterioration was detected. In two other patients, there was a clinical suspicion that seizures caused a transient neurological deterioration, see below.
The result of the EEG and video recording postoperatively is published before [13].

Delayed postoperative neurological deterioration and seizures
In the nine patients with a delayed deterioration of the neurological function, seizures were the proven or the probable cause. The deterioration occurred after 2 h, 3 h, 4 h, 12 h (n = 2), 36 h, 50 h, 3 days and 6 days postoperatively. In seven of these patients, there were EEG-verified seizures that accounted for the deterioration. In one patient, the neurological deterioration occurred after the EEG monitoring had finished, but the patients displayed focal seizures that were considered the cause of the deterioration. In another patient with an aggravated paresis in one leg after focal seizures 12 h postoperatively, no epileptic seizure activity could be recorded although the clinical picture even in this case favoured seizures as contributing factor to the worsening of the deficits.

Other complications:
In 14 patients (14%), postoperative MRI showed a new ischemic lesion and nine of them (64%) deteriorated neurologically after surgery. The ischemic lesion was considered a possible cause of the postoperative neurological deterioration in 5 In 4 of the 10 patients with postoperative ischemic lesions, intraoperative neurophysiological monitoring was used. There was a change of the intraoperative signals in one patient, described above. Table 4 describes the clinical characteristics of the patients with new postoperative ischemic lesion.
Three patients underwent a second surgery after the primary tumour resection: One patient showed a preoperative hemiparesis which was aggravated after surgery. An expanding haematoma in the surgical field was noted and was evacuated 24 h postoperatively. After the second surgery, the patient improved to the preoperative level. Another patient was reintubated directly after the primary surgery due to decreased level of consciousness and received an intraparenchymal pressure monitoring device. The third patient displayed a generalized seizure and decreased level of consciousness. A CT scanning revealed a distant haematoma in the posterior fossa, and the patient was subjected to an external intraventricular drainage procedure.

Summary of probable causes of postoperative neurological deterioration
To summarize, the probable causes of neurological deterioration in the 41 patients were EEG-verified seizures in seven patients, EEG-verified seizures + ischemia in one patient, clinical suspicion of seizures in two patients, resection of eloquent tissue in six patients, resection close to eloquent tissue in nine patients and resection of the SMA in two patients, ischemia in five patients (plus one described above with seizures) and postoperative haematoma in one patient. In eight patients, the reason for the transient deterioration was not clarified. We speculate that the probable reasons might have been a remaining effect of anaesthesia in two patients with a transient postoperative decline of a preoperative neurological deficit and multifactorial in four patients with a transient postoperative confusion. In two patients, no reasonable explanation could be found. In patients with no neurological deficits postoperatively, a higher grade of resection was achieved, median (IQR) 100  The parameters used in the simple regression analysis were age (continuous), sex (male/female), tumour grade (low/high/ other), preoperative neurological deficits (yes/no), presumed eloquent tumour location (yes/no) and tumour volume (continuous). The results are shown in Table 5. The variables chosen from the simple regression analysis to be examined in the multiple regression analysis were preoperative neurological deficits (p = 0.057) and presumed eloquent tumour location (p = 0.012) together with age and sex. In the multiple analysis did a presumed eloquent tumour location become a significant predictor of postoperative neurological deterioration p = 0.027 (see Table 5).

Prediction of remaining neurological deficits
For calculating risk factors of remaining neurological deficits, the following parameters were used: age, sex, tumour grade (high/low/other), preoperative neurological deficits, presumed eloquent tumour location, preoperative tumour volume, when the postoperative neurological deficits occurred in relation to surgery (continuous) and postoperative ischemic lesion on MRI (yes/no). The variables chosen to be examined in the multiple regression analysis were tumour grade (p = 0.08) and preoperative neurological deficits (p = 0.009) together with age and sex. In the multivariate analysis, preoperative neurological deficits became a significant predictor of remaining neurological deficits, p = 0.046 (see Table 5).

Discussion
To summarize this study, postoperative neurologic deterioration occurred in 41% of the patients, and patients with tumours in presumed eloquent areas showed more often new postoperative neurological deficits compared with patients with tumours in non-eloquent areas. The probable cause of postoperative neurologic deterioration was EEG-verified seizures in seven patients, a new ischemic lesion in five patients, both of these in one patient and postoperative haematoma in one patient. In 11 patients, tumour resection close to eloquent areas including the SMA was considered the probable cause of neurologic deterioration, and in six patients, the resection included eloquent tissue resulting in neurological deficits after surgery. In the majority of patients (78%), the deficits occurred directly after surgery, and in the nine patients with a delayed neurological deterioration, seizures were the proven (n = 5) or probable (n = 4) cause of the new deficits. In 66% of  the patients with postoperative deficit (27% of the whole group), there was complete regression of the postoperative deficits, and in another 15% of the patients with postoperative deficit (6% of the whole group), there was almost complete regression with a slight disturbance of the function remaining after 3 months. Remaining deficits were found in 6% of all patients, and all these patients showed preoperative neurological deficits and high-grade tumours with mainly eloquent locations. Eloquent tumour location became a predictor of postoperative neurological deterioration, and preoperative neurological deficits were a predictor that the deficits would remain.

Neurological deterioration
The incidence of any neurological deterioration after craniotomy for primary brain tumours in our study was 41%, which is higher than previously described [6,7,21,15,14,27,22,32]. However, the decline in the neurological function was in the majority of patients transient, with complete or almost complete regression of the symptoms in 81% of the patients and a complete regression in 66% of the patients. Thus, the incidence of permanent postoperative neurological deficits was lower, 6%, in the whole group of patients, and another 6% of the patients reported a slight remnant deficit which did not impair function. These numbers are comparable with the incidence, 7-20% [14,15,21,27,4,39,38] of postoperative neurological deficits, described earlier. A meta-analysis of outcome in glioma surgery showed that early deficits occurred in 30% of patients [9], and in a study by Gempt et al., transient postoperative neurological deficit was found in 17% of newly diagnosed, and 32% of recurrent gliomas and permanent neurological deficits were found in 7 and 16% respectively [15]. Sawaya et al. [27] described neurological complications in 8.5%, Lonjaret et al. [21] in 16% and Brell et al. [4] in 20.5% of patients after surgery for brain tumours. In a study by Berger et al., immediate motor deficits were found in 22% and speech deficits in 3% of patients [3].
Our study shows that the deficits occurred directly after surgery in 78% of the patients who developed deficits, which is in line with previous studies [21]. In Lonjaret's study, 85% of the patients showed neurologic complication during the first 2 h after surgery [21], and the central nervous system was the dominating location for postoperative complications within the first 24 h after brain tumour surgery [37].
The reason for neurological deterioration may be direct tissue damage after surgical manipulation or an effect of resection of eloquent tissue. Neurological deterioration may also occur secondary to tissue oedema, arterial ischemia, venous  [15,16], vasospasm after vessel tears [23], haematomas [16] or be an epileptic ictal or postictal phenomena. Symptoms due to surgical manipulation, resection or ischemia are expected to occur immediately after surgery, whereas postoperative haematomas, epileptic ictal phenomena and symptoms due to venous infarctions or vasospasm may occur after a delay [16]. A transient neurological deterioration could also be due to unmasking of an already existing deficit or borderline function which could be compensated for under fully alert and awake conditions but is revealed after anaesthesiology. In this study, 6% of the patients showed a new ischemic lesion on MRI as a plausible cause of the postoperative neurological deficits and more patients with tumours in presumed eloquent areas developed new postoperative neurological deficits. Two patients developed a hemiparesis with almost complete regression after surgery in the supplementary motor area, which is a well-known phenomenon after surgery in this area [17]. Plasticity due to reorganisation of the function is a plausible explanation of the improved function after a certain period of time in those cases [12].
In a previous work [13], we examined the occurrences of epileptic seizures with continuous EEG monitoring after surgery for presumed primary brain tumours. We found that 7% of the patients displayed postoperative epileptic seizures the first 24 h after surgery [13]. However, in all 9 patients (9%) in this study with a delayed neurological deterioration, seizures were the proven (n = 5) or probable (n = 4) cause of the new deficits.
In six patients (6%), there were remaining deficits after 3 months. These patients all showed preoperative neurological deficits and harboured tumours in or in close connection to the motor, sensor and language areas, and the diagnosis in all patients was high-grade gliomas which showed a very fast regrowth in at least two of them. The perioperative neurophysiological monitoring used in 3 of these patients did not show any warning signs of impaired motor function. Thus, the monitored motor function seemed to be neurophysiologically intact, and the plasticity and capacity of improvement may have been impaired by the aggressive growth of their high-grade tumours. When we compared the recovery of postoperative neurological deficits between low-and high-grade gliomas, we found some trends with a better recovery in the group of patients with lowgrade gliomas, in which no patients showed remaining deficits. This should be compared with the group of patients with highgrade gliomas in which 6/27 (22%) of patients showed remaining deficits and 5/6 (83%) of these patients had tumours in eloquent areas. These numbers are too small for statistic calculations, but we think this trend seems to be reasonable and favours the fact that patients with low-grade gliomas have a better plasticity due to the slow growth of the tumour.
The substantial variation of the incidence of postoperative neurological deterioration after brain tumour surgery found in the literature could be explained by the differences in the methodology of the studies, heterogenicity of the materials, i.e. if tumours in eloquent areas are included [16] and methods of detecting the complications [10] and definitions of neurological complications. In a prospective study, with the goal of reporting the incidence of neurological complications, a higher incidence is to be expected due to more meticulous schedules for detecting any decline of neurological deterioration compared with a retrospective analysis based on register data. Our study provides data on a detailed level of value for increasing the knowledge of short-term surgical outcome. This kind of data is of value for the preoperative information to the patients and could be helpful in the decision process regarding the indication for surgery when weighing possible benefits and risks and for optimizing the perioperative treatment of the individual patient [29].

Postoperative ischemic lesions
Our numbers of postoperative ischemic lesions (14%) are lower than in the study by Gempt et al. [15], who identified new postoperative ischemic lesions in 31% of newly diagnosed and in 80% of patients with recurrent gliomas. Other have described new ischemic lesions after glioma surgery in 23% [3], 64% [30] and 70% [34] of patients. Tumour location in proximity to perforating arteries [15], insular tumours [3] and recurrent gliomas [3,15] has been identified as a risk factor for postoperative ischemic lesions, and age was found to be an independent risk factor for stroke within 30 days after brain tumour surgery [2]. Vascular reorganization or vessel obliteration by brain irradiation was suggested as an explanation for the increased number of ischemic lesions after resection of recurrent gliomas [15]. A higher probability of new postoperative neurological deficit has been identified in patients with a new postoperative ischemic lesion [15]. In our study, there was no significant difference of the occurrence of neurological deficits in patients with or without ischemic postoperative lesions on MRI. None of the patients with remaining neurologic deficits showed ischemic postoperative lesions, and new postoperative ischemic lesions could neither be identified as a predictor of remaining neurological deficits in this small group of patients.

Postoperative haematoma
The incidence of haematomas requiring evacuation in this study was 1%. Even if the reported incidence of postoperative haematoma after craniotomy varies a lot [28], this is in line with previous studies [21,37,18,43].

Risk factors for neurological deterioration
We found that more patients with preoperative neurological deficits developed new postoperative neurological deficits compared with the patients with no preoperative neurological purpose. The advantage of the study is that the consecutively included patients are individually evaluated on a detailed level in the acute phase after surgery. In two patients, information regarding the postoperative course after discharge from hospital was lacking.

Conclusions
After surgery for primary brain tumours, neurological deterioration occurred in 41% and remaining deficits in 6% of patients. Only patients with preoperative neurological deficits and high-grade tumours mainly in eloquent areas showed persistent deficits. Epileptic seizures accounted for the majority of delayed neurological deterioration. Eloquent tumour location was a predictor of postoperative neurological deterioration, and the presence of preoperative neurological deficits was a predictor of remaining new postoperative neurological deficits.
Funding information Open Access funding provided by Uppsala University. The Selanders Cancer Foundation and ALF research funds at Uppsala University Hospital provided financial support. The sponsor had no role in the design or conduct of this research.

Compliance of ethical standards
Conflicts of interest The authors declare that they have no conflict of interest.
Ethical approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee (name of institute/committee) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent Informed consent was obtained from all individual participants included in the study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.