The evolution of epilepsy surgery in tuberous sclerosis in Sweden: A national registry study

Purpose: This study aimed to characterize the Swedish cohort of surgically treated patients with TSC and explore differences in preoperative investigation and outcome over time. Methods: Data on patient and seizure characteristics were retrieved from the Swedish National Epilepsy Surgery Register. Two-year follow-up results were compared between the years 1997 – 2010 and 2011 – 2018. Preoperative investigations were re-evaluated. Results: Eighteen tuberectomies and seven callosotomies were identified. Seizure freedom after tuberectomy was achieved in 11 % (1/9) 1997 – 2010 and 56 % (5/9) 2011 – 2018. The number of tuberectomies increased each decade. Patients operated on in 1997 – 2010 had higher seizure frequency (median 175 seizures/month vs. 102) and incidence of infantile spasms (4/9 vs. 1/9, none after 2011). There was a trend towards surgery at a younger age (median 86 months 1997 – 2010 vs. 48 months 2011 – 2018). None with > 200 seizure/month, SEGA, or history of infantile spasms achieved seizure freedom. Two patients underwent anterior callosotomy (1992 and 1994) and became free of drop attacks. Five callosotomies were performed between 2011 and 2013, one patient became free of drop attacks. Two complications with new neurological deficits were reported. The median age at surgery was higher in the callosotomy group (14 years) than in the tuberectomy group (5 years). Conclusion: Seizure freedom after tuberectomy in patients with TSC has increased over time in our cohort. Signs of a heavier disease burden were more frequently observed 1997 – 2010 and associated with worse outcomes. Callosotomy operations were prevalent at the beginning of the 2010s.


Introduction
Tuberous sclerosis complex (TSC) is a rare genetic disorder caused by pathogenic heterozygous variants in the mechanistic target of rapamycin (mTOR) pathway regulators, TSC1 or TSC2 [1].Disinhibition of mTOR leads to the growth of benign hamartomas in several organs.Epilepsy is the most prevalent neurological manifestation in patients with TSC, occurring in approximately 90 % of patients and 80 % <2 years of age, with about 40 % presenting with infantile spasms [2].Drug-resistant epilepsy occurs in two-thirds of patients with TSC compared with one-third of individuals in the general epilepsy population [1,2].Patients with drug-resistant epilepsy should be referred to a tertiary center offering advanced investigations and treatments, including the possibility of surgery.Early epilepsy surgery has shown a better outcome in terms of cognitive development [3].
During the last 15 years, ground-breaking discoveries regarding the Abbreviations: 2FU, follow-up 2 years of surgery; AMT, 11C-alpha-methyl-L-tryptophan; ASM, anti-seizure medication; CT, Computed Tomography; MRI, magnetic resonance imaging; mTOR, mechanistic target of rapamycin; PET, Positron emission tomography; SEGA, subependymal giant cell astrocytoma; SISCOM, Subtraction Ictal SPECT co-registered to MRI; SNESUR, Swedish National Epilepsy Surgery Register; SPECT, single-photon emission computed tomography; TSC, Tuberous Sclerosis Complex; vEEG, video-electroencephalogram.pathology and treatment of TSC-associated epilepsy have been made.Treatment with vigabatrin before the onset of clinical seizures reduces the incidence of infantile spasms and drug-resistant epilepsy as well as prolongs the time to the first seizure [4,5].Everolimus, an mTOR inhibitor, has also shown good efficacy in treating drug-resistant epilepsy in patients with TSC [6].However, the interest in surgical alternatives seems to remain [7,8].
Epilepsy surgery is an underutilized treatment, partially due to patient-and physician-related factors and misconceptions of surgical risk versus benefits [7,[9][10][11][12][13].Only 40 % of patients evaluated for epilepsy surgery will actually undergo surgery [14].The first report of epilepsy surgery in TSC cases was published as early as 1957 and included two patients [15].Small series of patients with TSC who underwent epilepsy surgery have been reported from different centres according to two types of interventions: (1) tuberectomy, which involves removal of the presumed epileptogenic tuber, and (2) callosotomy, which involves severing the corpus callosum to reduce drop attacks [16].
In the present study, we collected comprehensive data from the unique national population-based epilepsy surgery cohort in Sweden (Swedish National Epilepsy Surgery Register [SNESUR]) on patients with TSC who underwent epilepsy surgery between 1990 and 2018.Our primary aim was to describe the group and study the outcome after surgery in these patients, whereas our secondary aim was to explore differences and possible trends over time in investigations, procedures, and disease characteristics.

Design
This is a national register-based study focusing on seizure outcome at 2-year follow-up (2FU) after epilepsy surgery in individuals with TSC in the Swedish cohort and exploring investigational and treatment-related trends over time.The funding sources of this study had no involvement in study design, data collection, analysis, or interpretation.

Collection of patient demographic data
Patients with TSC who underwent epilepsy surgery between 1990 and 2018 were identified through SNESUR, a unique population-based national register of epilepsy surgery with complete national coverage.Data were collected retrospectively and prospectively between 1990 and 1995 and prospectively since 1995.Preoperative and 2FU data were collected from the register.
The SNESUR-related search criteria were as follows: surgical treatment of epilepsy performed between 1990 and 2018 and a diagnosis of TSC, a post-resection histological diagnosis of TSC, or any other mention of "tuberous sclerosis" or "TSC" in the database's free text field.Patients with a TSC diagnosis and previous epilepsy surgery (tuberectomy or corpus callosotomy) with available 2FU were included in this study, regardless of age.If a patient had a history of more than one surgery with available 2FU, both operations were included.
All six epilepsy surgery centers in Sweden were contacted and asked for confirmatory documentation to ensure complete national coverage and validate the register's information if needed.Finally, verbal information was given to the physicians and patients' families who attended the national TSC meeting in Stockholm in May 2019.Written information about the study was published in the magazine of the patients' organizations in two consecutive editions to inform them about the study and ensure that no individual eligible for inclusion was missed.Written informed consent was retrieved from eligible patients.
Medical records of all patients were retrieved from treating or operating hospitals to confirm the diagnosis of TSC and retrieve further information regarding treatment for previous infantile spasms, vigabatrin treatment (introduced in Sweden 1993), everolimus treatment (first Swedish patient with TSC began treatment in 2011), and preoperative investigations.Preoperative investigations extracted included magnetic resonance imaging (MRI), computed tomography (CT), extracranial or intracranial video-EEG (vEEG), single-photon emission computed tomography (SPECT), positron emission tomography (PET), and subtraction ictal SPECT co-registered to MRI (SISCOM) when performed within 3 years before surgery or mentioned in the epilepsy surgery conference report.

Outcome measure
The seizure outcome was classified according to the modified Engelscale adopted by SNESUR and is presented for 2FU.Long-term follow-up has been presented elsewhere, focusing on patient satisfaction [17].Class I included patients who achieved sustained seizure freedom or were seizure-free with auras only, had a few seizures at the early postoperative stage, and experienced atypical generalized seizures after withdrawal of anti-seizure medications (ASM).Class II, III, IV, and V comprised those with a >75 % decrease, a 50-75 % decrease, a 0-50 % decrease, and an increase in seizure frequency, respectively.
For comparison of outcomes after tuberectomy between time points (earlier surgery versus surgery in recent years), the data were dichotomized based on the year the tuberectomy was performed (1997-2010 vs. 2011-2018).These periods were chosen as 2011 was the year everolimus was introduced in Sweden.Furthermore, around this period, 3 Tesla MRI had become standard in epilepsy surgery investigations.This time point also allowed for comparison between equal-sized groups.Callosotomies were deemed too few to be dichotomized.

Radiology
For all participants, preoperative and postoperative MRI data were retrieved when available in interpretable quality.The latest preoperative and the first postoperative images of acceptable qualities and sequences were used.When available, CT images were retrieved for preoperative images to identify calcified tubers.All images were then reviewed by an experienced neuroradiologist (HS) and a PhD student (KP).Variables obtained for participants who underwent tuberectomy were: the site of resection, volume of the resected tuber on postoperative MRI, if it was the largest tuber overall, if the resected tuber was cystic or calcified, if the resected tuber somehow differed from most other tubers in terms of any quality (unique tuber), if a type A, B, or C tuber was excised [18], and presence of radial migration lines to the resected tuber.If more than one tuber was excised, the above-mentioned parameters were obtained for each tuber.All participants' variables obtained were: dominating type of tuber (A-C), number of tubers supratentorial and cerebellar, presence of subependymal giant cell astrocytoma (SEGA) at time of surgery, and field strength of MRI camera used preoperatively.The volume of the tuber was calculated on T2 weighted and fluid-attenuated inversion recovery (FLAIR) images if available preoperatively.The diagnosis of SEGA was made according to international guidelines for TSC [19].

Neurophysiology and functional imaging
Information about the type of preoperative neurophysiological investigations used (routine EEG, extracranial or intracranial vEEG, SPECT, PET, or SISCOM) was retrieved from SNESUR and collected EEG and vEEG reports.Interictal epileptiform activity was either mono-focal (interictal activity in only one lobe during vEEG) or multifocal.The locations on the lobar level of the major interictal focus and seizure onset were described when available. 11C-alpha-methyl-L-tryptophan (AMT) was introduced in Sweden in 2015 (personal communication T. Danfors 2023-04-20) and is used to study the synthesis of serotonin in the brain and has shown increased uptake on PET in the epileptogenic tuber in patients with TSC [20]. 18Fluoro-2-deoxyglucose PET is more readily available and is used to study metabolism to determine the localization of the functional deficit zone [21].Medical records on the above-mentioned investigations were retrieved and re-examined by an experienced clinical neurophysiologist (IR) for tuberectomies.Neurophysiological data for callosotomy cases were less rigorously reviewed because medical records overall did not indicate focal epilepsy, and the indication for surgery in these patients was to relieve drop attacks.

Statistics
Due to the small number of patients included, statistical analysis was not deemed possible.Descriptive statistics with median and range are presented in this study.IBM SPSS Statistics for Windows, version 27 (IBM Corp., Armonk, NY, USA) was used.

Ethics
The study was approved by the Regional Ethical Review Board in Lund (Diary # 2019-00518) regarding accessing medical records and contacting the patients directly and by the Regional Ethical Review Board in Gothenburg (Diary # 078-09) regarding accessing data from the SNESUR.

Demographics
A total of 18 tuberectomies with available 2FU were identified among 17 patients.Patient demographics, clinical presentations, and preoperative investigations are presented in Table 1.The surgeries were performed between 1997 and 2018.One female patient underwent two tuberectomies, 8 years apart (1998 and 2006), which were presented in this study separately because 2FU was available for both cases.Two patients had a history of one or more tuberectomies without 2FU; therefore, only the latest operation was indicated.Furthermore, most participants (13/18) had the surgery performed before 9 years of age, and none had surgery between 10 and 17 years of age.

Disease characteristics
Patients with infantile spasms were evenly distributed until 2011 (5/ 10), after which no patients were reported with this condition (supplementary material).The median number of tubers was 16.5 .All four patients with the largest resected tuber underwent surgery after 2011(supplementary material).In the 2011-2018 cohort, patients underwent surgery at a younger age, had a shorter duration of epilepsy before surgery, and had fewer preoperative seizures monthly (Table 4).
No patient with >200 preoperative seizures monthly, a SEGA detectable on MRI, or a history of infantile spasm achieved seizure freedom after surgery.More unique, cystic, calcified, and largest resected tubers were present in the Engel I group (Table 3).

Anti-seizure medication use
The median (range) number of previously tried and discontinued ASM before surgery in 1997-2010 was 4 (0-9) compared with 3 (0-5) in 2011-2018.Before 2010, three patients took more ASMs, four took fewer ASMs, and one became ASM-free after surgery.All patients in Engel II had an ASM reduction of 1-2 medications.After 2010, six patients took fewer ASM after surgery, and three were ASM-free.One patient took everolimus preoperatively, whereas four initiated everolimus postoperatively on epilepsy indication (1, 4, 11, and 12 years postoperatively).The effect on seizures of everolimus in these patients has been presented elsewhere [17], where three out of four achieved seizure freedom.

Demographics
Seven patients underwent corpus callosotomy.Two patients had a previous tuberectomy 14 and 16 years prior, respectively.Two anterior callosotomies at two different centers were performed in 1992 and 1994.Five total callosotomies were performed in 2011-2013 at the same

Preoperative investigations
No patient underwent preoperative PET.Among the five operations performed in the 2010s, all involved vEEG (one intracranial), whereas three involved ictal SPECT (Table 1).

Disease characteristics
All five patients in the 2011-2013 cohort had a history of infantile spasms.The two latest patients had a SEGA (supplementary material).The median number of tubers was 18.5 .Comparisons between outcomes were not performed due to the low number of seizure-free patients.

Table 2
Neurophysiological data and seizure outcome for each patient with a tuberectomy.

Outcome and complications
There was a trend towards a less favorable outcome in recent years; only the first four patients (1992, 1994, and 2011) achieved Engel I (n = 3) or II (n = 1).The remaining three had a worsening in drop attacks (Engel V).
Two complications were reported.One patient had a new-onset neglect of the right arm, and one had a new-onset urine incontinence and could not walk straight.

Anti-seizure medication use
The median (range) number of ASMs before, at, and after surgery were 2 (2-5), 2 (2-4), and 3 (1-4), respectively.No patient was medication-free at any time point.After surgery, four patients took more ASMs, one took the same number of ASMs, and two took fewer ASMs.The six latest patients had previously tried or were taking vigabatrin at the time of surgery.The information available in medical records to determine whether everolimus had a treatment effect was insufficient.

Comparison between callosotomy and tuberectomy
All patients who underwent callosotomy had at least moderate intellectual disability, and all with available data had autism (6/7) compared with a slight majority (10/18 and 11/18, respectively) in the tuberectomy group (Table 1 and supplementary material).No patient in the callosotomy group underwent PET compared with 24 % (4/17, one missing) in the tuberectomy group .In the callosotomy group, 14 % (1/ 7) underwent intracranial vEEG compared with 53 % (9/17, one missing) in the tuberectomy group .Patients in the callosotomy group had an earlier onset of seizures (median 4 vs. 7 months), underwent surgery at an older age (median 14 years 0 months vs. 5 years 2 months), and had a higher median number of seizures/month (median 200 vs. 150) (Table 1).No patient had cerebellar tubers in either group, and all had tuber type B as dominating tuber type.

Discussion
The present study revealed that the outcome of tuberectomy in patients with TSC in Sweden has improved over time, from approximately 10 % postoperative seizure freedom before 2010 to >50 % after 2010.Data was collected from a comprehensive national register (SNESUR) with complete national coverage, a unique trait in a field where publications usually only report small center-based cohorts.
Notably, several investigative modalities, such as AMT-PET, functional MRI, and increased field strength of MRI, have emerged or improved significantly over the period during which our patient cohort was treated; however, no modality or finding from preoperative investigations could clearly be associated with seizure freedom.This might be caused by intricate networks connecting tubers in patients with TSC, making it a particular challenge to define the epileptogenic zone [22].The advancement in preoperative investigations might also have been reflected in the resected tuber size, with the smallest and largest in recent years, indicating a more tailored approach.A larger resection is  associated with a higher rate of seizure freedom [23].
In recent years, patients proceeding to surgery seemed to have an overall milder TSC phenotype with fewer seizures.In addition, a trend toward surgery at an earlier age was observed.
In a previous Italian study that involved surgeries performed between 1996 and 2019 [24], the rate of seizure freedom was reported to be 55 %.However, only 10 out of 28 patients in the seizure-free group had a definite TSC diagnosis.The phenotype of the patients in the Italian study was less severe than that of those in our cohort.As in our cohort and other studies, the presence of SEGA or a history of infantile spasms was associated with a worse outcome [24,25].In a Chinese study that reviewed surgeries performed between 2006 and 2019, seizure freedom was reported to be comparable to that in the later years of our cohort [26].
Another large retrospective multicenter study from China by Liu et al. [27], which included 364 patients who underwent surgery between 2000 and 2017, showed 71 % and 60 % rates of seizure freedom at 1and 4-year follow-ups, respectively.In discordance with this, we recently published a paper on long-term follow-up and patient satisfaction in the same material as this study, showing the inverse relationship with increased seizure freedom over time, supposedly due to everolimus treatment for the most part [17].
The only previous study that evaluated trends in epilepsy surgery for patients with TSC was a review of children published in 2012 by Ibrahim et al. [28].This review compiled data from multiple small studies worldwide (3-25 patients per study) and showed increased seizure freedom in recent years.However, a good outcome was defined as Engel I-II.The results in this review are consistent with those in the present study, which included data after 2012.The present study's data is also consistent with general trends in epilepsy surgery, with an increased number of surgeries, particularly in the pediatric population [29,30].
The seizure outcome before 2011 in the present study is less favorable than that in the international literature from that time [16,28,31,32].Publication bias in favor of successful cases could be a possible explanation for this finding.The Swedish series included a small number of cases; however, it is unique in its long duration of included operations, the national coverage where all surgeries, not only successful or center-specific, are included, and the prospective collection of outcome data.
The number of epilepsy surgeries for patients with TSC increased per decade, despite a trend of fewer epilepsy surgeries per year in Sweden [33].A similar trend was seen in a previous study, with more patients with TSC having epilepsy surgery after 2000 [13].This could indicate an increased awareness of the possibility of epilepsy surgery in TSC and a changed perception of epilepsy surgery as not only a last-resort treatment for this patient group.The increase in epilepsy surgery cases is also surprising because everolimus has emerged as a compelling new treatment for refractory epilepsy in patients with TSC [34].In our material, three out of four patients achieved seizure freedom not until after initiating everolimus postoperatively, compared to 5-11 % in an extension of the original EXIST-3 study [35].We can only speculate whether epilepsy surgery prior to initiating everolimus might have an additional anti-seizure effect.
The younger age of first reported seizure after 2010 may be explained by the increased awareness of the importance of early seizure detection and the introduction of serial EEG before the onset of clinical seizures [4,5].This is further supported by a study from the international TuberOus SClerosis registry to increase disease Awareness (TOSCA) which showed that a diagnosis of epilepsy in TSC have increased over time [13].
Cognitive development is better among seizure-free patients after epilepsy surgery than among patients who achieve seizure-freedom without epilepsy surgery, especially after surgery at a younger age [3,14].In our study, there was a trend towards earlier surgery in patients with fewer seizures, which may reflect the increasing knowledge of this positive effect [3,7,36].A similar trend has been shown in a previous review article [28].
In the international literature, infantile spasm has previously been considered a contraindication for surgery [16].This is inconsistent with the findings in our study, in which infantile spasms frequently occurred initially, with a decreasing incidence in recent times and none reported after 2011.We can only speculate about the reason for this: one reason might be the addition of vigabatrin as a first-line treatment for epilepsy in patients with TSC.Vigabatrin has been shown to delay the onset of seizures, decrease the incidence of infantile spasms, and decrease the incidence of drug-resistant epilepsy [4].
In our study, 3/6 with mono-focal EEG achieved seizure freedom, in contrast to 1/10 in those who did not.This is in accordance with a previous review by Fallah et al. that showed that unifocal ictal scalp EEG abnormality was associated with good outcome [37].The review by Fallah et al. also showed that EEG/MRI concordance was associated with good outcome, which might explain the higher usage of SPECT in the non-seizure free group in our study [37].It might be reasoned that in individuals without EEG/MRI concordance, SPECT will be used in a further effort to find the epileptogenic tuber even though these patients might not be the optimal cases for surgery.
Callosotomies have been shown to have the best effect on drop attacks, with 55 % achieving seizure freedom [38][39][40].This is consistent with the findings in our study, as three out of seven patients achieved seizure freedom from drop attacks after callosotomy, a similar outcome to that in other studies on TSC [41,42].All patients who had a callosotomy had infantile spasms and often a long duration of epilepsy, which otherwise are poor predictors of good outcome [40].In contrast to previous studies, where partial callosotomy often has lower rates of seizure freedom compared to complete, both patients who had an anterior callosotomy became free from drop attacks [40].

Limitations
The main limitation of this study is the small cohort.However, several measures were taken to ensure complete national coverage.A prospective national register, information in the patient organization magazine, and phone calls to all epilepsy-surgery centers were the methods by which the patients were identified.This is one of the first national prospective studies to evaluate epilepsy surgery for TSC.However, the long duration of this study is also one of the limitations because it limits the ability to collect and re-evaluate preoperative investigations and medical records on clinical reasoning for surgery for the earliest patients, as these are no longer available.The long duration of the study period also makes comparisons over time more difficult as sudden changes in treatment, such as everolimus and preventive vigabatrin, may be hidden in continuous data.This might be mitigated by our approach of grouping at a time point of large changes, such as the time of approval of everolimus for TSC-associated seizures.The small cohort limited our ability to perform statistical analysis and search for predictors of outcomes.The small cohort and the self-reported outcome by patients and caregivers in terms of seizure frequency limited further sub-analysis of Engel classes.

Conclusion
The number of patients with TSC who underwent tuberectomy and achieved seizure freedom at 2FU increased per decade, despite the advancement of everolimus and preventive vigabatrin, which might have decreased the need for epilepsy surgery.The only investigative modality associated with seizure outcome was SPECT, which was overrepresented in those not achieving seizure freedom.
Signs of a more severe disease burden preoperatively in the form of infantile spasms, SEGA, and a high monthly seizure count were associated with a worse outcome and were more often observed before 2010.The introduction of everolimus in 2011 might have rendered surgery less compelling to patients with a more severe TSC-phenotype.The age at surgery was lower after 2010, indicating an increased knowledge of the benefit of surgery at a young age.
Callosotomy operations for patients with TSC were prevalent at the beginning of the 2010s.Anterior callosotomy had a better outcome compared to total callosotomy.

Ethical publication statement
We confirm that the authors have read the journals position on ethical publication and affirm that this study was conducted in accordance with those guidelines.Informed consent were obtained from patients or caregivers as applicable by the ethical review board.

Fig. 1 .
Fig. 1.Seizure outcome over time.The vertical line represents the two time periods.Blue bars represent seizure-free individuals, orange represents a > 75 % decrease in seizure frequency, and gray represents a < 75 % decrease in seizure frequency after epilepsy surgery.

Fig. 2 .
Fig. 2. Correlation between resected tuber size in mm 3 and the year of epilepsy surgery.Each circle represents one operation.N = 14.

Table 1
Patient demographics, clinical characteristics, and preoperative investigations.

Table 3
Comparison between Engel classes for tuberectomies at 2FU.
Mono-focal EEG foci were defined as an interictal activity in one lobe during vEEG.One individual had both types B and C tuber resected, and the resected tuber could not be determined as type A or B in one individual.*one missing,.** two missing, ¤four missing, # five missing AMT-PET: α-[11C]-Methyl-L-tryptophan Positron Emission Tomography; SEGA: Sub-ependymal Giant Cell Astrocytoma; SPECT: Single-photon emission computed tomography; vEEG: video electroencephalography.