Seizure freedom but not epilepsy surgery is associated with fewer neuropsychiatric difficulties in patients with tuberous sclerosis

Background: Drug-resistant epilepsy (DRE) in selected individuals with the rare tuberous sclerosis complex (TSC) may benefit from resective epilepsy surgery. Furthermore, associated neuropsychiatric disorders (TAND) are common in patients with TSC; however, long-term data on how surgery affects neuropsychiatric comorbidities are sparse. Materials and methods: Two retrospective approaches were used to identify children with TSC and DRE with onset at < 18 years of age. The study group (surgical) was identified through the Swedish National Epilepsy Surgery Registry (n = 17), a registry with complete national coverage since 1990 and prospective patient enrolment since 1995. The reference group (non-surgical) was identified by searching medical records retrieved from the tertiary hospital of Southern Sweden (n = 52). Eligible participants were invited to complete the validated TAND lifetime checklist. Those who did not complete the checklist, never had DRE, or were aged < 7 years old were excluded from the study. The reference group was balanced with the study group for putative confounders, in the following hierarchical order: DRE at the survey, age at seizure onset, age at follow-up, and sex. Results: After the balancing procedure, both groups comprised 13 participants. The median time from epilepsy onset to the survey was 18.5 (range: 7.75 – 40.25) and 16.0 (7.33 – 33.5) years in the study and reference groups, respectively. The median time from surgery to the survey was 13 years (range: 4 – 22). No significant differences were found in behavioural problems, autism spectrum disorder diagnosis or symptoms, or intellectual disability between the groups, regardless of surgery. Seizure-free individuals (n = 11) performed better in social skills (p = 0.016), intellectual skills (p = 0.029), and overall TAND scores (p = 0.005) than the non-seizure-free group (n = 15). Conclusion: This is the first study to evaluate TAND comorbidities during the long-term follow-up after epilepsy surgery in patients with TSC. We found no evidence of the adverse effects of TAND comorbidities after tuber-ectomy. However, a larger study that allows for a better adjustment for confounders is needed. Following previous studies, seizure-free individuals had fewer symptoms within most TAND domains compared with the group with uncontrolled epilepsy, indicating less severe symptomatology.


Introduction
Tuberous sclerosis complex (TSC) is a rare genetic disorder with an estimated prevalence of 5.4/100 000 in Sweden [1].In patients with TSC, a disinhibition of the protein kinase mechanistic target of rapamycin (mTOR) leads to hamartomas in several organs.The most prevalent manifestations occur in the central nervous system, including neuropsychiatric disorders, intellectual disabilities, and epilepsy [2].
Epilepsy affects 70-85 % of people with TSC, of which two-thirds of them develop drug-resistant epilepsy (DRE) [1,[5][6][7].Epilepsy is strongly associated with various aspects of TAND [8,9].Pereira et al. showed that the prevalence of symptoms within TAND was 94.4 % in people with epilepsy and only 47.8 % in people without epilepsy [8].In one study, intellectual disability occurred in 61 % of patients with TSC and epilepsy, compared with 12 % of those without a history of epilepsy [5].In a study from 2020 by Gupta et al [10], severe intellectual disability and autism were associated with seizure onset before 2 years of age.None of the patients without epilepsy had a diagnosis of autism.In contrast, the number of individuals with a diagnosis of ADHD was higher among individuals with epilepsy onset after 2 years of age.The authors found no association between epilepsy and anxiety or depression [10].Additionally, in TSC, people with epilepsy lag behind on cognitive testing, whereas people without epilepsy perform better and maintain stable performance over time [11].In line with these results, successful epilepsy surgery at a young age has been associated with better cognitive development in several studies [12][13][14].
Epilepsy surgery may be a treatment option for selected patients with TSC and DRE.However, identifying surgical candidates is challenging because multiple potential epileptic tubers are common.An extensive presurgical workup is required, which may be particularly difficult for patients with TSC and severe TAND problems.This challenge likely leads to reluctance among relatives and professionals to refer patients for surgery, resulting in the underutilisation of surgical treatment for DRE in patients with TSC [7,12].Therefore, a knowledge gap needs to be addressed regarding the long-term postoperative outcomes for patients with TSC.However, we have previously published national data on patient-related measurements and long-term seizure outcomes in a population-based TSC cohort [15].To our knowledge, no previous study has investigated the long-term neuropsychiatric outcomes after resective epilepsy surgery in TSC using a national cohort.Therefore, in this study, we primarily aimed to assess the long-term TAND outcomes in a national cohort of patients with TSC and surgically treated DRE.The secondary aim was to compare the TAND outcomes between seizure-free patients and those with continuing seizures.

Design
This was a national, register-based, cross-sectional cohort study investigating long-term TAND outcomes in individuals with TSC who underwent resective epilepsy surgery.We compared the surgical group to a reference group with no previous epilepsy surgery.

Patient selection
The surgical group included patients with epilepsy and TSC who underwent at least one resective surgical treatment between 1990 and 2018.Patients were identified using the Swedish National Epilepsy Surgery Register (SNESUR), a unique population-based registry with complete national coverage.Patient enrolment was retrospective and prospective between 1990 and 1995 and was exclusively prospective since 1995.The inclusion criteria were a definitive clinical diagnosis of TSC and previous resection surgery.The SNESUR search criteria and collection of variables were based on previous publications on this cohort [15,16].
The non-surgical group included individuals with epilepsy and TSC born between 1 January 1983 and 30 December 2020, identified through a medical record search at Skåne University Hospital.These individuals were followed up at the hospitals and habilitation centres of children and adolescents in southern Sweden, an area with approximately 1,750,655 inhabitants in 2023.The inclusion criteria for this group included a definitive diagnosis of TSC in medical records, no history of epilepsy surgery, and an epilepsy diagnosis.
The limited sample size in our study did not allow for confounding adjustments using multiple regression analysis.Instead, we focused on reducing the risk of confounders by balancing the nonsurgical and surgical groups.The first step involved excluding individuals without DRE before age 18 and those who did not respond to TAND assessments.Children aged < 7 years old were excluded because TAND symptoms may not be recognised early [3].The next step was a ranking in hierarchical order based on factors potentially influencing TAND outcomes: DRE at follow-up, age at seizure onset (cut-off 12 months) [17][18][19], age at survey completion at long-term follow-up, and sex.Fig. 1 illustrates this selection process.Through this balancing procedure, we established the study and reference groups.

Data collection
Patients or caregivers were contacted via telephone for information about the study and confirmation of previous or present epilepsy.Informed consent was obtained, and the Swedish TAND lifetime checklist was mailed to the non-surgical group between October and December 2023 and to the surgical group between September 2019 and January 2020.Two reminders were sent if no response was received within 1 month of the previous The study group were offered personal meetings.All epilepsy surgery centres were contacted and asked to provide confirmatory documentation and contact information from patients and caregivers to ensure complete national coverage of all patients with previous epilepsy surgery.Verbal information about the study was provided to patients and caregivers at the national TSC meeting in Stockholm in May 2019, and written information was published in the magazine of the National TSC Patient Organization on two consecutive issues.
The SNESUR data included sex, intellectual disability, attentiondeficit hyperactivity disorder (ADHD), autism, age at seizure onset, type of surgery (resective surgery or callosotomy), and age at surgery.Data on current DRE, presence of subependymal giant cell astrocytoma, history of infantile spasms, and treatment with everolimus were collected from medical records at the time of survey response.
Data for the non-surgical group were collected from medical records.They included age at seizure onset, intellectual disability, ADHD, autism, epilepsy treatment (including everolimus), presence of previous/current DRE, subependymal giant cell astrocytoma, and a history of infantile spasms.A previous publication provides detailed reporting of this data [7].Drug-resistant epilepsy was defined as continuous debilitating seizures despite treatment with at least two anti-seizure medications at adequate doses.
Intellectual disability was classified into three categories: normal, borderline/mild, and moderate-severe-profound (coherent with domain 5b on the TAND checklist).If the participant or caregiver did not respond, the information relied on cognitive testing in the medical records (applicable to two patients without intellectual disability in the surgery group).

TAND checklist
The TAND checklist has been validated in the English [20].The Swedish translation followed the standard translation procedure and authorisation of the TAND checklist.An agreement was reached between the clinicians and the national TSC patient organisation for translation.The TAND checklist was translated to Swedish using a forward and back translation procedure performed by the local team (physicians and the Swedish TSC patient organisation), followed by a forward and back discussion between the TAND checklist coordinator and the "local team" until a consensus was reached with the TANDem consortium.Once completed, the checklist was typeset and published online.
The TAND lifetime checklist comprises 12 questions (domains), each with several items of current and previous symptoms.The first two questions address general psychomotor development and current abilities (such as dressing, language, and mobility) to give the interviewer a broad understanding of the interviewed.Questions 3-8 measures previous and current problems and consist of yes/no items related to (3) behavioural problems, (4) psychiatric disorders (autism, ADHD, anxiety, and depression), (5) intellectual ability, (6) academic abilities, (7) specific cognitive difficulties, and (8) psychosocial problems (such as low self-esteem and stress within the family).Questions 9 and 12 are a numerical scale from 0 to 10 that summarise the problems within the TAND according to the interviewed (9) and the observer (12).Questions 10 and 11 are free formulation answers about priorities (10) and unaddressed TAND-related issues (11).Question 3 comprises 19 items, which can be summarised as a total score and a subscore concerning social communication problems within the autism spectrum [20].The total score of domain 3 has shown a good correlation to the behavioural screening test Strengths and Difficulties Questionnaire (SDQ) and the subscore of social communication problems to the Social Communication Questionnaire (SCQ), and may therefore be seen as a measure of severity for these problems [20].The TAND checklist is accessible from the TANDem Consortium website (https://tandconsortium.org/ch ecklists/).

Statistics
The Statistical Package for the Social Sciences (SPSS) version 27 (IBM Corp., Armonk, NY, USA.) was used to analyse and generate descriptive data.The data were considered non-parametric because of the small sample size.To compare the study and reference groups, we used Fisherś chi-square test for dichotomous variables and the Mann-Whitney U test for continuous variables.Due to the small sample size, adjusting for confounders using multiple regression was not feasible.Instead, a procedure of balancing the groups based on previously determined factors was employed, as described in section 2.3.Because of the limited sample size, we chose not to adjust for multiple testing which would increase the likelihood of type II errors.All statistical methods were predetermined in consultation with a medical statistician.Fig. 1 were created using BioRender and Fig. 2 RStudio 2023.12.1 + 402 for Windows, utilising the ggplot2 package.
To calculate a numerical value for domain 3 as an indication of severity, we summarised the number of items that answered ''yes'' (maximum 19) and the number of yes responses to item 3 h-3 m (maximum 6) to assess the social communication domain.This aligns with the method used to validate TAND and has shown good agreement with other questionnaires addressing the severity of behavioural problems [20].
We selected the following domains and items from the TAND checklist to compare the groups: the total score of behavioural and social communication subscore of domain 3, a diagnosis of autism spectrum disorder from domain 4, intellectual disability from domain 5, and the overall subjective experience of how much the difficulties within the TAND affect the family (domain 9).The total score in domain 3 was selected to reduce the risk of false positives from analysing each item.We calculated the social communication subscore to avoid missing potential symptoms of autism that may not warrant a full diagnosis of an autism spectrum disorder.

Ethics
The Regional Ethical Review Board in Lund (diary # 2019-00518) approved the study regarding access to medical records and direct contact with patients, and the Regional Ethical Review Board in Gothenburg (diary # 078-09) approved access regarding data from the SNESUR.The funding sources for this study were not involved in the study design, data collection, analysis, or interpretation.

Results
Table 1 presents comparative data of eligible individuals in the reference group who did (n = 17) or did not (n = 7) respond to the TAND checklist.DRE was present in all participants at some point.All individuals without intellectual disabilities completed the questionnaires.On average, younger males who responded were over-represented in the non-response group.Only one patient in the surgical group failed to respond; she did not achieve seizure freedom postoperatively and had a mild intellectual disability.All patients had previous or current focal seizures, five in the study and six in the reference group had infantile spasms.
Table 2 presents demographic data for all individuals included after balancing both groups for DRE, age at seizure onset, sex, and age at cross-section.Two individuals (15 %) in each group completed the survey independently.Further details regarding the outcome of epilepsy surgery have been presented in a previous study [15,16].Within the study group, 54 % experienced ongoing seizures, compared with 62 % in the reference group.The resection site in the study group comprised

Table 1
Comparison between those who responded to the TAND survey and those who did not in the non-surgical group.Data retrieved from the medical records.Age is given per the end of the inclusion period in the no-surgery cohort (30 December 2020).ADHD: Attention-Deficit Hyperactivity Disorder; IQR: Interquartile range; TAND: TSC-associated neuropsychiatric disorders.
nine frontal, two parietal, one multilobar, and one temporal including the hippocampus.Among individuals excluded from the surgical group due to lack of response to TAND or being too young, none experienced ongoing seizures at the time of survey response.Conversely, all individuals in the non-surgical group excluded due to no match in the study group had ongoing seizures.
Table 3 shows no statistically significant differences between the balanced groups in the studied TAND domains.The median score for behavioural difficulties was seven in the surgical group, compared with 13 (out of a maximum of 19) in the non-surgical group; however, this result was not statistically significant.The only instance of no TAND symptoms in domain 9 (overall rating of difficulties, ranging from 0 to 10 maximum) was found in the non-surgical group (n = 1).
The results did not change when all the individuals who responded to the TAND survey were included in the sensitivity analysis.None of the excluded individuals with previous resective surgery had a valid intelligence quotient (IQ) test.Among excluded patients, all (one missing) in the nonsurgical group and none in the surgical group experienced seizures during the survey.

Secondary aim
Table 4 presents the differences between seizure-free individuals and those with DRE during the survey response when pooling together the study and reference groups.The seizure-free group exhibited a lower social communication score in domain 3, more individuals without intellectual disabilities, fewer individuals with moderate-to-severe intellectual disabilities, and a lower score on the overall rating of TAND difficulties (domain 9).Further subgroup analyses of the study and reference groups were not feasible because of the small number of individuals in each group.

Discussion
This is the first nationwide long-term follow-up study of TAND outcomes in patients with TSC after epilepsy surgery.We found no statistically significant differences in TAND scores at long-term followup between the study group with previous resective surgery and the reference group without prior surgery.The secondary aim, which was to compare seizure-free individuals with those having persistent seizures, revealed a significant difference in social communication, intellectual disability, and overall TAND symptoms, with fewer TAND symptoms among seizure-free individuals.
TSC is complex, and multiple confounders were not included in our analyses.Our choice of DRE and age at epilepsy onset as the highestranked balancing factors was based on previous studies.Early seizure onset is strongly correlated with intellectual disability and autism [10,17,21] and in most studies, it is the strongest predictor of subsequent neurodevelopmental disorders [5,10,11,18,22].Seizures at 12 months of age are strong predictors of later cognitive function [11].Previous studies have found a slower gain of cognitive abilities from 8 months of age in patients with TSC, with a significant association between epilepsy severity and seizure onset in the first years of life [17,18].In a previous study, Tye et al. linked a higher tuber burden to increased seizure severity and low IQ [18].Unfortunately, we lacked sufficient data on individuals' tuber burden; therefore, we could not include this factor in our balancing procedure.However, it could be argued that early seizure onset, DRE, tuber burden, and genotype are probably overlapping and interdependent entities.
The TAND checklist was developed to better describe, diagnose, and screen for neurodevelopmental difficulties in patients with TSC [2].As presented in Table 3, over half of the participants had moderate-tosevere intellectual disabilities.These rates were similar in the study and reference groups.In previous studies, the rates of moderate to profound intellectual disability in TSC were 27-28 % [3,23].The higher rate of low intellectual performance in our study cohort could indicate a more severe impact of the disease on general brain function.Another aspect raised in previous surgical TSC series is the probability of IQ improvement at follow-up.In a large TSC study by Liu et al., 28 % of the participants had improved postoperative IQ, with a mean improvement of 6.1 IQ points.Multivariate analysis indicated that freedom from postoperative seizures was associated with postoperative IQ improvement [19].In a study by Grayson et al. [14], 19 children with TSC and epilepsy surgery before 2 years of age were compared to children with TSC and no epilepsy, controlled epilepsy, and drug-resistant epilepsy.The study found attenuated gains in neurodevelopment over time in all patients with epilepsy.The surgery group scored the lowest on all cognitive tests utilised.However, after surgery, a stabilisation on cognitive test scores was observed [14].In contrast to the study by Grayson et al, none in our study had surgery before 2 years of age, which may be one reason why we found no indication that the severity or prevalence of intellectual disability differed between the study group and the reference group at the long-term follow-up.However, our  findings support the association between seizure freedom and a normal IQ (Table 4).
Our long-term follow-up supports the finding that seizure freedom is associated with a higher intellectual level, independent of whether it is achieved by optimising pharmacological treatment or resective surgery.In contrast, Stomberg et al. demonstrated that patients who achieved seizure freedom after epilepsy surgery experienced a significant increase in their developmental level, a phenomenon not observed in patients who achieved seizure freedom without surgery [12].Stomberg et al. compared patients who underwent surgery with those who were not eligible for surgery, suggesting the possibility of selection bias.Patients deemed ineligible may have had more severe disease and multiple epileptogenic tubers.In our study, there were no significant differences in the TAND domains between the surgical and non-surgical groups at the long-term follow-up; however, we lacked longitudinal data, and our cohort was too small for subgroup comparisons.Another difference between our study and Stomberg's study was our attempt to adjust for confounders by balancing our surgical group with a reference group.In contrast, Stomberg et al. did not adjust for confounders.
Previous studies on this and other cohorts have observed that caregivers of patients with TSC sometimes declined referrals for evaluation and offers to perform surgery [7,12,24].The need for extensive preoperative investigations with prolonged in-hospital stays is a deterrent [7,25].Severe autism, intellectual disabilities, and behavioural problems present significant challenges.However, we found no significant difference in the rate of TAND symptoms between the groups.This could indicate that children with severe neuropsychiatric disorders were able to undergo extensive preoperative investigations and surgery, but it cannot be excluded that these symptoms developed later, although unlikely due to higher age at surgery in this cohort.These results, combined with our previous results of very high satisfaction after epilepsy surgery [15], may strengthen the decision of patients with TSC and their caregivers to undergo evaluations and possibly subsequent surgery.In a previous report on this national cohort of patients undergoing epilepsy surgery in Sweden since 1990, no complications were reported in the resective surgery group [16].In the present study, we found no significant differences in TAND between the study and reference groups, indicating that surgery does not precipitate the deterioration of TAND at the group level.In multiple studies, successful surgery has been associated with achieving seizure freedom, neurodevelopmental improvement, and improved quality of life [12,13,15,19,25].These reports, combined with the results of the present study, may provide caregivers with additional information when deciding whether to proceed with epilepsy surgery.

Strengths and limitations
This study has some limitations.First, it was a limited cohort, including patients with TSC and previous epilepsy surgery in Sweden, potentially resulting in an underpowered sample size to determine statistical differences.However, a notable strength is our reasonable certainty that all potential study participants in Sweden were approached, with all but one agreeing to participate.This unique approach enables a national-based analysis without selection bias in the study cohort.Second, the small sample size did not allow multiple regression analyses for confounder adjustments, which we attempted to address by balancing a population-based reference group.Third, only postoperative TAND data were available, with longitudinal data on outcome parameters lacking.A pre and postoperative comparison has been done previously, and it showed a good effect of seizure freedom on cognitive function.No previous study has compared balanced groups with or without previous epilepsy surgery in patients across all ages regarding TAND outcomes at long-term follow-up.Fourth, participants in the reference group were recruited from Southern Sweden, contrasting the national data on surgical treatment in the SNESUR.This discrepancy may have led to skewed results.However, this group has recently been extensively reported and appears to represent the general TSC population [7,26].Fifth, this study was conducted before the TAND self-evaluation instrument was developed [27]; therefore, the TAND-lifetime checklist, originally not intended for self-evaluation, was used.This methodology has been successfully used in previous studies [28].Participants were offered a personal visit and, if declined, were provided with the contact information of the study personnel for any questions.Sixth, we lacked data on the exact duration of seizure freedom.However, previous studies have shown that a younger age of seizure onset is the highest predictor of future intellectual disability [5,10,11,18,22], not the duration of epilepsy [10], and this was taken into account in the analysis.Also, in the surgery group, most individuals achieved seizure freedom at the time of surgery and thereafter remained seizure-free [15], although we lack this data for the reference group.Finally, to reduce the number of statistical analyses and the risk of false positives, domain 3 items were summarised into a composite score.This approach was necessary; however, it may lead to false-negative results owing to potential differences between the groups in specific items.In addition, we chose not to include a correction of multiple testing, as this would further reduce the risk of false negatives.

Conclusion
This is the first study with national data to evaluate TAND comorbidities during the long-term follow-up after epilepsy surgery in patients with TSC.We found no evidence of the adverse effects of TAND

Table 4
Comparison of TAND to seizure situation in the balanced groups.

Fig. 1 .
Fig. 1.Flowchart showing the selection process for each group.

Fig. 2 .
Fig. 2. Comparison of the chosen TAND domains between individuals with drug-resistant epilepsy and seizure-free individuals at the time of the survey.A higher score on the domain indicates more severe neuropsychiatric symptoms.DRE: Drug-resistant epilepsy.

Table 2
Comparison of demographic data between the two balanced groups.

Table 3
Comparison of TAND results for individuals in the two balanced groups.