Exit Strategy: Balancing the Risks and Rewards of Antiseizure Medication Withdrawal

The majority of people with epilepsy achieves long-term seizure-freedom and may consider withdrawal of their anti-seizure medications (ASMs). Withdrawal of ASMs can yield substantial benefits but may be associated with potential risks. This review critically examines the existing literature on ASM withdrawal, emphasizing evidence-based recommendations, where available. Our focus encompasses deprescribing strategies for individuals who have attained seizure freedom through medical treatment, those who have undergone successful epilepsy surgery, and individuals initiated on ASMs following acute symptomatic seizures. We explore state-of-the-art prognostic models in these scenarios that could guide the decision-making process. The review underscores the importance of a collaborative shared-decision approach between patients, caregivers, and physicians. We describe the subjective and objective factors influencing these decisions and illustrate how trade-offs may be effectively managed in practice.

Predictive models of epilepsy outcomes

PURPOSE OF REVIEW

Multiple complex medical decisions are necessary in the course of a chronic disease like epilepsy. Predictive tools to assist physicians and patients in navigating this complexity have emerged as a necessity and are summarized in this review.

RECENT FINDINGS

Nomograms and online risk calculators are user-friendly and offer individualized predictions for outcomes ranging from safety of antiseizure medication withdrawal (accuracy 65-73%) to seizure-freedom, naming, mood, and language outcomes of resective epilepsy surgery (accuracy 72-81%). Improving their predictive performance is limited by the nomograms' inability to ingest complex data inputs. Conversely, machine learning offers the potential of multimodal and expansive model inputs achieving human-expert level accuracy in automated scalp electroencephalogram (EEG) interpretation but lagging in predictive performance or requiring validation for other applications.

SUMMARY

Good to excellent predictive models are now available to guide medical and surgical epilepsy decision-making with nomograms offering individualized predictions and user-friendly tools, and machine learning approaches offering the potential of improved performance. Future research is necessary to bridge the two approaches for optimal translation to clinical care.

Early antiseizure medication withdrawal and risk of seizure recurrence in children after epilepsy surgery: A retrospective study

OBJECTIVE

The timing of antiseizure medication (ASM) withdrawal in children after epilepsy surgery remains controversial and lacks recognized standards. Given the various negative effects of ASM on development in children, this study aimed to evaluate the safety and feasibility of early ASM withdrawal after epileptic resection surgery.

METHODS

We retrospectively assessed the seizure outcomes and ASM profiles of children who had undergone epileptic resection surgery between August 2015 and August 2020 and attempted ASM reduction in the early postoperative phase. Tapering the dose of ASM was attempted when children were seizure-free with no interictal epileptiform discharges (IEDs) on electroencephalogram (EEG) for at least 6 months postoperatively.

RESULTS

This study included 145 children with a median follow-up duration of 40 months. Early ASM tapering was attempted postoperatively in 99 (68.3 %) children. Postoperative ASM discontinuation was attempted in 87 (60.0 %) children. Nine (9.1 %) children experienced seizure recurrence during the ASM reduction stage, and 10 (11.5 %) experienced recurrence after ASM discontinuation. Incomplete resection (P = 0.003) and postoperative seizures before ASM tapering (P = 0.003) were independent predictors of seizure recurrence during and after early ASM withdrawal.

SIGNIFICANCE

ASM withdrawal is viable and safe to be initiated in children who are seizure-free postoperatively and have no IEDs on the scalp EEG for at least 6 months. Children with incomplete resection and postoperative seizures before ASM withdrawal are at a higher risk of seizure recurrence and may need to continue ASM for a longer period.

Predicting seizure recurrence after an initial seizure-like episode from routine clinical notes using large language models: a retrospective cohort study

BACKGROUND

The evaluation and management of first-time seizure-like events in children can be difficult because these episodes are not always directly observed and might be epileptic seizures or other conditions (seizure mimics). We aimed to evaluate whether machine learning models using real-world data could predict seizure recurrence after an initial seizure-like event.

METHODS

This retrospective cohort study compared models trained and evaluated on two separate datasets between Jan 1, 2010, and Jan 1, 2020: electronic medical records (EMRs) at Boston Children's Hospital and de-identified, patient-level, administrative claims data from the IBM MarketScan research database. The study population comprised patients with an initial diagnosis of either epilepsy or convulsions before the age of 21 years, based on International Classification of Diseases, Clinical Modification (ICD-CM) codes. We compared machine learning-based predictive modelling using structured data (logistic regression and XGBoost) with emerging techniques in natural language processing by use of large language models.

FINDINGS

The primary cohort comprised 14 021 patients at Boston Children's Hospital matching inclusion criteria with an initial seizure-like event and the comparison cohort comprised 15 062 patients within the IBM MarketScan research database. Seizure recurrence based on a composite expert-derived definition occurred in 57% of patients at Boston Children's Hospital and 63% of patients within IBM MarketScan. Large language models with additional domain-specific and location-specific pre-training on patients excluded from the study (F1-score 0·826 [95% CI 0·817-0·835], AUC 0·897 [95% CI 0·875-0·913]) performed best. All large language models, including the base model without additional pre-training (F1-score 0·739 [95% CI 0·738-0·741], AUROC 0·846 [95% CI 0·826-0·861]) outperformed models trained with structured data. With structured data only, XGBoost outperformed logistic regression and XGBoost models trained with the Boston Children's Hospital EMR (logistic regression: F1-score 0·650 [95% CI 0·643-0·657], AUC 0·694 [95% CI 0·685-0·705], XGBoost: F1-score 0·679 [0·676-0·683], AUC 0·725 [0·717-0·734]) performed similarly to models trained on the IBM MarketScan database (logistic regression: F1-score 0·596 [0·590-0·601], AUC 0·670 [0·664-0·675], XGBoost: F1-score 0·678 [0·668-0·687], AUC 0·710 [0·703-0·714]).

INTERPRETATION

Physician's clinical notes about an initial seizure-like event include substantial signals for prediction of seizure recurrence, and additional domain-specific and location-specific pre-training can significantly improve the performance of clinical large language models, even for specialised cohorts.

FUNDING

UCB, National Institute of Neurological Disorders and Stroke (US National Institutes of Health).

Aberrant structural rich club organization in temporal lobe epilepsy with focal to bilateral tonic-clonic seizures

OBJECTIVE

The purpose of this study was to assess the differences of topological characteristic and rich club organization between temporal lobe epilepsy (TLE) patients with focal seizure (FS) only and those with focal to bilateral tonic-clonic seizures (FBTCS).

METHODS

We recruited 130 unilateral TLE patients, of which 57 patients with FS only and 73 patients with both FS and FBTCS, and 68 age- and gender-matched healthy controls (HC). Whole-brain networks were constructed based on diffusion weighted imaging data. Graph theory was applied to quantify the topological network metrics and rich club organization. Network-based statistic (NBS) analysis was administered to investigate the difference in edge-wise connectivity strength. The non-parametric permutation test was applied to evaluate the differences between groups. Benjamini-Hochberg FDR at the alpha of 5% was carried out for multiple comparations.

RESULTS

In comparison with HC, both the FS and FBTCS group displayed a significant reduction in whole-brain connectivity strength and global efficiency. The FBTCS group showed lower connectivity strength both in the rich club and feeder connections compared to HC. The FS group had lower connectivity strength in the feeder and local connections compared to HC. NBS analysis revealed a wider range of decreased connectivity strength in the FBTCS group, involving 90% of the rich club regions, mainly affecting temporal-subcortical, frontal-parietal, and frontal-temporal lobe, the majority decreasing connections were between temporal lobe and stratum. While the decreased connectivity strength in the FS group were relatively local, involving 50% of rich club regions, mainly concentrated on the temporal-subcortical lobe.

CONCLUSIONS

Network integration was reduced in TLE. TLE with FBTCS selectively disrupted the rich club regions, while TLE with FS only were more likely to affect the non-rich club regions, emphasizing the contribution of rich club organization to seizure generalization.

Comparison of the real-world effectiveness of vertical versus lateral functional hemispherotomy techniques for pediatric drug-resistant epilepsy: A post hoc analysis of the HOPS study

Objective

This study was undertaken to determine whether the vertical parasagittal approach or the lateral peri‐insular/peri‐Sylvian approach to hemispheric surgery is the superior technique in achieving long‐term seizure freedom.

Methods

We conducted a post hoc subgroup analysis of the HOPS (Hemispheric Surgery Outcome Prediction Scale) study, an international, multicenter, retrospective cohort study that identified predictors of seizure freedom through logistic regression modeling. Only patients undergoing vertical parasagittal, lateral peri‐insular/peri‐Sylvian, or lateral trans‐Sylvian hemispherotomy were included in this post hoc analysis. Differences in seizure freedom rates were assessed using a time‐to‐event method and calculated using the Kaplan–Meier survival method.

Results

Data for 672 participants across 23 centers were collected on the specific hemispherotomy approach. Of these, 72 (10.7%) underwent vertical parasagittal hemispherotomy and 600 (89.3%) underwent lateral peri‐insular/peri‐Sylvian or trans‐Sylvian hemispherotomy. Seizure freedom was obtained in 62.4% (95% confidence interval [CI] = 53.5%–70.2%) of the entire cohort at 10‐year follow‐up. Seizure freedom was 88.8% (95% CI = 78.9%–94.3%) at 1‐year follow‐up and persisted at 85.5% (95% CI = 74.7%–92.0%) across 5‐ and 10‐year follow‐up in the vertical subgroup. In contrast, seizure freedom decreased from 89.2% (95% CI = 86.3%–91.5%) at 1‐year to 72.1% (95% CI = 66.9%–76.7%) at 5‐year to 57.2% (95% CI = 46.6%–66.4%) at 10‐year follow‐up for the lateral subgroup. Log‐rank test found that vertical hemispherotomy was associated with durable seizure‐free progression compared to the lateral approach (p = .01). Patients undergoing the lateral hemispherotomy technique had a shorter time‐to‐seizure recurrence (hazard ratio = 2.56, 95% CI = 1.08–6.04, p = .03) and increased seizure recurrence odds (odds ratio = 3.67, 95% CI = 1.05–12.86, p = .04) compared to those undergoing the vertical hemispherotomy technique.

Significance

This pilot study demonstrated more durable seizure freedom of the vertical technique compared to lateral hemispherotomy techniques. Further studies, such as prospective expertise‐based observational studies or a randomized clinical trial, are required to determine whether a vertical approach to hemispheric surgery provides superior long‐term seizure outcomes.

Individual prediction of motor vehicle accidents for patients with epilepsy

The objective of the study was to design a clinically useful tool to predict the risk of seizure-related motor vehicle accidents (MVAs) for people with epilepsy (PWE). Participants were patients who visited our epilepsy center in West China Hospital from October 2012 to October 2019 and were divided into a primary cohort and a validation cohort. Ultimately, we included 525 patients in the primary cohort and 86 patients in the validation cohort. Proportional hazard regression was performed to measure the prognostic factors of car accidents. The outcome was used to create a nomogram model. The final model had 7 factors, with a C-index of 0.85 (95% CI, 0.80-0.91), to predict the possibility of non-MVA for PWE. For the validation cohort, the C-index was 0.83 (95% CI, 0.72-0.95). This nomogram model can offer more individualized advice to PWE who are still driving by estimating the risk of car accidents.

Underutilization of epilepsy surgery: Part II: Strategies to overcome barriers

Interventions focused on utilization of epilepsy surgery can be divided into groups: those that improve patients' access to surgical evaluation and those that facilitate completion of the surgical evaluation and treatment. Educational intervention, technological innovation, and effective coordination and communication can significantly improve patients' access to surgery. Patient and public facing, individualized (analog and/or digital) communication can raise awareness and acceptance of epilepsy surgery. Educational interventions aimed at providers may mitigate knowledge gaps using practical and concise consensus statements and guidelines, while specific training can improve awareness around implicit bias. Innovative technology, such as clinical decision-making toolkits within the electronic medical record (EMR), machine learning techniques, online decision-support tools, nomograms, and scoring algorithms can facilitate timely identification of appropriate candidates for epilepsy surgery with individualized guidance regarding referral appropriateness, postoperative seizure freedom rate, and risks of complication after surgery. There are specific strategies applicable for epilepsy centers' success: building a multidisciplinary setup, maintaining/tracking volume and complexity of cases, collaborating with other centers, improving surgical outcome with reduced complications, utilizing advanced diagnostics tools, and considering minimally invasive surgical techniques. Established centers may use other strategies, such as multi-stage procedures for multifocal epilepsy, advanced functional mapping with tailored surgery for epilepsy involving the eloquent cortex, and generation of fresh hypotheses in cases of surgical failure. Finally, improved access to epilepsy surgery can be accomplished with policy changes (e.g., anti-discrimination policy, exemption in transportation cost, telehealth reimbursement policy, patient-centered epilepsy care models, pay-per-performance models, affordability and access to insurance, and increased funding for research). Every intervention should receive regular evaluation and feedback-driven modification to ensure appropriate utilization of epilepsy surgery.

Models for predicting treatment efficacy of antiepileptic drugs and prognosis of treatment withdrawal in epilepsy patients

Although antiepileptic drugs (AEDs) are the most effective treatment for epilepsy, 30–40% of patients with epilepsy would develop drug-refractory epilepsy. An accurate, preliminary prediction of the efficacy of AEDs has great clinical significance for patient treatment and prognosis. Some studies have developed statistical models and machine-learning algorithms (MLAs) to predict the efficacy of AEDs treatment and the progression of disease after treatment withdrawal, in order to provide assistance for making clinical decisions in the aim of precise, personalized treatment. The field of prediction models with statistical models and MLAs is attracting growing interest and is developing rapidly. What’s more, more and more studies focus on the external validation of the existing model. In this review, we will give a brief overview of recent developments in this discipline.

Temporal lobe epilepsy surgery in children and adults: A multicenter study

OBJECTIVE

To assess seizure and cognitive outcomes and their predictors in children (<16 years at surgery) and adults undergoing temporal lobe epilepsy (TLE) surgery in eight Italian centers.

METHODS

This is a retrospective multicenter study. We performed a descriptive analysis and subsequently carried out multivariable mixed-effect models corrected for multiple comparisons.

RESULTS

We analyzed data from 511 patients (114 children) and observed significant differences in several clinical features between adults and children. The possibility of achieving Engel class IA outcome and discontinuing antiepileptic drugs (AEDs) at last follow-up (FU) was significantly higher in children (P = .006 and < .0001). However, percentages of children and adults in Engel class I at last FU (mean ± SD, 45.9 ± 17 months in children; 45.9 ± 20.6 months in adults) did not differ significantly. We identified different predictors of seizure outcome in children vs adults and at short- vs long-term FU. The only variables consistently associated with class I outcome over time were postoperative electroencephalography (EEG) in adults (abnormal, improved,odds ratio [OR] = 0.414, P = .023, Q = 0.046 vs normal, at 2-year FU and abnormal, improved, OR = 0.301, P = .001, Q = 0.002 vs normal, at last FU) and the completeness of resection of temporal magnetic resonance (MR) abnormalities other than hippocampal sclerosis in children (OR = 7.93, P = .001, Q = 0.003, at 2-year FU and OR = 45.03, P < .0001, Q < 0.0001, at last FU). Cognitive outcome was best predicted by preoperative performances in either age group.

SIGNIFICANCE

Clinical differences between adult and pediatric patients undergoing TLE surgery are reflected in differences in long-term outcomes and predictors of failures. Children are more likely to achieve sustained seizure freedom and withdraw AEDs after TLE surgery. Earlier referral should be encouraged as it can improve surgical outcome.

Algorithms in clinical epilepsy practice: Can they really help us predict epilepsy outcomes?

Significant technological advances have improved our ability to localize epilepsy and investigate the electrophysiology in patients undergoing preparation for epilepsy surgery. Conversely, our process of decision-making and outcome prediction has remained essentially restricted to subjective clinical judgment. This may have hindered our ability to improve outcomes. In this review, we highlight the cognitive biases that interfere with medical decision-making and present data on the use of algorithms and statistical models in general health care, before pivoting to discuss applications in the context of epilepsy.

Prediction of the recurrence risk in patients with epilepsy after the withdrawal of antiepileptic drugs

Many seizure-free patients who consider withdrawing from antiepileptic drugs (AEDs) hope to discontinue treatment to avoid adverse effects. However, withdrawal has certain risks that are difficult to predict. In this study, we performed a literature review, summarized the causes of significant variability in the risk of postwithdrawal recurrent seizures, and reviewed study data on the age at onset, cause, types of seizures, epilepsy syndrome, magnetic resonance imaging (MRI) abnormalities, epilepsy surgery, and withdrawal outcomes of patients with epilepsy. Many factors are associated with recurrent seizures after AED withdrawal. For patients who are seizure-free after treatment, the role of an electroencephalogram (EEG) alone in ensuring safe withdrawal is limited. A series of prediction models for the postwithdrawal recurrence risk have incorporated various potentially important factors in a comprehensive analysis. We focused on the populations of studies investigating five risk prediction models and analyzed the predictive variables and recommended applications of each model, aiming to provide a reference for personalized withdrawal for patients with epilepsy in clinical practice.

Predicting seizure freedom with AED treatment in newly diagnosed patients with MRI-negative epilepsy: A large cohort and multicenter study

OBJECTIVE

We developed and validated a prediction score for predicting the probability of 6-month and 12-month seizure freedom of antiepileptic drug (AED) treatment in newly diagnosed patients with magnetic resonance imaging (MRI)-negative epilepsy.

METHODS

The development cohort included 543 consecutive patients from the Epilepsy Center of Henan Provincial People's Hospital, while the validation cohorts included 493 consecutive patients in two independent cohorts. Univariate analysis and a forward and backward elimination of multivariate Cox regression analysis were used to select predictive factors. The performance of the score was evaluated with C-index, calibration plots, and decision curve analysis. The risk stratification was also performed.

RESULTS

The score included five routinely available predictors including Circadian rhythms, Electroencephalography before AED treatment, Neuropsychiatric disorders, Perinatal brain injury, and History of central nervous system infection (CENPH score). When applied to the external validation cohort, the score showed good discrimination with C-index (development group: 0.83; validation group: 0.78), and calibration plots indicated well calibration, as well as the decision curve analysis showed good predictive accuracy and clinical values in four cohorts. The points of the score were categorized to the following three probability levels for predicting seizure freedom: high probability (0-83.11 points), medium probability (83.11-122.71 points), and low probability (>122.71 points). And online calculator was established to make this score easily applicable in clinical practice.

CONCLUSIONS

We established a simple, practical, and evidence-based prediction score for predicting seizure freedom with AEDs to aid in the clinical consultation and treatment decision for the newly diagnosed patients with MRI-negative epilepsy.

Story Learning Test: Decelerated Learning and Accelerated Forgetting in Children with Epilepsy

Introduction

Increasing interest is seen for early and late memory consolidation and accelerated forgetting, but little is known about these phenomena in children with epilepsy. The present study analysed the trajectory of learning and retention in typically developing children and children with epilepsy on a story learning test.

Methods

285 children, 126 typically developing children and 159 children with epilepsy, in ages between 4 and 10 years and Full-Scale IQs ≥ 75, were given a specifically designed story learning test (iter-sein). The learning phase included Initial reading and a Free Recall trial with 10 Questions, and up to three repetition trials with Questions. Trials of Delayed Free Recall and Questions followed after half an hour, the next day and 1 week later. With several repeated measures analyses of variance, level of performance and gains or losses over time were analysed.

Results

Age-dependent learning was seen after repetitions. On the Questions, typically developing children outperformed children with epilepsy increasingly, due to smaller gains after the second trial. Learned information was similarly preserved. Free Recall showed similar performance for both groups up to day 2. A week later, a conspicuous loss of information was observed in the children with epilepsy, whilst typically developing children retained the information. On index scores, reliable cognitive loss of information was seen in epilepsy in 24.5% of the children. Semantic neuropsychological tasks and severity measures of epilepsy were associated with level of performance.

Discussion

The results provided evidence for early decelerated learning and late accelerated forgetting in children with epilepsy.

Epilepsy Education: Recent Advances and Future Directions

PURPOSE OF REVIEW

The goal of this review is to survey the current literature on education in epilepsy and provide the most up-to-date information for physicians involved in the training of future doctors on this topic. We intended to review what opportunities exist to enhance our current teaching practices that may not be well-known or widely used, but may be adapted to a broader audience.

RECENT FINDINGS

Many new techniques adopting principles of education (e.g., retrieval practice and spaced learning) or new technologies (e.g., pre-recorded lectures, computer-enhanced modules, and simulation practice) have been trialled to enhance medical education in epilepsy with some success. Many of these techniques are currently adaptable to a wider audience or may soon be available. The use of these opportunities more broadly may allow expansion of educational research opportunities as well as enhancing our ability to pass on information. As the knowledge base in epilepsy continues to dramatically expand, we need to keep evaluating our teaching techniques to ensure we are able to pass along this knowledge to our future providers.

Development and validation of a prediction model for the probability of responding to placebo in antidepressant trials: a pooled analysis of individual patient data

BACKGROUND

Identifying potential placebo responders among apparent drug responders is critical to dissect drug-specific and nonspecific effects in depression.

OBJECTIVE

This project aimed to develop and test a prediction model for the probability of responding to placebo in antidepressant trials. Such a model will allow us to estimate the probability of placebo response among drug responders in antidepressants trials.

METHODS

We identified all placebo-controlled, double-blind randomised controlled trials (RCTs) of second generation antidepressants for major depressive disorder conducted in Japan and requested their individual patient data (IPD) to pharmaceutical companies. We obtained IPD (n=1493) from four phase II/III RCTs comparing mirtazapine, escitalopram, duloxetine, paroxetine and placebo. Out of 1493 participants in the four clinical trials, 440 participants allocated to placebo were included in the analyses. Our primary outcome was response, defined as 50% or greater reduction on Hamilton Rating Scale for Depression at study endpoint. We used multivariable logistic regression to develop a prediction model. All available candidate of predictor variables were tested through a backward variable selection and covariates were selected for the prediction model. The performance of the model was assessed by using Hosmer-Lemeshow test for calibration and the area under the ROC curve for discrimination.

FINDINGS

Placebo response rates differed between 31% and 59% (grand average: 43%) among four trials. Four variables were selected from all candidate variables and included in the final model: age at onset, age at baseline, bodily symptoms, and study-level difference. The final model performed satisfactorily in terms of calibration (Hosmer-Lemeshow p=0.92) and discrimination (the area under the ROC curve (AUC): 0.70).

CONCLUSIONS

Our model is expected to help researchers discriminate individuals who are more likely to respond to placebo from those who are less likely so.

CLINICAL IMPLICATIONS

A larger sample and more precise individual participant information should be collected for better performance. Examination of external validity in independent datasets is warranted.

TRIAL REGISTRATION NUMBER

CRD42017055912.

Antiepileptic Drug Withdrawal after Surgery in Children with Focal Cortical Dysplasia: Seizure Recurrence and Its Predictors

Background and Purpose

This study investigated the seizure recurrence rate and potential predictors of seizure recurrence following antiepileptic drug (AED) withdrawal after resective epilepsy surgery in children with focal cortical dysplasia (FCD).

Methods

We retrospectively analyzed the records of 70 children and adolescents with FCD types I, II, and IIIa who underwent resective epilepsy surgery between 2004 and 2015 and were followed for at least 2 years after surgery.

Results

We attempted AED withdrawal in 40 patients. The median time of starting the AED reduction was 10.8 months after surgery. Of these 40 patients, 14 patients (35%) experienced seizure recurrence during AED reduction or after AED withdrawal. Half of the 14 patients who experienced recurrence regained seizure freedom after AED reintroduction and optimization. Compared with their preoperative status, the AED dose or number was decreased in 57.1% of patients, and remained unchanged in 14.3% after surgery. A multivariate analysis found that incomplete resection (p=0.004) and epileptic discharges on the postoperative EEG (p=0.025) were important predictors of seizure recurrence after AED withdrawal. Over the mean follow-up duration of 4.5 years after surgery, 34 patients (48.6% of the entire cohort) were seizure-free with and without AEDs.

Conclusions

Children with incomplete resection and epileptic discharges on postoperative EEG are at a high risk of seizure recurrence after drug withdrawal. Complete resection of FCD may lead to a favorable surgical outcome and successful AED withdrawal after surgery.