Risk of seizure recurrence after achieving initial seizure freedom on the ketogenic diet

Authors

  • Katherine S. Taub,

    Corresponding author
    1. Division of Neurology, Departments of Pediatrics and Neurology, The Children's Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A
    • Address correspondence to Katherine S. Taub, Division of Neurology, Children's Hospital of Philadelphia, 3501 Civic Center Blvd, 10th Floor, Philadelphia, PA 19104, U.S.A. E-mail:taubk@email.chop.edu

    Search for more papers by this author
  • Sudha Kilaru Kessler,

    1. Division of Neurology, Departments of Pediatrics and Neurology, The Children's Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A
    Search for more papers by this author
  • A. G. Christina Bergqvist

    1. Division of Neurology, Departments of Pediatrics and Neurology, The Children's Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania, U.S.A
    Search for more papers by this author

Summary

Objective

Few studies have examined the long-term sustainability of complete seizure freedom on the ketogenic diet (KD). The purpose of this study was to describe the risk of seizure recurrence in children who achieved at least 1 month of seizure freedom on the KD, and to assess clinical features associated with sustained seizure freedom.

Methods

Records of patients initiated on the KD at The Children's Hospital of Philadelphia (CHOP) from 1991 to 2009 were reviewed. Subjects who attained seizure freedom for at least 1 month within 2 years were included in the study. Seizure frequency was recorded based on caregiver-reported seizure diaries as unchanged, improved, or worse compared to baseline. Those patients with seizure freedom ≥1 year were compared to those with seizure freedom <1 year in terms of demographics, age of seizure onset, number of antiepileptic drugs (AEDs) prior to KD, and epilepsy classification.

Results

Of 276 patients initiated on the KD, 65 patients (24%) attained seizure freedom for a minimum of 1 month. The majority of these patients had daily seizures. The median time to seizure freedom after KD initiation was 1.5 months. Seizures recurred in 53 patients (82%), with a median time to seizure recurrence of 3 months. However, seizure frequency after initial recurrence remained far less than baseline. No clinical features were identified as risk factors for seizure recurrence.

Significance

Seizure recurrence on the KD after 1 month of seizure freedom most often occurred as occasional breakthrough seizures and not a return to baseline seizure frequency. This study provides evidence to support the continued use of the KD in patients with initial seizure freedom even after breakthrough seizures.

A PowerPoint slide summarizing this article is available for download in the Supporting Information section here.

image

Dr. Katherine S. Taub is an Assistant Professor of Neurology at the Perelman School of Medicine at the University of Pennsylvania, and an epileptologist at The Children's Hospital of Philadelphia.

Epilepsy affects nearly 1% of American children,[1] and it carries a substantial lifetime risk of medical and psychosocial morbidity,[2] particularly for the 20–30% of children with epilepsy who have treatment-resistant seizures.[3] The ketogenic diet (KD), in which fats comprise 90% of caloric intake, is an effective therapy for children with treatment-resistant epilepsy, offering a greater chance of seizure freedom than continued trials of antiepileptic drugs (AEDs). In short-term outcome studies, 15–20% of patients achieve seizure freedom and one third achieve >90% seizure reduction.[4-7] Less is known about the ability of the KD to sustain long-term seizure freedom. The purpose of this study was to describe the risk of seizure recurrence in children who achieved at least 1 month of seizure freedom on the KD, and to assess clinical features associated with sustained seizure freedom.

Methods

The Children's Hospital of Philadelphia (CHOP) Institutional Review Board approved this retrospective cohort study. All available records of patients initiated on the KD at CHOP between 1991 and 2009 were reviewed. Records for one third of the patients were not available for review because patients were initiated on the KD prior to 2000, were no longer active patients at CHOP, or had records archived off-site. Patients were started on the KD using standardized protocols.[8] After a randomized controlled trial in 2005 at CHOP of gradual versus fasting KD initiation, all subsequent patients have used the gradual initiation protocol. Subjects who were included in the study attained seizure freedom for at least 1 month within 2 years of initiating the KD. This prolonged period of time was considered because positive effects of the KD are not necessarily limited to the period immediately after initiation.[9] Early responders were defined as those who became seizure-free within 3 months of KD initiation. Seizure frequency was recorded at each patient encounter based on caregiver-reported seizure diaries. Seizure frequency after seizure recurrence was categorized as unchanged, improved, or worse compared to baseline. Because exact dates of recurrence were not always available, time to recurrence was categorized as <3 months, 3 to 6 months, 6 to 12 months, and >12 months after achieving seizure freedom. Descriptive statistics included means and standard deviations, or medians and ranges, as appropriate. Comparisons between patients with seizure freedom for ≥1 year and those with seizure freedom for <1 year were made using t-tests and Wilcoxon rank sum tests for parametric and nonparametric continuous variables, respectively, and by Fisher's exact tests for dichotomous variables. Kaplan-Meier curves were generated to examine the probability of remaining seizure-free after initial seizure freedom. Univariate Cox proportional hazards regression was used to compare differences in time to recurrence for clinical factors including early or late response to KD, seizure onset before age 1 year, a history of four or more unsuccessful AEDs prior to KD initiation, and lowering or discontinuing a concomitant AED before 3 months of KD treatment. Analyses were performed using STATA 10.1 (StataCorp LP, College Station, TX, U.S.A.).

Results

Between 1991 and 2009, 410 patients were initiated on the KD at CHOP and 276 had follow-up records readily available for review. Of these, 65 patients became seizure-free for a minimum of 1 month during KD treatment. Characteristics of the cohort are summarized in Table 1. Mean age ± standard deviation (SD) at KD initiation was 5 ± 3.2 years, and 37 (57%) were boys. The majority of patients had daily seizures (n = 45, 69%); almost all (n = 55, 85%) had seizures greater than once weekly at baseline and had been unsuccessfully treated with four or more AEDs (n = 50, 77%) prior to KD initiation. Epilepsy syndrome and etiology varied greatly, but the most common was focal epilepsy due to structural or genetic abnormalities (n = 24, 37%). Of the 33 patients with structural abnormalities, the most common finding was dysplasia (n = 12, 18%). Median time to seizure freedom after KD initiation was 1.5 months (range 0–20 months). Forty-seven patients (72%) were early responders, achieving seizure freedom within the first 3 months of KD treatment, whereas 18 (28%) were late responders.

Table 1. Clinical characteristics
 Total number of patients seizure-free for at least 1 month on KD (n = 65)Seizure-free <1 year (n = 46)Seizure-free ≥1 year (n = 19)p-Value
  1. IQR, interquartile range.

  2. a

    Fisher's exact test.

Male/Female, n37/2826/2011/180.57a
Age in years at KD initiation (mean ± SD)5.0 ± 3.24.81 ± 3.105.54 ± 3.420.41a
Time in years to KD initiation (median, IQR)3.18 ± 2.592.98 ± 1.923.68 ± 3.770.33a
Number of AEDs prior to KD (median, IQR)4 (4–4)4 (4–4)4 (3–4)0.75a
1. Mode of seizure onset, n (%)0.61a
Focal30 (46)22 (48)8 (42) 
Generalized18 (28)11 (24)7 (37) 
Both focal and generalized8 (12)5 (11)3 (16) 
Spasms9 (14)7 (16)2 (11) 
2. Etiology
Genetic3 (5) 
SCN1A3 (5)2 (4)1 (5) 
Structural33 (51) 
Dysplasia12 (18)9 (20)3 (16) 
Stroke4 (6)1 (2)3 (5) 
Intraventricular hemorrhage2 (3)1 (2)1 (5) 
Encephalomalacia3 (5)3 (7) 
Tumor3 (5)2 (4)1 (5) 
Periventricular leukomalacia5 (8)4 (9)1 (5) 
Unknown4 (6)3 (7)1 (5) 
Metabolic3 (5) 
GLUT 1 deficiency syndrome1 (2)1 (5) 
Methylenetetrahydrofolate reductase deficiency1 (2)1 (2) 
Pyruvate dehydrogenase deficiency1 (2)1 (2) 
Unknown26 (40)19 (41)7 (37) 
3. Electroclinical syndrome25 (39) 
West syndrome9 (14)7 (15)2 (10) 
Febrile seizures plus1 (2)1 (2) 
Severe myoclonic epilepsy of infancy1 (2)1 (2) 
Myoclonic astatic epilepsy9 (14)4 (9)5 (26) 
Lennox-Gastaut3 (5)3 (7) 
Landau-Kleffner syndrome1 (2)1 (5) 
Juvenile absence epilepsy1 (2)1 (2) 

Seizures recurred in 53 patients (82%) during the observed follow-up period. The rate of recurrence did not differ between early and late KD responders (hazard ratio 1.08, p = 0.8). In the Kaplan-Meier analyses, median time to seizure recurrence was 3 months or less, and the likelihood of remaining seizure-free at 18 months was 3% (Fig. 1). In all but six patients, seizure frequency after recurrence was less than prior to KD initiation. In the 45 patients who had daily seizures at baseline, a decline in seizure frequency to less than once monthly occurred in 26 (59%), and only 10 (23%) developed daily seizures again. This improvement in seizure frequency was sustained throughout the duration of KD treatment (Table 2). Nineteen patients (29%) experienced complete seizure remission lasting 1 year or longer while on the KD. Nine of these 19 patients had been weaned off the KD and remained seizure-free for a median of 2.5 years after KD completion. Of a total of nine patients with myoclonic astatic epilepsy (MAE), four patients achieved this sustained seizure freedom for 2.5 years after KD completion. Three of these MAE patients remained seizure-free and were successfully weaned off the KD with no further seizures at the time of last follow-up (13.5, 33, and 36 months after KD wean). The fourth MAE patient remained seizure-free for 2 years off the KD, but experienced a late breakthrough seizure so was then treated with a modified Atkins diet. No statistically significant differences were seen between the group with complete seizure remission lasting 1 year or longer and those with shorter remissions in age at KD initiation, time from seizure onset to KD initiation, baseline seizure frequency, or distribution of epilepsy syndromes (Table 1).

Table 2. Seizure frequency at baseline compared to years 1, 2, and 3 on KD
 Total number of patients seizure-free for at least 1 month on KD (n = 65)Seizure-free <1 year (n = 46)Seizure-free ≥1 year (n = 19)p-Value
Seizure Frequency before KD, n (%)0.73
Daily45 (69)30 (65)15 (79)
Weekly10 (15)9 (20)1 (5)
Monthly7 (11)4 (9)3 (16)
Less than once monthly3 (5)3 (6)0 (0)
Seizure Frequency on KD, n(%)
 Year 1Year 2Year 3Year 1Year 2Year 3Year 1Year 2Year 3
Improved55 (85)52 (80)44 (68)36 (78)33 (72)27 (59)19 (100)19 (100)17 (89)
Unchanged10 (15)6 (9)6 (9)10 (22)6 (13)5 (11)0 (0)0 (0)1 (5)
Worse0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
Figure 1.

Median time to seizure recurrence was 3 months or less, and the likelihood of remaining seizure-free at 18 months was 3%.

During KD treatment, the total number of AEDs was reduced in 50 patients (77%), and 10 patients (15%) were on the KD alone at their last follow-up appointments. Discontinuation of AEDs began no earlier than 3 months after KD initiation in 35 patients (54%). No difference in time to recurrence was observed between patients whose AEDs were weaned early (prior to 3 months) and those who discontinued AEDs later. In addition, time to seizure recurrence did not vary with the presence or absence of MRI abnormalities, seizure onset before or after 1 year of age, or number of AED failures prior to KD initiation.

Discussion

For many children starting the KD, seizure severity is so great that meaningful improvement is the goal, not necessarily seizure freedom. However, a substantial proportion of patients do achieve seizure freedom—rates of seizure freedom are 15–20% in meta-analyses of early studies and in larger case series from other centers.[4-7] Unfortunately, in prior studies of long-term KD efficacy, seizure freedom has been defined as no seizures within 28 days of the follow-up visit,[10, 11] or has not been well defined at all.[12, 13] An inexact understanding of what seizure freedom means in this context may complicate the treating clinician's and parental expectations of a child who achieves initial seizure freedom on the KD. This is the first study evaluating the likelihood of sustained seizure freedom in patients who were seizure-free for at least 1 month after KD initiation.

In our cohort, 24% of children initiated on the KD had at least 1 month of seizure freedom, a rate comparable to prior studies in other cohorts. Seizure recurrence was common—occurring in 82% of initially seizure-free patients, with only a 3% likelihood of remaining completely seizure-free at 18 months. Most seizure recurrences occurred early, within 4 months of attaining seizure freedom. However, seizure recurrence was most often represented as occasional breakthrough seizures and not a return to baseline seizure frequency.

Comparisons in long-term seizure freedom rates between cohorts are complicated by variability in assessment methods. Prior studies have reported percent seizure freedom at 1 year or longer, but have not always taken into account patient dropout, and have sometimes reported seizure frequency only in the 1 month preceding the outcome assessment time point—thus potentially overestimating seizure freedom rates. By taking into account seizure occurrence over the entire period of observation, not on a serial cross-sectional basis, our observations have greater clinical significance. Based on our findings, we recommend that families of children who become seizure-free on the diet be counseled that breakthrough seizures are likely to occur, but do not necessarily indicate the start of a decline to baseline seizure frequency.

Factors clearly influencing time to seizure recurrence after initial seizure freedom were not apparent in this study. Timing of AED discontinuation (early or late) did not affect time to seizure recurrence. Although not specifically answering questions about optimal timing of AED withdrawal after successful KD treatment, this observation does suggest that early AED withdrawal has no greater influence on risk of breakthrough seizures than later withdrawal.

It is hypothesized that the KD sufficiently alters the balance of excitatory and inhibitory forces in KD responders with resulting sustainable elevated seizure threshold.[14-17] Occasional perturbations still occur, pushing that balance temporarily over the threshold, but overall the bar of seizure susceptibility is raised. Many mechanisms of antiepileptic action have been proposed for the KD, and there may be multiple mechanisms occurring to a variable extent in each patient.

The retrospective nature of this study limits the evaluation of further details surrounding seizure recurrence. Although standardized seizure-reporting calendars were employed throughout the study period as part of our standard practice, the study may be subject to reporting or recall bias, an unfortunately common problem in epilepsy therapy studies.

This report may temper the outlook for sustained complete seizure freedom in KD responders; however, it should not temper enthusiasm for the KD as treatment for patients with medication-resistant epilepsy. Standardized measurement and reporting of seizure outcomes with KD therapy will further enhance progress in the field.

Acknowledgments

We thank the patients and their families enrolled in the KD program at CHOP, and the Dietary Treatment Team who are so dedicated to the care of these patients. We also thank the Patient Oriented Research Curriculum program in the Department of Neurology at the Perelman School of Medicine at the University of Pennsylvania for providing support for this project.

Disclosure

None of the authors have any conflicts of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary