The Outcome of Children with Intractable Seizures: A 3- to 6-Year Follow-up of 67 Children Who Remained on the Ketogenic Diet Less Than One Year

Epilepsia, 47(2):425 430, 2006 Blackwell Publishing, Inc. C 2006 International League Against Epilepsy The Outcome of Children with Intractable Seizures: A 3- to 6-Year Follow-up of 67 Children Who Remained on the Ketogenic Diet Less Than One Year Elisabeth B. Marsh, John M. Freeman, Eric H. Kossoff, Eileen P. G. Vining, James E. Rubenstein, Paula L. Pyzik, and Cheryl Hemingway John M. Freeman Pediatric Epilepsy Center, The Johns Hopkins Medical Institutions, Baltimore, Maryland, U.S.A. Summary: Purpose: To determine the long-term outcome of children with difficult-to-control seizures who remained on the ketogenic diet for <1 year. Methods: Between 1994 and 1996, 150 children with epilepsy, refractory to at least two medications, initiated the ketogenic diet according to the Hopkins protocol. Three to six years after diet initiation, all the families were contacted by telephone or questionnaire to assess their child s current seizure status, medications, and therapies. Results: Sixty-seven children discontinued the diet within 1 year of initiation. Follow-up data were available for 54 of these children. Ten subsequently had surgery, and three underwent VNS implantation. These operated-on children were significantly more likely to be >50% controlled at follow-up than were those managed with medications alone (p < 0.05). A statistically significant difference in long-term outcome was noted between those who responded while on the diet, even if they discontinued it before 1 year, and those who did not (p < 0.05), but no statistical correlation was found between length of time that they had remained on the diet and long-term prognosis. Conclusions: Almost half of the children who discontinued the diet during the first year had a decrease in seizures when assessed 3 6 years later. Twenty-two percent of these had become seizure free without surgery. We were unable to ascertain whether this may have been due to new medications. Those who saw some improvement while on the diet were more likely to have a favorable long-term outcome. Resective surgery, in children who were candidates, or vagal nerve stimulation (VNS) implantation, was more likely to result in significant seizure improvement than was management with medications alone. Whether or not the diet was effective, most families did not regret trying it and would recommend it to others. Key Words: Intractable epilepsy Children Ketogenic diet. Epilepsy is defined as two or more seizures of unknown cause (1). Ten percent to 30% of children with epilepsy are believed to continue to have seizures despite the appropriate use of multiple anticonvulsants (2 5). Attempts have been made to predict which children s epilepsy will become difficult to control (2,5 8); however, the consensus remains that once a child fails to respond to two antiepileptic medications (AEDs) used appropriately, he or she has only a small chance of responding to further medication trials (4,9). The ketogenic diet, originally developed in the 1920s, is high in fat, low in protein and carbohydrate, and has been shown to be an effective alternative treatment for children with difficult-to-control seizures (10 12). A 1998 prospective study of 150 children with intractable epilepsy who initiated the diet at Johns Hopkins found that 83 re- Accepted September 11, 2005. Address correspondence and reprint requests to Dr. J. M. Freeman at Meyer 2-147, Johns Hopkins Hospital, 600 N. Wolfe Street, Baltimore. MD 21287-7247, U.S.A. E-mail: jfreeman@jhmi.edu mained on the diet for >1 year, and most saw a substantial decrease in the frequency of their seizures (12). These children were reassessed 3 6 years after their diet initiation (13), and 78% continued to have a >50% reduction in seizure frequency as well as often a decrease in medications. This current study is an attempt to assess long-term outcomes in the 67 children who discontinued the diet before 1 year because of perceived lack of efficacy. What happens to those who are not fortunate enough to respond within 1 year? METHODS The original prospective study consisted of a consecutive series of 150 children ranging in age from 4 months to 16 years (mean, 5.3 years), with a baseline of 410 seizures a month despite the use of at least two AEDs. They had been exposed to an average of 6.2 medications before initiating the diet (11) (see Table 1.) 425

426 E. B. MARSH ET AL. TABLE 1. Comparison of the group who discontinued the diet prior to one year with the initial cohort Entire 150- patient cohort 67 who discontinued the diet before 1 year Male/Female 85:65 40:27 Age: average 5.3 yr 6.6 yr Range 4 mo 26 yr 6 mo 26 yr Seizure frequency 410 per mo 921 per mo Seizure type Multiple, varied a Multiple, varied a No. of medications 1.97 1.89 a See Freeman et al., 1998. The cohort was begun on the classic ketogenic diet at Johns Hopkins Hospital, following the Hopkins protocol (14), and reevaluated 3, 6, and 12 months after initiating the diet. The diet s efficacy was assessed by determining the percentage decrease in seizure frequency from a subject s baseline by using parental seizure calendars. A 50% decrease in seizure frequency is the same criterion used in assessing the efficacy of new AEDs (15). Medication changes and overall satisfaction also were noted. Three to six years after initiating the diet, a follow-up questionnaire was mailed to all of the families seeking information on their child s current seizure frequency, AEDs, and overall health status. The families also were asked about their overall satisfaction with the experience, their reasons for discontinuing the diet, and other therapies they had tried since discontinuing. They were asked whether they would try the diet again or recommend it to other families. One hundred seven of the 150 questionnaires were returned, and 31 of the 43 remaining subjects were contacted by telephone. Twelve were lost to follow-up. The protocol had received JCCI and GCRC approval. The data were analyzed by using χ 2 analysis. RESULTS Outcomes of those discontinuing the diet during the first year The 67 children who discontinued the diet during the first year ranged in age from 6 months to 15.7 years (mean, 6.6 years), with one 26-year-old patient. As shown in Table 1, those discontinuing the diet were similar in demographics to the original 150-patient cohort. They were slightly older and had a higher baseline seizure frequency ranging from <10 (n = 6) to >100 (n = 36) per month, with a mean of 921 per month. In the original study, seizure frequency at baseline did not appear to predict response to the diet; however, the mean was only 410 per month (11). No statistically significant correlation was found between age, sex, or seizure type and the likelihood that the diet would be effective and well tolerated; however, the diet s effectiveness was a strong predictor of the length of time a child remained on the diet (11). The outcomes of the 67 children who discontinued the diet within the first year of initiation are shown in Table 2. Three children were lost to follow-up before informing us of their seizure status while on the diet. Twenty-three (34%) discontinued the diet after <3 months, 20 (30%) continued for 3 6 months, and 21 (31%) continued for 6 12 months. Predictably, those with the least improvement discontinued first. Of those who had a <50% decrease in their seizures, almost half (42%) discontinued the diet within the first 3 months. Thirteen (30%) discontinued after 3 6 months, and the remaining 12 (28%) remained on the diet 6 12 months. Although a 50% reduction in seizure frequency is considered a successful response to the diet (or to a new medication), not all of the families were satisfied with this level of control, and a few still decided to discontinue the diet. Nine children were lost before long-term follow-up data could TABLE 2. Efficacy of the ketogenic diet in 67 patients who remained on it for <1 year (N = 67) % Seizure reduction at follow-up Time on the diet % Seizure reduction when discontinued Free >90% 50 90% <50% Lost a <3 mo(n= 23) <50% (n = 18) 5 (3 surgery) 1 10 (1 surgery) 2 50 90% (n = 3) 1 1 1 >90% (n = 1) 1 Free (n = 1) 1 3 6 mo (n = 20) <50% (n = 13) 3 (1 surgery) 1 (1 surgery) 2 6 (1 surgery) 1 50 90% (n = 5) 1 1 2 1 >90% (n = 2) 1 1 (1 surgery) Free (n = 0) 6 12 mo (n = 21) <50% (n = 12) 1 (1 surgery) 1 2 (1 VNS) 5 (1 surgery, b 1 VNS) 3 50 90% (n = 5) 1 1 (1 surgery) 1 1 1 >90% (n = 4) 2 2 (1 VNS) Free (n = 0) a Three unknown. b Patient had hemispherectomy before the diet. Surgery includes hemispherectomies and lobectomies.

KETOGENIC DIET AND INTRACTABLE SEIZURES 427 TABLE 3. Outcomes of those on the diet <1 year who did not undergo surgery or vagal nerve stimulator implantation (N = 41) % Seizure reduction at follow-up % Seizure reduction when discontinued Free >90% 50 90% <50% <50% (n = 26) 4 (15%) 2 (8%) 3 (12%) 17 (65%) 50 90% (n = 9) 2 (22%) 1 (11%) 2 (22%) 4 (44%) >90% (n = 5) 2 (40%) 1 (20%) 1 (20%) 1 (20%) Free (n = 1) 1 (100%) Total 9 4 6 22 be obtained, for a total of 12 with unknown long-term outcomes. Outcomes in those without surgery Thirteen of the 67 children subsequently underwent epilepsy surgery or received vagus nerve stimulators (VNSs). (One additional child who was put on the diet after a failed hemispherectomy for unilateral megalencephaly has been removed from data analysis). The seizure status of the 41 remaining children provided the opportunity to investigate the outcomes of those whose epilepsy was refractory to modern medications and, in the majority of cases, the ketogenic diet (Table 3). Although not shown, 16 of these 41 children had discontinued the diet within the first 3 months, 14 discontinued the diet 3 6 months after initiation, and 11 stopped just before 1 year. Their outcomes were independent of time spent on the diet (p > 0.05), and remaining on the diet without seeing a decrease in seizures did not improve long-term outcome. The effect of degree of seizure control at discontinuation of the diet on prognosis The level of seizure control at the time these 41 children stopped the diet was, however, predictive of their longterm outcome. Twenty-six of the 41 children had <50% control at the time they discontinued the diet. Two thirds of them remain <50% controlled with medication 3 6 years later (p < 0.05). Three have improved to 50 90%, two are >90% improved, and four are seizure free. At follow-up, the group is being managed with an average of 1.96 AEDs. In contrast, of the 15 children who had a >50% decrease in seizures while on the diet, only five have <50% control, three have 50 90% control, two have >90% control, and five are seizure free. They are taking an average of 1.54 AEDs. The majority of these 41 children have maintained the level of seizure control they achieved while on the diet. A statistically significant difference (p < 0.05) in longterm outcome is seen between those who discontinued the diet with a <50% decrease in seizures and those whose seizures had been >50% decreased. This difference was independent of how long they had remained on the diet. Of the 67 patients who discontinued the diet before 1 year (Table 2), follow-up data were available for 54. Twenty-six (48%) of the 54 still had <50% reduction in seizures 3 6 years later. Only four of them had undergone epilepsy surgery, and one had a VNS implanted. Twenty-eight of the 54 children are now >50% controlled at follow-up, and nine of them have undergone surgery (7) or VNS implantation (2). Improvement without surgery Most children who failed to respond to the ketogenic diet also failed to respond to newer medications. If they did not subsequently undergo a cortical resection or receive a VNS, they continue to have poor control of their seizures. However, multiple medications failed for 19 children, who discontinued the ketogenic diet before 1 year because of perceived lack of efficacy, who currently have >50% reduction in seizures without surgical intervention (Table 3). Nine are seizure free, four have a >90% decrease, and six have a 50 90% decrease. These improvements in seizure control were more common in children who had some response while on the diet. Seventeen of the 26 children who had a <50% response during their time on the diet continue to have uncontrolled seizures. Improvement with surgery Children who fail to respond to the ketogenic diet after multiple medications have failed are significantly more likely to see improvement in their seizures in the long term if they subsequently undergo epilepsy surgery or receive a VNS (p < 0.05). Those with a resectable seizure focus have an especially favorable prognosis. Of the 10 children who underwent surgical resection (Table 4), seven are >90% controlled at follow-up; five of them are seizure free. Two of the three children who received VNSs are currently 50 90% controlled at follow-up. Medications The 67 children were taking an average of 1.89 AEDs when they began the diet; 16 were taking three or more. Throughout the study, medications were decreased in a nonsystematic manner to a level deemed appropriate for each child, taking into account the balance between seizure control and the unwanted side effects. When they discontinued the diet, these same patients were taking an average of 1.38 medications, and only nine patients were taking three or more. TABLE 4. Long-term outcomes of the 13 who later underwent surgery or received a vagal nerve stimulator % Seizure reduction at follow-up % Seizure reduction when discontinued Free >90% 50 90% <50% <50% (n = 10) 5 1 VNS 2 + 1 VNS 50 90% (n = 1) 1 >90% (n = 2) VNS 1 Free (n = 0)

428 E. B. MARSH ET AL. TABLE 5. Comparison of outcomes at 3 6 years of the original 150 patients, of whom 83 remained on the diet 1 year, and those who discontinued the diet after <1 year No surgery ± Surgery % Of original 150 On diet >1 yr Off diet < 1 yr Off Diet <1 yr Seizure free 20/150 13% 20/83 24% 9/41 22% 14/54 26% >90% 21/150 14% 21/83 25% 4/41 10% 6/54 11% 50 90% 24/150 16% 24/83 29% 6 /41 15% 8/54 a 15% <50% 18/150 12% 18/83 22% 22 /41 54% 26/54 b 48% a Two had VNS. b One had VNS. Unlike the follow-up results for the subset of children who remained on the diet for the entire year (13), children who discontinued the diet early because of little perceived improvement had little to no change in their average number of AEDs. Three to 6 years later, they were still taking an average of 1.81 medications, and 14 were being managed with three or more AEDs. Only six were medication free. As expected, children with better long-term seizure control tended to be managed with fewer medications. See Table 5. Parental perceptions and reasons to discontinue the diet Although we asked families to try the diet for 6 weeks so that it could be adjusted and optimized, parents were free to discontinue the diet at any time they thought that the costs in time and effort outweighed the benefits. Twentythree of the 67 families discontinued the diet before 3 months primarily because of the parents perception that the diet was either ineffective (n = 33), or too restrictive (n = 19) for their child. Thirteen others stopped because of illness, and two discontinued for other reasons. A correlation was found between the efficacy of the diet in controlling the child s seizures and the length of time the parents were willing to continue. Eighteen of the 23 children discontinuing before 3 months had <50% improvement in their seizure control. By 6 months, another 13 children with a <50% reduction had discontinued. After 6 months, we saw a greater number of children who had >50% improvement also choosing to stop the diet. Of note, on the follow-up questionnaire, seven of the families who discontinued the diet before year reported that the diet had reduced their child s seizures by <50%, although their seizure calendars demonstrated a >50% reduction while on the diet. Satisfaction with the diet On the follow-up questionnaire, families were asked if they would consider trying the diet again, and if they would recommend the ketogenic diet to other families in similar situations. Although many were not willing to put their child back on the diet, they stated that it had been a good alternative option and were eager to recommend it to other families who had children with difficult-to-control seizures. Thirty-six percent were willing to give the diet another try, although more than half had been unsuccessful the first time. Sixty-one percent reported that they would not be willing to try the diet again (90% of those children had seen no effect). In general, families were pleased that they had tried the diet and were happy with the experience, even if they had not thought it to be effective in their child s treatment. Eighty-five percent reported that they would recommend the diet to other families. Does the diet affect outcome? Previous studies demonstrated that the ketogenic diet is more effective in controlling difficult-to-control seizures than are even modern medications (11). Many children have a dramatic and sustained decrease in their seizures and are able to taper or discontinue medications. Many are off medication and without seizures 3 6 years later. Have we uncovered a new approach to treating and perhaps curing epilepsy, or is this merely the natural history of children with difficult-to-control seizures? Table 5 compares the 3- to 6-year outcomes of children who remained on the diet for 1 year (13) with those from the same cohort who discontinued the diet after <1 year. With an intention-to-treat analysis, 27% of the whole original cohort of 150 children treated with the diet had a >90% decrease in their seizures 3 6 years later. Analyzing only those 83 children from the original cohort who remained on the diet for 1 full year, usually because the diet had been effective in decreasing their seizures, 49% of these previously intractable children had a 90% decrease in their seizures (13). At the time of follow-up, most of these children were no longer on the diet and had decreased or eliminated their medications. A surprise in our current study was the finding that even when the diet was of little or no benefit and had been discontinued during the initial year, 32% of children had a >90% decrease in their seizures and 22% had become seizure free, 3 6 years later, even without surgery. DISCUSSION Individuals whose seizures cannot be controlled with a single AED are said to have only a 20 30% chance of achieving control with a second medication, and those for

KETOGENIC DIET AND INTRACTABLE SEIZURES 429 whom a second drug fails have an even lower probability that any additional medication will control their epilepsy (6). In the past decade, the ketogenic diet has been increasingly used for children for whom multiple medications have failed. The diet is well tolerated and successful in diminishing or controlling the seizures in >50% of these children (11). Their seizure reduction may last for years after the diet is discontinued (13). Does the ketogenic diet, in addition to decreasing the number of seizures, alter the long-term outcome of children with difficult-to-control epilepsy, or is outcome simply the natural history of an intractable childhood seizure disorder? To evaluate the course of refractory epilepsy in childhood in the current era, we evaluated the 41 children from our series that had difficult-to-control seizures, who had discontinued the diet within the first year after initiation, and who had not undergone resective surgery or VNS implantation (Table 3). Of these, 26 (63%) had had a <50% decrease in their seizures while on the diet. Two thirds of them remained poorly controlled 3 6 years later. However, it is notable that even if the diet did not decrease a child s seizures, six (23%) of 26 had a marked decrease in their seizures 3 6 years later, and four of those had become seizure free. Whether this is due to the natural history of intractable seizures or to the introduction of new medications could not be established by this study. Fifteen (37%) of the 41 children had a >50% improvement in their seizures (Table 3) during the short time spent on the diet. Two thirds continued to have a >50% reduction in seizures at follow-up, and five (33%) had become seizure free. Even partial response to the diet appears to predict the future. The children who responded even mildly to the diet were more likely to be well controlled years later, whereas the children who did not respond were more likely to continue to have poorly controlled seizures. It would appear that most children who did not respond to the diet also failed to respond to newer AEDs. If medications and the diet fail, surgery should be considered, rather than continuing to try more medications. If a resectable lesion is found, these children have the best chance of doing well. Seven of the 10 undergoing resective surgery had a >90% decrease in their seizures, and five of them became seizure-free. In summary, this study shows that although the majority of children with difficult-to-control seizures do not outgrow them, and that children who do not respond to the ketogenic diet are likely to continue to have difficult-tocontrol seizures if they are not candidates for additional surgical treatment, some do well. Those for whom the ketogenic diet is effective are likely to remain better over the long term. However, to our surprise, we found that all intractable epilepsy in children is not intractable. Three to 6 years later, 22% of the children with what we had believed to be intractable seizures (refractory to more than six medications) were seizure free, and an additional 10% had a >90% reduction in their seizures even without surgical intervention (Table 5). If one adds those who had a resectable lesion or who were treated with the VNS, the percentage of children whose seizures were controlled is even higher. Thus the outcomes of children with very difficult-to-control seizures may not be as bleak as the literature would indicate. Even in children who have a mean of >900 seizures per month and for whom both multiple medications and the ketogenic diet have failed, some chance remains for a future decrease in their seizures. These data indicate that within the population of children with difficult-to-control seizures, two populations may exist, those who respond to the ketogenic diet and those who do not. The long-term prognosis for the two groups is different, and it is possible that response to the diet may actually be a predictor of future outcome independent of the diet s duration. How long should children remain on the diet? Most children who had substantial improvement continued the diet for 12 months, but we do not know whether this is necessary. Historically, the diet had been continued for 2 years, and then gradually tapered the following year (16). But is the duration of the diet important in determining seizure control? Thirty-one percent of those remaining on the diet for <3 months had a >50% decrease in their seizures at follow-up 3 6 years later. The majority of these children had a <50% decrease in their seizures while on the diet. Those on the diet for 3 6 months had a 50% chance of being well controlled at follow-up, whereas those on the diet for 6 months to 1 year had a 64% chance. Most of these children had improvement in their seizure control while on the diet. All children were likely to remain at the level of seizure control they achieved while on the diet, regardless of whether they discontinued the diet after 3, 6, or <12 months. Even those children who were on the diet for only a few months were able to retain the long-term benefits if they had some improvement. Likewise, even if a child remained on the diet for almost a year, if it failed to decrease their seizures significantly, they were likely to be poorly controlled at follow-up. Whether a few short weeks on the diet is enough to improve seizure control in the long term, or whether these results represent the natural history of this group s difficult-to-control seizures, remains to be determined. It should be noted that this population was evaluated as a part of our routine treatment protocol for children on the ketogenic diet. The studies were performed without outside support. Had it been a funded study, we might have had EEGs before diet initiation and at follow-up. We might have obtained MRI scans and better delineation of the seizure types, the IQs before and after the diet, and many other laboratory parameters. We would like

430 E. B. MARSH ET AL. to have had more information about the effects of new medications on seizure control. We were constrained by the parameters that we thought were necessary for the care of the child and because the information for this study was obtained by questionnaire and phone followup. CONCLUSION To determine the long-term outcome of children with difficult-to-control seizures, we evaluated 67 children for whom multiple medications had previously failed and were on the ketogenic diet for <1 year. Without surgery, the majority of those who did not respond to the diet continued to have seizures 3 6 years later. However, even in this very intractable group, 22% became seizure free, and an additional 10% had a >90% reduction in their seizures. Whether this surprising improvement in seizure control is due to the natural history of the condition or to new medications is unclear; however,these numbers may provide some hope to families struggling with intractable epilepsy. REFERENCES 1. Engel J Jr. Introduction: what is epilepsy? In: Engel J Jr, Pedley TA, eds. Epilepsy: a Comprehensive Textbook. Philadelphia: Lippincott- Raven, 1997:2. 2. Berg AT, Shinnar S, Levy SR, et al. Early development of intractable epilepsy in children: a prospective study. Neurology 2001;56:1445 1452. 3. Camfield CS, Camfield PR, Gordon KE, et al. Predicting the outcome of childhood epilepsy: a population based study yielding a simple scoring system. J Pediatr 1993;122:861 868. 4. Annegers JF, Hauser WA, Elveback LR. Remission of seizures and relapse in patients with epilepsy. Epilepsia 1979;20:729 737. 5. Sillanpaa M. Remission of seizures and predictors of intractability in long-term follow-up. Epilepsia 1993;34:930 936. 6. Kwan P, Brodie MJ. Early identification of refractory epilepsy. N Engl J Med 2000;342:314 319. 7. Camfield PR, Camfield CS. Antiepileptic drug therapy: when is epilepsy truly intractable? Epilepsia 1996;37:S60 S65. 8. Huttenlocher PR, Hapke JR. A follow-up study of intractable seizures in childhood. Ann Neurol 1990;28:699 705. 9. Sander JWAS. Some aspects of prognosis in the epilepsies: a review. Epilepsia 1993;34:1007 1016. 10. Swink T, Vining EPG, Freeman JM. The ketogenic diet: 1997. Adv Pediatr 1997;44:297 329. 11. Freeman JM, Vining EPG, Pillas DJ, et al. The efficacy of the ketogenic diet: a prospective evaluation of intervention in 150 children. Pediatrics 1998;102:1358 1363. 12. Stafstrom CE, Rho JM, eds. Epilepsy and the Ketogenic Diet. Totowa: Humana Press, 2004. 13. Hemingway C, Freeman JM, Pillas DJ, et al. The ketogenic diet: a 3 6 year follow-up of 150 children enrolled prospectively. Pediatrics 2001;108:898 905. 14. Freeman JM, Kelly MT, Freeman JB. The Epilepsy Diet Treatment: An Introduction to the Ketogenic Diet. New York: Demos, 1994. 15. Walker MC, Sander JWAS. The impact of new antiepileptic drugs on the prognosis of epilepsy: seizure freedom should be the ultimate goal. Neurology 1996;46:912 914. 16. Livingston S. Comprehensive Management of Epilepsy in Infancy, Childhood, and Adolescence. Springfield: Charles C Thomas, 1972:400.