Will seizure control improve by switching from the Modified Atkins Diet to the traditional Ketogenic Diet?

SUMMARY

It has been reported that children can maintain seizure control when the ketogenic diet (KD) is transitioned to the less restrictive modified Atkins diet (MAD). What is unknown, however, is the likelihood of additional seizure control from a switch from the MAD to the KD. Retrospective information was obtained from 27 patients who made this dietary change from 4 different institutions. Ten (37%) had ≥10% additional seizure reduction with the KD above the MAD, of which 5 became seizure-free. The 5 children who did not improve on MAD failed to improve when transitioned to KD. A higher incidence of improvement with the KD occurred for those with myoclonic-astatic epilepsy (70% vs. 12% for all other etiologies, p = 0.004), including all who became seizure-free. These results suggest the KD probably represents a “higher dose” of dietary therapy than the MAD, which may particularly benefit those with myoclonic-astatic epilepsy.

The modified Atkins diet (MAD) is a variation of the ketogenic diet (KD) that was designed to mimic the effects on ketosis and seizures of the KD but with less restrictiveness (Kossoff & Dorward, 2008). Evidence suggests that the KD can be successfully transitioned to the MAD without loss of seizure control if there is difficulty with fat tolerance, hunger, or to help maintain compliance (Kossoff et al., 2003). Results from the opposite switch, namely from the MAD to the traditional KD, are unknown, however.

Recognizing that information to date suggests that there may be slightly less likelihood of a >50% seizure reduction with the MAD than the KD (Henderson et al. 2006; Kossoff et al, 2006; Kang et al., 2007; Kossoff et al 2007; Weber et al., 2009), and that urinary ketosis appears to be both higher and more stable with the KD as well (Kang et al., 2007), a practical question exists of whether additional seizure control would occur if the MAD is switched to the KD. There are unfortunately little modifications that can be made to fine tune the MAD other than encouraging fat and lowering the daily carbohydrate limit, the latter of which has been reported in at least one study as ineffective (Kossoff et al., 2007). This is not an inconsequential decision for all caregivers, as this change typically involves an inpatient hospitalization, possible fasting period, protein limitation, teaching families to weigh and measure foods, and utilizing only specific dietitian-provided recipes. For many of these families, avoiding these restrictions and inconveniences led to the choice of the MAD originally, and they may be reluctant to change.

METHODS

Children were identified from four institutions in four countries with large KD experience, but additionally an identified experience using the MAD. Children included were any who had been treated with the modified Atkins diet but then subsequently switched within several months to the traditional KD. The basic KD and MAD protocols were similar between institutions with some minor differences in the KD initiation only in regards to fasting and choice of ketogenic ratio (Kossoff & Dorward, 2008; Wirrell 2008). One child (age 4 years) was primarily started on the MAD in order to further improve possible behavioral abnormalities related to his polyol pathway abnormality (van der Knaap et al, 1999) and subsequently switched to the KD, but was not included in this cohort as he remained seizure-free on both treatments. In all other cases the switch was made with the goal of improving seizure control. All children were treated with both diets at the original institution and had at least 1 month of follow-up with both diets. No patients were included in this study that were treated with the MAD alone, KD alone, or switched from the KD to the MAD (within the past 6 months).

Some of the children in this cohort have been published previously as part of other clinical studies of the MAD, but additional subsequent information was obtained about results of their switch to the KD, which was not included in study results (Kossoff et al., 2003; Kossoff et al 2006; Kang et al., 2007; Kossoff et al., 2007; Kossoff et al, 2010). Patient information was sent anonymously to the primary institution (JHH) via email from August 2009 – March 2010 and incorporated into a single database for analysis. Records were then reviewed and analyzed by the PI (EK) from March to May 2010.

The Johns Hopkins Committee on Clinical Investigation approved the analysis and informed consent was obtained from parents/caregivers in the JHH cohort for MAD prospective studies and again at KD onset. Categorical data were analyzed with Fisher exact test for independence of rows and columns. Numerical data were analyzed using the Wilcoxon rank-sum test to compare medians and paired t-test to compare means. Significance level for all tests was p = 0.05.

RESULTS

Overall demographics

Twenty-seven children were identified from the participating institutions (USA (n=17), Denmark (n=5), Germany (n=3), Korea (n=2))(Table 1). There were no clear differences in any demographics or outcomes between institutions. The median age at seizure onset was 2.3 years (range: 0.1–15.0 years), MAD onset 5.5 years (range: 2.2–19.0 years), and KD onset 6.0 years (range: 2.4–20.0 years). Fifteen (56%) were fasted at KD onset and 24 (89%) were started on a 4:1 ratio (fat: protein and carbohydrate grams combined) using the standard KD protocol during a 3–4 day inpatient hospitalization.

Table 1.

Results from switches from the modified Atkins diet (MAD) to the ketogenic diet (KD). Patients in order of increasing age (years) at time of MAD onset.

Subject/Country	Gender	Underlying etiology	Seizure type	Age at MAD (years)	MAD duration (months)	KD duration to date (months)	Fasted?	Best MAD response	Best KD response	Current status
1 / USA	Female	Tuberous sclerosis	Partial	2.2	4.0	1.0	No	75%	75%	Off diets
2 / USA	Male	Lennox Gastaut syndrome	Generalized	2.5	6.0	2.0	Yes	50%	25%	Off diets
3 / USA	Male	MAE	Generalized	2.7	4.0	15.0	Yes	90%	Seizure free	KD
4 / Denmark	Female	Multifocal malformations	Partial	2.8	3.0	14.0	No	50%	50%	MAD
5 / USA	Male	MAE	Generalized	2.9	4.0	9.0	Yes	95%	95%	KD
6 / USA	Male	MAE	Generalized	3.1	8.0	12.0	Yes	90%	Seizure free	MAD
7 / USA	Female	Dravet syndrome	Generalized	3.3	10.0	8.0	Yes	50%	90%	Off diets
8 / Denmark	Female	Dravet syndrome	Generalized	4.4	3.0	3.0	Yes	50%	50%	Off diets
9 / USA	Female	Idiopathic	Partial	4.5	3.0	10.0	No	50%	95%	KD
10 / USA	Female	MAE	Generalized	4.7	4.0	18.0	Yes	75%	Seizure free	KD
11 / Germany	Male	MAE	Generalized	5.2	2.0	2.5	No	0%	0%	KD
12 / USA	Male	MAE	Generalized	5.5	24.0	6.0	Yes	50%	75%	Off diets
13 / USA	Male	MAE	Generalized	5.5	7.0	30.0	Yes	90%	Seizure free	Off diets
14 / Denmark	Male	Bilateral strokes	Partial	5.8	11.0	9.0	Yes	90%	90%	Off diets
15 / Denmark	Female	Idiopathic	Partial	6.5	1.5	1.0	No	25%	25%	KD
16 / Germany	Female	Lissencephaly	Partial	6.5	3.0	3.0	Yes	99%	99%	MAD
17 / USA	Male	MAE	Generalized	7.0	3.0	6.0	No	75%	90%	Unknown
18 / Korea	Female	Herpes encephalitis	Partial	8.0	2.0	2.0	No	25%	25%	Off diets
19 / USA	Female	MAE	Generalized	8.5	6.0	30.0	Yes	50%	Seizure free	KD
20 / Korea	Female	Head trauma	Partial	9.0	1.0	2.0	No	0%	0%	Off diets
21 / Denmark	Female	Idiopathic	Generalized	11.2	2.0	9.0	No	0%	0%	Off diets
22 / USA	Female	Lennox Gastaut syndrome	Generalized	13.0	4.0	1.0	Yes	0%	0%	Unknown
23 / USA	Female	Idiopathic	Generalized	13.0	2.0	3.0	Yes	25%	50%	Off diets
24 / USA	Female	Lennox Gastaut syndrome	Generalized	13.0	30.0	24.0	Yes	75%	75%	KD
25 / Germany	Female	Idiopathic	Partial	14.3	1.5	2.2	No	0%	0%	Off diets
26 / USA	Female	Absence	Generalized	18.0	4.0	6.0	No	90%	90%	Unknown
27 / USA	Female	Idiopathic	Partial	19.0	13.0	3.0	No	90%	50%	MAD

Abbreviations: KD=ketogenic diet; MAD=modified Atkins diet; MAE=myoclonic-astatic epilepsy; USA=United States of America.

Seizure outcomes

During treatment with the MAD, 19 subjects (70%) had at least a 50% seizure reduction, of which 8 reported 90–99% improvement. Only 5 children in this cohort had no discernible improvement with the MAD. The KD resulted in an overall >50% improvement in an identical 19 (70%) of subjects. Overall, an improvement of ≥10% seizure reduction over the MAD was reported by 10 (37%). Five children became seizure-free. Only children with seizure reduction on the MAD subsequently improved with the KD.

During the period of time on the MAD, 3 patients were able to discontinue 1 anticonvulsant, 2 successfully discontinued 2 (Patients 9 and 11), and no patient had medications added. On the KD, 6 were able to reduce anticonvulsants, including 4 (Patients 3, 6, 10, and 13) with MAE who became seizure-free and then shortly afterwards medication-free.

Five (Patients 7, 12, 22, 23, and 26) had anticonvulsants added to improve seizure control while on the KD due to insufficient benefit. These medications included clobazam, clorazepate (2), topiramate, and zonisamide. None of these children had additional seizure reduction with these anticonvulsants and all are either off the KD or lost to follow-up.

Factors of those who improved with the KD

Demographic factors for those 10 children who had improvement after switching to the KD were compared to the 17 who did not (Table 2). There was an increased likelihood of an improvement if the patient had myoclonic-astatic epilepsy compared to all other etiologies combined, 70% vs. 12%, p = 0.004. Additionally, all 5 children who became seizure-free after transitioning to the KD had myoclonic-astatic epilepsy. A trend towards greater likelihood of improvement if a child was fasted at KD onset was identified (80% vs. 41%, p = 0.06).

Table 2.

Demographic factors of the 10 children who improved (≥10% seizure reduction) with the switch from the MAD to the KD compared to those 17 who did not.

Demographic	Improvement (n=10)	No improvement (n=17)	p value
Gender (female)	5 (50%)	13 (76%)	0.13
Age at seizure onset, years, mean (SEM)	3.8 (1.0)	3.4 (1.0)	0.80
Age at MAD onset, years, mean (SEM)	5.8 (1.0)	8.5 (1.3)	0.11
MAD duration, months, mean (SEM)	7.1 (2.0)	5.6 (1.7)	0.58
Generalized epilepsy	9 (90%)	10 (59%)	0.10
Myoclonic-astatic epilepsy	7 (70%)	2 (12%)	0.004
Higher reported urinary ketosis after switching to KD	8 (80%)	12 (71%)	0.48
Fasted at KD onset	8 (80%)	7 (41%)	0.06
KD ratio (4:1)	9 (90%)	15 (88%)	0.70
Prior KD use	0 (0%)	3 (18%)	0.23
Concurrent valproate use	7 (70%)	7 (41%)	0.15

KD=ketogenic diet; MAD=modified Atkins diet; SEM=standard error of the mean.

Ketosis and adverse effects

Urinary ketosis was evaluated overall and was reported by parents as higher with the KD in 20 (74%). Nearly all of these subjects were reported as having consistently large (80–160 mg/dl) urinary ketosis with the KD, compared to inconsistently large or declining ketosis with the MAD. Serum ketosis was not routinely measured.

The mean highest total cholesterol on the MAD was 220.3 mg/dl (SEM = 13.6), which was similar to the mean highest value for the KD, 207.3 mg/dl (SEM = 12.8), p = 0.49. There was also no difference in the mean highest triglyceride levels between diets (110.9 mg/dl (SEM = 12.3) vs. 111.2 mg/dl (SEM = 10.6), p = 0.98). No child in either group had kidney stones, bone fractures, or any other clear adverse effect of dietary therapy requiring its discontinuation.

DISCUSSION

These results indicate that the KD is likely to improve seizure control in approximately one-third of children previously treated with the MAD. Results are tempered by the retrospective nature of this study. This modest additional improvement may be similar to adjusting the dose of anticonvulsant drugs, and therefore the KD may mimic a higher drug “level” than the MAD, but not a truly independent therapy. If these diets were truly dissimilar, one would expect twice as many to have improved, with results similar to the 60% seizure improvement reported for patients started on the KD or MAD without prior dietary treatments (Henderson et al., 2006; Kossoff et al., 2006; Kang et al., 2007). Additionally, no child improved with the KD who had no improvement with the MAD previously, which would have been expected with a unique therapy.

What was striking was that, of those 5 who became seizure-free with the KD, all had myoclonic-astatic epilepsy. These results may be affected by a selection bias as a large number of children with myoclonic-astatic epilepsy were switched to the KD in this cohort and this could represent knowledge of the high published seizure-free rates with the traditional KD (Oguni et al., 2002; Kilaru & Bergqvist, 2007). In addition, for unclear reasons only those treated in the USA had improvement with the switch. These results suggest that a switch to the KD in this particular subpopulation may be reasonable to consider if seizure freedom has not occurred with the MAD after perhaps 6–12 months. Additionally, the chances of an improvement were nearly twice as likely if the KD was started with a fast, despite most of these children already being ketotic due to the MAD, although this did not reach statistical significance.