Authors :
Presenting Author: Emily Grew, MD – NYU Langone
Dana Amaya, MSRD – NYU Langone
Parmpreet Dhillon, MD – NYU Langone Health
Rationale:
The ketogenic diet is a well-established treatment for intractable epilepsy. Higher diet ratios are traditionally thought to produce higher ketosis levels and better seizure outcomes. However, diet ratio and ketosis levels may be limited by nutrient needs, medical conditions, and side effects. We aim to describe the relationship between diet ratio, ketosis level, seizure frequency, side effects, and subjective benefit in patients on the ketogenic diet for epilepsy.Methods:
We conducted a retrospective analysis of patients on the ketogenic diet for epilepsy at our institution since 2010 using longitudinal Epic data entered into RedCap. Primary outcome was change in seizure frequency (increased, no change, decreased) at 3, 6, 12, and 24 months since diet initiation. Seizure frequency was modeled as a three-level ordinal outcome accounting for repeated measures in generalized estimating equations (GEE) with a cumulative logit link. Ketosis level (low, medium, high), diet ratio (5 levels, from Modified Atkins to 4:1+), and time were evaluated for impact on seizure frequency. Secondary outcomes included ketosis level (predicted by ratio and time), side effects, and subjective benefit e.g improved seizure duration or severity, alertness, sleep), analyzed in GEE models. Data were analyzed in SPSS v28.Results:
We analyzed 107 patients attending at least 3 visits across 4 potential timepoints, totaling 302 discrete observations. Demographic data are available in Table 1.
Averaged across time points, seizure frequency decreased in 66.6% of patients and ASMs were downtitrated in 27.7%; both rates increased over time. 80.7% reported subjective benefit. 49.8% experienced side effects (Table 2).
Higher ketosis level predicted decreased seizure frequency (χ²=7.117, p=0.028). Seizure was significantly more likely to decrease in patients with high ketones compared to medium (B=1.083, p=0.03) and low (B=1.209, p=0.026) ketones. The difference between low and medium ketones was not significant (B=0.120, p=0.776).
Diet ratio approached significance for predicting ketosis level (χ²=8.935, p=0.063) but did not predict seizure frequency (χ²=3.515, p=0.476). Subjective benefit was associated with diet group at 3 (p=0.027) and 6 months (p=0.033), but the effect was not present longitudinally (p=0.963). Subjective benefit was not associated with ketosis level (p=0.120). Side effects were not significantly associated with diet ratio (p=0.388) or ketosis level (p=0.543).
Conclusions:
Our data show that seizure frequency was significantly predicted by ketosis level but not by diet ratio. Subjective benefit and side effects were not associated with diet ratio or ketosis level. This suggests that patients at lower diet ratios may experience the benefits of the ketogenic diet for epilepsy comparably to those at high ratios, and that patients may be advanced to higher ratios without significant increase in side effects.Funding: None