Highlights
Key points
– In the Human Epilepsy Project 2 (HEP2), a prospective observational cohort of focal treatment‑resistant epilepsy (FTRE), 68.3% of participants experienced lower seizure frequency in the second half of follow‑up compared with the first half.
– Modeled mean monthly seizure frequency fell by 68.7% (95% CI, 52.9%–84.5%) overall across 18–36 months of follow‑up; reductions were observed across follow‑up epochs but exhibited variable precision.
– Adding an antiseizure medication (ASM) was common (54.7%) and associated with seizure reduction in two‑thirds of those treated, but seizure freedom remained uncommon; participants with implantable devices did not have statistically different seizure trajectories than those without devices.
Background: disease context and unmet needs
Focal treatment‑resistant epilepsy (FTRE) imposes substantial morbidity: increased risk of injuries, impaired cognition and psychosocial function, and reduced quality of life. The International League Against Epilepsy (ILAE) consensus defines drug‑resistant epilepsy as failure of adequate trials of two tolerated, appropriately chosen ASMs to achieve sustained seizure freedom; many patients labeled “refractory” have failed far more agents. Understanding how seizure burden changes over time in these patients is critical for clinical decision‑making, counseling, and the design and interpretation of interventional trials that claim disease‑modifying effects for ASMs, neuromodulation devices, or other therapies. The HEP2 study addresses this gap by tracking seizure frequency longitudinally in a contemporary, multisite cohort of FTRE patients receiving real‑world care.
Study design and methods
This was a prospective, observational, multi‑center cohort conducted at 10 comprehensive epilepsy centers in the United States between May 2018 and September 2021 (Human Epilepsy Project 2). Eligible participants were aged 16–65 years with focal epilepsy and had failed 4 or more ASMs, at least two of which were due to lack of seizure control. Participants were a volunteer sample and were followed for 18–36 months with intensive seizure capture using daily electronic diaries, monthly check‑ins, review of medical records, and case report forms. Treatment was at the treating physician’s discretion and included ASM adjustments, implantable devices, and, infrequently, surgery.
Primary outcome: longitudinal trends in seizure frequency quantified as seizure freedom rates and percent reductions in modeled monthly seizure frequency across the follow‑up period and within prespecified epochs (0–12 months, 12–24 months, and >24 months). Secondary assessments included responses to ASM additions and analyses comparing seizure trajectories in device‑treated versus non‑device participants.
Analysis: Of 196 approached, 146 met eligibility; 128 provided sufficient seizure diary data for analysis (2 outliers were excluded). Analyses employed statistical modeling to estimate percentage reductions and 95% confidence intervals and compared trajectories across epochs and subgroups.
Key findings
Cohort characteristics
– N=146 enrolled; mean (SD) age 40 (12) years; mean age at epilepsy diagnosis 19.8 (13.6) years; 57.5% female.
– Device use: 35 participants had implantable devices at baseline or during follow‑up; only 1 participant underwent epilepsy surgery during the study.
Primary outcome: seizure trends
– Of 128 participants with analyzable seizure diaries, 86 (68.3%) had reduced seizure frequency in the second half of participation compared with the first half.
– Modeled overall mean monthly seizure frequency percentage reduction was 68.73% (95% CI, 52.92%–84.54%), P 24 months (cohort 3) 66.03% (95% CI, 48.25%–83.80%) (all P < .001). Note that wide confidence intervals for some epochs and percent reductions exceeding 100% reflect modeling features and variability in short epochs.
Effects of treatment changes
– An ASM was added in 69 participants (54.7%); among these, 46 (66.7%) experienced reductions in seizure frequency, including some instances of seizure freedom, although seizure freedom was described as uncommon overall.
– Participants with implantable devices did not demonstrate significantly different seizure trajectories compared with those without devices in adjusted analyses, suggesting that device‑treated participants did not fare better, on average, than others in this cohort over the follow‑up period.
Safety and adverse events
– The provided summary did not detail safety outcomes or device‑related adverse events; as an observational study, systematic adverse event adjudication may be incomplete.
Interpretation and clinical implications
The HEP2 data show that many patients meeting stringent criteria for FTRE (≥4 ASM failures) nevertheless experience substantial reductions in seizure frequency over time. These findings have several practical implications:
1. Natural history and clinical management
– Seizure frequency in FTRE can improve materially over months to years even in patients labeled as highly treatment resistant. Clinicians should recognize that long‑term trajectories may include meaningful spontaneous or treatment‑associated improvement, which should temper prognosis discussions and inform sequential treatment choices.
2. Role of ASM additions
– Adding ASMs contributed to seizure reduction in a sizable proportion of treated participants, but complete seizure freedom remained rare. That underscores the potential utility of rational polytherapy or alternative ASMs for reduction of seizure burden even when prior trials failed, while maintaining realistic expectations about achieving complete remission.
3. Devices and claims of disease modification
– Device‑treated participants (including neuromodulation) showed similar trajectories to those without devices in this cohort. The observational nature of HEP2 cannot definitively adjudicate device efficacy, but the findings advise caution in attributing long‑term seizure reductions observed in open‑label device trials to disease modification without randomized controls.
4. Implications for trial design and interpretation
– These results highlight the importance of randomized, controlled, and ideally blinded designs for claims of disease modification. Open‑label, single‑arm studies are vulnerable to regression to the mean, placebo and Hawthorne effects, selection bias (volunteer samples), and confounding by concurrent treatments and care intensity. Trials aiming to demonstrate disease‑modifying effects should include concurrent controls, prespecified long‑term endpoints, and objective outcome measures when possible.
Methodological considerations and limitations
– Observational design: Without randomization or an internal control arm, causality cannot be inferred; improvements may reflect natural history, enhanced adherence and monitoring, regression to the mean, or cointerventions rather than specific treatment effects.
– Volunteer sample and center selection: Recruitment at comprehensive epilepsy centers and reliance on volunteers may limit generalizability to broader community populations.
– Seizure ascertainment: Daily electronic diaries improve capture relative to infrequent clinic recall but retain risks of under‑ or over‑reporting.
– Treatment heterogeneity: Interventions were provided at clinician discretion, producing confounding by indication; the frequency and timing of ASM changes and device programming adjustments varied across participants.
– Small device subgroup and limited surgical exposure: Only 35 device recipients and a single surgery reduce power to detect device or surgical effects.
– Modeling artifacts: Some confidence intervals were wide or exceeded plausible bounds (>100% reduction), reflecting statistical modeling limits in short epochs or small subgroups and emphasizing the need to interpret point estimates within their precision bounds.
Expert commentary and biological plausibility
Several mechanisms could underlie longitudinal improvement in seizure burden despite prior multiple ASM failures: natural fluctuation and eventual remission in some epilepsy syndromes; late response to a newly introduced ASM or optimized polytherapy; behavioral changes, improved adherence, and lifestyle modifications; or unmeasured disease‑modifying influences (e.g., anti‑epileptogenic processes or neuromodulatory effects of chronic device therapy). However, discerning true disease modification from symptomatic seizure suppression or non‑specific effects requires controlled experimental designs and biomarkers (e.g., changes in interictal epileptiform activity on EEG) that were not systematically reported in HEP2.
Conclusions and future directions
The HEP2 cohort demonstrates that substantial seizure reduction over 18–36 months is common among patients with focal treatment‑resistant epilepsy, and that ASM additions can provide meaningful incremental benefit while rarely producing full seizure freedom. Device recipients in this cohort did not show clearly superior trajectories compared with non‑device participants. These observations counsel caution when interpreting open‑label claims of disease modification from single‑arm trials. Future research priorities include randomized controlled trials with long‑term blinded follow‑up to test disease‑modifying hypotheses, development and validation of objective biomarkers of disease activity and remission, comparative effectiveness research to identify patient subsets most likely to benefit from device or pharmacologic strategies, and investigations into biological mechanisms underpinning late remission in refractory epilepsy.
Funding and clinicaltrials.gov
Funding and trial registry details were not provided in the study summary supplied here. Readers should consult the original publication for specific funding sources, declarations of interest, and protocol registration information.
References
1. Potnis O, Biondo G, Sukonik R, Grzeskowiak C, Cutter G, Altalib H, Kuzniecky R, Lowenstein D, French J; HEP2 Investigators. Seizure Frequency Trends Over Time in Treatment‑Resistant Focal Epilepsy. JAMA Neurol. 2025 Oct 20. doi:10.1001/jamaneurol.2025.4085. Epub ahead of print. PMID: 41114972.
2. Kwan P, Arzimanoglou A, Berg AT, Brodie MJ, Allen Hauser W, Mathern G, Moshé SL, Perucca E, Wiebe S, French JA; ILAE Commission on Therapeutic Strategies. Definition of drug‑resistant epilepsy: Consensus proposal by the ad hoc Task Force of the ILAE Commission on Therapeutic Strategies. Epilepsia. 2010 Jun;51(6):1069–1077.
3. Kwan P, Brodie MJ. Early identification of refractory epilepsy. N Engl J Med. 2000 Feb 3;342(5):314–319.
