Using Seizure Dynamotypes to Predict Response to Antiseizure Medications
Abstract number :
2.347
Submission category :
7. Anti-seizure Medications / 7C. Cohort Studies
Year :
2025
Submission ID :
1161
Source :
www.aesnet.org
Presentation date :
12/7/2025 12:00:00 AM
Published date :
Authors :
Presenting Author: Ashley Zachery-Savella, PhD – University of Utah
Kishore Jay, BS – University of Utah
Joshua Wooley, BS – University of Sydney
Michelle Le, BS – University of Utah
Sally Scofield, BS – University of Utah
Kyle Thomson, PhD – University of Utah
Gerald Saunders, PhD – University of Utah
Ben Erekson, BS – University of Utah
Peter West, PhD – University of Utah
Daria Daria Nesterovich Anderson, PhD – University of Sydney
Karen Wilcox, PhD – University of Utah
Rationale: Epilepsy treatment is complicated by the fact that patients with similar phenotypes and seizure classifications do not respond similarly to the same treatments, suggesting subtler differences exist that cannot be accounted for by our currently-limited classification system. Investigating seizure onset and offset patterns in models of epilepsy and their response to anti-seizure medications may provide additional metrics to quantify drug efficacy and could lead to improvements in how epilepsy is classified and treated.
Methods: We analyzed seizure frequency and dynamotype in six treatment cohorts: 294 seizures/21 animals in the carbamazepine (30 mg/kg, t.i.d.) cohort, 820 seizures/17 animals in the phenobarbital (PB, 50 mg/kg, b.i.d.) cohort, 848 seizures/22 animals in the valproate (VPA, 240 mg/kg, t.i.d.) cohort, and 825 seizures/21 animals in the phenytoin (PHT, 20 mg/kg, b.i.d.) cohort. We analyzed two additional cohorts with surface telemeters: 1008 seizures/23 animals that received PBl, and 1447/16 animals that received PHT. Onset and offset patterns, or “dynamotypes,” of each seizure were visually categorized in a randomized, blinded fashion by trained human raters and a custom machine learning script. There are three onset dynamotypes: SupH, characterized by increasing amplitude; SNIC, characterized by increasing frequency; and SubH, characterized by arbitrary patterns in amplitude and frequency. There are three offsets: SupH—decrease in amplitude, SNIC—decrease in frequency, and SubH—arbitrary amplitude/frequency (Figure 1).
Results: Figure 2: PHT significantly reduced the proportion of SNIC onset and offset patterns compared to predrug baseline (Dirichlet-multinomial Bayesian; OLS p = 0.005). In the PHT telemeter cohort, PHT significantly reduced SupH and SNIC offset patterns with a subsequent increase in SubH (LMM, SupH: p = 6.1e-8; SNIC: p = 1.5e-9; SubH: p = 2.6e-10). VPA increased SupH onset (Dirichlet-multinomial Bayesian). VPA decreased SubH offset with a subsequent increase in SNIC offset (weighted least squares, FLC: p = 0.016; SNIC: p = 0.023). PB significantly decreased SupH and SNIC onset with a corresponding significant increase in SubH onset proportion (Dirichlet-multinomial Bayesian).
Conclusions: Dynamotype analysis reveals drug-specific modulation of seizure onset and offset patterns. PHT did not reduce seizure frequency but significantly decreased SNIC onset and offset proportions, suggesting altered dynamics without clinical efficacy. In contrast, VPA reduced seizure frequency, and remaining seizures had a higher proportion of SNIC offset while seizures with SubH offset were decreased. PB also shifted onset dynamics by reducing SupH and SNIC with a reciprocal increase in SubH. These results support dynamotypes as potential predictors of response in the IAK model.
Funding: This project has been partly funded by Federal funds from NINDS Epilepsy Therapy Screening Program, NIH, and Department of Health and Human Services, under Contract No. HHS 75N95022C00007. Also: Undergraduate Research Opportunities Program at the University of Utah, NIH NINDS: F32 NS114322 awarded to Daria Anderson and University of Utah Skaggs Fellowship awarded to Ashley Zachery-Savella
Anti-seizure Medications