Authors :
Presenting Author: Bovornpat Suriyapakorn, PharmD – University of Minnesota
Shen Cheng, PhD – University of Minnesota
Eric Rosenthal, MD – Massachusetts General Hospital
James Cloyd, PharmD – University of Minnesota
Jaideep Kapur, MBBS, PhD – University of Virginia
Robert Silbergleit, MD – University of Michigan
James Chamberlain, MD – Children's National Hospital
Shahriar Zehtabchi, MD – Downstate Health Sciences University
Thomas Bleck, MD – Northwestern University
Mark Quigg, MD – University of Virginia
Lisa Coles, PhD – University of Minnesota
Rationale:
Established status epilepticus (ESE) is a life-threatening condition with high morbidity. Current ESE treatments are inadequate and effective in less than 50% of patients. Ketamine, an NMDA antagonist, is promising as an adjunct to levetiracetam for treating ESE. Understanding the dose-exposure-response relationship through pharmacokinetic (PK) studies is crucial for successfully translating ketamine to treat ESE, but traditional intensive sampling design, requiring over 12 blood samples per participant, is impractical in clinical care, including emergency settings. This study aims to develop a practical PK sampling strategy for KESETT by evaluating the feasibility of sparse sampling using pharmacometric simulation and re-estimation.
Methods:
A ketamine PK model based on published report1 was used to simulate rich concentration-time profiles for 1 and 3 mg/kg ketamine single-dose in 400 virtual patients, with demographics simulated according to existing studies2. Simulations were performed assuming a 20% sampling time error. Three sampling strategies were assessed:
- Intensive sampling
- Sparse sampling with three-time points
- Sparse sampling with two-time points
Simulated data were used to re-estimate model parameters, and partial area under the curve (pAUC) was calculated using empirical Bayes estimates (EBEs). Sampling schemes were evaluated by comparing the following:
- The alignment of PK parameters and pAUCs with intensive sampling.
- Percent prediction error (PPE).
Modeling and simulations were conducted with NONMEM v7.5 and the R package mrgsolve v1.4.1.
Results:
PK profiles were generated for the two ketamine doses. Three sampling strategies were evaluated:
- Intensive sampling: Provided PK estimates and pAUC, serving as the reference standard.
- 3-sample method: Demonstrated the most aligned PK parameters and pAUC, with a PPE of 15%.
- 2-sample method: Showed slightly greater deviation in PK parameters and pAUC, with a median PPE of 16%.
Validation with 100 simulated datasets confirmed the robustness of both the 3- and 2-sample method, maintaining a PPE within ±20% of intensive sampling in 87% and 86% o
f cases, respectively.Conclusions:
Both the 3- and 2-sample methods effectively estimate ketamine PK, with the 3-sample method showing slightly better alignment. However, the 2-sample method remains within acceptable PPE limits. Therefore, the 2-sample method is a more practical and effective alternative to intensive sampling for the PK study in KESETT.
Funding: None