A joint ketamine and esketamine exposure-response model of MADRS score in major depressive disorder to enable informed trial design for candidate drugs with similar mode of action
Alan Faraj (1), Anders Kristoffersson (1), Michael Desch (2), Ronald Niebecker (2)
(1) Pharmetheus AB, Uppsala, Sweden (2) Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany
Objectives: Ketamine and esketamine have been shown to have positive effects on Montgomery-Åsberg Depression Rating Scale (MADRS), an endpoint commonly utilized to measure drug effects in major depressive disorder (MDD). This work aimed to develop a pharmacokinetic-pharmacodynamic (PKPD) model based on digitized MADRS score data and to perform simulations to demonstrate the utility of the model for dosage optimization of ketamine, and drugs with similar mode of action.
Methods: Individual MADRS scores were digitized using WebPlotDigitizer (1) after twice and thrice weekly administration of ketamine (0.5 mg/kg IV, ~4 weeks of treatment plus >2 weeks of follow up, n=18 and n=16 respectively) (2), and twice weekly dosing of esketamine (0.2, 0.4 mg/kg IV, 7 days, n=9 and n=11 respectively) (3). Mean change from baseline placebo data was digitized following twice and thrice weekly administration up to day 15 (each n=16) (2), and individual placebo data was digitized up to day 4 (n=10) (3). MADRS scores were treated as continuous type data. A population PKPD model built on a ketamine dose titration study, with wash out between dose escalations, (4) was evaluated but could not describe the combination of short and long-term (4 weeks) data including follow-up. An indirect response model with positive feedback, suitable for long lasting effects after short treatment interventions with slow return to baseline, was tested (5). The drug effects tested stimulating the feedback rate (KFB ) were linear (slope) and Emax models. The model was extended with a proportional placebo component implemented on the positive feedback compartment at baseline. A tolerance model and differences in potency between ketamine and esketamine were tested. As no PK data was available, all PK parameters were fixed to literature values for ketamine (4) and esketamine (6), and the individual PK parameters were informed by the PKPD data. Discrimination of models were mainly based on visual predictive checks (VPC), individual plots and change in objective function value (OFV). Based on the model, MADRS scores over 6 weeks were simulated after 0.125, 0.5 and 2 mg/kg ketamine as I. a single dose, II. 4 weeks of dosing every fourth day, and III. 6 weeks of dosing with the same frequency. Subsequently, a 2 mg/kg/week dosing was simulated given once daily, every fourth day or once weekly.
Results: The final model was an indirect response model with positive feedback and predicted the digitized MADRS score data well following placebo, ketamine and esketamine treatment. Both the rapid decrease within 2h after the first dose, and subsequent slower decrease over 4 weeks of treatment were predicted well. In addition, the stable or slowly increasing MADRS score was also predicted by the model during the follow-up period (~18 days) after end of treatment. The drug effect was found to be linearly concentration dependent stimulating washout of MADRS. Esketamine was found to be ~5x more potent than ketamine, but the estimate should be regarded with caution as study effects cannot be ruled out. The observed twice weekly ketamine (0.5 mg/kg) showed similar or slightly better effect than the same dose administered thrice weekly, but a tolerance model was not successful in improving the description of the data. Simulations showed the average long-term effect to be insensitive to dosing interval, and was rather governed by the total weekly dose. However, higher doses given less often created a faster onset of effect, while dividing the dose into more frequent administration smoothed the response profile, suggesting that treatment of MDD with Ketamine may benefit from a loading dose followed by daily maintenance dosing. The results should however be interpreted with caution as the once daily and once weekly dosing is extrapolated from dosing twice and thrice weekly.
Conclusions: Overall, the model performed well describing ketamine, esketamine and placebo data. Although the dataset was limited by narrow dose range and no PK data, the model can be used for evaluating alternative dosing regimens for ketamine/esketamine, and as a framework to prospectively inform design of trials for candidate drugs with similar mode of action against MDD.
 Singh, J. et al., A Double-Blind, Randomized, Placebo-Controlled, Dose-Frequency Study of Intravenous Ketamine in Patients With Treatment-Resistant Depression. Am J Psychiatry 173:8, 2016. 173:8
 Singh et al., Intravenous Esketamine in Adult Treatment-Resistant Depression: A Double-Blind, Double-Randomization, Placebo-Controlled Study. Biol. Psychiatry 424:431, 2016.
 Perez, Ruixo., Population Pharmacokinetics of Esketamine Nasal Spray and its Metabolite Noresketamine in Healthy Subjects and Patients with Treatment‑Resistant Depression. Clinical Pharmacokinetics, 60:501-516, 2021