Chenyao Liu1, Ms Mercy Atieno2, Mr Alessandro Di Deo1, Karin Kipper3, Josemir Sander2, Professor Charles Newton4, Professor Oscar Della Pasqua1
1Clinical Pharmacology & Therapeutics, University College London, 2Epilepsy Pathway Innovations in Africa (EPInA) study group, 3Chalfont Centre for Epilepsy, University College London,, 4University of Oxford
Introduction: Epilepsy is one of the most frequent chronic neurological conditions in adults and children, and its impact is enormous, particularly in resource-limited settings, such as many of the countries in sub-Saharan Africa[1]. In addition to disease progression and refractoriness, treatment outcome can be affected by variable adherence to anti-epileptic drugs (ADEs). The SMS Trial (NCT05321641) was set up as a multicentre, randomized clinical study in adult and paediatric patients with epilepsy to investigate the impact of a mobile messaging reminder service on adherence to ADEs. Patients were randomly assigned to four cohorts, namely text SMS, graphic/audio SMS, both text and graphic/audio SMS, and non-reminder SMS as control. Here we present the implementation of a model-based approach for evaluation of the effect of different patterns of adherence[2] as assessed by self-reporting, on the exposure to carbamazepine (CBZ) and phenobarbital (PB). Method: The pharmacokinetic data available for this analysis consisted of therapeutic drug monitoring samples from 223 patients receiving CBZ (median age 27.2 (range 2.5 – 79.4) yrs) and 189 patients receiving PB (median age 25.3 (range 1.9- 80.2) yrs). Given the sparse Pk data collected, a Bayesian approach has been used to incorporate prior parameter distributions from published population PK models[3–5]. Model refinement and evaluation were implemented in NONMEM v.7.5 using the FOCE-I estimation method. Standard diagnostic criteria (i.e., visual predictive checks, goodness-of-fit, decrease in log-likelihood) were used for model selection and evaluation. Predicted individual CBZ and PB concentration vs time profiles were derived and used to calculate systemic exposure (AUC0-24h) and average steady-state concentrations (Css). Exposure was summarised as medians (5th-95th percentiles) and compared across SMS groups to evaluate the implications of variable adherence, taking into account accepted target ranges for therapeutic drug monitoring. In addition, simulations were conducted to explore the effect of different patterns of adherence (i.e., random missingness, drug holiday, treatment termination) on systemic exposure to each drug. All simulation procedures were implemented using the final model parameter estimates. Statistical and graphical summaries were performed with R. Result: Baseline demographics of subjects taking CBZ (median dose 400 mg/day (range 100 – 1200)) were not significantly different from those receiving PB (median dose 60 mg/day (range 15 – 120)). Of note, the median weight was 56.4 kg (range 8.2 – 122.6) vs 52.6 kg (range 7.1 – 122.6) for CBZ and PB, respectively. The demographics were also balanced across SMS groups. Both CBZ and PB were described with a one-compartment model with first-order absorption and first-order elimination. Body weight was included as a covariate on clearance and volume of distribution according to an allometric function. The effect of co-medication known to affect CBZ and PB clearance was also considered. Phenytoin (PHT), valproic acid (VPA) and PB were included as covariates into the CBZ model, whilst PHT and VPA were added to the PB model. Final population estimates for CBZ were CL = 0.06 L/h (19.5% IIV) and Vd = 1.91L (10.1% IIV); for PB, CL = 0.17 L/h (32.1%) and Vd = 14.78 L. Predicted AUC0-24h for CBZ was not significantly different across SMS groups. In contrast, exposure to PB in text group and audio group was significantly higher than control (p = 0.005 and 0.0039, respectively), indicating improved adherence in text SMS and audio SMS groups for this AED. Despite the lack of differences in the observed exposure to CBZ, variable patterns of treatment adherence did cause a significant reduction in systemic concentrations, as indicated by the simulation scenarios. Randomly missing 20% of the prescribed doses would lead to a 21% reduction in median AUC0-24h. Similarly, drug holidays (no dose intake over 3 days) and treatment interruptions (miss doses for 1 week) caused, respectively, a 15% and 22% reduction in median AUC0-24. Conclusion: In summary, monitoring patient adherence to treatment appears to be associated with increased exposure in patients receiving PB, but this effect was not observed in those receiving CBZ, despite a balanced distribution of the patients across the SMS cohorts. A more comprehensive simulation-based evaluation of patients on stable and initial therapy, is required to establish the forgiveness of different AEDs and the clinical relevance of mobile messaging service.
1. Shawel B, Berhane Y. Adherence to anti-seizure medications and self-reported availability and affordability of the medications in Addis Ababa, Ethiopia. PLoS One. 2024;19(10):e0299964. doi:10.1371/journal.pone.0299964 2. Piana C. Adherence to Antiretroviral Combination Therapy in Children : What a Difference Half a Day Makes… Division of Pharmacology, Leiden Academic Center for Drug Research (LACDR), Faculty of Science, Leiden University; 2013. Accessed March 10, 2025. https://hdl.handle.net/1887/22077 3. Goto S, Seo T, Murata T, et al. Population estimation of the effects of cytochrome P450 2C9 and 2C19 polymorphisms on phenobarbital clearance in Japanese. Ther Drug Monit. 2007;29(1):118-121. doi:10.1097/FTD.0b013e318030def0 4. Jiao Z, Shi XJ, Zhao ZG, Zhong MK. Population pharmacokinetic modeling of steady state clearance of carbamazepine and its epoxide metabolite from sparse routine clinical data. J Clin Pharm Ther. 2004;29(3):247-256. doi:10.1111/j.1365-2710.2004.00557.x 5. van Dijkman SC, Rauwé WM, Danhof M, Della Pasqua O. Pharmacokinetic interactions and dosing rationale for antiepileptic drugs in adults and children. British Journal of Clinical Pharmacology. 2018;84(1):97-111. doi:10.1111/bcp.13400
Reference: PAGE 33 (2025) Abstr 11545 [www.page-meeting.org/?abstract=11545]
Poster: Drug/Disease Modelling - Other Topics