Franziska Schaedeli[1] Stark, Caterina Bissantz[2], Sian Lennon-Chrimes[1,3], Valérie Cosson[1], Neil Parrott[2], Nicolas Frey[1]
Roche Pharmaceutical Research and Early Development, Clinical Development[1], Pharmaceutical Sciences[2], Roche Innovation Center Basel[1,2], Switzerland; Roche Innovation Center Welwyn[3], UK
Objectives: DrugA is a small molecule being developed for a neurologic disorder in a pediatric target population. Dose selection for a POC study targeted adult-equivalent exposure. While PBPK modeling is an accepted tool for pediatric dose prediction, health authorities also requested pre-defined criteria for early dose confirmation or adjustment (EDC), and sufficient confidence in the key PK parameters (CL, Vss) in adolescents before dosing younger children. We propose a method for EDC with a minimum number of subjects based on combined PBPK and population PK (PopPK) modeling and bootstrap resampling techniques.
Methods: A PBPK model was developed for children aged 5-17 yr based on an adult model for DrugA utilizing the pediatric module in SimCYP for scaling to children. Drug exposure (AUCss) and CL were derived from 4000 simulated PK profiles, and initial selection of pediatric age-adjusted doses was guided by the ratio of pediatric vs. adult predicted CL (relCL) to achieve acceptable variability in exposure with a limited number of dose adjustment age groups (DAG). Criteria for EDC were set as follows: the dose is confirmed if observed relCL in a DAG is within 70-140% of predicted, otherwise it needs to be adjusted. The impact of sample size on the confidence for EDC was assessed by bootstrapping relCL from the pediatric PBPK database. For each sample size (N=4…40), the median relCL and a 90% CI was obtained from 1000 bootstrapped datasets. The minimum sample size with the 90% CI falling within the pre-defined limits was recommended for EDC. The performance of the approach with respect to EDC decision and PK parameter precision was assessed using PopPK model based clinical trial simulation (CTS).
Results: Median pediatric simulated CL approximated adult CL at the age of 16, but was reduced to 27% at 5 yr. Age and weight were equal predictors for relCL, justifying age-based pediatric dosing, which is preferred over weight-based dosing by clinicians. Four DAG (5-7, 8-10, 11-14, and 15-17 yr) were proposed, targeting 30, 45, 65 and 95% of the adult doses, based on predicted relCL and variability, available dose strengths and exposure limits for safety. EDC criteria was met with the bootstrapped 90% CI for N>=10 subjects per DAG. Results of CTS demonstrating the value and limitations of the approach will be presented.
Conclusions: PBPK modeling together with bootstrap techniques offer a tool to assess the sample size and criteria for EDC in a pediatric POC study.
Reference: PAGE 24 (2015) Abstr 3553 [www.page-meeting.org/?abstract=3553]
Poster: Methodology - Study Design