Alan Faraj (1), Rami Ayoun Alsoud (1), Joakim Nyberg (2), Ulrika S.H. Simonsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden, (2) Department of Pharmacy, Uppsala University, Uppsala, Sweden
Introduction: In most drug development programs, sponsors are recommended to submit a pediatric plan with the purpose of identifying the need for pediatric studies and to plan potential trials early. The plan can be referred to as a pediatric study plan (PSP [FDA](1) or a pediatric investigation plan (PIP [EMA])(2) and intends to make pediatric drug development programs efficient. In early development, adult pharmacokinetic (PK) data is collected to inform different aspects in future adult and pediatric trials. Population PK models offer advantages over non-compartmental analyses and are often applied to analyze the adult PK data. Based on the adult model, first-in-pediatric dose selection can be informed through allometry. Prior to initiation of a pediatric PK trial, sample size justification needs to be provided according to precision criteria(3) to ensure the estimation of important PK parameters with sufficient certainty. In rare disease areas, recruitment can be challenging and may lead to limited sample sizes, introducing risk of imprecise estimation of key PK parameters such as clearance (CL) and volume of distribution (V). This work aimed at optimizing the sampling schedules for two hemostatic drugs and to apply different approaches to prospectively power pediatric PK trials with limited sample size.
Methods: Adult population PK models(4,5) for marzeptacog alfa (MarzAA) and dalcinonacog alfa (DalcA) were used. Initially, three and four candidate sampling schedules were compared in pediatric subjects (60 µg/kg for MarzAA and 125 or 150 IU/kg for children between 11 and 6 years and 6 to 2 years of age, respectively, for DalcA) by simulation using the adult model followed by re-estimation with a simplified pediatric model. The design with the lowest uncertainty and bias in the PK parameters was taken forward to compare different approaches to prospectively power the pediatric trial. PK data in 1000 pediatric clinical trials (n=24 for both drugs) were simulated according to the sampling schedule chosen in the previous step. Each dataset was analyzed using the following approaches:
i. adult models
ii. simplified adult models
iii. adult models, fixed parameters except CL and V
iv. adult models, estimation of CL and V deviation in the pediatric population
v. adult informative Bayesian priors
vi. v with 2x inflated priors
vii. v with 5x inflated priors
viii. v with 10x inflated priors
Approaches i-iii involved both standalone and pooled analysis of pediatric and adult data, whereas the other approaches only included pediatric data. For each trial and analysis method, uncertainty estimates were used to calculate the number of trials with 95% confidence intervals of CL and V that fall within 60% and 140% of the geometric mean estimates. The test was performed for CL and V separately and in combination. The number of trials fulfilling the criteria was divided by the total number of trials to obtain the power.
Results: All designs provided similar relative standard errors and relative absolute bias estimates and therefore the designs with highest logistical potential were chosen. The power calculations showed that V was more difficult to estimate than CL for both drugs given the study designs, indicating that the chosen sampling schedules were more informative for CL. For i-iii, pooling data instead of standalone analysis improved the results, in particular for V (76, 100, 100 % versus 7, 7 and 52 % for MarzAA and 54, 97, 100 % versus 2, 8 and 0 %, for DalcA). Approach iv was similar to iii but in general performed worse than iii for MarzAA. However, for DalcA, iv performed better when the power calculations included V. The Bayesian priors approaches performed well in general with decreased power with decreased informativeness (vi-viii). Approaches v and vi provided >80 % power for both drugs regardless of parameters included in the calculations. Approaches vii and viii provided power ~80 % for MarzAA. For DalcA, the power to estimate CL alone was 82 and 83 % for vii and viii, respectively, but was reduced to <80 % when the power calculations included V.
Conclusions: The preferable approach depends on whether data can be pooled or not. The Bayesian priors approaches do not require pooling and performed well in general. Other simpler approaches were shown to be good alternatives when pooling of data can be made. Although the results are for two hemostatic drugs, they may be applicable to other drugs but remain to be confirmed.
References:
(1) Pediatric Study Plans: Content of and Process for Submitting Initial Pediatric Study Plans and Amended Initial Pediatric Study Plans. Available from: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/pediatric-study-plans-content-and-process-submitting-initial-pediatric-study-plans-and-amended
4
(2) EMA. European Medicines Agency. 2018 [cited 2023 Mar 8]. Paediatric investigation plans. Available from: https://www.ema.europa.eu/en/human-regulatory/research-development/paediatric-medicines/paediatric-investigation-plans
(3). Wang Y, Jadhav PR, Lala M, Gobburu JV. Clarification on Precision Criteria to Derive Sample Size When Designing Pediatric Pharmacokinetic Studies. J Clin Pharmacol.
(4). Faraj A, Knudsen T, Desai S, Neuman L, Blouse GE, Simonsson USH. Phase-3 Dose Selection of Marzeptacog Alfa (Activated) Informed by Population Pharmacokinetic Modeling: A Novel Hemostatic Drug. CPT Pharmacomet Syst Pharmacol.
(5). Faraj A, Le Moan N, Gorina E, Blouse GE, Knudsen T, Simonsson USH. Model-Informed Support of Dose Selection for Prophylactic Treatment with Dalcinonacog Alfa in Adult and Paediatric Hemophilia B Patients. Adv Ther.
Reference: PAGE 32 (2024) Abstr 10763 [www.page-meeting.org/?abstract=10763]
Poster: Methodology - Other topics