Jingxian Chen, Brian Maas, Luzelena Caro, Wei Gao
Merck & Co., Inc., Kenilworth, NJ, USA
Objectives: MK-1654 is a monoclonal antibody being developed for the prevention for Respiratory Syncytial Virus (RSV) in infants. A Phase 2b/3 pivotal study is actively recruiting eligible healthy full-term and pre-term infants (N=3300). In order to minimize sampling burden in vulnerable infants, a sparse pharmacokinetic (PK) collection of no more than 3 samples per infant is planned in this study. The objective of this analysis is to inform the PK sampling timepoints and sample size to facilitate robust PK analysis in infants.
Methods: MK-1654 PK was described using a two-compartment model with first-order absorption. Four post-dose time points were chosen for PK sampling optimization: time to maximal concentration (Tmax), Day 150 (D150) representing intended duration for RSV protection, a time between Tmax and Day 150 (Tmid) and a time at the terminal phase (Tterm) of the PK curve. A sample on Day 150 was collected in all infants for primary endpoint evaluation. PK optimization was conducted using R (v3.5.3) package, PopED (v0.4.0). PK sampling scenarios were created with varying number of active subjects, PK schemes and PK time points. Sample sizes were chosen incrementally up to the total sample size receiving active treatment of the study (N=2200). A combination of two or three PK timepoints were assigned to either two or three sequences as below
- 2+2: (Tmax, D150) or (D150, Tterm)
- 2+2+2: (Tmax, D150) or (Tmid, D150) or (D150,Tterm)
- 3+2: (Tmax, D150, Tterm) or (D150, Tterm)
These PK schemes were explored with different sample sizes. The predictive performance of the optimal sampling strategy was evaluated based on bias and precision estimates of selected PK parameters: clearance (CL), central volume of distribution (Vc), absorption rate constant (Ka) and their respective inter-individual variabilities. Due to the sparse sampling, inter-compartmental clearance (Q) and peripheral volume of distribution (Vp) could not be estimated and were largely informed by the MK-1654 PK data from adults; [1] therefore, they were not selected for sampling optimization. Parameter precision, represented as relative standard error (RSE), was used to guide PK sampling selection. The cutoff for precise parameter estimation used was %RSE < 35%.
Results: The PK optimization analysis revealed that for PK sampling schemes with 2 sampling times, 1500 infants in the active group provided reasonable PK parameter precision. PK schemes with and without Tmid yielded similar precision in selected parameters. 3+2 PK schemes yielded slightly higher parameter precision than 2+2 PK scheme given the same sample size of 1500 active subjects. The 3+2 scheme only required 400 active subjects in the 3-sample arm and 750 active subjects in the 2-sample arm to achieve similar parameter precision to the 2+2 scheme. That is, collecting one additional sample from 400 subjects in the 3-sample arm reduced the number of subjects required by 350 from a total of 1500 active subjects using 2+2 PK schemes.
Conclusions: Considering the convenience of clinical conduct, the 2+2 PK sampling strategy in at least 1500 subjects with 1:1 randomization to sampling is considered as a preferable option to provide adequate parameter precision while reducing sampling burden. These PK data from Phase 2b/3 infant study will be included in the infant population PK model development. The developed model will be used to provide PK exposure predictions for the studied population to further support exposure-response analysis.
[1] Maas, BM, et al. Open Forum Infectious Diseases, 2018, Nov, Volume 5, Issue suppl_1, S424–S425.
Reference: PAGE 29 (2021) Abstr 9651 [www.page-meeting.org/?abstract=9651]
Poster: Study Design