002: Model-based Investigation on Allometric Scaling in Pediatric PK: The Effects of Sample Size, Sampling Scheme & Age Range

Statement of Purpose: Understanding PK in pediatric patients is a regulatory requirement in drug development. Allometric scaling is commonly used to extrapolate PK of monoclonal antibodies (mAbs) from adult to children. It has been shown previously that PK in children is predictable for mAbs exhibiting linear dynamics based on adult data and allometry. However, the estimates and precision of the exponents used in allometric scaling can be impacted by how the PK data were collected in pediatric studies. This work aimed to investigate the key study design factors that could potentially affect the estimation of the allometric scaling exponents.

Description of Methods & Materials: A virtual pediatric population was constructed, with various body weight distribution using data from the CDC growth charts. The dataset was segmented into four age groups (2-5, 6-11, 12-17, and 2-17 y.o) and body weight values were randomly generated for each group. The generated weights were pooled together and sampled randomly assuming different sample sizes for simulations of PK concentrations up to 56 days (approximately 5 half-lives) with a single 2 mg/kg dose of a hypothetical mAb. The simulations used a realistic set of population PK parameters for mAb, with inter- and intra-individual variabilities. The allometric scaling exponents of the volume of distribution and clearance were fixed at 1.0 and 0.75, respectively, during the simulations. To investigate the impact of different sample sizes on the model estimates of allometric exponents, the simulated PK concentration data were next used in NONMEM model fitting with the allometric scaling exponents assumed unknown and to be estimated. Also, different hypothetical sampling schemes were explored, ranging from sparse to rich scenarios. Furthermore, a realistic clinical scenario was tested involving pooling intensively sampled Ph1 data in adult with sparsely sampled pediatric data, and the results were compared with the estimates when only sparsely sampled pediatric data were used in modeling.

Data & Results: Generally, the accuracy of estimates of allometric scaling exponents increases with larger sample size, broader range of age groups, and a sampling scheme combining adult data. With a sample size of ≥60, median values of the exponent estimates were comparable or close to the true allometric values. Also, a wider body weight distribution from a broader age range (2-17 y.o) is associated with higher accuracy of estimates when compared to individual age groups. Combining sparsely sampled pediatric data (14, 28 or 4, 14, 28 days) with intensively sampled adult data (1, 2, 4, 7, 14, 28, 42, 56 days) improved the accuracy of estimates, except for age group 12-17 y.o.

Interpretation, Conclusion or Significance: Pediatric data with a sample size of ≥ 60 would be adequate for pediatric allometric scaling. Also, combining sparsely sampled pediatric data with intensively sampled adult data improves the accuracy of estimates and also reduces the data requirement from a pediatric population. With the tested hypothetic mAb PK profile, sampling on days 4 (peak), 14 and 28 is sufficient to determine pediatric allometric exponents for both the apparent clearance and volume of distribution.

Disclosures: All authors have disclosed that they are employees of Janssen R&D, LLC. MNS has Johnson & Johnson stocks.