Conducting a clinical study in a pediatric patient population requires careful planning, as providing treatment to a child without causing harm is a weighty responsibility. The first tasks in striking this balance are dose selection and dosing interval determination, and this blog post describes how PK modeling and simulation can be applied to optimize the safety and efficacy of products for pediatric populations in clinical trials.

First steps in pediatric clinical studies

Data collection from preclinical and adult studies: Initiating pediatric clinical research typically begins with gathering data in preclinical in vivo studies and in healthy adult volunteers or patients. Population PK (popPK) modeling is used to extrapolate the pharmacokinetic (PK) and pharmacodynamic (PD) information to children, since the approach allows for the analysis of sparse and unbalanced datasets. PD modeling supports efficacy estimates in pediatric patients and contributes to efficacious dosing in PD design through simulation. A good, informative model explores the influence of covariates such as body weight and age to explain the variability in drug response and uses these covariates to translate PK information to children.

Exploratory analysis and non-compartmental analysis (NCA): The popPK model selection process starts with exploratory analyses (plotting and visual examination of PK profiles and descriptive statistics) and NCA. These are used to estimate exposure to a drug by estimating the area under the curve of a concentration-time graph, half-lives, and clearance patterns.

Non-linear mixed effect modeling (NLMEM): The next step is the NLMEM approach to developing and optimizing the PK model structure, which includes searching for the most statistically and clinically significant covariates. The process begins with a simple model, with complexity then being added (e.g., additional compartments or covariate relationships), and each addition is tested for statistical significance, physiological meaning, and clinical significance. The population characteristics that define the distribution of the PK/PD parameters, and their link to the covariates of interest, are then analyzed.

This approach works with both rich and sparse PK/PD data. Sparse sampling is typically used in patient studies to reduce burden and promote recruitment, and in such cases NLMEM is especially useful in providing information about efficacy and disease effect on the PK and PD of a drug. The covariates that are thus identified can link the model to pediatric populations based on developmental differences (e.g., body weight, creatinine clearance, and the activity of specific enzymes). If the model does not contain covariates relevant to young patients, they are added to the model based on the known information.

Allometric scaling for pediatric body weight estimates

Converting the newly created model from preclinical or adult studies typically starts with allometric scaling, an empirical approach connecting the clearance and volume of drug distribution to the body weight of a patient. For body weight-related estimates of PK parameters, the WHO growth chart is recommended for patients up to 2 years old. The CDC growth chart is the recommended source for demographic information on children older than 2. This data also can be sampled from the National Health and Nutrition Examination Survey database, which contains information about real patients and related characteristics.

One additional consideration, especially for longer-term pediatric studies, is changes in body weight due to natural growth. It is possible that patients may need to be transitioned from one weight and dosing category to another during a study, a situation that can be anticipated and planned for with the popPK model.

Additional developmental parameters for pediatric populations

Other parameters to account for include the maturation of organs and specific enzymes or transporters that participate in drug absorption, distribution, metabolism, and excretion. For instance, changes in enzyme expression in the liver as a child ages are well-characterized, but changes in the expression of enzymes and transporters by other organs have not been studied as extensively. Premier Consulting conducts thorough literature searches on a regular basis to update its models, increase the precision of predictions, and mitigate dosing risks.

The expression of blood proteins such as albumin is important as well in describing the PK of a drug in children, due to the effect of drug binding by blood proteins on PK. Physiological factors like body water content, fat tissue content, stomach pH, and liver maturation also must be incorporated into the model, according to the age of the children and the type of the drug.

Special care is required when dealing with very young children — preterm babies, neonates, and children up to age 2 — as these are the ages when changes in maturation are the most drastic and fast-evolving. This is true both for small-molecule drugs and biologics.

The effects of disease status on pediatric popPK modeling

In addition to developmental changes as children age, the disease status must be considered as a factor in PK modeling. Pediatric studies are done predominantly in patients, and data from healthy children are rarely available. In addition, the clinical presentation of a disease in children may be different from that in adults. For example, in some conditions like osteogenesis imperfecta, patient body weight may be lower than that of healthy children of the same age. In such a case, the demographics for the patient population should be included, rather than information from the standard growth chart.

On the other hand, pediatric patients with edema have higher body weights than average due to water retention, which is especially significant in neonatal patients. This may mean that lean body weight is a more beneficial datapoint than measured body weight for some drugs, and in some cases the volume of distribution estimates may need to be adjusted to account for the higher water content.

Pediatric studies for products with no adult clinical trials

Adult clinical trials are sometimes bypassed when a drug would be deleterious to healthy subjects and no adult patients are available due to high disease mortality at an earlier age. In this situation, when pediatric studies are initiated without data from adult trials, modeling is of special importance and requires creative approaches and up-to-date knowledge of the condition.

The model typically provides initial estimates for dosing in children, and the lowest dose predictions are tested with caution in older children in a staggered manner. Real-time PK analysis of the first patients allows real-time model refinement to reduce the risk of testing for a more efficacious dose and in younger children. This ability to quickly analyze the results and refine the dosing model is crucial for the safety and success of the study and helps streamline the delivery of the new medications to children.

There are numerous advantages to using a model-based approach for PK studies in children. NLMEM facilitates the analysis of sparse, unbalanced datasets, which are common in neonatal and pediatric research settings, where each participant may contribute only a small number of samples, and where sample timing and the number of available samples can vary among patients.

All models must be validated in the clinical trial to determine if the model and the covariates are correct and thus provide an accurate dose and dosing interval information. The goal of PK modeling and simulation is to explore the behavior of medicines in children and provide safe and effective doses to treat their diseases.

Premier Consulting has extensive experience in using PK modeling to conduct dose selection and dosing interval determination for pediatric clinical trials. Contact us to find out how we can advance your pediatric study planning and execution.

Co-Authors:

Barry Mangum, PharmD, FCP
Senior Vice President, Pediatrics
Premier Research

Serge Guzy, PhD
President and CEO
Pop-Pharm