Multiple drugs to treat coronavirus disease 2019 (COVID-19) are currently under investigation. As of May 1, 2020, ClinicalTrials.gov lists 409 active studies evaluating various therapeutics to treat COVID-19. Most of these studies enroll exclusively adults, providing limited data on children. However, if efficacy of COVID-19 therapeutics is established in adults, these drugs will also be widely prescribed to children for whom appropriate dosing has not been established. We know from observational studies across multiple therapeutics that inappropriate drug dosing places children at risk for treatment failure, toxicities, and even death.
Dosing of drugs is different in children compared with adults primarily owing to age-related changes in anatomy and physiology, such as tissue composition, relative organ weight, blood flow rates, and the maturity of clearance processes, such as kidney and enzymatic function.2 Many of these processes mature nonlinearly, limiting direct weight-based extrapolations from adults to children.
To determine optimal dosing in children, dedicated pediatric pharmacokinetic trials are needed. Despite an increase in pediatric drug trials, owing in large part to the Best Pharmaceuticals for Children Act and research consortiums such as the Pediatric Trials Network, pediatric drugs trials remain challenging. Pediatric trials are hampered by low study consent rates, difficulty in obtaining blood, limited blood volume available for biologic specimens, and heterogeneity in drug exposure across the age range. As a result, pediatric pharmacokinetic trials take significant time and resources. A 2019 analysis of pediatric drug trials submitted to the US Food and Drug Administration found that only 55.9% of pharmacokinetic studies were completed within 5 years. During a rapidly evolving pandemic, we urgently need a more efficient approach to determine rational dosing of established therapeutics for children affected by COVID-19.
In the absence of lengthy pediatric pharmacokinetic trials, modeling and simulation can leverage existing data to determine first-in-pediatric dosing in the face of urgent need. Broadly, pharmacokinetic models use mathematical equations to characterize drug disposition in the body. The processes that govern drug disposition in children can be built into the model (eg, maturation function that reflects age-related change in glomerular filtration). Once a model is established, simulation is used to predict concentrations for various dosing regimens, usually in a large number of simulated individuals reflecting population variability. Based on the population variability, these simulations will generate a range of exposures for each dosing regimen. If a specific concentration (eg, trough) or exposure (eg, area under the curve) is associated with efficacy or toxicity, the simulation results can estimate the number of children who would achieve the target concentration or remain below a toxic threshold for a given dose. Based on the risk-benefit balance for a particular therapeutic, an optimal dosing regimen across the pediatric age spectrum can be determined.
In this issue of JAMA Pediatrics, Maharaj et al8 use modeling and simulation to determine optimal dosing regimens for hydroxychloroquine and remdesivir in children with COVID-19. The authors used different modeling and simulation approaches for each drug based on the data available to build the models. For hydroxychloroquine, they used a physiologically based pharmacokinetic approach; for remdesivir, they scaled an adult-population pharmacokinetic model to children using allometry with age-dependent exponents. In both cases, they were able to establish pediatric dosing that achieved drug exposures that were comparable with the exposures achieved with standard adult dosing.
The axiom attributed to George Box that all models are wrong but that some are useful highlights the need to understand a model’s assumptions and limitations. Both of the modeling approaches used in the study (population pharmacokinetic and physiologically based pharmacokinetic) have strengths and limitations that are well described.9 Regardless of the technical approach, for modeling and simulation to be effective in determining first-in-pediatric dosing, there must be an established target drug exposure. Often, this is simply matching the simulated pediatric exposure with the exposure observed in the adult efficacy trial. In other instances, there is an effective concentration identified through in vitro studies. In the current study, Maharaj et al8 highlight the challenges in these assumptions. For hydroxychloroquine, they used the pediatric model to match the exposure following a standard dosing regimen in adults (400 mg twice a day for 1 day or 200 mg twice a day for 4 days). However, these matched exposures resulted in concentrations below the threshold needed for viral suppression that was identified in an in vitro study.10 Although hydroxychloroquine may have an additional immunomodulatory effect, these results suggest that hydroxychloroquine dosing recommendations for both adults and children may be inadequate to achieve appropriate viral suppression.
Mathematical equations by themselves cannot replace prospective clinical trials. Once model-based recommendations are made, they will need to be evaluated. One innovative approach to collecting pharmacokinetic data in children is through opportunistic studies. Opportunistic studies enroll participants who are already prescribed the drug of interest per standard of care. Ideally, pharmacokinetic samples are also collected at times of routine laboratory blood draws, eliminating the need for study-specific phlebotomy. There are multiple advantages to this approach compared with traditional pharmacokinetic trials, mostly notably that the population of interest can be quickly enrolled. Limitations of opportunistic studies center on the sparse sampling and limited control over timing of samples. As a result, opportunistic studies require larger numbers of children and specialized expertise in modeling.
The Pediatric Trials Network11 has been at the forefront of pediatric drug research. The Pediatric Trials Network is a contract sponsored by the Eunice Kennedy Shriver National Institute of Child Health and Human Development through the Best Pharmaceuticals for Children Act that is tasked with coordinating efforts and creating the infrastructure to conduct safe and effective pediatric clinical drug trials. The investigators in the current study are the leaders of this network and have pioneered pediatric modeling and simulation and innovative approaches to pediatric trial design. In addition to the modeling efforts in this issue of JAMA Pediatrics,8 the Pediatric Trials Network funds the Pharmacokinetics, Pharmacodynamics, and Safety Profile of Understudied Drugs Administered to Children Per Standard of Care (POPS) study (NCT04278404) that is currently enrolling children who are prescribed certain drugs per standard of care, including drugs being studied to treat COVID-19. The combined approach of modeling and simulation with confirmatory opportunistic pharmacokinetic trials provides a powerful and efficient way to determine optimal dosing in children during emergency situations such as the current COVID-19 pandemic.