Using Maturation Models to Select Infant Doses

Why Infant Dose Selection Needs Both Age and Size

A dose scaled only by body weight assumes that clearance per unit of the chosen body-size scale is similar across ages. That assumption can fail in infants: clearance processes may still be maturing, so two infants of the same weight but different developmental ages can have different clearance and exposure. Conversely, age alone is not enough, because infants of the same age can differ in body size. A population pharmacokinetic (PK) model can represent these influences together and use the resulting clearance predictions to compare candidate doses.

The aim here is a transparent, hypothetical example of dose selection, not a dosing recommendation for a real drug. The target exposure and model parameters below are illustrative. In an actual development decision, the target would need scientific justification, and the model and predictions would need evaluation against relevant data.

A Clearance Model With Size and Maturation

Let \(WT\) be body weight in kilograms and \(PMA\) be postmenstrual age in weeks. The model is

Here, \(CL\) is clearance in litres per hour; \(CL_{70}\) is the mature reference clearance for a 70 kg individual; the exponent 0.75 is an illustrative allometric size exponent; \(PMA_{50}\) is the age at which the maturation fraction is one-half; and \(H\) controls the steepness of the maturation curve. The units used for \(PMA\) and \(PMA_{50}\) must match. The maturation fraction is

For positive ages and \(H>0\), \(M(PMA)\) is between zero and one. It equals one-half when \(PMA=PMA_{50}\), and it approaches one as \(PMA\) increases without bound. Thus the model's mature reference clearance is recovered at the 70 kg reference size as maturation approaches its plateau. The numerical choices \(CL_{70}=10\ \mathrm{L/h}\), \(PMA_{50}=55\) weeks, and \(H=3\) are hypothetical, not universal values.

This multiplicative form assumes that the specified size effect and age-related maturation effect can be represented separately. It also assumes the model is suitable over the ages and weights where it will be used. The exponent, maturation parameters, and any between-individual variability should be estimated or justified for the drug and population, not selected merely because a familiar form is available.

Connecting Clearance to an Exposure-Based Dose

For a linear PK model at steady state, the area under the concentration-time curve over a dosing interval is the amount of drug reaching systemic circulation during that interval divided by clearance. For oral dosing, if bioavailability \(F\) is constant and the administered amount is \(D\), then

Here, \(D_{24}\) is the total oral dose administered in 24 hours, and \(AUC_{ss,24}\) is the steady-state AUC over 24 hours. The second equation is obtained by rearranging the first. It is valid for this calculation when PK is linear over the relevant concentration range and \(F\) is known and does not change across the infants being compared. If the formulation, route, dosing interval, or bioavailability differs, the exposure calculation must reflect that difference.

Worked Comparison Across Developmental Ages

The next comparison holds the model parameters and exposure target fixed but considers three hypothetical age-and-weight combinations. This illustrates why the predicted dose per kilogram can change with age even though weight is included in the clearance model.

Why the Model Form Produces This Dose Calculation

The key justification is that the dose calculation follows directly from the exposure definition under linear PK. If clearance is constant over the interval and the system is at steady state, the total systemic amount entering during 24 hours is \(F D_{24}\). Clearance removes drug at a rate proportional to concentration; over the interval, the relationship between systemic amount and total exposure is \(F D_{24}=CL\times AUC_{ss,24}\). Dividing by \(CL\) gives the AUC equation, and multiplying by \(CL/F\) gives the target-based dose. No claim about a particular concentration-time profile is required for this AUC relationship.

The important model-based step is predicting \(CL\) for the infant. The model's multiplicative structure means that, holding all else fixed, a higher weight raises predicted clearance through the size term, while a higher maturation fraction raises it through the age term. Consequently, the target-based dose changes in the same direction as predicted clearance when \(AUC_{\mathrm{target}}\) and \(F\) are held fixed.

Check a Weight-Based Alternative

A fixed milligram-per-kilogram regimen is simpler to prescribe, but it does not automatically match exposure across ages. Consider a hypothetical fixed regimen of \(15\ \mathrm{mg/kg/24\ h}\). For the 4.0 kg infant, the dose is \(60\) mg/24 h. The predicted AUC is \(0.80(60)/0.3247=147.8\ \mathrm{mg\cdot h/L}\). For the 7.0 kg infant, the dose is \(105\) mg/24 h, and predicted AUC is \(0.80(105)/1.107=75.9\ \mathrm{mg\cdot h/L}\). Under this model, the same mg/kg regimen gives notably different predicted AUCs.

This comparison does not prove that an age-adjusted regimen is clinically superior. It shows that, given this model and these assumptions, a single mg/kg amount does not attain the same target exposure in the two cases. Alternatives include a fixed dose, body-surface-area scaling, weight allometry without an explicit maturation term, or a more mechanistic physiological model. Each makes different assumptions and may be more or less appropriate depending on the drug, data, biological knowledge, and intended use. A more complex model is not automatically more reliable.

Assessing Uncertainty and Model Performance

A dose calculated from a typical clearance prediction is not a guarantee that each infant will attain the target. Parameter uncertainty, between-individual variability, measurement error, and uncertainty in bioavailability all contribute to prediction uncertainty. The maturation curve is especially influential when the available data do not cover the ages at which the recommendation will be used.

Useful diagnostics and checks include comparing observed and model-predicted concentrations or exposure summaries across age and weight; checking whether residual patterns vary systematically with age; and examining whether the model is being used within the range of ages and sizes represented in its evidence. For a population model, also examine whether estimated parameter uncertainty and between-individual variability are plausible and whether the data can distinguish age effects from size effects. Age and weight often change together, so a dataset with little overlap in weight across ages may provide weak information to separate their contributions.

A systematic underprediction or overprediction in the youngest infants, for example, could indicate an unsuitable maturation function, a missing covariate, incorrect assumptions about bioavailability, or data problems. It should prompt investigation rather than automatic adjustment of the dose. Extrapolation beyond the observed age range is particularly dependent on the assumed curve shape and plateau.

Common Pitfalls and Limitations

• Treating the maturation function as a dose rule. The function predicts one component of clearance; the dose follows only after combining it with size, the exposure target, bioavailability, and the PK assumptions.
• Confusing postmenstrual age with postnatal age. These age definitions are not interchangeable. The model must use the age variable on which its parameters were defined.
• Assuming an allometric exponent is universal. The value 0.75 here is illustrative. Its suitability depends on the drug, population, and model evidence.
• Ignoring exposure variability. A dose that matches a target for typical predicted clearance may not do so for every infant. A point prediction is not an individual guarantee.
• Reporting a calculated dose without considering administration. Practical doses may need rounding to available strengths or volumes. The resulting exposure should be recalculated, and the formulation and route must be appropriate.

The calculation also leaves out any age-related change in bioavailability, absorption, protein binding, or other processes that affect concentration or response. It assumes a common exposure target is appropriate across the ages being compared. Those assumptions may not hold. Exposure matching is a model-informed basis for comparing doses, not proof of equal efficacy or safety.