Mapping Parent and Metabolite Species to PK-PD Compartments

Why the Species-to-Compartment Map Matters

A parent drug can form one or more metabolites, and a measured concentration is not automatically the amount in every modeled compartment. A PK-PD model therefore needs an explicit map: which chemical species are represented, where their amounts are tracked, how formation and elimination connect them, and which model quantity corresponds to each assay result. Ambiguity at this stage can create unit errors, double-counted losses, or a mismatch between a predicted quantity and the laboratory measurement.

The map describes model structure, not necessarily a literal anatomical picture. A “central” compartment may represent a well-mixed distribution space rather than a single physical location. A parent and its metabolite are distinct chemical species, even if both are measured in the same plasma sample. Conversely, one species may be represented in multiple modeled compartments.

Amounts, Concentrations, and Formation Links

An amount is an extensive quantity, commonly expressed in mass units such as mg or in molar units such as mmol. A concentration is amount per volume. For a well-mixed compartment with volume \(V_P\), the parent concentration is \(c_P=A_P/V_P\); the corresponding metabolite relation is \(c_M=A_M/V_M\). State the compartment and units whenever using a concentration. The volumes \(V_P\) and \(V_M\) need not be equal, even when both species are measured in plasma.

A simple linear parent–metabolite map can use first-order parent formation of metabolite \(M\), other parent elimination, and first-order metabolite elimination. In the equations below, \(k_f\), \(k_{e,P}\), and \(k_{e,M}\) have units of inverse time. \(k_f\) is the rate constant for parent disappearance through the specified metabolite-forming pathway; \(k_{e,P}\) is the rate constant for other parent loss; and \(k_{e,M}\) is the rate constant for metabolite loss. For now, take both amounts in mass units and define \(f_{M/P}\) as the mass of \(M\) formed per mass of \(P\) consumed on that pathway.

The input term has units of amount per time; for an oral dose, absorption or bioavailability may need its own representation. The parent’s loss through formation is \(k_fA_P\), while the amount of metabolite formed per time is \(f_{M/P}k_fA_P\). These rates are equal only if the mass yield is one. The yield is not automatically one because the metabolite may have a different molecular mass, and the pathway may not convert every parent molecule into the specified metabolite.

If formation is specified on a molar basis, define the amounts and yield on that basis instead. For a one-to-one molar conversion, one mole of parent consumed forms one mole of metabolite, but the resulting masses can differ. The yield basis must be consistent with the units of the state variables. For multiple pathways, each pathway’s parent loss and product formation should be represented separately, with the appropriate yield for each product.

Deriving the Balance Equations

The equations follow from amount accounting. Over a short interval \(\Delta t\), a first-order formation pathway consumes approximately \(k_fA_P\Delta t\) of parent, and other first-order elimination consumes approximately \(k_{e,P}A_P\Delta t\). Adding the parent input and dividing the net amount change by \(\Delta t\), then taking the limit as \(\Delta t\) approaches zero, gives the parent balance. The pathway forms \(f_{M/P}k_fA_P\Delta t\) of metabolite and eliminates approximately \(k_{e,M}A_M\Delta t\), yielding the metabolite balance.

This derivation assumes the pathway rates and yields are defined for the current model state, and that the first-order rate expressions are appropriate over the modeled range. A nonlinear or time-varying pathway requires a different rate expression, but the same accounting principle applies: identify each input, loss, and formation contribution once, with compatible units.

Yield Basis and Species Identity

A formation link must specify whether it transfers mass, moles, or a fraction of parent loss. The following example uses a one-to-one molar conversion, so a mass yield greater than one is possible without creating molecules or violating the stated stoichiometry.

Observed Analytes and Observation Mapping

The assay result is an observation of an analyte, while the model state is an amount or concentration in a specified compartment. A simple direct observation model might equate a measured plasma concentration to a predicted central-compartment concentration, possibly with measurement error. That equality is an assumption about sampling, assay specificity, and the compartment representation; it is not a universal identity.

An assay can report total concentration, unbound concentration, or a signal converted to concentration by calibration. The model quantity must match that definition. If a sample contains parent and metabolite, a parent-specific assay generally reports parent rather than the sum of both species; a nonspecific assay may instead combine signals. State which situation applies. Also check whether the model predicts plasma concentration while the assay uses whole blood, or whether the reported quantity is an amount per sample rather than a concentration.

Assumptions, Diagnostics, and Alternatives

A compartment map is interpretable only when its assumptions are explicit. The simple equations above assume well-mixed represented compartments, first-order formation and elimination, constant rate constants over the interval, and a fixed yield. They also assume that amounts are defined on a consistent basis and that every represented pathway is included once. For concentration predictions, the volume and the correspondence between model compartment and sampled matrix must be specified.

Useful checks are often structural before they are statistical. Confirm that all terms in each balance have units of amount per time; verify that parent formation loss and metabolite formation gain are connected by the declared yield; and confirm that an assay is matched to its analyte and measurement matrix. If a metabolite amount falls despite a stated formation rate greater than elimination, recheck the signs and arithmetic. If the parent equation includes the formation pathway in \(k_f\) and also includes it in a separate total-clearance term, inspect for double counting.

Common warning signs include a predicted concentration compared with a measured amount, a mass yield applied to molar state variables, a metabolite assay mapped to parent concentration, or an unexplained sum of parent and metabolite concentrations. Residual patterns can also indicate a poor structural map, but they do not by themselves establish which biological pathway is missing. The mapping should be checked against assay definitions and study design before adding complexity.

The main alternative to separate parent and metabolite states is a lumped model that represents a combined or summary quantity. A lumped model can be useful when the data cannot distinguish species or when the scientific question concerns only an aggregate measurement. It cannot, without additional assumptions, provide species-specific amounts or formation rates. An explicit species-and-compartment model makes those quantities available but requires more structural and measurement information. A further alternative is to track each species in multiple distribution compartments; this can represent distribution more fully, but it increases the number of states and connections that must be supported by data.

These choices have limits. A compartment is a mathematical representation, not proof of a physical space. A fixed yield may be inappropriate when pathways vary with concentration or time. An apparent volume should not be confused with an anatomical volume. Finally, if the observation mapping is uncertain, a precise model prediction can still be the wrong quantity to compare with the assay. The map is therefore part of the model specification, not merely a drawing convention.