Precursor-Product Models From Biochemical Mass Balance

Why Link a Precursor to a Measured Product?

A measured biomarker may not be formed directly from its source. It may first arise as a precursor, then be converted into the product that is assayed. If drug action changes precursor formation or conversion, the product response can be delayed even when the product itself has simple loss kinetics. A mass-balance model makes this pathway explicit: material entering, leaving, or being converted in each pool determines the rate of change of that pool.

The key is to write a separate balance for each biochemical pool. Conversion removes material from the precursor and supplies material to the product; it is not an independent input to the product that can be added without accounting for precursor disappearance. The equations below describe an irreversible precursor-to-product pathway. They do not assume that the measured product is the only destination of precursor material.

Constructing the Amount Equations

For a one-way pathway, the conversion flux \(v_c\) is an exit from the precursor pool. The same conversion generates product at rate \(\nu v_c\). Other losses, such as degradation or elimination, are separate exits; they should not be counted as product formation unless the biochemical pathway actually produces the measured product.

Each term in an amount equation must have units of amount per time. For example, if amounts are in nmol and time is in hours, \(R_P\), \(v_c\), \(L_P\), \(R_M\), and \(L_M\) must be in nmol/h. The stoichiometric factor \(\nu\) converts the precursor conversion rate into the corresponding product formation rate. For a one-to-one molar conversion, \(\nu=1\); if one precursor molecule yields two product molecules, \(\nu=2\). If the model uses mass units rather than molar units, the conversion factor must account for the relevant molecular weights.

A common simple specification uses constant precursor formation, first-order conversion, and first-order losses. In that case \(v_c=k_cA_P\), \(L_P=k_PA_P\), and \(L_M=k_MA_M\), where each rate constant has units of inverse time. With no direct product input, the equations become:

These equations encode pathway direction: \(A_P\) supplies the conversion flux, and that flux supplies \(A_M\). The equations do not imply that conversion is the only precursor exit; \(k_PA_P\) represents other precursor loss. Nor do they imply that product loss returns material to the precursor pool.

From Amounts to Measured Concentrations

Biomarkers are often reported as concentrations, but the biochemical balance is naturally written for amounts. If the precursor and product occupy constant volumes \(V_P\) and \(V_M\), define \(C_P=A_P/V_P\) and \(C_M=A_M/V_M\). The volumes need not be equal. Dividing each amount equation by its own pool volume gives:

where the first-order model and zero direct product input are assumed. The factor \(V_P/V_M\) is essential: conversion is driven by precursor amount, but the measured product rate is a change in product concentration. Omitting this factor is correct only under a compatible equal-volume convention. These concentration equations also assume fixed pool volumes. If a volume changes over time, differentiating \(C=A/V\) adds a volume-change term.

Steady State and Nonlinear Fluxes

For constant inputs and first-order rates, a steady state is a pair of amounts for which both derivatives are zero. Setting the precursor equation to zero gives \(A_P^*=R_P/(k_c+k_P)\). Substituting this into the product equation gives the corresponding product amount:

provided \(k_c+k_P>0\) and \(k_M>0\). This result shows that the product baseline depends not only on product loss but also on how much precursor formation is diverted into conversion rather than other precursor loss. It is a consequence of the stated model, not a general formula for every biochemical pathway.

Mass balance does not require every rate to be first order. For example, precursor loss might be saturable, while conversion remains first order. The balance structure is unchanged; only the specified loss function changes. The following calculation illustrates that distinction without treating the particular loss law as a universal assumption.

Assumptions, Diagnostics, and Model Choice

The equations are interpretable only when their biological and measurement assumptions are made explicit. They assume that each amount represents a well-mixed pool, that the listed fluxes account for the relevant entries and exits, and that conversion proceeds from precursor to product in the stated direction. The first-order version further assumes the specified proportional rate laws. The concentration form assumes constant volumes, and the stoichiometric yield must be appropriate for the units and pathway being modeled. If the product also has a direct source, \(R_M\) belongs in its balance rather than being silently attributed to precursor conversion.

Several checks can reveal a construction error before fitting. Every term in an amount balance must have amount/time units; every term in a concentration balance must have concentration/time units. Conversion must decrease the precursor and increase the product, with the correct yield. At zero pool amount, the modeled derivative should not point toward a negative amount when the physical input and loss laws are nonnegative. Predicted concentrations should also be compared with the actual measurement definition: a plasma concentration may not represent the total amount in a tissue pool.

A linked-pool model is useful when precursor dynamics or pathway structure matter. A simpler lumped product-turnover model, such as \(dA_M/dt=R_M-k_MA_M\), can be a reasonable approximation when precursor behavior is not of interest and product formation can be treated as an effective input. But that equation cannot distinguish a changing upstream precursor from a changing conversion rate. Conversely, adding a precursor state costs parameters and may not be supported by data that observe only the product. A good mechanistic description is not automatically identifiable or estimable from a particular sampling design.

A frequent mistake is to add conversion as product formation while leaving it out of the precursor equation. That construction creates product from nowhere: it violates the stated pathway because the precursor is not consumed. The correction is to use the same conversion flux on both sides of the link, with a negative sign for precursor departure and a positive, stoichiometrically scaled sign for product appearance. Another common error is to multiply by a concentration when the rate constant was defined for amount, or to divide by the wrong pool volume. Writing units next to every flux is a practical defense against both errors.