Pooling Organs With Similar Kinetics

Why Pooling Is a Kinetic Choice

A whole-body model does not need a separate state for every organ to make useful predictions. If several tissues have sufficiently similar blood supply and drug partitioning, grouping them can reduce the number of states while preserving a prediction of interest. The important question is not simply whether the organs have similar names or functions. It is whether their drug concentrations change in a sufficiently similar way under the model assumptions and inputs.

The earlier material on exact lumping established that aggregation is exact only when the grouped system has a closed description in terms of the pooled state. Here, the focus is how perfusion and partitioning determine whether tissues can share a kinetic trajectory, how to calculate the pooled quantities, and what information is lost when organ states are combined.

This compact model assumes that each tissue is adequately represented by one well-mixed concentration, that tissue and venous concentrations are related by the stated partition coefficient, and that blood entering each tissue has the same arterial concentration. It does not represent diffusion gradients within an organ, separate cell types, or organ-specific metabolism. Those omissions matter if the prediction requires such detail.

Substituting \(C_{t,i}=K_iC_{v,i}\) into the tissue equation gives \(dC_{v,i}/dt=\lambda_i(C_a-C_{v,i})\). Thus, under the shared arterial input, the rate \(\lambda_i\) governs how quickly the venous concentration approaches the arterial concentration. Tissues with similar \(K_i\) but very different \(\lambda_i\) need not have similar time courses; similarly, matching rates alone does not make their tissue concentrations equal if their partition coefficients differ.

Conditions for a Shared Trajectory

Suppose two or more tissues start with the same venous concentration and experience the same arterial input. If their \(\lambda_i\) values are equal, their venous concentrations remain identical. This follows because each concentration solves the same differential equation with the same input and initial value. Their tissue concentrations can nevertheless differ when their \(K_i\) values differ, since \(C_{t,i}=K_iC_{v,i}\).

For a shared trajectory that is not stationary, equal rates are also necessary whenever \(C_a-C_v\ne 0\). If all venous concentrations equal \(C_v\), then each derivative must be the same, but \(dC_{v,i}/dt=\lambda_i(C_a-C_v)\). When \(C_a-C_v\ne 0\), this equality requires the \(\lambda_i\) values to match.

There is an important equilibrium exception. If \(C_a\) is constant and every tissue starts at \(C_{v,i}=C_a\), then every derivative is zero, so all tissues remain at \(C_a\) even if their rates differ. Equal rates are not necessary for this stationary shared state. In practice, pooling should be justified for the changing conditions and initial states relevant to the prediction, not inferred from a single equilibrium condition.

Why the Pooled Quantities Have These Forms

The pooled tissue concentration is defined from total tissue amount divided by total tissue volume. Under the partition relation, the total amount is \(\sum_i V_iC_{t,i}=\sum_i V_iK_iC_{v,i}\). If all grouped tissues share \(C_v\), this becomes \(C_v\sum_i V_iK_i\). Dividing by \(V_{\mathrm{pool}}\) gives \(C_{t,\mathrm{pool}}=K_{\mathrm{pool}}C_v\), with the volume-weighted coefficient shown above.

The blood-flow-weighted venous concentration leaving the pooled tissues is \(\sum_i Q_iC_{v,i}/Q_{\mathrm{pool}}\). When each \(C_{v,i}=C_v\), this equals \(C_v\). The summed amount-change rate is likewise \(\sum_i Q_i(C_a-C_{v,i})=Q_{\mathrm{pool}}(C_a-C_v)\). Together, these relations show why the aggregate can be represented by one perfused tissue state when a shared venous trajectory is maintained.

If every grouped tissue has the same rate \(\lambda\), then \(Q_i=\lambda V_iK_i\) for each tissue, so the pooled rate is also \(\lambda\). This confirms that the shared trajectory is consistent with the pooled equation. The argument depends on the shared venous state; if tissue venous concentrations diverge, the summed outflow generally cannot be recovered from the pooled tissue amount alone.

Different Partition Coefficients Can Still Be Pooled

A shared venous trajectory does not require identical partition coefficients. When \(K_i\) values differ, however, the grouped tissue concentration is not simply the common venous concentration multiplied by an arbitrarily selected organ’s \(K_i\). It uses the volume-weighted \(K_{\mathrm{pool}}\). This distinction matters because a pooled tissue concentration summarizes total amount per volume, while venous concentration describes blood leaving the tissues.

When the Rates Differ

If rates differ, an initially shared venous concentration will generally separate once the system is away from equilibrium. A single pooled state then may fail to determine the pooled venous output: distinct combinations of organ concentrations can have the same total tissue amount but different flow-weighted venous concentrations. That is a failure of closure, not an arithmetic error in the pooled volume or amount.

Practical Checks, Alternatives, and Limitations

Before pooling tissues for a changing exposure, check whether their expected rate constants are close, whether their initial states are compatible with a shared trajectory, and whether the common arterial-input assumption is reasonable. The earlier material on organ blood-flow allocation is relevant here: a pooled flow should be the sum of flows for the same physiological scenario, not an average of flows or a value that silently changes total cardiac output.

Useful diagnostics include comparing \(Q_i/(V_iK_i)\) across candidate tissues, plotting organ-specific predictions from a more detailed model over the time range of interest, and comparing pooled venous output as well as total tissue amount. Large differences in rates, substantial separation in simulated organ concentrations, or divergent predictions for the target output are warnings against pooling. Matching one scalar rate does not prove that all assumptions are appropriate, especially if arterial inputs or partition behavior differ.

Pooling is most useful when the prediction target concerns aggregate tissue amount, a shared outlet concentration, or a whole-body profile for which organ-specific detail is not needed. An organ-resolved model is preferable when the question concerns a particular organ’s exposure, toxicity, or concentration-time course. A lumped model with separate groups may be a practical intermediate choice when some tissues are compatible but others are not. These alternatives trade simplicity for organ-level resolution; none is automatically best for every prediction.

The main limitation is that a pooled state cannot identify its component states. Even with a correctly calculated \(C_{t,\mathrm{pool}}\), one cannot infer each organ’s concentration from that average without additional information. Near-equal rates may justify an approximation for a specified prediction, but they do not make the model exactly lumped. The acceptable approximation should be judged against the intended use and the consequences of prediction error.