Testing Extended Clearance Models Against In Vivo Evidence

Why Test the Pathways, Not Just the Total Clearance?

A model can reproduce plasma concentrations or total hepatic clearance while assigning drug to the wrong routes. For example, an overestimated uptake process might be offset by an overestimated return of drug from hepatocytes to blood. Agreement in one aggregate measure would not establish that the pathway assumptions are sound. Testing an extended clearance model therefore means comparing its predictions with several kinds of in vivo evidence: hepatic extraction, biliary recovery, and unbound intracellular exposure.

The central question is not whether one calculated number exactly matches one measurement. It is whether the model, with stated assumptions and uncertainty, can account for the observed evidence across pathways. A discrepancy can indicate an incorrect parameter, an unreliable measurement, or a structural omission such as an unrepresented elimination route. These possibilities need to be distinguished before changing the model.

Model and Assumptions

Consider a linear hepatic model with uptake from blood into hepatocytes and three competing intracellular fates: return to blood, biliary efflux, and metabolism. Define \(CL_{\mathrm{uptake}}\) as the uptake clearance, \(CL_{\mathrm{return}}\) as the clearance for return to blood, \(CL_{\mathrm{bile}}\) as biliary efflux clearance, and \(CL_{\mathrm{met}}\) as metabolic clearance. These clearances act on the relevant unbound concentrations. Let \(C_{\mathrm{u,b}}\) be unbound drug concentration in blood and \(C_{\mathrm{u,h}}\) unbound intracellular concentration in hepatocytes.

For the basic calculations below, assume hepatic blood is well mixed, the liver is at steady state for the clearance calculation, and the intracellular pathways are linear and time-invariant. Assume the measured clearances apply to the same population and conditions as the in vivo evidence. The well-stirred liver relation further assumes that drug leaving the liver is represented by a single hepatic outflow concentration. These are testable simplifications, not universal properties of hepatic disposition.

Let \(Q_{\mathrm{h}}\) be hepatic blood flow and \(f_{\mathrm{u,b}}\) the unbound fraction in blood. All clearances and concentrations must use compatible blood or plasma conventions; the equations here use blood flow and blood-referenced concentrations. If the underlying data use plasma concentrations, the conversion between blood and plasma must be handled consistently.

From Intracellular Pathways to Hepatic Extraction

At steady state, the rate entering the hepatocyte equals the total rate leaving it. Uptake is \(CL_{\mathrm{uptake}}C_{\mathrm{u,b}}\), while total intracellular loss is \((CL_{\mathrm{return}}+CL_{\mathrm{bile}}+CL_{\mathrm{met}})C_{\mathrm{u,h}}\). Equating these rates gives the predicted intracellular concentration:

Only biliary efflux and metabolism are irreversible removal from the body in this simplified liver model. Return to blood is not counted as hepatic elimination: the returned drug can circulate and reach the liver again. Thus the effective hepatic intrinsic clearance is uptake multiplied by the fraction of intracellular loss that follows either elimination route.

Under the well-stirred liver assumption, hepatic clearance and extraction ratio are:

This formulation makes the model’s structure visible. Increasing uptake can raise effective intrinsic clearance, but that increase is reduced when a larger share of intracellular drug returns to blood. The extraction ratio is bounded between zero and one under positive finite clearances and flow.

Testing Biliary Recovery and Pathway Fractions

Under the linear competing-pathway assumption, the fraction of intracellular drug leaving by a particular route equals that route’s clearance divided by the sum of all intracellular exit clearances. For a known amount of drug entering hepatocytes, this gives the expected partition among blood return, bile, and metabolism. This comparison requires care: biliary recovery as a fraction of an administered dose is not automatically the same as biliary fraction of drug taken up by the liver. The administered dose may include drug that never entered hepatocytes, and recovery may be incomplete.

Testing Intracellular Exposure

Intracellular exposure tests a different implication of the model. At steady state, the ratio of unbound intracellular to unbound blood concentration is determined by uptake relative to the sum of intracellular exit clearances. A high predicted intracellular concentration can coexist with modest hepatic extraction if uptake is high but return to blood is also substantial. Therefore, total clearance alone cannot determine intracellular exposure.

How to Judge Agreement and Locate Failure

A strong evaluation checks multiple predictions and their conditions rather than applying a single pass/fail threshold. Compare model predictions with observations on the same scale and at compatible times, and carry uncertainty in measured quantities and parameters into the comparison. If the evidence is imprecise, a numerical difference may be uninformative; if uncertainty is small and predictions repeatedly miss in distinct measures, structural assumptions deserve scrutiny.

Several patterns are especially informative. Correct extraction but excessive predicted intracellular concentration may point to an incorrect uptake-to-exit balance. Correct intracellular exposure but too much predicted biliary recovery may implicate the bile-versus-metabolism partition. Extraction and biliary recovery that are both underpredicted could indicate an omitted elimination route, under-scaled uptake, or incompatible blood-flow assumptions. These patterns guide diagnosis but do not uniquely identify a parameter; identifiability must be considered.

Alternatives, Diagnostics, and Limitations

The extended model can be compared with a simpler well-stirred model using a single effective intrinsic clearance. The simpler model may be adequate for predicting hepatic clearance, but it cannot independently predict intracellular exposure or the split between bile and metabolism. Conversely, a compartmental hepatic PBPK model can represent changing concentrations, zonation, and additional compartments, but requires more parameters and more informative data. A detailed model is not automatically better: it is useful only when the evidence can constrain its added structure.

Look for practical warning signs: predicted extraction outside the physically possible range; route fractions that do not sum to one; a mismatch between calculated clearance units or concentration bases; sensitivity to a single poorly measured parameter; and disagreement that disappears only after inconsistent definitions of recovery are used. The steady-state formulas are not appropriate when pathways are saturable over the tested range, when concentrations change too rapidly for the assumed balance, or when transport direction and intracellular binding are misrepresented. Time-course modeling or nonlinear pathway descriptions may then be needed.

A common mistake is to treat agreement in hepatic extraction as proof of the transporter mechanism. Extraction is an aggregate outcome: different uptake and intracellular loss combinations can yield the same effective clearance. The correction is to test additional, mechanistically distinct observations—especially pathway-specific recovery and unbound intracellular exposure—and to ask whether the measurements can actually distinguish competing explanations.