Why Microsomal IVIVE Can Underpredict In Vivo Clearance

Why a Microsomal Prediction Can Be Too Low

Microsomal in vitro to in vivo extrapolation (IVIVE) converts a measured rate of drug loss in a liver microsomal incubation into an estimate of hepatic clearance. The calculation can be internally correct and still underpredict clearance in vivo: an important enzyme-mediated contribution may not be represented by the assay, the preparation may have lower activity than the relevant in vivo enzyme system, or a correction or scaling input may be inappropriate. The important task is to locate a plausible source of the gap, rather than to multiply the prediction by an arbitrary factor until it matches an observed value.

The introductory material on the IVIVE chain and the derivation of the microsomal clearance scaling equation establish the general calculation. Here the focus is diagnosing underprediction: how each candidate source enters the chain, how to quantify its effect, and what the calculation does and does not establish. Comparisons should be made on the same clearance basis. A microsomal estimate of hepatic clearance should not be compared uncritically with total systemic clearance if renal or other extrahepatic elimination contributes to the latter.

The Calculation and Its Assumptions

Let \(CL_{int,app}\) be apparent intrinsic clearance measured in a microsomal incubation, \(f_{u,inc}\) the unbound fraction in that incubation, MPPGL the microsomal protein per gram of liver, and \(W_{liver}\) liver mass. In the usual binding-corrected scaling calculation, the estimated unbound intrinsic clearance per unit microsomal protein is \(CL_{int,app}/f_{u,inc}\). Scaling it to the whole liver gives:

If \(CL_{int,app}\) is in mL/min/mg microsomal protein, MPPGL in mg microsomal protein/g liver, and \(W_{liver}\) in g, the result is in mL/min. The hepatic well-stirred model then gives:

Here \(CL_h\) is hepatic clearance, \(Q_h\) is hepatic blood flow, and \(f_{u,p}\) is the unbound fraction in plasma. This equation assumes the well-stirred liver representation and compatible blood/plasma conventions for its inputs. The scaling equation assumes that the measured clearance is an appropriate estimate of the microsomal metabolic activity being scaled; that the binding correction is applicable; and that MPPGL and liver mass represent the target population and species. The assay should also measure metabolic loss under conditions in which a first-order clearance estimate is appropriate. If these assumptions fail, numerical precision in the multiplication does not make the prediction reliable.

Underprediction can enter at different points. An unmeasured pathway or enzyme can make the measured intrinsic clearance incomplete. Low activity of the relevant enzyme in the microsomal preparation can depress the assay result. Overestimating \(f_{u,inc}\) makes the binding correction too small. Choosing too small an MPPGL or liver mass also lowers the scaled intrinsic clearance. These effects can coexist, but evidence is needed to distinguish them.

How the Sources Change the Prediction

The direction of the binding effect follows directly from the correction. Since \(0<f_{u,inc}\leq 1\), dividing apparent clearance by \(f_{u,inc}\) increases or leaves unchanged the estimated unbound intrinsic clearance. If incubation binding is ignored, the calculation uses \(CL_{int,app}\) in place of \(CL_{int,app}/f_{u,inc}\), which can understate intrinsic clearance when binding is appreciable. A measured fraction should be used under the relevant incubation conditions; a generic value or a value measured in a different matrix is not a substitute.

A pathway may be missed if the assay substrate, analytical endpoint, or microsomal preparation does not capture all relevant metabolic turnover. If the missing activity contributes to hepatic elimination, its omission lowers the intrinsic clearance presented to the well-stirred model. This possibility does not justify simply adding a guessed pathway rate. One needs evidence that the activity exists in the target system and an appropriate way to quantify it.

Low enzyme activity is distinct from a pathway being entirely absent. The enzyme may be represented in the microsomes but have lower abundance or activity than is appropriate for the intended in vivo prediction. Where suitable enzyme-specific or inter-system activity information is available, an established relative activity or inter-system extrapolation factor may be applicable. Such a factor is an evidence-based adjustment, not a universal correction for every microsomal assay.

Scaling factors have a direct multiplicative effect on \(CL_{int,scaled}\). If MPPGL or liver mass is too small, the scaled intrinsic clearance is too small; if too large, it is too large. The effect on hepatic clearance is not necessarily proportional, because the well-stirred equation is nonlinear. A change in a scaling factor should therefore be propagated through the hepatic model rather than reported as an equal percentage change in \(CL_h\).

A Short Justification: Why More Intrinsic Clearance Raises Hepatic Clearance

The well-stirred equation can be written as a function of scaled intrinsic clearance \(C\), with \(Q_h\) and \(f_{u,p}\) fixed. Differentiating with respect to \(C\) gives:

Thus, within this model and for positive flow and unbound fraction, increasing scaled intrinsic clearance increases predicted hepatic clearance. The derivative also explains why the response becomes less steep as intrinsic clearance grows: the denominator increases, so the same increment in intrinsic clearance produces a smaller clearance change. This establishes the direction of an adjustment, not whether the adjustment is justified by biological evidence.

Diagnosis, Alternatives, and Limitations

A useful diagnosis follows the calculation chain from the assay result to the clearance being compared. Check that the units cancel correctly, that the binding fraction corresponds to the incubation conditions, and that the MPPGL and liver mass match the species and scaling target. Then ask whether the microsomal assay represents the relevant metabolic activity and whether the in vivo comparator is specifically hepatic clearance rather than total systemic clearance. These checks help identify where an underestimate could arise; none alone proves a mechanism.

A missed pathway is plausible when independent evidence indicates an additional contribution not captured by the assay. Low enzyme activity is plausible when enzyme-specific activity information supports an applicable correction. A scaling-factor explanation is plausible when a more appropriate target-specific input is available and the original input was mismatched. If no such evidence is available, report the prediction and its assumptions rather than inventing an adjustment. In particular, do not attribute every gap to an omitted pathway: mismatched clearance definitions, binding inputs, or scaling assumptions can produce the same direction of discrepancy.

Microsomes are not the only in vitro system. Hepatocytes can represent a broader set of cellular processes, while S9 fractions and recombinant enzymes answer different questions about metabolic capacity and enzyme-specific activity. These alternatives may help investigate a suspected gap, but they are not interchangeable measurements: each has its own scaling requirements and assumptions. Choosing another system does not automatically resolve a mismatch unless the relevant activity is measured and the resulting estimate is compared on a consistent basis.

Even a carefully corrected microsomal estimate remains conditional on the assay, binding measurements, scaling factors, and hepatic model. The well-stirred model translates intrinsic clearance into hepatic clearance under its assumptions; it does not identify which enzyme caused a discrepancy. A numerical fit to an in vivo value is not independent validation if the inputs or adjustment factors were selected to produce that fit.

Common Pitfall