Calculating Intrinsic Clearance from Substrate Depletion

From Substrate Disappearance to Clearance

In a substrate-depletion assay, the concentration of a parent compound is measured at several times after incubation with a metabolic preparation. The decline contains information about how rapidly the preparation removes substrate. To use that information for IVIVE, the time-course rate must be expressed as a clearance per amount of preparation, rather than only as a rate constant for the particular incubation.

The calculation rests on a specific model: over the time interval used, substrate disappearance is first order in substrate concentration, and the effective metabolic activity is constant. Under that model, the natural logarithm of concentration decreases linearly with time. The negative of the fitted slope is the depletion rate constant; multiplying it by incubation volume per amount of protein converts it to an apparent intrinsic clearance.

Definition: The apparent in vitro intrinsic clearance, \(CL_{\mathrm{int,app}}\), is the substrate-disappearance clearance normalized to the amount of in vitro preparation. For a microsomal assay with total microsomal protein amount \(m_{\mathrm{prot}}\), incubation volume \(V\), and first-order depletion rate constant \(k_{\mathrm{dep}}\), \(CL_{\mathrm{int,app}}=k_{\mathrm{dep}}V/m_{\mathrm{prot}}\). Its units are volume per time per mass of protein, such as mL/min/mg protein. “Apparent” identifies the assay-based estimate; it does not by itself apply a binding correction or establish an in vivo clearance.

Deriving the Log-Linear Relationship

Let \(C(t)\) be the measured substrate concentration at time \(t\). If the assay follows first-order disappearance, then the rate of concentration decline is proportional to the concentration:

\frac{dC(t)}{dt}=-k_{\mathrm{dep}}C(t), \qquad k_{\mathrm{dep}}>0.

Separate variables and integrate from time zero, when the concentration is \(C_0\), to time \(t\):

\int_{C_0}^{C(t)}\frac{dC}{C} = -k_{\mathrm{dep}}\int_0^t du \quad\Longrightarrow\quad \ln C(t)-\ln C_0=-k_{\mathrm{dep}}t.

Thus \(\ln C(t)=\ln C_0-k_{\mathrm{dep}}t\). If the log concentrations are fitted by the linear model \(y_i=a+bt_i+\epsilon_i\), where \(y_i=\ln C_i\), then the estimated depletion rate is \(k_{\mathrm{dep}}=-b\). The intercept estimates the log concentration at time zero when zero is included in the fitted time range; it need not equal the observed initial log concentration exactly because measurements have error.

For ordinary least squares, the fitted slope is

\hat b= \frac{\sum_i(t_i-\bar t)(y_i-\bar y)} {\sum_i(t_i-\bar t)^2}, \qquad \hat k_{\mathrm{dep}}=-\hat b.

The clearance conversion follows from the same rate model. If \(m_{\mathrm{prot}}\) is the protein amount in volume \(V\), then its concentration is \(m_{\mathrm{prot}}/V\). A clearance \(CL_{\mathrm{int,app}}\) per mass of protein produces the first-order rate \(CL_{\mathrm{int,app}}m_{\mathrm{prot}}/V\). Equating that rate to \(k_{\mathrm{dep}}\) gives:

\hat{CL}_{\mathrm{int,app}} = \hat k_{\mathrm{dep}}\frac{V}{m_{\mathrm{prot}}}.

The units verify the conversion: \((\mathrm{min}^{-1})(\mathrm{mL})/(\mathrm{mg})=\mathrm{mL/min/mg}\). Equivalently, if protein concentration \(P=m_{\mathrm{prot}}/V\) is known in mg/mL, then \(CL_{\mathrm{int,app}}=\hat k_{\mathrm{dep}}/P\).

Worked Calculation by Log-Linear Regression

Worked Example: Fitting a Microsomal Depletion Time Course

Suppose a hypothetical incubation contains 0.500 mg microsomal protein in a volume of 2.00 mL. Measured substrate concentrations are 8.000, 6.550, 5.363, and 4.390 µM at 0, 15, 30, and 45 minutes. Assume, for this calculation, that the four measurements are within a first-order depletion range and that the assay conditions remain constant.

Take natural logarithms of the displayed concentrations. To four decimal places, the values are 2.0794, 1.8795, 1.6795, and 1.4793. Fit these rounded log values against time. Their means are \(\bar t=22.5\) min and \(\bar y=1.779425\). The slope numerator is

\sum_i(t_i-\bar t)(y_i-\bar y) = (-22.5)(0.299975)+(-7.5)(0.100075) +(7.5)(-0.099925)+(22.5)(-0.300125) =-15.00225.

The slope denominator is \(\sum_i(t_i-\bar t)^2=1125\ \mathrm{min}^2\), so

\hat b=\frac{-15.00225}{1125} =-0.0133353\ \mathrm{min}^{-1}, \qquad \hat k_{\mathrm{dep}}=0.0133353\ \mathrm{min}^{-1}.

There are \(V/m_{\mathrm{prot}}=2.00/0.500=4.00\ \mathrm{mL/mg}\). Therefore,

\hat{CL}_{\mathrm{int,app}} = 0.0133353\ \mathrm{min}^{-1} \times 4.00\ \mathrm{mL/mg} =0.0533412\ \mathrm{mL/min/mg\ protein} \approx 0.05334\ \mathrm{mL/min/mg\ protein}.

This is the apparent clearance normalized to microsomal protein for the assumed assay conditions. The fit estimates the slope from all four time points; it is not simply the ratio of the initial and final concentrations.

Endpoint Estimates and Their Trade-Offs

With only an initial and a final measurement, the first-order equation gives \(k_{\mathrm{dep}}=\ln(C_0/C_t)/t\). This is a useful calculation when an endpoint estimate is intended and the first-order assumption is credible. With multiple measurements, regression uses the whole time course and allows the pattern to be inspected. A single endpoint cannot reveal curvature or identify a time point that departs from the trend.

Worked Example: Converting an Endpoint Estimate

In a hypothetical assay, substrate concentration falls from 10.0 µM at zero time to 3.70 µM at 40.0 min. The incubation contains 0.250 mg protein in 1.00 mL. If first-order depletion is assumed over the full interval, the endpoint estimate is

\hat k_{\mathrm{dep}} = \frac{\ln(C_0/C_t)}{t} = \frac{\ln(10.0/3.70)}{40.0\ \mathrm{min}} = \frac{0.994252}{40.0\ \mathrm{min}} =0.0248563\ \mathrm{min}^{-1}.

The volume-to-protein ratio is \(1.00/0.250=4.00\ \mathrm{mL/mg}\). Hence

\hat{CL}_{\mathrm{int,app}} = 0.0248563\ \mathrm{min}^{-1}\times4.00\ \mathrm{mL/mg} =0.0994252\ \mathrm{mL/min/mg\ protein} \approx0.09943\ \mathrm{mL/min/mg\ protein}.

The arithmetic is conditional on the full 40-minute interval behaving as one first-order phase. Because only two measurements enter, this calculation provides no internal check of that condition.

Approach	What is calculated	Main strength	Main limitation
Two-point endpoint	Log concentration ratio divided by elapsed time	Simple when only two suitable time points are available	Cannot diagnose curvature or point-specific departures
Log-linear regression	Slope of log concentration against time	Uses multiple measurements and makes the fitted trend inspectable	Depends on a defensible fitted range and an appropriate error pattern
Nonlinear fitting of the exponential model	Parameters of \(C(t)=C_0e^{-k_{\mathrm{dep}}t}\) on the concentration scale	Can represent the original measurement scale directly	Requires an explicit error model; the result can differ from log-scale fitting when errors are not multiplicative

Checking the Fitted Range

The log-linear model is useful only when the data support it over the selected time range. Plot or inspect log concentration against time, check the residuals around the fitted line, and compare early and late portions of the course. A systematic bend or a run of residuals with the same sign suggests that one constant depletion rate may not describe the whole interval. An isolated departure can also reflect measurement variability, so a point should not be removed solely because it changes the estimate.

Worked Example: Detecting a Change in Apparent Rate

Consider a hypothetical time course with concentrations 10.0, 7.41, 5.49, 4.07, and 3.70 µM at 0, 10, 20, 30, and 40 min. The early segment from 0 to 30 min gives

k_{\mathrm{early}} = \frac{\ln(10.0/4.07)}{30\ \mathrm{min}} = \frac{0.898942}{30\ \mathrm{min}} \approx0.0299647\ \mathrm{min}^{-1}.

For the final 10-minute interval, the rate calculated from its endpoints is

k_{\mathrm{late}} = \frac{\ln(4.07/3.70)}{10\ \mathrm{min}} = \frac{0.0953}{10\ \mathrm{min}} \approx0.00953\ \mathrm{min}^{-1}.

These segment estimates differ substantially. A single rate calculated from the full 40-minute endpoints would be \(\ln(10.0/3.70)/40=0.02486\ \mathrm{min}^{-1}\), which conceals that change. The appropriate conclusion is not automatically that the early estimate is the true clearance: first investigate whether the selected range and assay measurements support a constant first-order rate. If no range is defensible, a single \(CL_{\mathrm{int,app}}\) should not be presented as a well-supported estimate.

Other practical checks include confirming that time is expressed in the unit used for the reported rate, that concentrations are positive before taking logarithms, and that the protein amount and incubation volume refer to the same incubation. A factor-of-1000 unit mismatch or use of protein concentration where protein amount is required can change the reported clearance even when the fitted slope is correct. Replicate incubations can help assess reproducibility; the individual time courses should be examined rather than relying only on a pooled average that may hide different patterns.

Interpretation, Limitations, and Common Mistakes

The depletion slope is a rate constant for the particular in vitro conditions. The volume-to-protein conversion expresses that rate per amount of preparation. In later IVIVE calculations, this normalized estimate may be combined with appropriate scaling information, but the estimate itself is not a whole-liver clearance or a human systemic clearance. It is also not automatically an unbound intrinsic clearance.

Assuming a straight log plot without checking it. A line is justified by the first-order model, not guaranteed by the assay. Inspect the observed log concentrations and residual pattern.
Using the slope with the wrong sign. Concentration decline gives a negative fitted slope; the positive depletion rate is its negative, \(\hat k_{\mathrm{dep}}=-\hat b\).
Using protein concentration as though it were protein amount. The conversion \(V/m_{\mathrm{prot}}\) uses the total amount in the incubation. If protein concentration is used instead, divide \(k_{\mathrm{dep}}\) by that concentration.
Calling an endpoint rate a regression estimate. The endpoint formula and a multi-point fitted slope are different calculations. State which was used and do not imply that two points diagnose linearity.
Reporting excessive precision. The fitted clearance is conditional on measured concentrations, time points, and assay conditions. Report a sensible number of significant figures and retain units.

The principal limitation is that a single first-order estimate summarizes only the chosen incubation and fitted time window. If the log-linear pattern is poor, the analyst should not force a clearance calculation merely because the formula is available. The data may require a justified alternative analysis or further assay evaluation.

Check Your Understanding

Answer each question, then compare your reasoning with the solutions.

A fitted log-concentration slope is \(-0.0180\ \mathrm{min}^{-1}\). What is the depletion rate constant?
An incubation has \(k_{\mathrm{dep}}=0.0200\ \mathrm{min}^{-1}\), volume 1.50 mL, and 0.300 mg protein. Calculate \(CL_{\mathrm{int,app}}\).
Why can a two-point endpoint estimate not establish that depletion was first order throughout the interval?
What does a curved pattern in log concentration versus time imply for reporting one depletion rate?
If a protein concentration is 0.200 mg/mL, show how to calculate apparent intrinsic clearance from \(k_{\mathrm{dep}}=0.0160\ \mathrm{min}^{-1}\).

Solutions. (1) \(\hat k_{\mathrm{dep}}=-\hat b=-(-0.0180)=0.0180\ \mathrm{min}^{-1}\). (2) \(V/m_{\mathrm{prot}}=1.50/0.300=5.00\ \mathrm{mL/mg}\), so \(CL_{\mathrm{int,app}}=0.0200\times5.00=0.100\ \mathrm{mL/min/mg\ protein}\). (3) Two endpoints determine one average log decline, but provide no observations between them with which to detect curvature or a changing rate. (4) Curvature is evidence against one constant first-order rate across the full range; assess the data and fitted interval before reporting a single estimate. (5) \(CL_{\mathrm{int,app}}=k_{\mathrm{dep}}/P=0.0160\ \mathrm{min}^{-1}/(0.200\ \mathrm{mg/mL})=0.0800\ \mathrm{mL/min/mg\ protein}\).