Why the Assay Pattern Matters
An inhibitor can reduce CYP-mediated metabolism without producing the same change at every substrate concentration. For a competitive inhibitor, sufficiently high substrate concentration can overcome much of the inhibition. For other reversible inhibition forms, increasing substrate does not restore the uninhibited maximum rate in the same way. Choosing the model from that pattern matters: a victim–perpetrator model that uses the wrong inhibition form can misrepresent the victim’s metabolic rate as substrate and inhibitor concentrations change.
The earlier material on parameterizing reversible CYP inhibition addresses potency estimates from in vitro data. Here the question is different: which kinetic form do the substrate-dependent assay observations support? The key evidence is how the apparent \(K_m\) and \(V_{\max}\) change across inhibitor conditions, not whether one measured rate is lower than another.
Assumptions and Diagnostic Patterns
This equation is used under conditions where the assay measures initial rates: product formation is small enough that substrate depletion and product effects are negligible, and the enzyme–substrate reaction is adequately represented by a single Michaelis–Menten pathway. The inhibitor is reversible, its concentration is effectively constant during the measurement, and the assay conditions do not introduce substantial time-dependent enzyme loss. The simple factors \(\alpha\) and \(\alpha'\) also assume that the relevant free inhibitor concentration is known or adequately represented by the assay concentration. When these conditions fail, a pattern that looks like a different inhibition form may instead reflect an assay or model misspecification.
At a fixed inhibitor concentration, the equation can be written in Michaelis–Menten form using apparent parameters:
These apparent parameters summarize the rate curve at that inhibitor condition. Their changes across conditions provide the familiar kinetic patterns. Competitive inhibition is the special case \(\alpha'=1\), corresponding to no inhibitor binding to the enzyme–substrate complex. In that case \(V_{\max,\mathrm{app}}\) is unchanged while \(K_{m,\mathrm{app}}\) increases. Pure noncompetitive inhibition has \(\alpha=\alpha'>1\): \(V_{\max,\mathrm{app}}\) decreases while \(K_{m,\mathrm{app}}\) stays unchanged. Mixed inhibition has \(\alpha\ne\alpha'\): both apparent parameters generally change. If \(\alpha>\alpha'\), apparent \(K_m\) increases; if \(\alpha<\alpha'\), it decreases. Thus, a decreasing apparent \(K_m\) does not by itself rule out inhibition.
The word “noncompetitive” is sometimes used loosely for inhibition that lowers activity. In this tutorial, pure noncompetitive means equal inhibitor interaction with free enzyme and enzyme–substrate complex, represented by \(\alpha=\alpha'\). A mixed model with unequal factors is not pure noncompetitive, even if its \(V_{\max,\mathrm{app}}\) decreases.
A Short Derivation of the Assay Pattern
Taking the reciprocal of the rate equation shows why reciprocal plots have different intercept and slope patterns:
The slope in a plot of \(1/v\) against \(1/S\) is \(\alpha K_m/V_{\max}\), and the vertical intercept is \(\alpha'/V_{\max}\). These are not independent clues: together they imply the apparent parameters above. For competitive inhibition, the intercept is unchanged and the slope increases. For pure noncompetitive inhibition, slope and intercept increase in the same proportion, leaving their ratio—and therefore apparent \(K_m\)—unchanged. For mixed inhibition, the slope and intercept need not change proportionally.
This derivation explains the patterns; it does not make reciprocal plotting the preferred estimation method. Inverting noisy rates changes the error structure and gives disproportionate influence to low-rate observations. A global nonlinear fit to the untransformed rates is generally more appropriate for model comparison.
Worked Example: A Competitive Pattern
Suppose a hypothetical CYP assay has \(V_{\max}=100\) pmol/min/mg and \(K_m=10\) µM without inhibitor. At \(I=5\) µM, the data are consistent with \(K_i=5\) µM and no binding to the enzyme–substrate complex. Then \(\alpha=1+5/5=2\) and \(\alpha'=1\). Calculate rates at substrate concentrations of 5, 10, and 20 µM.
The model is \(v=100S/(2(10)+S)=100S/(20+S)\). At \(S=5\) µM, \(v=100(5)/(25)=20.00\) pmol/min/mg. At \(S=10\) µM, \(v=1000/30=33.33\) pmol/min/mg. At \(S=20\) µM, \(v=2000/40=50.00\) pmol/min/mg.
The apparent parameters are \(V_{\max,\mathrm{app}}=100/1=100\) pmol/min/mg and \(K_{m,\mathrm{app}}=2(10)/1=20\) µM. The lower rates at the tested substrate concentrations coexist with an unchanged maximum rate and an increased apparent \(K_m\), the competitive pattern. The data should include sufficiently high substrate concentrations to estimate whether the same maximum rate is plausible.
Worked Example: A Pure Noncompetitive Pattern
Consider a separate hypothetical assay with the same uninhibited \(V_{\max}=100\) pmol/min/mg and \(K_m=10\) µM. At \(I=5\) µM, suppose \(K_i=K_i'=5\) µM. Thus \(\alpha=\alpha'=2\). Find the apparent parameters and rate at \(S=10\) µM.
The apparent maximum rate is \(100/2=50.00\) pmol/min/mg, and the apparent Michaelis constant is \(2(10)/2=10.00\) µM. Direct substitution gives \(v=100(10)/(2(10)+2(10))=1000/40=25.00\) pmol/min/mg. Without inhibitor, the rate at \(S=10\) µM would be \(100(10)/(10+10)=50.00\) pmol/min/mg.
The maximum rate is halved, whereas the apparent \(K_m\) is unchanged. That proportional change in both \(\alpha\) and \(\alpha'\) supports pure noncompetitive inhibition under the stated model; a reduced rate at one substrate concentration alone would not establish it.
Worked Example: A Mixed Pattern and Parameter Recovery
In a third hypothetical assay, let the uninhibited parameters again be \(V_{\max}=100\) pmol/min/mg and \(K_m=10\) µM. At \(I=5\) µM, suppose \(K_i=5\) µM and \(K_i'=10\) µM. Then \(\alpha=2\) and \(\alpha'=1.5\). Calculate the apparent parameters and rate at \(S=20\) µM.
Here \(V_{\max,\mathrm{app}}=100/1.5=66.67\) pmol/min/mg and \(K_{m,\mathrm{app}}=2(10)/1.5=13.33\) µM. At \(S=20\) µM, \(v=100(20)/(2(10)+1.5(20))=2000/50=40.00\) pmol/min/mg. Both apparent parameters have changed, and their changes are not those of pure noncompetitive inhibition.
The factors can also be recovered from apparent parameters if the uninhibited parameters are known: \(\alpha'=V_{\max}/V_{\max,\mathrm{app}}=100/66.67\approx1.50\), and \(\alpha=\alpha'K_{m,\mathrm{app}}/K_m=1.50(13.33)/10\approx2.00\). At this inhibitor concentration, those factors imply \(K_i=I/(\alpha-1)=5/(2-1)=5.00\) µM and \(K_i'=I/(\alpha'-1)=5/(1.5-1)=10.0\) µM. These are hypothetical values inferred under the model, not universal enzyme constants.
Choosing Among Models and Checking the Fit
A practical analysis fits several candidate forms to rates measured over substrate concentrations and inhibitor conditions. The uninhibited data help anchor \(V_{\max}\) and \(K_m\); inhibited conditions then inform the inhibition factors. A competitive model constrains \(\alpha'=1\), a pure noncompetitive model constrains \(\alpha=\alpha'\), and a mixed model estimates the two factors separately through \(K_i\) and \(K_i'\). These are alternative constraints on the same general rate equation, not three unrelated descriptions.
Prefer a global nonlinear fit to the measured rates, using an error model suitable for the assay, rather than estimating a separate curve at each inhibitor concentration and comparing noisy estimates by eye. Assess residuals across the substrate and inhibitor ranges, parameter uncertainty, and whether the added flexibility of the mixed model is supported. A mixed model will usually fit at least as well because it has fewer constraints; improved fit alone does not show that its extra parameter is meaningfully identified. Information criteria or an appropriate nested-model comparison can help, but should be considered alongside uncertainty and residual patterns.
Useful diagnostics include systematic residual trends with substrate concentration, implausible estimates, and large uncertainty or strong dependence between \(K_i\) and \(K_i'\). A narrow substrate range, especially one without observations near saturation, may make \(V_{\max,\mathrm{app}}\) difficult to estimate. If rates at the highest tested substrate concentrations still rise substantially, the experiment may not identify the maximum rate well enough to distinguish competitive from other forms. Replicate variation and inhibitor concentrations should also be checked; a single anomalous point should not determine the chosen mechanism.
| Supported form | Constraint | Expected apparent pattern | Victim pathway representation |
|---|---|---|---|
| Competitive | \(\alpha'=1\) | Unchanged \(V_{\max,\mathrm{app}}\); increased \(K_{m,\mathrm{app}}\) | Use \(\alpha=1+I/K_i\), with the inhibitor multiplying the \(K_m\) term |
| Pure noncompetitive | \(\alpha=\alpha'\) | Reduced \(V_{\max,\mathrm{app}}\); unchanged \(K_{m,\mathrm{app}}\) | Use the same factor on both terms, reducing the rate maximum |
| Mixed | \(\alpha\ne\alpha'\) | Maximum rate changes; apparent \(K_m\) may rise or fall | Retain separate \(K_i\) and \(K_i'\) effects in the rate equation |
Encoding the Supported Form in a Victim–Perpetrator Model
In a victim model, the assay-supported form belongs in the CYP-mediated pathway rate, alongside the victim’s other disposition processes. For a competitive form, set \(\alpha'=1\) and use the inhibitor effect on the \(K_m\) term. For pure noncompetitive inhibition, constrain the two factors to equality. For mixed inhibition, retain both factors separately. Do not choose among these forms merely to improve a DDI exposure prediction; the assay pattern and model adequacy should support the structure.
At a specified substrate concentration, the inhibited-to-uninhibited rate ratio implied by the model is
This expression makes clear why a single constant “percent inhibition” may be unsuitable: the ratio depends on substrate concentration as well as inhibitor concentration. In a PBPK implementation, evaluate the pathway rate using the model’s relevant enzyme-site substrate and inhibitor concentrations, with compatible units, rather than applying the assay rate reduction unchanged at all times and concentrations. The assay supplies the inhibition form and potency information; it does not by itself establish the in vivo concentrations at the enzyme site.
The form also has limits. A pattern that changes with incubation time, reflects enzyme loss, or cannot be described by a single Michaelis–Menten pathway may require a different model. In particular, do not force time-dependent activity loss into a reversible inhibition equation. If the mixed model is weakly identified, report that uncertainty and consider whether the assay design can distinguish the two binding interactions before encoding separate effects as well determined.
Common Mistake
Wrong reasoning: “The inhibitor lowers the rate at every measured substrate concentration, so it must be noncompetitive.” A lower rate at a finite set of concentrations is compatible with competitive, pure noncompetitive, and mixed inhibition. Correction: compare the substrate-dependent curves and their apparent parameters. Competitive inhibition is specifically the case \(\alpha'=1\); pure noncompetitive inhibition requires \(\alpha=\alpha'\); unequal factors indicate mixed inhibition. If the substrate range or precision cannot distinguish those patterns, the model choice remains uncertain.
Check Your Understanding
Answer each question, then compare with the concise solutions below.
- If an inhibitor condition has unchanged apparent \(V_{\max}\) and increased apparent \(K_m\), which inhibition form is supported by the simple model, and what is its defining constraint?
- A pure noncompetitive assay has \(V_{\max}=80\) pmol/min/mg, \(K_m=8\) µM, and \(\alpha=\alpha'=2\). Find both apparent parameters.
- For mixed inhibition with \(\alpha=1.5\), \(\alpha'=2\), \(K_m=12\) µM, and \(V_{\max}=60\) pmol/min/mg, calculate \(K_{m,\mathrm{app}}\) and \(V_{\max,\mathrm{app}}\).
- Why is a reciprocal plot not usually the preferred primary fitting method for noisy rate data?
- What should be concluded if the assay has no substrate concentrations near saturation and cannot distinguish whether apparent \(V_{\max}\) changes?
Solutions. (1) Competitive inhibition, with \(\alpha'=1\), meaning no inhibitor binding to the enzyme–substrate complex in this model. (2) \(V_{\max,\mathrm{app}}=80/2=40.00\) pmol/min/mg and \(K_{m,\mathrm{app}}=2(8)/2=8.00\) µM. (3) \(K_{m,\mathrm{app}}=1.5(12)/2=9.00\) µM and \(V_{\max,\mathrm{app}}=60/2=30.00\) pmol/min/mg. (4) Taking reciprocals changes the error structure and gives low-rate observations disproportionate influence; fit the untransformed rates with an appropriate error model. (5) The inhibition form is not adequately identified by those data. Do not treat an uncertain maximum-rate pattern as proof of competitive inhibition; obtain more informative substrate coverage or carry model uncertainty forward.