Common Errors in TMDD Model Interpretation

Why TMDD Interpretation Requires Care

A target-mediated drug disposition model separates drug that is free from drug bound to its target. That distinction is mechanistic, not merely a choice of labels: free drug and drug–target complex can have different distribution, clearance, and pharmacological consequences. A measured concentration may combine both forms, while a model parameter describing an effective affinity may combine binding and complex loss. Treating these quantities as interchangeable can produce an internally consistent calculation with the wrong biological interpretation.

This tutorial focuses on three recurring errors: interpreting total drug as free drug, reading a quasi-steady-state affinity parameter as a microscopic binding constant, and using a rapid-equilibrium reduction without checking whether binding is fast enough. The calculations below use concentrations in a shared compartment and make the assumptions explicit.

What Total Drug Does—and Does Not—Tell You

A standard TMDD model includes reversible binding, complex internalization, and nonspecific elimination of free drug. For clarity, write the concentration equations as

Here \(u(t)\) is drug input expressed as concentration per time; \(k_{\mathrm{el}}\), \(k_{\mathrm{deg}}\), \(k_{\mathrm{on}}\), \(k_{\mathrm{off}}\), and \(k_{\mathrm{int}}\) are rate constants; and \(k_{\mathrm{syn}}\) is target synthesis. The equations assume a well-mixed compartment, 1:1 reversible binding, and the specified first-order loss processes. More elaborate models may add compartments or other loss pathways.

Add the free-drug and complex equations to obtain the total-drug balance. The binding terms cancel because association removes free drug as it creates complex, while dissociation does the reverse:

This cancellation is the key interpretation check. Reversible binding alone redistributes drug between free and bound forms; it does not destroy total drug. But the two loss terms act on different forms. In general, total drug cannot be substituted for \(D\) in the nonspecific elimination term, and it cannot be assumed to track the concentration available to bind target. If an assay dissociates complexes during measurement, its reported concentration may be closer to total drug than to free drug. If it preserves complexes, the result depends on the assay's selectivity and calibration.

Effective Affinity Is Not Necessarily Microscopic Affinity

The microscopic equilibrium dissociation constant for reversible binding is \(K_D=k_{\mathrm{off}}/k_{\mathrm{on}}\). It describes the binding reaction when association and dissociation balance and complex loss does not alter that balance. In a TMDD model with complex internalization, setting the complex derivative to zero instead gives a different parameter combination.

This is a quasi-steady-state relation: it follows from approximating \(dC/dt\) by zero, not from assuming that binding and dissociation alone are at equilibrium. Since \(K_{\mathrm{ss}}=K_D+k_{\mathrm{int}}/k_{\mathrm{on}}\), it equals \(K_D\) only when complex internalization is negligible relative to dissociation in this expression. With \(R_{\mathrm{tot}}=R+C\), the quasi-steady-state relation also yields \(C=R_{\mathrm{tot}}D/(K_{\mathrm{ss}}+D)\), provided the stated concentration definitions and model assumptions apply.

When Is Rapid Equilibrium a Defensible Reduction?

A rapid-equilibrium reduction assumes binding and dissociation adjust much faster than the processes that appreciably change free drug, free target, or complex. Under that assumption, binding is approximated by \(DR=K_D C\). In contrast, the quasi-steady-state approximation sets \(dC/dt\approx0\) and gives \(DR=K_{\mathrm{ss}}C\). The two relations are not interchangeable: the rapid-equilibrium relation requires binding equilibrium to be a good approximation, whereas the quasi-steady-state relation accounts for internalization in the complex-loss term.

A useful local check for the speed of reversible binding is the relaxation rate \(k_{\mathrm{on}}(D+R)+k_{\mathrm{off}}\), obtained from the binding-only system near fixed concentrations. Its reciprocal gives a characteristic binding-relaxation time. This check is not, by itself, proof that a reduction is valid: internalization and changes in drug or target must also be slow enough for the chosen approximation. In particular, rapid equilibrium is doubtful when binding relaxation is not much faster than the rates that drive those changes, or when internalization competes materially with dissociation.

Choosing and Diagnosing an Interpretation

A reduced model can be useful, but the interpretation must match the reduction's assumptions. The comparison below concerns the binding relation, not a guarantee that any model will fit a given data set.

Several diagnostic questions help expose interpretation errors before they enter a scientific conclusion:

• What does the assay measure? Check whether sample processing preserves complexes, dissociates them, or detects only a selected form. Do not assign a free-drug meaning to a total-drug result without evidence.
• Which state variable appears in each equation? In the model above, nonspecific elimination acts on \(D\), while internalization acts on \(C\). Replacing either with \(D_{\mathrm{tot}}\) changes the model.
• Which affinity parameter was estimated? Check the defining equation. A parameter entering a quasi-steady-state relation is not automatically a microscopic binding constant.
• Are the timescales separated? Compare binding relaxation with internalization, target turnover, and drug concentration changes over the relevant range, not just at one convenient concentration.
• Are reduced and full models consistent with observations? Systematic disagreement in concentration-time patterns, or parameter estimates that shift markedly when a reduction changes, warns that the simplified interpretation may be inadequate. Predictive checks assess model behavior but do not, on their own, prove a microscopic mechanism.

The full dynamic model is the direct alternative when binding and complex loss cannot safely be collapsed into an equilibrium relation, though it may demand richer data and can be difficult to estimate. A quasi-steady-state model can reduce dynamic complexity while retaining an effective loss term. Rapid equilibrium is more restrictive about binding timescales. Neither reduction should be selected solely because it has fewer parameters or yields a familiar affinity value. Reduced models also have limits: if their assumptions fail over part of the dose range or time course, a single effective parameter may conceal changing mechanisms.

Common Mistakes and Corrections

• Wrong: “The assay reports 10 nM drug, so free drug is 10 nM.” Correction: If the assay reports \(D+C\), it reports total drug; free drug requires a measurement or model that distinguishes \(D\) from \(C\).
• Wrong: “The fitted affinity is 10 nM, so the molecular \(K_D\) is 10 nM.” Correction: Establish whether the model parameter is \(K_D\) or \(K_{\mathrm{ss}}\); internalization can make the latter larger.
• Wrong: “Binding is reversible, therefore it is at equilibrium.” Correction: Reversibility does not imply rapid equilibration. Compare binding relaxation with internalization and other concentration-changing timescales.
• Wrong: “A good fit proves the reduced model's affinity has a microscopic meaning.” Correction: A fit supports compatibility with the observations under that model; it does not change the parameter's mathematical definition or establish a unique mechanism.