The Effect Site as a Latent Equilibration State

Why Introduce an Effect Site?

A measured plasma concentration and a pharmacodynamic response do not always rise and fall together. A response may peak after plasma concentration has begun to decline, or the same plasma concentration may correspond to different responses at different times. A useful model for this delay separates the concentration that is measured or predicted in plasma from a latent concentration state that drives the response.

The effect compartment is a modeling device for representing this separation. It does not require that a distinct anatomical compartment has been sampled or even identified. Instead, it introduces an unobserved state, called the effect-site concentration, that approaches plasma concentration at a specified rate. A pharmacodynamic model then uses that state as its concentration input. The earlier material on receptor theory describes concentration–response relationships; the effect compartment provides one way to represent a delay between plasma exposure and the concentration supplied to such a relationship.

Definition: Let $C_p(t)$ denote plasma concentration at time $t$, and let $C_e(t)$ denote effect-site concentration at time $t$. In the first-order effect-compartment model, $C_e(t)$ is a latent state governed by $$ \frac{dC_e(t)}{dt}=k_{e0}\bigl(C_p(t)-C_e(t)\bigr), $$ where $k_{e0}>0$ is the equilibration rate constant, with units of inverse time. The model treats the pharmacodynamic response at time $t$ as depending on $C_e(t)$, through a specified response model, rather than necessarily depending directly on $C_p(t)$.

The equation defines the state’s instantaneous rate of change. If plasma concentration is greater than effect-site concentration, the rate is positive; if it is lower, the rate is negative. The state therefore moves toward the plasma input rather than instantaneously matching it, except when the two are already equal. The notation $k_{e0}$ is often pronounced “K-e-zero.” It is a rate, not a concentration and not a direct measure of response magnitude.

“Latent” means that $C_e(t)$ is not directly observed in the usual data set. Plasma samples and response measurements may be observed, while $C_e(t)$ is calculated as part of the model’s evolving internal state. Its value depends on the plasma input, the initial state, and $k_{e0}$. Thus, a reported effect-site concentration is a model-based quantity, not a substitute for a measured tissue concentration.

Assumptions and Interpretation

The model requires a time-indexed plasma concentration input $C_p(t)$ that is appropriately measured or modeled. That input may be constant or time-varying. A constant input is valid: if $C_e$ initially differs from it, the effect-site state changes; if the two are equal, the state stays at equilibrium for as long as the input remains constant. The model also requires a defined initial value for $C_e$, a positive $k_{e0}$ that is treated as constant over the modeled conditions, and a response model whose relevant concentration input is $C_e$.

The first-order assumption summarizes equilibration using one state and one rate constant. It says that the rate of change is proportional to the current difference between plasma and effect-site concentrations. It does not specify a detailed anatomical transport pathway. For the response link, the basic interpretation is that the response model acts on the current $C_e(t)$. Any additional delay between $C_e(t)$ and measured response, or any response change that persists after $C_e(t)$ changes, would require further structure rather than being automatically explained by this concentration state.

Concentrations used in the difference $C_p(t)-C_e(t)$ must have the same units, and time units must be consistent with $k_{e0}$. For example, when time is measured in hours, $k_{e0}$ is expressed in $\mathrm{h}^{-1}$. A response model may have its own parameters and units, such as a baseline response and a maximum response, but these do not change the units or definition of $k_{e0}$.

Why the State Moves Toward Plasma

A short sign argument explains the “equilibration” interpretation without solving the state equation. Suppose the plasma input is held at a constant value $C_p$, and define the difference from equilibrium as $D(t)=C_e(t)-C_p$. Since $C_p$ is constant, the state equation gives

\frac{dD(t)}{dt} =\frac{dC_e(t)}{dt} =k_{e0}\bigl(C_p-C_e(t)\bigr) =-k_{e0}D(t).

Because $k_{e0}>0$, the rate of change of $D(t)$ has the opposite sign from $D(t)$. If $C_e(t)$ is below $C_p$, the difference is negative and the state rises; if it is above $C_p$, the difference is positive and the state falls. If the state equals $C_p$, its rate of change is zero. This justifies calling $C_p$ the equilibrium value under a constant input. The argument establishes direction and equilibrium, not a formula for the full time course.

A larger $k_{e0}$ means a larger instantaneous rate for the same concentration difference. It does not mean that the response is larger: response magnitude is determined by the pharmacodynamic response model. Nor does $k_{e0}$ describe how quickly plasma concentration changes, which is determined by the pharmacokinetic input and model.

Worked Example: Constant Plasma Input With a Different Initial State

Consider a hypothetical situation in which plasma concentration is held at $8.0\ \mathrm{mg/L}$, the initial effect-site concentration is $2.0\ \mathrm{mg/L}$, and $k_{e0}=0.60\ \mathrm{h}^{-1}$. What does the model say about the state’s initial direction and rate of change?

Use the state equation at the initial time:

\left.\frac{dC_e}{dt}\right|_{t=0} =k_{e0}\bigl(C_p(0)-C_e(0)\bigr) =(0.60\ \mathrm{h}^{-1})(8.0-2.0\ \mathrm{mg/L}) =3.6\ \mathrm{mg/L/h}.

The positive rate means that $C_e$ is initially increasing. The input is constant, but the state is not initially at equilibrium, so constancy of plasma concentration does not imply a constant effect-site concentration. If the input stays fixed, the sign argument above shows that the state moves toward $8.0\ \mathrm{mg/L}$. This calculation gives the initial rate; it does not give a later value of $C_e$.

Worked Example: A State Above the Plasma Input

In another hypothetical case, $C_p(t)=4.0\ \mathrm{mg/L}$, $C_e(t)=7.0\ \mathrm{mg/L}$, and $k_{e0}=0.25\ \mathrm{h}^{-1}$ at a particular time. Calculate the instantaneous change in effect-site concentration and interpret its sign.

\frac{dC_e}{dt} =(0.25\ \mathrm{h}^{-1})(4.0-7.0\ \mathrm{mg/L}) =-0.75\ \mathrm{mg/L/h}.

The negative rate means that $C_e$ is decreasing at that time. The state is above the plasma input, so the model moves it downward toward the input. This is a local rate, not a claim that the concentration will decrease at exactly this rate indefinitely: as the difference changes, the rate given by the equation changes too.

How the Latent State Enters the Response Model

The effect compartment changes the concentration supplied to the pharmacodynamic relationship. It does not, by itself, specify that relationship. For example, a hypothetical saturating response model might be written as $E(t)=E_0+E_{\max}C_e(t)/(EC_{50}+C_e(t))$, where $E(t)$ is the response, $E_0$ is baseline response, $E_{\max}$ is the maximum increase above baseline, and $EC_{50}$ is the effect-site concentration producing half of that maximum increase under this model. The model’s response parameters and the equilibration parameter have distinct roles.

Worked Example: Same Plasma Concentration, Different Latent States

Suppose a hypothetical response model is $E=E_0+E_{\max}C_e/(EC_{50}+C_e)$, with $E_0=10$ response units, $E_{\max}=90$ response units, and $EC_{50}=3.0\ \mathrm{mg/L}$. At two times, plasma concentration is $6.0\ \mathrm{mg/L}$, but the modeled effect-site concentrations are $2.0\ \mathrm{mg/L}$ and $6.0\ \mathrm{mg/L}$, respectively. Calculate the response at each state.

For $C_e=2.0\ \mathrm{mg/L}$:

E=10+90\frac{2.0}{3.0+2.0} =10+90(0.400) =46.0\ \text{response units}.

For $C_e=6.0\ \mathrm{mg/L}$:

E=10+90\frac{6.0}{3.0+6.0} =10+90(0.667) \approx70.0\ \text{response units}.

The same plasma concentration is compatible with different modeled responses because the latent states differ. The values illustrate how state history can matter in the model; they do not establish that either effect-site concentration is directly observed or that this particular response model is suitable for a real drug. Replacing $C_e$ with $C_p$ would instead impose an instantaneous concentration–response relationship.

Diagnostics, Alternatives, and Limitations

A coherent interpretation depends on the full model, not merely on adding a state variable. Plasma concentrations must be measured or predicted adequately at the times relevant to response. The initial value of $C_e$ must be compatible with the dosing and observation history; an arbitrary starting value can make early predictions misleading. The response model must also represent the measured endpoint appropriately, including its baseline and observation process where needed.

Useful warning signs include systematic mismatch between predicted and observed responses during rising versus falling plasma concentrations, residual patterns that persist after accounting for the effect-site state, and estimates that change markedly when the input history or initial condition is altered. Such patterns do not uniquely identify the cause. They may indicate an inadequate plasma input, a poor response model, additional delay or turnover, or failure of the single-rate equilibration assumption. A fitted effect compartment is not proof of a particular anatomical site or transport mechanism.

The main alternatives answer different modeling questions. A direct-effect model links response to current plasma concentration and is simpler when delay is negligible or not supported by the data. An effect-compartment model adds a latent concentration state but retains a response relationship to concentration at the current time. An indirect-response model represents delayed response through dynamics in the response-generating process itself. These approaches are not interchangeable by definition: the appropriate choice depends on the observed time course and the scientific question. A state-based model may improve predictions while still leaving the underlying biological mechanism uncertain.

The effect-compartment model is limited by its compact assumptions: one latent state, one positive constant equilibration rate, and a specified response link. It does not automatically account for multiple distribution delays, changing equilibration rates, tolerance, active metabolites, or response persistence independent of effect-site concentration. When such features matter, the model needs additional structure and evidence rather than a broader interpretation of $C_e$ or $k_{e0}$.

Common pitfall: It is incorrect to say that a constant plasma concentration forces effect-site concentration to be constant. The state equation shows that $C_e$ changes whenever $C_p-C_e$ is nonzero; only equality gives zero rate under a constant input. It is also incorrect to interpret $k_{e0}$ as response potency or as a measured tissue concentration. It governs the rate of movement of the latent state toward the plasma input.

Before interpreting a fitted effect compartment, check that concentration and time units are consistent, the initial state is defined, the plasma input may be constant or time-varying as appropriate, and the response model explicitly uses $C_e$. Then assess whether the observed responses are adequately represented without treating the latent state as direct evidence of a physical compartment.

Check Your Understanding

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

At one time, $C_p=5.0\ \mathrm{mg/L}$, $C_e=3.0\ \mathrm{mg/L}$, and $k_{e0}=0.40\ \mathrm{h}^{-1}$. What is the instantaneous rate of change of $C_e$, and in which direction is it moving?
If plasma concentration is constant and $C_e=C_p$, what does the state equation say about the state’s rate of change?
Why is $C_e$ described as latent, and what does the response model assume about its role?
Name two model features that an effect compartment does not automatically represent.

Solutions. (1) $(0.40\ \mathrm{h}^{-1})(5.0-3.0\ \mathrm{mg/L})=0.80\ \mathrm{mg/L/h}$, so $C_e$ is increasing. (2) The rate is $k_{e0}(C_p-C_e)=0$, so the state remains at that value while the input stays constant. (3) It is latent because it is an internal modeled state rather than a directly observed concentration; the basic response model uses $C_e(t)$ as its concentration input at time $t$. (4) Examples include multiple equilibration processes, changing $k_{e0}$, tolerance, active metabolites, and response persistence not explained by $C_e$.