Why Inhibiting Loss Changes Both Level and Timing
In an indirect response model, response is governed by the balance between production and loss. When a drug inhibits loss, the production rate is unchanged, but each unit of response is removed more slowly. The response therefore tends toward a higher equilibrium than its untreated baseline, and it approaches that equilibrium at a slower rate. These two consequences follow from the same altered loss-rate constant.
The focus here is the exact trajectory while the degree of loss inhibition is constant. That restriction makes it possible to solve the response equation in closed form; it is also a substantive assumption, not merely a mathematical convenience.
The model assumes constant production, first-order loss proportional to the current response, and a constant inhibition level over the interval being modeled. The response and production rate must use compatible units: for example, if \(R\) is measured in mg/L, then \(k_{\mathrm{in}}\) has units mg/L/day and both rate constants have units day\(^{-1}\). The inhibition fraction is dimensionless. For a finite, stable drug-on equilibrium, inhibition must be partial: \(I_{\mathrm{out}}<1\).
Equilibrium and Exact Trajectory
Write the effective loss-rate constant as \(\lambda=k_{\mathrm{out}}(1-I_{\mathrm{out}})\). For partial inhibition, \(\lambda>0\). At equilibrium, the rate of change is zero, so production equals inhibited loss. Solving that balance gives
Thus, when \(I_{\mathrm{out}}=0\), the equilibrium equals baseline, \(R_{\mathrm{ss}}=R_0\). For positive partial inhibition, \(0<I_{\mathrm{out}}<1\), the equilibrium is above baseline. As inhibition increases toward 1, this model predicts a progressively higher equilibrium.
To derive the trajectory, subtract the equilibrium from both sides of the differential equation. Since \(k_{\mathrm{in}}=\lambda R_{\mathrm{ss}}\),
Let \(D(t)=R(t)-R_{\mathrm{ss}}\), the signed gap between the response and its equilibrium. Then \(dD/dt=-\lambda D\). When \(D(0)\ne0\), separating variables and integrating from time 0 to \(t\) gives \(\ln|D(t)|-\ln|D(0)|=-\lambda t\), so \(D(t)=D(0)e^{-\lambda t}\). If \(D(0)=0\), the differential equation gives \(D(t)=0\) for all \(t\), which is the same formula. Substituting back yields the exact solution for any initial response \(R(0)=R_{\mathrm{init}}\):
The exponential multiplier is positive and decreases toward zero. Consequently, the response moves monotonically toward \(R_{\mathrm{ss}}\) from either side and does not cross it under the stated model. The signed gap has half-life \(\ln(2)/\lambda\): after that time, the gap between response and equilibrium is half its initial size. This is not generally the time for \(R(t)\) itself to halve.
Worked Example: Response Starting at Baseline
Suppose a hypothetical system has \(k_{\mathrm{in}}=12\) response units/day and \(k_{\mathrm{out}}=0.30\) day\(^{-1}\). The system starts at baseline, and a constant drug effect inhibits loss by \(I_{\mathrm{out}}=0.40\). Calculate the equilibrium, the response after 5 days, and the half-life of the gap to equilibrium.
First, baseline is \(R_0=k_{\mathrm{in}}/k_{\mathrm{out}}=12/0.30=40.0\) response units. The effective loss rate and drug-on equilibrium are $$ \lambda=0.30(1-0.40)=0.180\ \text{day}^{-1}, \qquad R_{\mathrm{ss}}=\frac{12}{0.180}=66.67\ \text{response units}. $$
The initial response is \(R_{\mathrm{init}}=40.0\) response units. At 5 days, $$ R(5)=\frac{12}{0.180}+\left(\frac{12}{0.300}-\frac{12}{0.180}\right)e^{-0.180(5)} \approx 55.82\ \text{response units}. $$
The gap half-life is $$ t_{1/2,\mathrm{gap}}=\frac{\ln(2)}{0.180} =\frac{0.6931}{0.180}=3.851\ \text{days}. $$ At day 5 the response is above baseline but remains below the drug-on equilibrium, as expected for a trajectory starting at baseline and moving upward toward \(66.67\) response units.
Interpreting the Approach Rate
The equilibrium determines where the response tends; \(\lambda\) determines how quickly the remaining gap shrinks. Because \(\lambda=k_{\mathrm{out}}(1-I_{\mathrm{out}})\), increasing inhibition reduces the effective loss rate and lengthens the gap half-life. For example, with \(k_{\mathrm{out}}=0.30\) day\(^{-1}\), no inhibition gives a gap half-life of \(0.6931/0.30=2.310\) days. At 40% inhibition, it is 3.851 days. The slower approach and higher equilibrium are linked consequences of reducing the same loss rate.
The initial slope is also determined by the starting response and effective loss rate: $$ \left.\frac{dR}{dt}\right|_{t=0} =k_{\mathrm{in}}-\lambda R_{\mathrm{init}} =\lambda(R_{\mathrm{ss}}-R_{\mathrm{init}}). $$ If the system starts at untreated baseline, then \(R_{\mathrm{init}}=R_0\), and the initial slope is \(k_{\mathrm{out}}I_{\mathrm{out}}R_0\). It is zero when inhibition is zero and positive when inhibition is positive. This initial slope is not a separate drug effect; it follows from the difference between the starting response and the new equilibrium.
Worked Example: Starting Below Baseline
Consider a hypothetical system with \(k_{\mathrm{in}}=8.00\) mg/L/day, \(k_{\mathrm{out}}=0.200\) day\(^{-1}\), and constant \(I_{\mathrm{out}}=0.250\). Its initial response is \(R_{\mathrm{init}}=30.0\) mg/L, below its untreated baseline. Find the response after 4 days and determine the initial direction of change.
The baseline is \(8.00/0.200=40.0\) mg/L. The effective loss rate and inhibited-loss equilibrium are $$ \lambda=0.200(1-0.250)=0.150\ \text{day}^{-1}, \qquad R_{\mathrm{ss}}=\frac{8.00}{0.150}=53.33\ \text{mg/L}. $$
Using the initial condition in the exact solution, $$ R(4)=\frac{8.00}{0.150}+\left(30.0-\frac{8.00}{0.150}\right)e^{-0.150(4)} \approx 40.53\ \text{mg/L}. $$
The initial slope is \(8.00-(0.150)(30.0)=3.50\) mg/L/day, so the response initially increases. Since \(30.0<53.33\), the solution approaches the equilibrium from below. It passes the untreated baseline on its way upward, but it does not cross the inhibited-loss equilibrium.
Assumptions, Diagnostics, and Limits
The closed-form expression is exact for this specified differential equation, but its validity for a particular application depends on whether the equation is a reasonable description over the interval of interest. The key conditions are constant \(k_{\mathrm{in}}\), constant \(k_{\mathrm{out}}\), constant fractional inhibition, and first-order response loss. The initial response must also be specified on the same response scale used in the model.
Useful checks are direct consequences of the solution. For \(0\leq I_{\mathrm{out}}<1\), the predicted effective loss rate is positive, the exponential term decays, and the equilibrium is finite. The fitted or calculated trajectory should move toward that equilibrium without crossing it. If the response starts at baseline and inhibition is positive, the initial slope should be positive. If inhibition is zero, the model reduces to untreated turnover, with equilibrium \(R_0\) and approach rate \(k_{\mathrm{out}}\). Violations of these checks may indicate an arithmetic or unit error, a misspecified initial condition, or failure of the assumed model.
Systematic deviations in observed data can also signal that the assumptions do not hold. For example, an effect that changes over time is not represented by a constant \(I_{\mathrm{out}}\); changing production or non-first-order loss likewise falls outside this solution. Measurement noise can obscure the ideal monotone trajectory, so a small observed fluctuation does not by itself disprove the model. The relevant diagnostic is whether deviations are persistent or structured rather than plausibly attributable to observation variability.
At complete inhibition, \(I_{\mathrm{out}}=1\), the loss term vanishes and the equation becomes \(dR/dt=k_{\mathrm{in}}\). Its solution is \(R(t)=R_{\mathrm{init}}+k_{\mathrm{in}}t\), not a finite-equilibrium exponential. The formula \(R_{\mathrm{ss}}=R_0/(1-I_{\mathrm{out}})\) is undefined there. This boundary case is why the finite-equilibrium derivation requires partial inhibition.
Worked Example: Time to Reach a Fraction of the Equilibrium Change
A hypothetical response starts at its baseline of 50.0 units, with \(k_{\mathrm{out}}=0.400\) day\(^{-1}\) and constant inhibition \(I_{\mathrm{out}}=0.500\). How long does it take to complete 90% of the change from baseline to the inhibited-loss equilibrium?
The effective loss rate is \(0.400(1-0.500)=0.200\) day\(^{-1}\). The equilibrium is \(R_{\mathrm{ss}}=50.0/(1-0.500)=100\) units. The fraction of the total equilibrium change completed at time \(t\) is $$ \frac{R(t)-R_0}{R_{\mathrm{ss}}-R_0} =1-e^{-\lambda t}. $$
Set this fraction equal to 0.900 and solve: $$ 0.900=1-e^{-0.200t}, \qquad e^{-0.200t}=0.100, \qquad t=\frac{-\ln(0.100)}{0.200} =\frac{2.3026}{0.200} =11.51\ \text{days}. $$
This is the time to complete 90% of the 50-unit increase, not the time to reach 90% of the numerical equilibrium value. The distinction matters whenever the initial response is not zero.
How This Solution Differs From Alternatives
The exact solution for inhibited production describes the neighboring mechanism in which the drug reduces \(k_{\mathrm{in}}\) while leaving the loss rate unchanged. Under constant production inhibition, the equilibrium is below baseline for positive partial inhibition, and the exponential approach rate remains \(k_{\mathrm{out}}\). Under inhibited loss, positive partial inhibition instead raises the equilibrium and reduces the approach rate to \(k_{\mathrm{out}}(1-I_{\mathrm{out}})\). With zero inhibition, both mechanisms return the response to baseline; this boundary should not be described as an increase or decrease.
| Feature | Constant inhibition of loss | Constant inhibition of production |
|---|---|---|
| Altered rate | Loss rate becomes \(k_{\mathrm{out}}(1-I_{\mathrm{out}})\) | Production rate is reduced |
| Equilibrium with positive partial inhibition | Above baseline | Below baseline |
| Exponential approach rate | Reduced from \(k_{\mathrm{out}}\) | Remains \(k_{\mathrm{out}}\) |
| At zero inhibition | Equilibrium equals baseline | Equilibrium equals baseline |
Numerical integration is an alternative when a model cannot be solved in closed form or when its rates vary over time. For the constant-inhibition equation here, numerical integration is unnecessary for calculating the trajectory and can introduce avoidable approximation error. A numerical method should reproduce the exact solution under the same constant-rate assumptions, within its numerical tolerance. Neither approach fixes a poor structural assumption: if inhibition is not constant, this particular closed-form trajectory is not the appropriate solution.
Common Mistakes and Key Takeaways
- Calling the equilibrium higher when inhibition is zero. At \(I_{\mathrm{out}}=0\), \(R_{\mathrm{ss}}=R_0\); it is higher only for positive partial inhibition.
- Using the untreated loss rate in the exponential. Under inhibited loss, the exponent contains \(k_{\mathrm{out}}(1-I_{\mathrm{out}})\), not \(k_{\mathrm{out}}\).
- Confusing response half-life with gap half-life. The quantity \(\ln(2)/\lambda\) halves the distance from equilibrium; it does not generally halve \(R(t)\).
- Assuming the finite-equilibrium formula applies at complete inhibition. At \(I_{\mathrm{out}}=1\), loss is zero and the response increases linearly under constant positive production.
- Applying a constant-effect solution to a changing effect. The derivation requires a constant inhibition fraction over the interval.
The central result is that partial inhibition of loss raises the steady-state response by the factor \(1/(1-I_{\mathrm{out}})\) relative to baseline, while slowing the exponential approach in proportion to the remaining fraction of loss activity. Both predictions depend on the same effective loss-rate constant.
Check Your Understanding
Answer each question, then compare with the solutions below.
- A system has \(R_0=24\) units and constant \(I_{\mathrm{out}}=0.20\). What is the inhibited-loss equilibrium?
- If \(k_{\mathrm{out}}=0.50\) day\(^{-1}\) and \(I_{\mathrm{out}}=0.60\), what is the effective loss rate and the half-life of the gap to equilibrium?
- With positive partial inhibition, a response starts above its inhibited-loss equilibrium. Does the model predict an increase or decrease, and can the trajectory cross the equilibrium?
- What trajectory applies when \(I_{\mathrm{out}}=1\), and why is there no finite equilibrium under positive constant production?
- For \(R_0=40\) units, \(k_{\mathrm{out}}=0.25\) day\(^{-1}\), and \(I_{\mathrm{out}}=0.50\), starting at baseline, calculate the response after 2 days.
Solutions. (1) \(R_{\mathrm{ss}}=24/(1-0.20)=30.0\) units. (2) \(\lambda=0.50(1-0.60)=0.200\) day\(^{-1}\), so \(t_{1/2,\mathrm{gap}}=0.6931/0.200=3.466\) days. (3) It decreases toward the equilibrium because \(dR/dt=\lambda(R_{\mathrm{ss}}-R)<0\) when \(R>R_{\mathrm{ss}}\). The exponential solution approaches without crossing. (4) Loss is zero, so \(dR/dt=k_{\mathrm{in}}\) and \(R(t)=R_{\mathrm{init}}+k_{\mathrm{in}}t\). For positive \(k_{\mathrm{in}}\), the response keeps increasing and cannot balance production at a finite response. (5) The equilibrium is \(40/(1-0.50)=80\) units and \(\lambda=0.25(0.50)=0.125\) day\(^{-1}\). Therefore \(R(2)=80+(40-80)e^{-0.125(2)}=80-40e^{-0.250}\approx80-31.15=48.85\) units.