Why Add a Second Peripheral Compartment?
A two-compartment model represents a central space and one peripheral distribution space. Some concentration-time profiles may require a richer structural description: after the initial decline, concentrations can show more than one subsequent change in slope before reaching a slow terminal decline. A three-compartment model adds a second peripheral space, allowing two distinct reversible exchange processes to influence concentrations in the central compartment.
The added space is a model feature, not automatically a particular organ or tissue. It can represent a distribution process that differs in its apparent capacity or rate of exchange from the first peripheral process. The model is useful when that extra process helps describe the observed concentration-time shape and is supported by informative data. It does not, by itself, establish a biological explanation for the shape.
Amount-Balance Model
Let \(A_1(t)\), \(A_2(t)\), and \(A_3(t)\) be the amounts of drug, in mg, in the central compartment and the two peripheral compartments. Let \(V_1\), \(V_2\), and \(V_3\) be their apparent volumes, in L, and define \(C_i(t)=A_i(t)/V_i\), in mg/L. The central concentration \(C_1(t)\) is the model concentration associated with the measured plasma concentration. Let \(CL\) be central clearance, and let \(Q_2\) and \(Q_3\) be the intercompartmental clearances linking the central compartment to peripheral compartments 2 and 3. Each clearance has units of L/h.
The standard model assumes constant volumes and clearances, well-mixed compartments, linear transfers proportional to the concentration difference between connected compartments, and elimination only from the central compartment. For an IV bolus, the dose is placed instantaneously in the central compartment; after that input, the model has no further dose input. The differential equations are:
For example, \(Q_2C_1\) is the rate, in mg/h, of transfer from the central compartment to compartment 2, while \(Q_2C_2\) is the reverse rate. Their difference is \(Q_2(C_1-C_2)\). Thus transfer is reversible: when \(C_2>C_1\), the net transfer is from compartment 2 toward the central compartment. The same reasoning applies to compartment 3. Summing the three amount equations cancels both pairs of transfer terms, leaving \(d(A_1+A_2+A_3)/dt=-CL C_1\). Intercompartmental transfer redistributes drug; it does not eliminate it.
Using \(A_i=V_iC_i\), the concentration equations follow by dividing each amount equation by its compartment volume. For example, \(dC_1/dt=-(k_{10}+k_{12}+k_{13})C_1+k_{21}(V_2/V_1)C_2+k_{31}(V_3/V_1)C_3\). Equivalently, because \(k_{21}V_2/V_1=Q_2/V_1=k_{12}\), the coefficient on \(C_2\) is \(k_{12}\). Keeping track of whether an equation describes amount or concentration prevents a common parameterization error: \(Q_2\) multiplies a concentration difference in the amount equation, whereas \(k_{12}\) is a rate constant in a rate-based parameterization.
Worked Example: Converting Clearances and Volumes to Rate Constants
Consider a hypothetical IV bolus model with \(V_1=12\) L, \(V_2=30\) L, \(V_3=60\) L, \(CL=3\) L/h, \(Q_2=6\) L/h, and \(Q_3=2\) L/h. Calculate the five first-order rate constants.
For elimination, \(k_{10}=CL/V_1=3/12=0.2500\ \text{h}^{-1}\). For transfer between the central compartment and compartment 2, \(k_{12}=Q_2/V_1=6/12=0.5000\ \text{h}^{-1}\) and \(k_{21}=Q_2/V_2=6/30=0.2000\ \text{h}^{-1}\). For the other exchange path, \(k_{13}=Q_3/V_1=2/12=0.1667\ \text{h}^{-1}\) and \(k_{31}=Q_3/V_3=2/60=0.03333\ \text{h}^{-1}\).
The two transfer directions for a given connection use the same intercompartmental clearance but different volumes in their denominators. Consequently, \(k_{12}\) need not equal \(k_{21}\), and \(k_{13}\) need not equal \(k_{31}\). A small reverse rate constant does not mean that the reverse process is absent; it can reflect a large peripheral volume.
From Amount Equations to a Three-Exponential Curve
The three amount equations form a linear system with constant coefficients. Write the amount vector as \(\mathbf{A}(t)=(A_1(t),A_2(t),A_3(t))^\mathsf{T}\), and write the system as \(d\mathbf{A}/dt=M\mathbf{A}\), where \(M\) is the matrix of transfer and elimination rates after substituting \(C_i=A_i/V_i\). For a constant matrix, the solution is determined by its eigenvalues and eigenvectors. With positive volumes and clearances, and with elimination present, the system is stable: its modes decay over time. When the three decay rates are distinct, the central concentration after an IV bolus has the form:
Here the \(\lambda_i\) are the positive macro-rate constants, in inverse time, and the \(B_i\) are concentration coefficients whose values depend on the dose and model parameters. This is the three-compartment counterpart of the exponential macroconstant description introduced in the earlier material on microconstants and macroconstants. The microconstants specify individual transfer and elimination processes; the macroconstants describe the combined decay modes visible in the concentration curve. The three terms do not correspond one-to-one with three anatomical locations.
A short justification comes directly from the linear-system solution. If \(M\mathbf{v}_i=-\lambda_i\mathbf{v}_i\), the amount vector \(\mathbf{v}_i e^{-\lambda_i t}\) satisfies \(d\mathbf{A}/dt=M\mathbf{A}\), since its derivative is \(-\lambda_i\mathbf{v}_i e^{-\lambda_i t}=M\mathbf{v}_i e^{-\lambda_i t}\). When three independent eigenvectors span the three-dimensional amount space, the general solution is a sum of these three modes. Taking the central amount and dividing by \(V_1\) gives the stated concentration form. Repeated eigenvalues can require a modified mathematical form, but three distinct modes are the usual conceptual description.
For an IV bolus of dose \(D\), the initial conditions are \(A_1(0^+)=D\), \(A_2(0^+)=A_3(0^+)=0\). Therefore \(C_1(0^+)=D/V_1\), while the peripheral concentrations initially equal zero. The initial central concentration slope is:
The initial decline reflects both elimination and outward distribution, because both peripheral compartments initially have lower concentrations than the central compartment. Later, the peripheral concentrations rise and drug can return to the central compartment. At sufficiently late times, the slowest decay mode dominates the curve, provided its coefficient is nonzero. Its rate, rather than the initial slope, characterizes the terminal exponential decline.
Worked Example: Initial Concentration and Initial Decline
Use the hypothetical parameters from the previous example and suppose the IV bolus dose is 120 mg. Calculate the initial central concentration and its initial rate of change.
Immediately after the bolus, \(C_1(0^+)=D/V_1=120\ \text{mg}/12\ \text{L}=10.00\ \text{mg/L}\). Since both peripheral amounts are initially zero, \(C_2(0^+)=C_3(0^+)=0\). The amount rates at that instant are \(dA_2/dt=Q_2C_1=6(10)=60\ \text{mg/h}\) and \(dA_3/dt=Q_3C_1=2(10)=20\ \text{mg/h}\). Elimination is \(CL C_1=3(10)=30\ \text{mg/h}\).
The central amount is therefore decreasing at \(60+20+30=110\ \text{mg/h}\), so \(dC_1/dt=-110/12=-9.167\ \text{mg/L/h}\). The same result follows from the initial-slope formula: \(-[(3+6+2)/12](10)=-9.167\ \text{mg/L/h}\). Across all three compartments the total amount decreases at 30 mg/h, not 110 mg/h, because the 80 mg/h leaving the central compartment by distribution enters the peripheral compartments.
What the Additional Distribution Space Represents
The two peripheral spaces allow the model to represent two central-to-peripheral exchange processes with different combinations of apparent capacity and exchange rate. In the model, a peripheral compartment with a large volume and a relatively slow return rate can contribute to prolonged redistribution back toward the central compartment. Another peripheral compartment may exchange more rapidly. These features can produce a concentration curve with a rapid early decline, an intermediate redistribution pattern, and a slower terminal decline.
This interpretation is structural rather than anatomical. Compartments are well-mixed mathematical spaces. A fitted peripheral volume does not necessarily equal the measured volume of a tissue or organ, and a fitted exchange clearance does not prove a specific transport mechanism. The model also does not assert that molecules move through a literal sequence of three physical containers. Its exchanges represent the net behavior captured by the chosen structure.
Worked Example: Describing a Hypothetical Prolonged Tail
Suppose a hypothetical IV bolus profile has central concentrations of 10.0 mg/L at time zero, 4.2 mg/L at 1 h, 1.50 mg/L at 6 h, and 0.550 mg/L at 24 h. These values are illustrative observations, not results from a real study. Calculate the apparent first-order decline rate between 6 and 24 h, assuming the late interval is approximately log-linear.
For an exponential decline over an interval, the rate is estimated from the log concentration ratio divided by elapsed time: \(\lambda=\ln[C(6)/C(24)]/(24-6)\). Substitution gives \(\lambda=\ln(1.50/0.550)/18=\ln(2.727)/18=1.003/18=0.05574\ \text{h}^{-1}\). The corresponding apparent half-life is \(\ln(2)/\lambda=0.6931/0.05574=12.43\) h.
This calculation describes the observed late-interval slope under the stated approximation. A three-compartment model can represent a prolonged terminal tail alongside earlier changes in slope, because it has three decay modes. These four observations alone do not establish that the three-compartment model is required, nor do they identify which physical tissues might contribute to the tail.
Assumptions, Diagnostics, and Limitations
The model is most directly applicable when the dose is an IV bolus and concentration changes are reasonably represented by linear, time-invariant disposition over the studied range. The compartments are assumed to mix instantaneously, volumes and clearances are constant, transfers are first order, and elimination is linear from the central compartment. If the data show concentration-dependent elimination, time-varying clearance, or an input pattern not represented by an IV bolus, this structural model may not describe the process adequately without modification.
Diagnostics should ask whether the model's assumptions and structure are compatible with the observed data. Systematic residual patterns across early, intermediate, or late times can indicate that the predicted curve misses an important feature. A predicted initial decline that is too shallow or too steep points to a mismatch in the combined initial elimination and distribution effects. A persistent late-time mismatch suggests that the predicted terminal behavior does not capture the observations. Estimates that change markedly with small data or model changes, or that have large uncertainty, can signal that the available observations do not strongly distinguish the two exchange processes.
Interpretability depends on when and how concentrations were sampled. Early observations inform the rapid part of the profile, while later observations are needed to characterize a prolonged tail. A dataset concentrated in one time region may provide weak information about other phases. The added compartment also creates additional parameters, so a numerically successful fit is not automatically evidence that every parameter is precisely estimated or biologically meaningful.
| Approach | What it represents | Main limitation |
|---|---|---|
| One-compartment model | A single well-mixed disposition space with elimination | Cannot represent separate reversible distribution processes |
| Two-compartment model | A central space linked to one peripheral distribution space | Has one exchange path and fewer distinct decay modes |
| Three-compartment model | A central space linked to two peripheral distribution spaces | Additional parameters can be difficult to estimate and do not identify literal anatomy |
| Empirical exponential description | A sum of exponential terms describing concentration over time | Describes curve shape without necessarily specifying a compartmental transfer structure |
The two-compartment model is a simpler mechanistic alternative when one peripheral exchange process is adequate. An empirical sum of exponentials can summarize profile shape without assigning each term a transfer path. The three-compartment model provides a mechanistic bookkeeping structure for two exchange paths, but its additional parameters should be interpreted cautiously. The question of how many compartments are justified by particular data requires a separate model-comparison assessment.
Common Mistakes
- Treating each compartment as a known organ. The model's peripheral compartments are abstract distribution spaces; their parameter values alone do not identify anatomical tissues.
- Calling distribution elimination. Drug leaving the central compartment may enter a peripheral compartment and later return. Only the clearance term removes drug from the modeled system.
- Interchanging rate constants and clearances. \(Q_2\) has units of L/h, whereas \(k_{12}=Q_2/V_1\) has units of inverse time. The units and the volume used in the denominator identify the quantity.
- Reading the terminal slope as the initial decline. The initial slope includes outward transfer into both empty peripheral compartments as well as elimination. The late slope is governed by the slowest remaining decay mode.
- Assuming three exponentials prove three physical spaces. The exponential shape is a mathematical consequence of this model structure, not by itself proof of a particular anatomical mechanism.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- For \(V_1=10\) L and \(Q_2=5\) L/h, calculate \(k_{12}\). State its units and distinguish it from \(Q_2\).
- At a particular time, \(C_1=3\) mg/L and \(C_2=4\) mg/L, with \(Q_2=2\) L/h. What is the net transfer rate between the central compartment and compartment 2, and in which direction does it act?
- Why does the initial central concentration slope after an IV bolus reflect distribution as well as elimination?
- What does a prolonged terminal decline allow a three-compartment model to represent, and what does it not establish about anatomy?
Solutions. (1) \(k_{12}=Q_2/V_1=5/10=0.5000\ \text{h}^{-1}\); \(Q_2\) is an intercompartmental clearance in L/h, whereas \(k_{12}\) is a first-order rate constant. (2) The net amount rate into compartment 2 is \(Q_2(C_1-C_2)=2(3-4)=-2\ \text{mg/h}\). The negative sign means the net transfer is 2 mg/h from compartment 2 to the central compartment. (3) Immediately after the bolus, the peripheral concentrations are zero, so drug leaves the central compartment by both outward transfer pathways as well as central elimination. (4) The model can represent a slow decay mode following earlier distribution phases; the shape does not identify a particular tissue or prove that the compartment corresponds to a literal anatomical space.