Why a Return Pathway Matters
A plasma concentration profile may fall after a dose and then rise again. One possible explanation is enterohepatic recirculation: drug leaves the systemic circulation in bile, passes into the intestine, and some is reabsorbed into the systemic circulation. In a structural model, this is not the same as simply slowing elimination. It is a transfer out of the central compartment followed by a possible return of drug later.
A secondary peak can be a useful clue, but it does not by itself establish recirculation. Delayed absorption, distribution processes, changing physiology, and measurement variation can also create non-monotone profiles. The model should therefore encode a plausible pathway and be assessed against the timing and shape of the observed concentrations.
A Compartment Model for Biliary Transfer and Return
Let \(A\) be the amount of drug in the systemic central compartment, \(B\) the amount in a biliary-transit compartment, and \(R\) the amount in a return compartment representing drug available for intestinal reabsorption. Let \(k_e\) be the first-order rate constant for irreversible systemic elimination, \(k_b\) the rate constant for biliary transfer, \(k_t\) the rate constant for passage from biliary transit into the return compartment, \(k_r\) the reabsorption rate constant, and \(k_f\) the rate constant for loss from the return compartment, such as fecal loss. All rate constants have units of inverse time. Let \(V\) be the central volume, assumed constant, so that the measured concentration is \(C=A/V\).
Here \(u(t)\) is drug input into the systemic compartment, in amount per unit time. For an intravenous bolus, the post-dose equations use \(u(t)=0\), with the dose represented by the initial value of \(A\). The biliary route removes drug from \(A\), but \(k_bA\) is a transfer into \(B\), not necessarily irreversible elimination. Drug is lost irreversibly through \(k_eA\) and \(k_fR\) in this model.
This structure assumes constant first-order rates, a well-mixed amount in each compartment, and no unmodelled delay between the compartments beyond the transit represented by \(B\) and \(R\). It also assumes that reabsorbed drug enters the same central compartment used to describe systemic concentrations. These are modelling assumptions, not consequences of observing a second peak. If timing is poorly represented, additional transit structure or a different model may be needed.
How Return Can Produce a Secondary Peak
Because \(V\) is constant, the direction of change in concentration is the direction of change in central amount. After an input has ended, the central amount rises precisely when return flux exceeds the combined systemic elimination and biliary-transfer flux:
Initially, the central concentration may fall because \(A\) is large and little drug has yet reached \(R\). Later, drug transferred into \(B\) can pass into \(R\), increasing \(k_rR\). If that return flux becomes larger than the current loss flux from the central compartment, the concentration turns upward. As the return pool is depleted, the concentration can fall again. The compartment model therefore gives a mechanistic way to represent a fall, a later rise, and another decline.
A short mass-balance derivation clarifies why biliary transfer should not automatically be counted as total elimination. Add the three amount equations. The internal transfer terms cancel:
Thus, once input has stopped, total modelled drug amount decreases through irreversible elimination and return-compartment loss; transfer into bile alone does not reduce the total amount represented in these compartments. This accounting depends on the model boundary: if drug has left the represented system, that loss must appear as an irreversible outflow.
Worked Example: Probability of Reabsorption on the First Biliary Passage
Suppose, hypothetically, that \(k_e=0.25\ \mathrm{h}^{-1}\), \(k_b=0.50\ \mathrm{h}^{-1}\), \(k_r=0.30\ \mathrm{h}^{-1}\), and \(k_f=0.20\ \mathrm{h}^{-1}\). Assume the first-order competing-rate model above, and assume all drug reaching \(B\) eventually passes into \(R\). What fraction of drug in the central compartment is expected to reach the return compartment and then be reabsorbed on its first passage?
From the central compartment, biliary transfer and irreversible elimination are competing first-order routes. The fraction taking the biliary route is
In the return compartment, reabsorption and loss are competing first-order routes. The fraction reabsorbed, conditional on reaching \(R\), is
The first-passage fraction that follows both routes is their product: \(0.667\times0.600=0.400\). Therefore, under these assumptions, 40.0% of drug leaving the central compartment is expected to reach the return compartment and be reabsorbed on its first biliary passage. This is not the fraction that returns eventually after repeated cycles; that would require accounting for subsequent passages as well.
Reading a Concentration-Time Profile
A secondary peak is a local feature of the concentration-time profile: concentrations first decrease, later increase, and then decrease again. In sampled data, the exact turning time is usually unknown. A local increase between observation times shows that the profile rose over that interval; it does not locate the precise time of the maximum. A sequence of measurements around the rise and subsequent decline is more informative than one isolated high value.
Worked Example: Quantifying a Secondary Rise in a Hypothetical Profile
Suppose a hypothetical intravenous study gives the following concentrations in a participant. Assume the listed values are correctly measured plasma concentrations in \(\mathrm{mg/L}\).
| Time (h) | Concentration (mg/L) |
|---|---|
| 0 | 10.0 |
| 1 | 6.80 |
| 2 | 4.90 |
| 4 | 3.40 |
| 6 | 4.10 |
| 8 | 4.80 |
| 10 | 3.90 |
| 14 | 2.10 |
The finite-difference slope from 4 to 6 hours is
From 6 to 8 hours, it is \((4.80-4.10)/2=0.350\ \mathrm{mg\,L^{-1}\,h^{-1}}\). From 8 to 10 hours, it is \((3.90-4.80)/2=-0.450\ \mathrm{mg\,L^{-1}\,h^{-1}}\). The sampled profile therefore includes a clear secondary rise followed by a decline, with its highest sampled concentration in that later region at 8 hours. These calculations establish the observed pattern, not its cause. A return pathway is one mechanistic explanation to evaluate, rather than a conclusion proven by the profile alone.
Worked Example: Checking Whether Return Flux Can Reverse the Slope
At a hypothetical time after input has ended, suppose \(k_e=0.20\ \mathrm{h}^{-1}\), \(k_b=0.50\ \mathrm{h}^{-1}\), \(k_r=0.40\ \mathrm{h}^{-1}\), and \(V=5.0\ \mathrm{L}\). First consider \(A=20\ \mathrm{mg}\) and \(R=30\ \mathrm{mg}\). The central amount rate is
The concentration rate is \((-2.00\ \mathrm{mg/h})/(5.0\ \mathrm{L})=-0.400\ \mathrm{mg\,L^{-1}\,h^{-1}}\), so concentration is still falling. Now consider a later hypothetical state with \(A=15\ \mathrm{mg}\) and the same \(R=30\ \mathrm{mg}\). Then
The corresponding concentration rate is \(1.50/5.0=0.300\ \mathrm{mg\,L^{-1}\,h^{-1}}\), so concentration is rising at that instant. The change in direction follows from the changing balance of return and loss fluxes. These instantaneous calculations illustrate the peak condition; they are not a fitted trajectory or evidence that those particular compartment amounts occurred in a study.
Diagnostics, Alternatives, and Limits
Evidence for a return pathway should be evaluated across the profile and against the model's predicted timing. Useful checks include whether concentrations are sampled densely enough around the secondary rise, whether the fitted return flux can exceed central losses near the upturn, and whether the estimated transit and reabsorption rates are identifiable from the available data. A model that fits only because the return rates or compartment amounts take implausible values is not persuasive. Residual plots and predictions across the full profile can reveal systematic underprediction around a secondary peak or compensating overprediction elsewhere.
A single secondary high measurement may be assay variation or a recording problem. A broad, reproducible rise is more consistent with a structural feature, but is still not specific to enterohepatic recirculation. Delayed absorption can produce later input from a dosing depot; a distribution model can describe exchanges between systemic spaces; and a time-dependent process can alter the rate at which drug disappears. Each explanation makes different assumptions about the source of later concentration changes and should be compared using the dosing route, sampling schedule, physiological plausibility, and the complete concentration-time pattern.
| Model explanation | What supplies the later concentration change? | Key distinction |
|---|---|---|
| Enterohepatic recirculation | Drug transferred into bile and later reabsorbed | Requires a plausible return pathway and transit timing |
| Delayed absorption | Drug still entering from the administration depot | Depends on the input route and absorption process |
| Distribution exchange | Drug returning from another modelled systemic space | Does not require biliary transfer or intestinal reabsorption |
| Measurement variation | No systematic biological input is required | An isolated rise may not recur or follow a coherent time pattern |
The model is a simplification. It does not separately represent bile formation, intestinal transit details, binding, or the fraction of drug available for reabsorption unless those processes are added explicitly. Sparse sampling can leave return timing and rate constants weakly determined; different structures may then describe the observed concentrations similarly. More compartments are not automatically better: each added process needs data that can inform it and a clear physiological interpretation.
Common pitfall: “Biliary excretion is elimination, so a later peak must mean clearance changed.” In this model, biliary transfer \(k_bA\) is an outflow from the central compartment but an inflow to the biliary pathway. The correction is to distinguish transfer from irreversible loss: drug can return through \(k_rR\), and a secondary rise is possible when that return flux exceeds current central losses. Conversely, a secondary peak alone is not proof that such a return occurred.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions.
- If \(k_r=0.24\ \mathrm{h}^{-1}\) and \(k_f=0.16\ \mathrm{h}^{-1}\), what fraction of drug that reaches \(R\) is reabsorbed under the stated competing-rate model?
- At a time after input has stopped, \(k_e+k_b=0.60\ \mathrm{h}^{-1}\), \(A=10\ \mathrm{mg}\), \(k_r=0.30\ \mathrm{h}^{-1}\), and \(R=25\ \mathrm{mg}\). Is the central amount rising or falling, and at what rate?
- Why is \(k_bA\) not counted as an irreversible loss in the total amount balance?
- A profile has one high concentration between two lower measurements, but no other samples around that time. What can and cannot be concluded?
Solutions. (1) The reabsorbed fraction is \(k_r/(k_r+k_f)=0.24/(0.24+0.16)=0.600\), or 60.0%. (2) The central rate is \(-0.60(10)+0.30(25)=-6.0+7.5=1.50\ \mathrm{mg/h}\), so the central amount is rising at that instant. (3) Biliary transfer leaves \(A\) but enters \(B\); it cancels when the amount equations are summed. Only the model's irreversible outflows, \(k_eA\) and \(k_fR\), reduce the total represented amount after input stops. (4) The measurements show a sampled local increase if the high value exceeds its neighbours, but do not establish a reproducible secondary peak, its timing, or its mechanism. Additional observations and evaluation of alternative explanations are needed.