Why Model an Absorption Lag?
After an oral or depot dose, drug may not begin entering the systemic circulation immediately. A delay can arise from the dosage form, the route, or the time required for drug to become available for absorption. A structural model can represent this delay with an absorption lag time, while keeping the absorption process after the delay distinct from the delay itself.
The lag time is not the absorption rate constant. A lag shifts the start of absorption later; the rate constant \(k_a\) describes how quickly absorption proceeds once it has started. This distinction matters when interpreting profiles: a later onset and a slower rise are not necessarily the same feature.
Single-Dose Model and Its Derivation
Consider a one-compartment model with first-order elimination and first-order absorption from an absorption depot. Assume a dose \(D\) is given at time zero, the fraction \(F\) is available to the systemic circulation, the apparent volume is \(V\), and \(k_a\) and \(k_e\) are the absorption and elimination rate constants. Assume these parameters are constant over the dose, the dose enters the depot at once, and the lag is fixed. The model treats the dose as remaining unavailable for absorption until \(t_{\mathrm{lag}}\).
Before the lag has elapsed, this dose contributes no systemic drug. After the lag, let \(u=t-t_{\mathrm{lag}}\) be the time since absorption began. The depot amount then declines exponentially, so the input rate is \(F D k_a e^{-k_a u}\). The central amount follows from the balance of input and first-order elimination:
Solving this linear differential equation gives the central amount \(A(u)\), and dividing by \(V\) gives concentration. Thus, for \(k_a\ne k_e\):
The formula is the usual first-order oral concentration profile evaluated at the shifted time \(u\), not at \(t\). Its exponential terms are dimensionless; \(F D/V\) has concentration units, and the rate-constant ratio is dimensionless. If \(k_a=k_e\), the expression has a finite limit: \(C(t)=(F D/V)k_a u e^{-k_a u}\) for \(u>0\).
Worked Example: Concentration After a Single Delayed Dose
Suppose a hypothetical dose has \(F D/V=4.8\) mg/L, \(k_a=1.2\) h\(^{-1}\), \(k_e=0.20\) h\(^{-1}\), and \(t_{\mathrm{lag}}=2.0\) h. Calculate the concentration at 3.0 h after dosing.
The observation is after the lag, with elapsed absorption time \(u=3.0-2.0=1.0\) h. The two rate constants differ, so the single-dose equation applies directly:
At any time \(t\le 2.0\) h, the modeled concentration contribution from this dose is zero. The calculated value at 3.0 h reflects one hour of absorption and disposition after the lag, not three hours.
The peak time follows by differentiating the post-lag profile with respect to \(u\). Setting the derivative to zero yields \(k_a e^{-k_a u}=k_e e^{-k_e u}\), and hence, when \(k_a\ne k_e\):
The lag shifts the peak time by \(t_{\mathrm{lag}}\); it does not change the peak concentration in this ideal single-dose model, because the profile after absorption begins is otherwise unchanged. This result assumes the same dose, disposition, absorption rate, and bioavailability, with no other dose contributing.
Worked Example: Peak-Time Shift
Use \(k_a=1.2\) h\(^{-1}\), \(k_e=0.20\) h\(^{-1}\), and \(t_{\mathrm{lag}}=2.0\) h. Find the modeled peak time and compare it with the no-lag peak time.
The time from absorption onset to peak is:
With the lag, \(t_{\max}=2.0+1.792=3.792\) h after dosing. With no lag and the same other parameters, the peak is at 1.792 h. The model therefore predicts a 2.0 h shift in peak time, while retaining the same peak concentration for an isolated dose.
Repeated Doses: Apply the Lag to Each Dose
For repeated dosing, each administration has its own delayed absorption start. Under linear disposition and identical dose-specific parameters, the total concentration is the sum of the contributions from all doses whose lag times have elapsed. A convenient expression for doses \(D_j\) given at times \(t_j\) is:
A dose with \(t\le t_j+t_{\mathrm{lag}}\) contributes zero at time \(t\). Once its lag has elapsed, its contribution depends on the time since that dose's delayed absorption start. The equation assumes the same lag for every dose; if the lag varies across administrations, each dose needs its own lag parameter. The superposition also requires linear pharmacokinetics over the dose range being modeled.
Worked Example: Two Repeated Doses With a Lag
Suppose hypothetical doses are given at 0 and 12 h, with \(t_{\mathrm{lag}}=2.0\) h for each. Let \(F D/V=4.8\) mg/L per dose, \(k_a=1.2\) h\(^{-1}\), and \(k_e=0.20\) h\(^{-1}\). Calculate the concentration at 15 h, assuming no other doses.
At 15 h, the dose at time 0 has been absorbing for \(15-0-2=13\) h, and the dose at 12 h has been absorbing for \(15-12-2=1\) h. Both contributions are included:
The first contribution is small because the earlier dose has been undergoing elimination for longer. The second contribution is larger because it has only been absorbing for one hour. Omitting the earlier dose would understate the total concentration; treating both doses as if absorption began at administration would use the wrong elapsed times.
Interpreting, Checking, and Comparing Lag-Time Models
A lag-time model is useful when the data support a distinct delay before drug input. It can separate that delay from the subsequent absorption rate, which is especially valuable when early samples are available. But the parameter is a structural assumption, not a label that automatically explains any late rise or unexpectedly low early concentration.
Useful checks include plotting concentrations against the recorded time since dose, reviewing dose and sample timestamps, and comparing early observations with the model-predicted zero-input period. If measured concentrations attributable to the dose reliably precede the fitted lag, the assumed onset is inconsistent with the data. Conversely, if all samples occur well after the lag, the data may provide little information about its precise value. Sparse early sampling, assay limits, uncertain dose times, and variability in actual absorption onset can all make \(t_{\mathrm{lag}}\) weakly identified or misleading.
A common pitfall is to interpret a larger \(t_{\mathrm{lag}}\) as slower absorption. The correction is to distinguish the onset shift from the post-onset rate: \(t_{\mathrm{lag}}\) moves the profile along the time axis, whereas \(k_a\) controls its shape after input begins. In a single-dose profile, the formula for \(t_{\max}\) makes this separation explicit.
| Modeling choice | What it represents | Important distinction |
|---|---|---|
| No lag, first-order absorption | Absorption begins at the dose time and then proceeds at rate \(k_a\). | Does not represent a positive onset delay. |
| First-order absorption with a lag | Absorption begins after a fixed delay, then proceeds at rate \(k_a\). | Shifts the profile without changing its post-onset shape under the stated assumptions. |
| Alternative input structures | For example, transit-compartment absorption or zero-order drug input. | Represent different input patterns; they are not interchangeable interpretations of a fixed lag. |
A fitted lag can also absorb errors in recorded dose time or sampling time. Such a fit may improve the numerical match while assigning the delay to the wrong cause. Before adding or interpreting the parameter, verify the time origin and dosing records and inspect whether the model captures both the onset and the subsequent profile. A single fixed lag is also limited when delays vary substantially between doses or individuals; that limitation calls for a model suited to the variation, not an assumption that one common delay explains every profile.
Check Your Understanding
Try each question, then check the solutions below.
- A hypothetical dose has \(F D/V=3.0\) mg/L, \(k_a=0.80\) h\(^{-1}\), \(k_e=0.10\) h\(^{-1}\), and \(t_{\mathrm{lag}}=1.5\) h. Calculate its concentration at 2.5 h.
- For \(k_a=1.2\) h\(^{-1}\), \(k_e=0.20\) h\(^{-1}\), and a lag of 0.5 h, calculate the single-dose peak time.
- Using \(F D/V=4.8\) mg/L, \(k_a=1.2\) h\(^{-1}\), and \(k_e=0.20\) h\(^{-1}\), two doses occur at 0 and 10 h with a 2 h lag each. Calculate the total concentration at 13 h.
- Does increasing the lag change the modeled peak concentration after an isolated dose if all other model parameters remain fixed? Explain.
- Why can a lag estimate be unreliable when all samples are collected long after dose administration?
Solutions. (1) The elapsed absorption time is \(u=2.5-1.5=1.0\) h. Therefore, \(C=3.0(0.80/0.70)(e^{-0.10}-e^{-0.80})=3.429(0.904837-0.449329)=1.562\) mg/L. (2) The time from absorption onset to peak is \(\ln(6)/1.0=1.792\) h, so \(t_{\max}=0.5+1.792=2.292\) h after the dose. (3) The two elapsed absorption times at 13 h are 11 h and 1 h. Their contributions are \(5.76(e^{-2.2}-e^{-13.2})\approx0.6382\) mg/L and \(5.76(e^{-0.2}-e^{-1.2})\approx2.9810\) mg/L. The total is approximately \(3.619\) mg/L. (4) No. Under the isolated-dose assumptions, increasing the lag shifts the time of the peak but leaves the post-onset profile, including its maximum, unchanged. (5) Late samples may describe the profile after the onset delay without locating when absorption began. With little information near the onset, different lag values can fit similarly, so the estimate may be weakly identified.