Initial Values for Absorption Parameters

Why Absorption Starting Values Need a Different Calculation

Back-calculating clearance and volume uses information about elimination and exposure. Oral absorption adds further parameters: the absorption rate, a possible delay before absorption begins, and the fraction of the dose that reaches systemic circulation. These parameters affect different features of a concentration-time profile, but those features may be weakly observed when samples are sparse. The goal is therefore to construct plausible, internally consistent starting values, not to treat a sample peak as a precise measurement of the true peak or to claim that every parameter is identified by the data.

The calculations below use a single-dose, one-compartment model with first-order absorption and elimination. It is a transparent source of initial values when its assumptions are reasonably compatible with the dosing and concentration data. It is not a substitute for checking the model against the full observed profile.

Model and assumptions: For an oral dose $D$, let $F$ be absolute bioavailability, $k_a>0$ the first-order absorption rate constant, $k>0$ the first-order elimination rate constant, $V$ the apparent volume of distribution, and $t_{\mathrm{lag}}\geq 0$ the lag time. Assume linear, time-invariant one-compartment disposition; first-order absorption beginning after the lag; no relevant dose-dependent kinetics; and concentrations measured on a consistent time scale. For $t\geq t_{\mathrm{lag}}$ and $k_a\ne k$, $$ C(t)=\frac{F D k_a}{V(k_a-k)} \left[e^{-k(t-t_{\mathrm{lag}})}-e^{-k_a(t-t_{\mathrm{lag}})}\right]. $$ Before the lag, the model predicts no absorption-related concentration contribution from that dose. The rate constants have units of inverse time, $V$ has units of volume, and $C(t)$ has units of amount per volume.

The peak time is determined by the shape of this curve. Write $u=t-t_{\mathrm{lag}}$ for time since absorption begins. For unequal rates, differentiating the concentration with respect to $u$ and setting the derivative to zero gives the post-lag peak time. The multiplicative constant in the derivative is nonzero when $k_a\ne k$; it need not be positive.

\frac{dC}{du} = \frac{F D k_a}{V(k_a-k)} \left[-k e^{-ku}+k_a e^{-k_a u}\right]=0 \quad\Longrightarrow\quad k_a e^{-k_a u}=k e^{-ku}.

Taking logarithms and rearranging yields $u=\ln(k_a/k)/(k_a-k)$. Thus the modeled time of peak concentration after the dose is the lag plus this post-lag time. The peak equation is a relation among $k_a$, $k$, and $t_{\mathrm{lag}}$: peak timing alone does not independently determine all three.

Peak-time result: For $k_a\ne k$, the model’s time of peak concentration is $$ t_{\max}=t_{\mathrm{lag}}+\frac{\ln(k_a/k)}{k_a-k}. $$ When $k_a=k$, the continuous limit of the concentration curve is $C(t)=FDk u e^{-ku}/V$ for $u\geq0$, whose peak occurs at $u=1/k$. The peak-time formula therefore has the limit $t_{\mathrm{lag}}+1/k$.

Constructing an Initial Absorption Rate

If an elimination-rate starting value $k$ is available, and lag time and peak time can be provisionally specified, the peak-time relation can be solved numerically for $k_a$. In practice, this is a one-dimensional root-finding problem: choose a candidate $k_a>0$, calculate the post-lag peak time, and adjust $k_a$ until it matches the assumed or estimated post-lag peak time. A numerical solver should be constrained to positive rates.

A useful check follows directly from the formula. With fixed $k$, if $k_a>k$, the post-lag peak occurs before $1/k$; if $k_a<k$, it occurs after $1/k$. When the candidate absorption and elimination rates are very close, the peak time approaches $1/k$. These checks can reveal a rate entered with the wrong units or a numerical solution on the wrong side of $k$.

Worked Example: Using Peak Timing to Seed the Absorption Rate

Hypothetical data: Suppose a single oral dose has a reported peak at 4.0 h after dosing. Use a provisional lag time of 0.50 h and an elimination-rate starting value $k=0.150\ \mathrm{h}^{-1}$. Treat the reported peak time as an approximate target, not as an exact continuous-time peak.

The target time since absorption begins is $4.0-0.50=3.50$ h. We seek $k_a$ satisfying $$ \frac{\ln(k_a/0.150)}{k_a-0.150}=3.50\ \mathrm{h}. $$

Try $k_a=0.485\ \mathrm{h}^{-1}$. Substitution gives $$ \frac{\ln(0.485/0.150)}{0.485-0.150} = \frac{\ln(3.233)}{0.335} \approx \frac{1.174}{0.335} \approx 3.50\ \mathrm{h}. $$

Thus $k_a=0.485\ \mathrm{h}^{-1}$ is a reasonable initial value under the stated assumptions. Since $k_a>k$, the post-lag peak should occur before $1/k=6.67$ h, which is consistent with the calculated 3.50 h. The result depends on the provisional lag and elimination rate; changing either generally changes the absorption-rate value that matches the same total peak time.

Lag Time and the Limits of Sampled Peak Times

The model’s $t_{\max}$ is a continuous-time property. A study usually observes concentrations only at scheduled times, so the time of the largest recorded concentration is a sampled maximum, not necessarily the continuous-time peak. If the largest observation is at 8 h, for example, the true peak could occur before or after 8 h. Sparse observations may not even establish which side of the peak a particular sample lies on.

A lag value should therefore be based on relevant design or formulation information and early concentration-time behavior, rather than selected solely to force agreement with one sampled maximum. A known dosing-to-absorption delay can provide a defensible initial value. If the early observations do not resolve the onset, lag and absorption rate may trade off: a longer delay followed by faster absorption can resemble a shorter delay followed by slower absorption at the observed times.

Worked Example: A Sparse Profile Gives Only a Conditional Peak-Based Value

Hypothetical data: Concentrations are sampled at 4, 8, and 16 h after dosing, and the largest of these three values is at 8 h. Suppose $k=0.100\ \mathrm{h}^{-1}$ is used as a provisional value and, only for illustration, assume $t_{\mathrm{lag}}=0$. Taking the sampled maximum as if it were the continuous peak gives a post-lag target of 8 h.

A candidate rate $k_a=0.154\ \mathrm{h}^{-1}$ gives $$ \frac{\ln(0.154/0.100)}{0.154-0.100} = \frac{\ln(1.54)}{0.054} \approx \frac{0.4318}{0.054} =8.00\ \mathrm{h}. $$

This produces a conditional starting value, not evidence that the true absorption rate is known to three decimal places. The data have no observations between 4 and 8 h or between 8 and 16 h, so they do not locate the continuous peak precisely. A nonzero lag would also change the post-lag target: for example, a 1 h lag would make it 7 h rather than 8 h. The sparse profile therefore supports a range of plausible starting values more honestly than a falsely precise point estimate.

Using Exposure to Construct Bioavailability

For the stated linear model, integrating concentration over time gives the area under the curve (AUC). The expression simplifies because the difference of exponentials integrates to a difference of reciprocal rates. With $CL=kV$, the result is dose times bioavailability divided by clearance. A lag shifts the concentration curve in time but does not change its total AUC, provided the full profile is captured.

\begin{aligned} \mathrm{AUC}_{\mathrm{oral}} &= \frac{F D k_a}{V(k_a-k)} \left(\frac{1}{k}-\frac{1}{k_a}\right)\\ &= \frac{F D}{V k} = \frac{F D}{CL}. \end{aligned}

Consequently, if clearance is available independently, an initial absolute bioavailability value can be calculated as $F=\mathrm{AUC}_{\mathrm{oral}}CL/D$, with dose, AUC, and clearance expressed in compatible units. Oral exposure by itself does not generally separate $F$ from $CL$: the AUC identifies their ratio $F/CL$. If both are unknown and only oral data are available, assigning a value to one does not make the other independently identified.

Worked Example: Seeding Absolute Bioavailability From AUC

Hypothetical data: An oral dose is $D=200$ mg, its estimated AUC is $12.0\ \mathrm{mg\,h/L}$, and an independently supported clearance starting value is $CL=12.0\ \mathrm{L/h}$. Assume linear disposition and an AUC that adequately represents the full exposure.

Then $$ F_{\mathrm{start}} = \frac{\mathrm{AUC}_{\mathrm{oral}}CL}{D} = \frac{(12.0\ \mathrm{mg\,h/L})(12.0\ \mathrm{L/h})}{200\ \mathrm{mg}} = \frac{144\ \mathrm{mg}}{200\ \mathrm{mg}} =0.720. $$

The units cancel, as required for a fraction. The starting value corresponds to 72.0% absolute bioavailability under the assumed model and clearance. If this is intended to be absolute bioavailability, a value above 1 would prompt a review of dose and AUC units, the clearance assumption, and the model. It would not be appropriate simply to truncate an implausible result without investigating its cause.

Checks, Alternatives, and Failure Modes

Before using these values in an estimation run, check that the assumptions behind each calculation are plausible and that the units are consistent. The following distinctions are particularly important:

Observed maximum versus modeled peak: The largest sampled concentration is not automatically the continuous $t_{\max}$. Sampling times around the peak determine how informative that observation is.
Lag versus absorption rate: Both can shift the apparent peak. Sparse early samples may not distinguish a longer delay from a different absorption rate.
Bioavailability versus clearance: Oral AUC supplies $F/CL$, not $F$ alone. An independently informed clearance or suitable reference is needed to construct an absolute $F$ value.
Rate units: If peak times are in hours, $k$ and $k_a$ must be in $\mathrm{h}^{-1}$. Mixing minutes and hours can create rates that are wrong by a factor of 60.
Model mismatch: Delayed release, multiple absorption phases, dose-dependent kinetics, or multicompartment disposition may produce a profile that this simple model cannot describe adequately.

When a suitable intravenous reference is available under comparable linear conditions, dose-normalized oral and intravenous AUCs can be used to estimate absolute bioavailability. Relative bioavailability can instead compare two extravascular formulations; it is a ratio relative to the reference formulation and need not be at most 1. Those comparisons still depend on compatible conditions and reliable exposure estimates. If only oral data are available, an independently supported clearance is the key information that permits a separate absolute $F$ starting value.

Starting values should be judged by whether they are plausible and allow estimation to proceed, not by whether they exactly reproduce a noisy sampled peak or AUC. A value near a numerical boundary, unstable estimates under small changes to the assumed lag, or a fitted absorption rate that repeatedly moves far from its initial value can indicate weak information, poor scaling, or model mismatch. Inspect the time units, dose history, timing of early samples, and exposure calculation before making the starting value more elaborate.

Key takeaway: Use peak timing to construct a conditional $k_a$ value only after specifying $k$ and a plausible lag. Use AUC to construct $F$ only when clearance is independently informed. Sparse sampling weakens both calculations; a starting value is a numerical aid, not proof of identifiability.

Check Your Understanding

Answer each question, then compare with the solutions below.

If $k=0.20\ \mathrm{h}^{-1}$ and $k_a>k$, should the post-lag peak occur before or after $1/k$?
An oral profile has a sampled maximum at 6 h. Why is it not generally valid to set the model’s continuous $t_{\max}$ equal to 6 h?
An oral AUC is $5.0\ \mathrm{mg\,h/L}$, the dose is 50 mg, and clearance is independently estimated as $8.0\ \mathrm{L/h}$. Calculate the absolute bioavailability starting value.
With only oral AUC data and no independent information on clearance, which combination of $F$ and $CL$ is identified under the linear model?

Solutions. (1) Before $1/k=5.00$ h, because $k_a>k$ implies $\ln(k_a/k)/(k_a-k)<1/k$. (2) The observed maximum is only the largest concentration at scheduled sample times; the continuous peak may fall between samples, and sparse sampling may not locate it. (3) $F_{\mathrm{start}}=(5.0\ \mathrm{mg\,h/L})(8.0\ \mathrm{L/h})/(50\ \mathrm{mg})=40\ \mathrm{mg}/50\ \mathrm{mg}=0.800$, or 80.0%. (4) AUC identifies $F/CL$, since $\mathrm{AUC}_{\mathrm{oral}}/D=F/CL$; it does not separate $F$ and $CL$ without additional information.