Modeling Nonlinear Formulation-Dependent Absorption

Why Model Absorption Nonlinearly?

A formulation can change not only how quickly drug enters the systemic circulation, but also how much reaches it. A constant first-order absorption rate may be inadequate when uptake pathways have finite capacity, when drug is lost before absorption, or when the effective absorption rate changes over time. These distinctions matter for model-based comparisons: a change in the concentration-time profile does not necessarily imply a change in total exposure, and a change in total exposure requires an explanation beyond absorption timing alone.

The models here describe absorption separately from disposition. Assume a one-compartment systemic model with volume $V$, clearance $CL$, and elimination rate constant $k_e=CL/V$. The central concentration $C(t)$ changes according to the systemic input rate $R_{\mathrm{abs}}(t)$:

\frac{dC(t)}{dt}=\frac{R_{\mathrm{abs}}(t)}{V}-k_eC(t).

This equation assumes linear, time-invariant elimination over the interval being modeled. It does not assume that absorption is linear. Formulation-specific parameters can describe differences in uptake capacity, uptake affinity, or time course. Those parameters should be interpreted as model features, not as proof of a particular biological mechanism.

Saturable Absorption With Competing Loss

Definition: In a saturable absorption model, the rate at which drug is absorbed approaches a finite maximum as the amount available for absorption increases. Let $A_g(t)$ be the amount in an absorption compartment, $V_{\max,f}$ the formulation-specific maximum absorption rate, and $K_{m,f}$ the amount at which the absorption rate is half of $V_{\max,f}$. A model with presystemic loss at rate constant $k_{\mathrm{loss}}$ is $$ R_{\mathrm{abs},f}(t)=\frac{V_{\max,f}A_g(t)}{K_{m,f}+A_g(t)},\qquad \frac{dA_g(t)}{dt}=-R_{\mathrm{abs},f}(t)-k_{\mathrm{loss}}A_g(t). $$

Here $A_g$ and $K_{m,f}$ are amounts, $V_{\max,f}$ is amount per unit time, and $k_{\mathrm{loss}}$ is inverse time. The model assumes that the dose begins in the absorption compartment, the parameters are constant during the interval, and loss is first-order. Saturation alone limits the instantaneous uptake rate; the loss term is what allows some drug to disappear without reaching systemic circulation.

The total amount absorbed can be derived without solving for $A_g(t)$ explicitly. Since $A_g$ decreases from the initial amount $D$ toward zero, divide the absorption rate by the rate of decrease in $A_g$, then integrate over the amount remaining:

A_{\mathrm{abs}} =\int_0^\infty R_{\mathrm{abs}}(t)\,dt =\int_0^D \frac{V_{\max,f}}{V_{\max,f}+k_{\mathrm{loss}}(K_{m,f}+a)}\,da =\frac{V_{\max,f}}{k_{\mathrm{loss}}} \ln\!\left( \frac{V_{\max,f}+k_{\mathrm{loss}}(K_{m,f}+D)} {V_{\max,f}+k_{\mathrm{loss}}K_{m,f}} \right).

In this derivation, $a$ is the amount remaining at an instant. The integrand is the fraction of the instantaneous amount leaving the compartment that enters systemic circulation rather than being lost. Under linear systemic elimination and complete concentration observation, the resulting AUC is $A_{\mathrm{abs}}/CL$. Thus, changing absorption parameters can change AUC when loss competes with uptake.

Worked Example: Comparing Saturable Uptake Between Formulations

Suppose two hypothetical formulations are given a 100 mg dose. Both are described by the saturable model with $K_m=20$ mg and $k_{\mathrm{loss}}=0.10$ h$^{-1}$. Let $V_{\max}=20$ mg/h for formulation A and $30$ mg/h for formulation B. Assume $CL=5.0$ L/h for both. Calculate predicted absorbed amount and AUC for each.

The model assumptions are the same for both formulations: constant parameters, first-order loss, saturable uptake, and common linear systemic clearance. For formulation A:

A_{\mathrm{abs,A}} =\frac{20}{0.10} \ln\!\left(\frac{20+0.10(20+100)}{20+0.10(20)}\right) =200\ln\!\left(\frac{32}{22}\right) =74.94\ \mathrm{mg}.

Therefore $AUC_A=74.94/5.0=14.99$ mg·h/L. For formulation B:

A_{\mathrm{abs,B}} =\frac{30}{0.10} \ln\!\left(\frac{30+0.10(20+100)}{30+0.10(20)}\right) =300\ln\!\left(\frac{42}{32}\right) =81.58\ \mathrm{mg}.

Thus $AUC_B=81.58/5.0=16.32$ mg·h/L, and the predicted AUC ratio is $16.32/14.99=1.089$. Under these hypothetical assumptions, formulation B has about 8.9% higher predicted AUC because its greater uptake capacity reduces the fraction lost before absorption. This model-based ratio is not, by itself, a bioequivalence decision.

Time-Varying Absorption

A different model allows the effective first-order absorption rate to vary with time or formulation. For example, a release process may produce an early phase with one absorption rate and a later phase with another. With a competing first-order loss process, let $k_{a,f}(t)$ be the absorption rate constant:

\frac{dA_g(t)}{dt} =-\{k_{a,f}(t)+k_{\mathrm{loss}}\}A_g(t), \qquad R_{\mathrm{abs},f}(t)=k_{a,f}(t)A_g(t).

This model assumes the stated rates apply to the whole absorption compartment at each time, loss remains first-order, and systemic elimination remains linear. If $k_a$ is constant during an interval of length $\Delta t$, then $q=k_a+k_{\mathrm{loss}}$, the amount remaining at the interval end is $A_0e^{-q\Delta t}$, and the amount absorbed during that interval is

A_{\mathrm{abs,interval}} =A_0\frac{k_a}{q}\left(1-e^{-q\Delta t}\right).

This follows by integrating $k_aA_0e^{-qt}$ from $0$ to $\Delta t$. The formula also makes clear that faster absorption can increase the fraction absorbed before competing loss, not merely advance its timing.

Worked Example: Piecewise Absorption Rates

Consider a hypothetical 100 mg dose with $k_{\mathrm{loss}}=0.05$ h$^{-1}$. Formulation A has $k_a=0.20$ h$^{-1}$ for the first 4 h, then $k_a=0.05$ h$^{-1}$. Formulation B has $k_a=0.10$ h$^{-1}$ for the first 4 h, then $k_a=0.10$ h$^{-1}$. Calculate total absorbed amounts, treating the second phase as continuing until the remaining amount is absorbed or lost. Assume $CL=5.0$ L/h.

For A in the first 4 h, $q=0.20+0.05=0.25$ h$^{-1}$. Its absorbed amount is

100\frac{0.20}{0.25}(1-e^{-0.25(4)}) =80(1-e^{-1}) =50.57\ \mathrm{mg}.

The amount remaining is $100e^{-1}=36.79$ mg. In the second phase, the eventual fraction of that amount absorbed is $k_a/(k_a+k_{\mathrm{loss}})=0.05/0.10=0.5$. The additional absorbed amount is $36.79(0.5)=18.39$ mg. Total absorption for A is $68.96$ mg.

For B, the first-phase $q=0.10+0.05=0.15$ h$^{-1}$:

100\frac{0.10}{0.15}(1-e^{-0.15(4)}) =66.67(1-e^{-0.6}) =30.08\ \mathrm{mg}.

The amount remaining is $100e^{-0.6}=54.88$ mg. In the second phase, the eventual absorbed fraction is $0.10/0.15=2/3$, giving $54.88(2/3)=36.59$ mg. Total absorption for B is $66.67$ mg. The predicted AUCs are $68.96/5.0=13.79$ and $66.67/5.0=13.33$ mg·h/L, respectively. A has faster early uptake but only slightly greater total exposure; the later rates and competing loss also determine the total.

When Nonlinear Timing Does Not Change AUC

A central interpretive point is that nonlinear absorption does not automatically mean nonlinear AUC. If there is no presystemic loss and the entire dose is eventually absorbed, then the total systemic input is the dose, regardless of its timing. For the saturable model with $k_{\mathrm{loss}}=0$, $dA_g/dt=-R_{\mathrm{abs}}$, so integration gives $A_{\mathrm{abs}}=D$. By integrating the systemic concentration equation from time zero until concentrations return to zero, the elimination term balances total input:

0=\frac{A_{\mathrm{abs}}}{V}-k_e AUC \quad\Longrightarrow\quad AUC=\frac{A_{\mathrm{abs}}}{CL}.

Worked Example: Saturation Changes Timing but Not Total Exposure

Suppose a hypothetical 100 mg dose undergoes saturable absorption with $V_{\max}=20$ mg/h and $K_m=20$ mg, with no competing loss. Compare the time for half the dose to be absorbed with the AUC when $CL=5.0$ L/h.

At half absorption, $A_g=50$ mg. Separating variables in $dA_g/dt=-V_{\max}A_g/(K_m+A_g)$ gives

t=\frac{D-A_g}{V_{\max}}+ \frac{K_m}{V_{\max}}\ln\!\left(\frac{D}{A_g}\right) =\frac{100-50}{20}+\frac{20}{20}\ln(2) =3.193\ \mathrm{h}.

Although uptake is saturable, the amount remaining approaches zero and the whole 100 mg is eventually absorbed under the no-loss assumption. Therefore $AUC=100/5.0=20.00$ mg·h/L. Saturation changes the input-time profile; it does not change AUC in this example. In practice, observation windows and competing loss can invalidate the assumption that all of the dose reaches systemic circulation.

Model Choice, Diagnostics, and Limitations

A saturable uptake model and a time-varying first-order model can both produce formulation-dependent concentration profiles, but they encode different explanations. Saturable uptake imposes a maximum rate as the amount available increases. A time-varying rate describes a changing effective absorption process without requiring a capacity limit. A model with a formulation-specific release function can also represent drug becoming available gradually before uptake; that release process should not be conflated with absorption unless the data support doing so.

Model feature	What it represents	Key implication
Saturable uptake	Uptake rate approaches $V_{\max,f}$ at high available amounts	May change total absorbed amount when loss competes
Time-varying $k_{a,f}(t)$	Effective uptake rate changes over time	Can alter timing and, with loss, total absorbed amount
Constant first-order uptake without loss	Uptake rate is proportional to amount remaining	If absorption is complete, AUC depends on dose and clearance, not $k_a$

Diagnostics should focus on whether the model describes both the observed profile and the intended comparison. Plot observed and predicted concentrations by formulation and time, examine residuals for systematic early-versus-late patterns, and inspect whether estimates are stable across reasonable starting values and model variants. A persistent early underprediction followed by late overprediction can indicate a misspecified input-time profile. Implausibly extreme or highly correlated $V_{\max,f}$ and $K_{m,f}$ estimates can signal that the data do not identify the saturation curve well. A good fit alone does not establish that one mechanistic explanation is uniquely supported.

Common failure modes include fitting a complex absorption function when sampling is too sparse to resolve its shape; treating a change in peak concentration as proof of a change in absorbed amount; and attributing an AUC difference to absorption when clearance also differs. Comparisons should use a common disposition model when that is justified by the design and data, and uncertainty in estimated parameters should be carried into predicted contrasts. If the alternative absorption structures make materially different predictions, sensitivity analysis is more informative than selecting a single structure solely because it fits marginally better.

These models also have limitations. A single absorption compartment may not represent transit through multiple regions, and a fitted time-varying rate may summarize rather than explain the underlying release and uptake processes. Parameters such as $V_{\max}$, $K_m$, and $k_a(t)$ may not be separately identifiable from concentration data over a limited sampling window. These considerations constrain interpretation: model-based exposure predictions are conditional on the structural model, its assumptions, and the information in the data.

Common Mistake

Wrong reasoning: “The absorption rate is saturable, so AUC must be lower.” Saturation limits an instantaneous rate, but it does not by itself remove drug from the system. If all drug is eventually absorbed and systemic clearance is linear, total AUC is dose divided by clearance. Correction: determine whether competing loss, incomplete absorption, or a finite observation window changes the total systemic input. Then calculate AUC from that input and clearance; do not infer an AUC change from slower or nonlinear timing alone.

Key takeaway: Formulation-dependent absorption can change the shape of a concentration-time curve, the total amount absorbed, or both. The decisive distinction is whether the model changes total systemic input; under linear disposition, AUC is total systemic input divided by clearance.

Check Your Understanding

Answer each question, then compare your reasoning with the short solutions below.

In the saturable model, what does $V_{\max,f}$ represent, and what are its units?
A 60 mg dose is completely absorbed with no competing loss. If $CL=3.0$ L/h, what is AUC, regardless of the absorption-time profile?
For a piecewise interval, $A_0=40$ mg, $k_a=0.15$ h$^{-1}$, $k_{\mathrm{loss}}=0.05$ h$^{-1}$, and $\Delta t=2$ h. Calculate the amount absorbed during the interval.
Why can a saturable uptake model with competing loss predict a different AUC from one without loss?
Name one diagnostic pattern that could indicate the assumed absorption-time profile is inadequate.

Solutions. (1) $V_{\max,f}$ is the limiting uptake rate at large available amounts; its units are amount per time, such as mg/h. (2) Complete absorption gives $A_{\mathrm{abs}}=60$ mg, so $AUC=60/3.0=20.0$ mg·h/L. (3) Here $q=0.15+0.05=0.20$ h$^{-1}$, so $A_{\mathrm{abs,interval}}=40(0.15/0.20)(1-e^{-0.20(2)})=30(1-e^{-0.4})=9.89$ mg. (4) Competing loss removes some amount before absorption, so total systemic input can be less than the administered dose and can depend on uptake parameters. (5) For example, systematic underprediction at early times followed by overprediction later suggests that the model’s predicted input timing may be wrong.

Model feature	What it represents	Key implication
Saturable uptake	Uptake rate approaches \(V_{\max,f}\) at high available amounts	May change total absorbed amount when loss competes
Time-varying \(k_{a,f}(t)\)	Effective uptake rate changes over time	Can alter timing and, with loss, total absorbed amount
Constant first-order uptake without loss	Uptake rate is proportional to amount remaining	If absorption is complete, AUC depends on dose and clearance, not \(k_a\)