Sequential And Parallel Absorption

Why the Pathway Structure Matters

An absorption model describes how drug reaches the systemic circulation, not just how quickly the concentration rises. In a sequential structure, drug must pass through one input stage before reaching the next. In a parallel structure, the dose is divided among distinct pathways, and each pathway supplies drug to the systemic circulation independently. These structures can produce different concentration-time profiles even when they eventually deliver the same total amount of drug.

To isolate the effect of pathway structure, this tutorial uses a one-compartment disposition model with first-order elimination. Let \(A_c(t)\) be the amount of drug in the central compartment, \(C(t)=A_c(t)/V\) its concentration, \(V\) the apparent volume, and \(k_e=CL/V\) the elimination rate constant. Assume linear disposition, a single dose \(D\), no elimination from the absorption compartments, and complete eventual delivery of the dose to the central compartment. The absorption rate constants are positive and do not change over time.

Definition: Sequential absorption sends drug through two or more stages in series before systemic entry. Parallel absorption divides the dose into fractions that enter separate absorption pathways; each pathway supplies its own input to the systemic compartment. The fractions in a complete parallel model are nonnegative and sum to 1.

A Two-Step Sequential Model

Consider a dose initially in a first absorption compartment \(A_1\), which transfers drug to a second compartment \(A_2\) at rate constant \(k_1\). Drug leaves \(A_2\) and enters the central compartment at rate constant \(k_a\). The central amount is eliminated at rate constant \(k_e\). The corresponding equations are

\frac{dA_1}{dt}=-k_1A_1,\qquad \frac{dA_2}{dt}=k_1A_1-k_aA_2,\qquad \frac{dA_c}{dt}=k_aA_2-k_eA_c,

with \(A_1(0)=D\), \(A_2(0)=0\), and \(A_c(0)=0\). Solving the first two equations gives \(A_1(t)=De^{-k_1t}\) and, when \(k_1\ne k_a\),

A_2(t)=D\frac{k_1}{k_a-k_1}\left(e^{-k_1t}-e^{-k_at}\right).

The rate entering the central compartment is \(k_aA_2(t)\). The central amount is the accumulated input, with each past input reduced by elimination between its arrival time and the time of measurement. This convolution gives

C_{\mathrm{seq}}(t)=\frac{D}{V}\frac{k_1k_a}{k_a-k_1} \left[ \frac{e^{-k_1t}-e^{-k_et}}{k_e-k_1} -\frac{e^{-k_at}-e^{-k_et}}{k_e-k_a} \right].

This expression assumes the rate constants in its denominators are distinct. If any two rates coincide, the corresponding limiting expression is used; the model itself remains well defined. For example, equal successive rates produce terms involving \(t e^{-kt}\), rather than a singular concentration. More serial stages can be represented by adding a compartment and a transfer equation for each stage. Transit-compartment models use this kind of serial structure to create a gradual input profile.

Worked Example: Concentration After Two Serial Absorption Steps

Suppose a hypothetical 100 mg dose has \(D/V=5.00\ \mathrm{mg/L}\), with \(k_1=1.20\ \mathrm{h}^{-1}\), \(k_a=0.600\ \mathrm{h}^{-1}\), and \(k_e=0.200\ \mathrm{h}^{-1}\). Calculate the concentration at \(t=1.00\ \mathrm{h}\). The rates are positive and distinct, and the model assumes all dose passes through both stages before entering a linearly eliminated central compartment.

Substitute into the sequential formula:

C_{\mathrm{seq}}(1)=5.00\frac{(1.20)(0.600)}{0.600-1.20} \left[ \frac{e^{-1.20}-e^{-0.200}}{0.200-1.20} -\frac{e^{-0.600}-e^{-0.200}}{0.200-0.600} \right]\ \mathrm{mg/L}.

Using \(e^{-1.20}=0.301194\), \(e^{-0.600}=0.548812\), and \(e^{-0.200}=0.818731\), the two bracket terms are \((-0.517537)/(-1.00)=0.517537\) and \((-0.269919)/(-0.400)=0.674798\). Thus the bracket is \(0.517537-0.674798=-0.157261\), and the rate multiplier is \((0.720)/(-0.600)=-1.20\ \mathrm{h}^{-1}\). Therefore,

C_{\mathrm{seq}}(1)=5.00(-1.20)(-0.157261)=0.9436\ \mathrm{mg/L}.

The negative bracket and negative multiplier yield a positive concentration, as required. The intermediate stages delay systemic input relative to a direct single-step input.

Parallel First-Order Pathways

In a parallel model, fractions \(f_j\) of the dose enter separate absorption depots. For pathway \(j\), the initial amount is \(f_jD\), and drug leaves that depot at first-order rate \(k_{aj}\). With \(m\) pathways, \(\sum_{j=1}^{m}f_j=1\). The total rate entering the central compartment is the sum of the pathway-specific rates:

R_{\mathrm{par}}(t)=D\sum_{j=1}^{m} f_jk_{aj}e^{-k_{aj}t}.

For \(k_{aj}\ne k_e\), convolution with first-order central elimination gives

C_{\mathrm{par}}(t)=\frac{D}{V}\sum_{j=1}^{m} f_j\frac{k_{aj}}{k_{aj}-k_e} \left(e^{-k_et}-e^{-k_{aj}t}\right).

Each term represents one pathway’s contribution to concentration. A pathway with a larger absorption rate constant tends to supply its fraction earlier; a slower pathway can contribute for longer. These are tendencies, not direct guarantees about the observed terminal slope, which also depends on disposition and the other pathways.

Worked Example: Concentration From Two Parallel Pathways

For a hypothetical dose with \(D/V=5.00\ \mathrm{mg/L}\), suppose \(f_1=0.600\), \(k_{a1}=1.20\ \mathrm{h}^{-1}\), \(f_2=0.400\), \(k_{a2}=0.400\ \mathrm{h}^{-1}\), and \(k_e=0.200\ \mathrm{h}^{-1}\). The fractions are nonnegative and sum to 1; both absorption rates differ from the elimination rate. Calculate \(C(1.00\ \mathrm{h})\).

The pathway-specific rate ratios are \(1.20/(1.20-0.200)=1.20\) and \(0.400/(0.400-0.200)=2.00\). Thus,

C_{\mathrm{par}}(1)=5.00\left[ 0.600(1.20)(e^{-0.200}-e^{-1.20}) +0.400(2.00)(e^{-0.200}-e^{-0.400}) \right]\ \mathrm{mg/L}.

The first bracket contribution is \(0.720(0.818731-0.301194)=0.372626\); the second is \(0.800(0.818731-0.670320)=0.118729\). Their sum is \(0.491355\), so

C_{\mathrm{par}}(1)=5.00(0.491355)=2.457\ \mathrm{mg/L}.

This is higher at this time than the sequential example’s \(0.9436\ \mathrm{mg/L}\). That comparison reflects these particular parameter choices, not a rule that all parallel models produce higher concentrations.

Compare the Systemic Input Rates

Concentration is shaped by both input and elimination. Looking directly at the input rate can help distinguish the structures conceptually: sequential input depends on the amount reaching the final absorption stage, whereas parallel input is the sum of contributions from separately dosed pathways.

Worked Example: Input Rate at One Hour

Use the parameter values from the preceding examples and a hypothetical dose of \(D=100\ \mathrm{mg}\). For the sequential model, \(A_2(1)/D=[1.20/(0.600-1.20)](e^{-1.20}-e^{-0.600})=0.495235\). The systemic input rate is therefore

R_{\mathrm{seq}}(1)=k_aA_2(1) =(0.600)(100)(0.495235)=29.71\ \mathrm{mg/h}.

For the parallel model, the pathway rates add:

R_{\mathrm{par}}(1)=100\left[ (0.600)(1.20)e^{-1.20} +(0.400)(0.400)e^{-0.400} \right]\ \mathrm{mg/h}.

The two contributions are \(100(0.216860)=21.69\ \mathrm{mg/h}\) and \(100(0.107251)=10.73\ \mathrm{mg/h}\). Their sum is \(32.41\ \mathrm{mg/h}\). The calculations show how these specific structures and parameters yield different input rates at one hour; they do not establish which structure generated any real concentration data.

Assumptions, Diagnostics, and Model Choice

The equations above require first-order transfer or absorption, constant rate parameters, linear one-compartment disposition, known dose and volume scaling, and complete eventual absorption. They also treat each dose fraction as entering its assigned pathway at time zero. A transit chain is appropriate when the structural hypothesis is that input proceeds through successive stages; a parallel model is appropriate when the dose is hypothesized to split across distinct routes of input. The parameters do not have the same interpretation across these structures: \(k_1\) and \(k_a\) govern successive transfers, while \(f_j\) and \(k_{aj}\) describe allocated fractions and their pathway rates.

In practice, a delayed rise, a shoulder, or more than one apparent input phase may motivate testing richer structures than single first-order absorption. Such features do not prove a transit chain or parallel pathways: disposition, sampling times, between-subject variation, and measurement error can also affect the observed profile. Useful checks include plotting observed and predicted concentrations on linear and logarithmic scales, examining residuals over time, and checking whether fitted fractions remain between 0 and 1 and sum to 1. Systematic early- or late-time residual patterns suggest that the model may not capture the observed shape. Strongly correlated pathway fractions and rate constants, unstable estimates, or large uncertainty are signs that the data may not distinguish the proposed pathways.

The main alternatives include a single first-order absorption process, transit-compartment absorption, and a parallel model. A single process is simpler but may not represent a distributed input pattern. Transit compartments impose serial passage and can smooth or delay input. Parallel pathways allow a dose split and can represent a mixture of faster and slower input. More complex models can fit more shapes, but complexity is not evidence that each modeled pathway corresponds to a separately established biological route. Compare plausible structures using fit, diagnostics, parameter precision, and scientific interpretability rather than selecting a model from the visual profile alone.

Key result: For the stated linear disposition assumptions, both structures deliver the full dose eventually, so their total exposure is the same: \(AUC_{0-\infty}=D/CL\). They can nevertheless predict different concentrations and input rates at particular times because the timing and organization of systemic input differ.

To justify the exposure result, integrate the central-compartment balance \(dA_c/dt=R(t)-CL\,C(t)\) from time zero to infinity. Since \(A_c(0)=A_c(\infty)=0\), the integrated change is zero. Complete absorption gives \(\int_0^\infty R(t)\,dt=D\), so \(CL\int_0^\infty C(t)\,dt=D\), and therefore \(AUC_{0-\infty}=D/CL\). This equality depends on complete absorption and linear, time-invariant clearance; it does not imply that the concentration-time profiles are interchangeable.

Common mistake: Treating two sequential rate constants as if they described two fractions of the dose. In a serial model, the same drug molecules pass through both stages; in a parallel model, the fractions split at the outset.
Another pitfall: Assuming that a shoulder in observed concentrations proves parallel absorption. A shoulder is a feature to investigate, not a unique signature; competing structures and sampling limitations must be considered.

Check Your Understanding

Answer each question, then compare with the brief solutions below.

In a model where all dose passes from \(A_1\) to \(A_2\) and then into the central compartment, is the structure sequential or parallel? Why?
A parallel model has three fractions \(f_1=0.25\), \(f_2=0.50\), and \(f_3=0.25\). Does it satisfy the fraction constraint? What does each fraction represent?
Why can two complete-absorption models have the same \(AUC_{0-\infty}\) but different concentrations at one hour?
Name two signs in a model fit that could indicate the data do not support estimating separate parallel pathways reliably.

Solutions. (1) Sequential: each molecule must pass through \(A_1\) and \(A_2\) in order. (2) Yes; the fractions are nonnegative and sum to \(1.00\). Each is the share of dose allocated to its pathway. (3) Under linear disposition and complete absorption, the integrated input and clearance determine AUC, but different input timing changes the concentration at a particular time. (4) Examples include strongly correlated fractions and rate constants, unstable estimates, large parameter uncertainty, or systematic residual patterns.