Concentration Profiles in the Parallel-Tube Model

Why Concentration Changes Along the Liver

The parallel-tube model represents hepatic blood flow as many similar sinusoidal paths arranged in parallel. Blood entering each path encounters eliminating tissue as it moves toward the outlet. Consequently, concentration is not assumed to be uniform along the path: it declines progressively as drug is removed. This spatial profile is useful for interpreting what the model means by an outlet concentration and for distinguishing it from the well-stirred model’s single, mixed hepatic concentration.

The profile below describes total drug concentration in hepatic blood. It is a steady or quasi-steady description: the inlet concentration is treated as stable over the transit being considered. The model is an idealization, not a claim that every real sinusoid has the same flow or geometry.

Here \(x\) is a fraction of the path, not a distance in centimetres. The parameter \(\lambda\) is dimensionless: both \(f_uCL_{\mathrm{int}}\) and \(Q\) have units of volume per time. The exponent therefore has no units. If the model is parameterized using a clearance that already incorporates binding, it must not be multiplied by \(f_u\) a second time.

Deriving the Exponential Profile

The profile follows from a local mass balance. Assume that intrinsic clearance is distributed uniformly along the normalized path, that unbound concentration is \(f_uC(x)\), and that elimination is linear in unbound concentration. A small path segment of length \(dx\) then has intrinsic clearance \(CL_{\mathrm{int}}dx\). Under the parallel-tube representation, the resulting differential balance is

The negative sign indicates that elimination lowers concentration in the direction of flow. Dividing by \(Q C(x)\), and using \(\lambda=f_uCL_{\mathrm{int}}/Q\), gives a separable equation. Integrating from the inlet, where \(C=C_{\mathrm{in}}\) at \(x=0\), to any position \(x\) yields

Exponentiating gives \(C(x)=C_{\mathrm{in}}e^{-\lambda x}\). This derivation requires positive concentration and constant parameters along the path. It also assumes no drug is added within the liver, no axial mixing between positions, and no appreciable accumulation in a path segment during the steady transit described by the equation. With \(CL_{\mathrm{int}}=0\), \(\lambda=0\) and concentration stays at its inlet value; with positive clearance, the profile decreases monotonically.

Reading the Profile and Its Consequences

The exponential form has several direct interpretations. The fraction of inlet concentration remaining at position \(x\) is \(C(x)/C_{\mathrm{in}}=e^{-\lambda x}\). The fractional decrease from inlet to that position is therefore \(1-e^{-\lambda x}\). For a segment from \(x=a\) to \(x=b\), the concentration ratio is

Thus, equal path lengths have equal proportional decreases under the model, even though their absolute concentration decreases differ: a given proportional loss removes more concentration when the segment begins at a higher concentration. If the path volume is uniform per unit \(x\), the spatial average concentration is \(C_{\mathrm{in}}(1-e^{-\lambda})/\lambda\) for \(\lambda>0\). This average is neither the inlet nor the outlet concentration. It is a separate summary of the concentration distribution along the path.

Progressive Depletion Versus Uniform Mixing

The well-stirred liver model assumes uniform mixing: at steady state, one hepatic concentration represents the liver and is also the outlet concentration. The parallel-tube model instead assigns a different concentration to each position, with a continuous decrease from inlet to outlet. The two descriptions can be compared under the same \(Q\), \(f_u\), \(CL_{\mathrm{int}}\), and inlet concentration, while recognizing that they are distinct model assumptions.

For example, use the values from the first worked example. The parallel-tube outlet is \(8.00e^{-1.50}=1.785\ \mathrm{mg/L}\). For the well-stirred model, the steady-state mass balance is \(Q(C_{\mathrm{in}}-C_{\mathrm{out}})=f_uCL_{\mathrm{int}}C_{\mathrm{out}}\); solving gives \(C_{\mathrm{out}}=C_{\mathrm{in}}/(1+\lambda)\). Thus, its outlet prediction is \(8.00/(1+1.50)=3.200\ \mathrm{mg/L}\). The values differ because one model represents progressive depletion along a path and the other represents uniform mixing.

For positive \(\lambda\), the parallel-tube model predicts a lower outlet concentration than the well-stirred model at matched inputs. The comparison follows from \(e^\lambda>1+\lambda\) for \(\lambda>0\); taking reciprocals gives \(e^{-\lambda}<1/(1+\lambda)\). In this example, the fraction removed from inlet to outlet is \(1-1.7850413/8.00=0.7769\) for the parallel-tube model and \(1-3.200/8.00=0.6000\) for the well-stirred model. These are model-based predictions, not evidence that one model must fit every liver better.

Diagnostics, Limitations, and Common Pitfalls

The model predicts a smooth, positive, monotonically decreasing profile. A proposed profile that rises downstream, becomes negative, or has an abrupt jump is inconsistent with the stated equation unless an additional source, a change in conditions, or a measurement problem is present. Under the ideal model, a plot of \(\ln[C(x)/C_{\mathrm{in}}]\) against \(x\) is a straight line through zero with slope \(-\lambda\). Curvature or substantial position-dependent departures from that pattern can indicate that constant linear elimination or uniform path assumptions are inadequate; noisy measurements can also produce apparent departures.

A common mistake is to use the well-stirred outlet concentration as though it were the concentration at every point in a parallel-tube liver. That erases the progressive depletion the parallel-tube model is designed to represent. A second mistake is to treat \(x\) as a physical distance without converting it to the model’s normalized path coordinate. A third is to multiply by \(f_u\) twice when the clearance parameter already represents unbound-drug effects.

The profile is limited by its assumptions. It does not represent unequal sinusoidal transit times, mixing along the path, changing binding, saturable elimination, or concentration-dependent changes in clearance. When those features matter, the simple exponential profile may be a poor description. The well-stirred model is a useful alternative when uniform mixing is the intended approximation; neither idealization should be selected solely because its equation is convenient. In practice, the choice should be consistent with the biological question and the information available to assess the assumptions.