Neutral-Drug Partitioning Across Tissue Phases

Why Water and Lipid Contributions Matter

A neutral drug can distribute into tissue water and into lipid-like tissue material. The tissue-to-plasma partition coefficient, \(K_p\), reflects the combined amount in these tissue phases relative to plasma concentration. When the drug has little preference for lipid, tissue water may account for much of the predicted tissue concentration. As lipophilicity increases, the lipid contribution can become much larger, even when the tissue lipid fraction is modest.

This tutorial builds a simple, explicit two-contributing-phase calculation. It is useful for understanding how composition and lipophilicity affect a prediction, not for claiming that octanol is identical to tissue lipid or that every tissue can be represented by only two phases. The introductory material on tissue composition provides the physiological context for the fractions used here.

Definition: In the two-contributing-phase approximation, \(f_{w,t}\) and \(f_{l,t}\) are the tissue-volume fractions assigned to water and an effective lipid-like phase, respectively. \(P_{l:w}\) is the dimensionless concentration ratio of neutral drug in that lipid-like phase to its concentration in tissue water. The model first predicts tissue concentration relative to unbound plasma concentration, then converts to a total-plasma \(K_p\) using the plasma unbound fraction \(f_{u,p}\).

The convention matters. Let \(C_{p,u}\) be the unbound plasma concentration and \(C_{p,\mathrm{total}}\) the total plasma concentration. The plasma unbound fraction is \(f_{u,p}=C_{p,u}/C_{p,\mathrm{total}}\). In this tutorial, \(K_p\) means total drug concentration per tissue volume divided by total drug concentration per plasma volume. The corresponding ratio referenced to unbound plasma is denoted \(K_{p,u}\). Both are dimensionless concentration ratios.

Deriving the Two-Phase Expression

Assume the neutral drug is at equilibrium between plasma water and tissue water, so tissue-water concentration equals \(C_{p,u}\). Assume also that the concentration in the effective tissue lipid-like phase is \(P_{l:w}\) times the tissue-water concentration. A tissue volume contains a water fraction \(f_{w,t}\) and a lipid-like fraction \(f_{l,t}\); the calculation adds the amount contributed by each phase per unit tissue volume.

C_{t,\mathrm{total}} \approx f_{w,t}C_{p,u}+f_{l,t}C_{l} =f_{w,t}C_{p,u}+f_{l,t}P_{l:w}C_{p,u}

Divide by \(C_{p,u}\) to obtain \(K_{p,u}\). Since \(C_{p,u}=f_{u,p}C_{p,\mathrm{total}}\), division by total plasma concentration introduces the factor \(f_{u,p}\):

K_{p,u}\approx f_{w,t}+f_{l,t}P_{l:w} \qquad\text{and}\qquad K_p\approx f_{u,p}\left(f_{w,t}+f_{l,t}P_{l:w}\right)

Every fraction and partition ratio in these expressions is dimensionless, so the resulting \(K_p\) is dimensionless. Tissue fractions are expressed per unit tissue volume. The model includes only the contributions assigned to water and lipid-like material; it does not separately represent binding to tissue proteins, phospholipid subcompartments, or other constituents.

For a neutral compound, the octanol-water partition coefficient \(P_{ow}\) is related to the base-10 logarithm of partitioning by \(\log P=\log_{10}(P_{ow})\). A simple teaching calculation may use \(P_{l:w}\approx P_{ow}=10^{\log P}\) as a proxy. This is an approximation: octanol is not a direct measurement of the partition ratio into every tissue’s lipid-like material. Crucially, the model uses the ratio \(10^{\log P}\), not the logarithm \(\log P\) itself.

Assumptions: The drug is neutral over the relevant conditions; the unbound drug is at equilibrium between plasma water and tissue water; partitioning is linear over the concentration range considered; the supplied tissue fractions represent the tissue of interest; and the lipid-like phase can be represented by a single effective \(P_{l:w}\). The calculation also assumes that omitted tissue constituents do not make a separately modeled contribution. Violations can make this estimate inaccurate even when the arithmetic is correct.

Worked Calculations

Worked Example: A Low-Lipophilicity Neutral Drug

Hypothetical inputs. Consider a neutral compound with \(\log P=0.30\), \(f_{w,t}=0.70\), \(f_{l,t}=0.05\), and \(f_{u,p}=0.80\). The two tissue fractions are plausible illustrative inputs and sum to 0.75; the remaining tissue volume is not assigned a separate contribution in this approximation.

First convert the logarithm to a partition ratio:

P_{l:w}\approx 10^{0.30}=1.995

The water contribution to \(K_{p,u}\) is \(0.70\). The lipid-like contribution is \(0.05\times1.995=0.09975\). Therefore:

K_{p,u}\approx0.70+0.05(1.995)=0.7998

Converting to a total-plasma reference gives:

K_p\approx0.80(0.7998)=0.6398

Under these assumptions, the predicted total tissue concentration is about \(0.640\) times the total plasma concentration. The lipid-like contribution is smaller than the water contribution, consistent with this compound’s relatively low lipophilicity and the small lipid-like fraction.

Worked Example: A Highly Lipophilic Neutral Drug

Hypothetical inputs. Consider a neutral compound with \(\log P=2.50\), \(f_{w,t}=0.65\), \(f_{l,t}=0.08\), and \(f_{u,p}=0.90\). As before, the fractions are illustrative and only the specified water and lipid-like contributions are included.

The partition ratio is much larger than the log value:

P_{l:w}\approx10^{2.50}=316.2

The lipid-like contribution to \(K_{p,u}\) is \(0.08\times316.2=25.30\), compared with \(0.65\) from water. Thus:

K_{p,u}\approx0.65+0.08(316.2)=25.95

After converting to the total-plasma reference:

K_p\approx0.90(25.95)=23.35

The model predicts a total tissue concentration about \(23.4\) times the total plasma concentration. This large estimate is driven by the product of the lipid-like fraction and the high partition ratio, not by the fraction alone. It is a model result under the stated assumptions, not evidence that a real tissue will necessarily attain this value.

Worked Example: The Same Drug in Two Hypothetical Tissues

Hypothetical inputs. A neutral compound has \(\log P=1.00\) and \(f_{u,p}=0.85\). Compare tissue A, with \(f_{w,t}=0.75\) and \(f_{l,t}=0.03\), with tissue B, with \(f_{w,t}=0.25\) and \(f_{l,t}=0.60\). These fractions are illustrative; the respective unassigned fractions are not separately modeled.

Here \(P_{l:w}\approx10^{1.00}=10.00\). For tissue A:

K_{p,u,A}\approx0.75+0.03(10.00)=1.05,\qquad K_{p,A}\approx0.85(1.05)=0.8925

For tissue B:

K_{p,u,B}\approx0.25+0.60(10.00)=6.25,\qquad K_{p,B}\approx0.85(6.25)=5.313

The compound and plasma binding are held constant, yet the estimated \(K_p\) differs because the tissue fractions differ. Tissue B’s much larger lipid-like fraction gives a larger lipid contribution. This comparison illustrates why a drug property alone cannot determine tissue partitioning.

When Does the Lipid-Like Term Dominate?

The two contributions to \(K_{p,u}\) are \(f_{w,t}\) and \(f_{l,t}P_{l:w}\). The lipid-like term is larger than the water term precisely when:

f_{l,t}P_{l:w}>f_{w,t} \quad\Longleftrightarrow\quad P_{l:w}>\frac{f_{w,t}}{f_{l,t}} \quad\Longleftrightarrow\quad \log P>\log_{10}\left(\frac{f_{w,t}}{f_{l,t}}\right)

The last inequality uses \(P_{l:w}\approx10^{\log P}\) and assumes \(f_{l,t}>0\). For the inputs in the low-lipophilicity example, \(f_{w,t}/f_{l,t}=0.70/0.05=14\), so the lipid-like term exceeds the water term only when \(\log P>\log_{10}(14)\approx1.146\). The example’s \(\log P=0.30\) is below this threshold, whereas a sufficiently lipophilic compound could be above it.

This comparison is about contributions to the model, not a universal classification of compounds or tissues. A small lipid-like fraction can still matter greatly if the partition ratio is large, while a large lipid-like fraction may contribute little for a compound that partitions weakly into it.

Checks, Failure Modes, and Alternatives

Before interpreting a prediction, check that the inputs and reference concentration are consistent. The tissue fractions should be dimensionless volume fractions for the intended tissue, and \(f_{u,p}\) should correspond to the same drug and conditions as the plasma concentration used in \(K_p\). Convert \(\log P\) to \(10^{\log P}\) before multiplying by a fraction. Because the two terms are nonnegative for valid fractions and a positive partition ratio, the model’s \(K_{p,u}\) should also be nonnegative.

A particularly useful diagnostic is to report the separate water and lipid-like contributions, not only their sum. If a high predicted \(K_p\) is almost entirely due to \(f_{l,t}P_{l:w}\), verify both the tissue lipid fraction and the lipophilicity measurement. A one-unit increase in \(\log P\) multiplies the proxy partition ratio by ten, so modest uncertainty in logP can produce a large change when the lipid-like term dominates. Also check the magnitude against the intended application: an extreme estimate may signal an input or model mismatch, though an extreme value by itself does not prove an arithmetic error.

Common failure modes include using \(\log P\) directly in the equation, substituting a distribution coefficient measured at a pH where the compound is ionized, and omitting the plasma unbound-fraction conversion while labeling the result a total-plasma \(K_p\). The estimate also becomes unreliable if equilibrium between plasma water and tissue water is not appropriate, if partitioning is concentration-dependent, or if a neutral-compound approximation is applied to a drug with substantial ionization. Real tissues contain additional binding environments, and octanol-water partitioning is only a proxy for lipid-like partitioning.

The main alternatives answer related but different needs. Directly measured tissue-to-plasma ratios can capture the combined effects present in the measurement, but require suitable data and may depend on sampling and concentration basis. A simple water-plus-lipid estimate is transparent and helps identify which inputs drive a prediction, but omits other tissue interactions. More detailed mechanistic approaches represent additional phases or binding processes; they may be more appropriate when the compound’s charge state or tissue interactions make the two-phase assumptions inadequate. No approach should be treated as reliable merely because it produces a numerical value: assumptions and input provenance still need review.

Common pitfall: A learner sees \(\log P=2\) and inserts \(2\) as the lipid-to-water ratio. That gives \(f_{l,t}\times2\), which uses the logarithm instead of the partition ratio. The correction is \(P_{l:w}\approx10^2=100\), followed by multiplication by \(f_{l,t}\). A separate reference-basis check is then needed to decide whether to multiply by \(f_{u,p}\) to obtain total-plasma \(K_p\).

Key takeaway: In this approximation, neutral-drug \(K_p\) is the plasma-unbound fraction multiplied by the sum of a tissue-water contribution and a lipid-like contribution. Lipophilicity enters through \(10^{\log P}\), so the lipid term can rise rapidly as logP increases; the result remains conditional on the phase, equilibrium, and reference-basis assumptions.

Check Your Understanding

Answer each question, then compare with the solutions below.

For \(\log P=1.5\), what partition ratio is used as the lipid-like proxy?
If \(f_{w,t}=0.60\), \(f_{l,t}=0.04\), \(P_{l:w}=10\), and \(f_{u,p}=0.75\), calculate \(K_{p,u}\) and total-plasma \(K_p\).
For the same fractions in question 2, is the lipid-like contribution greater than the water contribution?
Why is it incorrect to call \(f_{w,t}+f_{l,t}P_{l:w}\) the total-plasma \(K_p\) when \(f_{u,p}\ne1\)?
Name one assumption whose failure could make the two-contributing-phase estimate misleading.

Solutions. (1) \(P_{l:w}\approx10^{1.5}=31.62\), not \(1.5\). (2) \(K_{p,u}\approx0.60+0.04(10)=1.00\), and \(K_p\approx0.75(1.00)=0.750\). Both ratios are dimensionless. (3) No. The lipid-like contribution is \(0.04(10)=0.40\), which is less than the water contribution \(0.60\). (4) The sum is referenced to unbound plasma concentration; converting to total plasma concentration requires multiplication by \(f_{u,p}\). (5) For example, the estimate may mislead if the drug is ionized under the relevant conditions, if tissue-water equilibrium is not appropriate, or if omitted tissue binding makes a substantial contribution.