Rodgers-Rowland Kp Prediction for Ionizable Drugs

Why Predict Kp From Composition?

A tissue-to-plasma partition coefficient, \(K_p\), is the ratio of total drug concentration in a tissue to total drug concentration in plasma at equilibrium, using consistent concentration units and reference conditions. For a PBPK model, this ratio helps relate the amount of drug in a tissue to its plasma concentration. When direct tissue measurements are unavailable, a composition-based prediction offers a mechanistic estimate from tissue volumes and drug properties.

The Rodgers–Rowland method represents tissue as water, neutral lipid, and phospholipid-containing phases. Drug can distribute differently among these phases: ionization affects its aqueous distribution, while lipophilicity and interactions with tissue constituents affect its partitioning into lipid phases. Plasma protein binding matters because a tissue’s free drug is related to unbound, rather than total, plasma drug. These inputs make the method useful for ionizable compounds, but also mean that a prediction depends on the quality and compatibility of the inputs.

Definition: In a Rodgers–Rowland composition-based prediction, \(K_p\) is calculated by summing tissue-phase contributions referenced to unbound plasma drug, then multiplying by the plasma unbound fraction \(f_{u,p}\) to express the result relative to total plasma drug. The phase contributions incorporate tissue composition and the drug’s partitioning and ionization properties.

Calculation Framework and Assumptions

Write the calculation in a compact form as

K_p = f_{u,p}\left(V_{\mathrm{aq}}Q_{\mathrm{aq}}+ V_{\mathrm{NL}}Q_{\mathrm{NL}}+ V_{\mathrm{PL}}Q_{\mathrm{PL}}\right).

Here, \(V_{\mathrm{aq}}\) is the tissue aqueous volume, including the aqueous compartments represented in the chosen tissue-composition input; \(V_{\mathrm{NL}}\) and \(V_{\mathrm{PL}}\) are neutral-lipid and phospholipid volume fractions or volumes on the same tissue-volume basis. Each \(Q\) is an effective phase concentration ratio relative to unbound plasma concentration. It includes the relevant partitioning and, where applicable, the ionization correction. Consequently, each product \(V_jQ_j\) is dimensionless when tissue volumes are expressed as fractions.

The effective phase terms are not interchangeable with raw measured logP or logD values. For example, neutral-lipid partitioning is informed by lipophilicity and the amount of neutral drug available to partition. The aqueous term depends on pKa and the relevant pH values. The phospholipid term reflects the model’s treatment of phospholipid association, which can be particularly important for some ionizable compounds. The method’s exact parameterization should be used consistently; do not substitute a value from a different partitioning model without checking its definition.

Assumptions: The prediction represents equilibrium partitioning; drug concentrations are on a consistent total-tissue and total-plasma basis; tissue composition and pH are represented by the selected inputs; the pKa and lipophilicity describe the relevant drug species; and the plasma unbound fraction is appropriate to the conditions of interest. Phase contributions are treated as additive in the model. A prediction is conditional on these assumptions, not a direct measurement of tissue concentration.

For an ionizable drug, an aqueous correction can be expressed as a ratio of total dissolved concentrations at tissue and plasma pH, assuming the same neutral-species activity in the two aqueous environments. For a monoprotic weak acid, the total-to-neutral factor at pH is \(1+10^{\mathrm{pH}-\mathrm{p}K_a}\); for a monoprotic weak base, it is \(1+10^{\mathrm{p}K_a-\mathrm{pH}}\). Thus, the aqueous tissue-to-plasma ratio is the tissue factor divided by the plasma factor. This relation follows from the Henderson–Hasselbalch species ratio; it does not by itself specify the lipid-phase contributions.

Worked Predictions

Worked Example: Weak Acid in a Hypothetical Tissue

Suppose a hypothetical weak acid has \(\mathrm{p}K_a=4.5\), \(\log P=3.38\), and \(f_{u,p}=0.80\). Assume plasma pH is 7.4 and tissue aqueous pH is 7.0. For illustration, let \(V_{\mathrm{aq}}=0.75\), \(V_{\mathrm{NL}}=0.020\), and \(V_{\mathrm{PL}}=0.050\). Use the model’s effective phase terms \(Q_{\mathrm{NL}}=3.015\) and \(Q_{\mathrm{PL}}=2.393\), calculated from the hypothetical drug and tissue inputs. The example illustrates how to combine terms; it is not a real compound or study.

For the aqueous term, the weak-acid ratio is

Q_{\mathrm{aq}}= \frac{1+10^{7.0-4.5}}{1+10^{7.4-4.5}} = \frac{1+10^{2.5}}{1+10^{2.9}} \approx 0.3989.

Using the unrounded powers gives a ratio of approximately \(0.398864\). The aqueous contribution is therefore \(0.75(0.398864)=0.299148\), or \(0.2991\) to four significant figures. The neutral-lipid contribution is \(0.020(3.015)=0.06030\), and the phospholipid contribution is \(0.050(2.393)=0.11965\). The sum before applying plasma binding is \(0.299148+0.06030+0.11965=0.479098\). Thus,

K_p=0.80(0.479098)=0.383278\approx 0.3833.

The predicted total tissue concentration is about \(0.3833\) times the total plasma concentration at the stated conditions. The corrected aqueous ratio rounds to \(0.3989\), and the weighted aqueous contribution rounds to \(0.2991\); retaining the unrounded ratio avoids a rounding-induced change in the final result.

Worked Example: Weak Base and an Aqueous pH Effect

Consider a hypothetical weak base with \(\mathrm{p}K_a=8.2\), plasma pH 7.4, and tissue aqueous pH 7.0. Suppose its tissue has \(V_{\mathrm{aq}}=0.70\), \(V_{\mathrm{NL}}=0.030\), and \(V_{\mathrm{PL}}=0.060\), and its plasma unbound fraction is \(0.25\). For this illustration, assume the model inputs provide \(Q_{\mathrm{NL}}=1.20\) and an effective phospholipid term \(Q_{\mathrm{PL}}=12.0\).

For a weak base, the aqueous ratio is

Q_{\mathrm{aq}}= \frac{1+10^{8.2-7.0}}{1+10^{8.2-7.4}} = \frac{1+10^{1.2}}{1+10^{0.8}} \approx \frac{16.8489}{7.3096} \approx 2.305.

The weighted aqueous contribution is \(0.70(2.305)=1.6135\). The neutral-lipid and phospholipid contributions are \(0.030(1.20)=0.0360\) and \(0.060(12.0)=0.7200\), respectively. Therefore, the sum of the phase contributions is \(1.6135+0.0360+0.7200=2.3695\), and

$$ K_p=0.25(2.3695)=0.5924. $$

The predicted total tissue concentration is \(0.5924\) times total plasma concentration under these hypothetical conditions. The aqueous ratio exceeds one because this weak base is more ionized in the more acidic tissue water under the stated equilibrium calculation. The phospholipid term is an input to this worked calculation, not a derivation of its drug-specific parameterization.

Worked Example: Effect of Plasma Unbound Fraction

Suppose a hypothetical tissue and drug combination gives a sum of phase contributions of \(1.20\), with all tissue, pH, and drug partition inputs held fixed. If \(f_{u,p}=0.10\), then \(K_p=0.10(1.20)=0.120\). If instead \(f_{u,p}=0.50\), then \(K_p=0.50(1.20)=0.600\).

The second prediction is five times the first because the plasma unbound fraction is five times larger and the bracketed tissue term is unchanged. This is a consequence of the model’s total-plasma reference: the comparison does not imply that changing plasma binding also changes the tissue’s intrinsic phase affinities. In real applications, binding and other inputs may not vary independently.

Why the Structure Works

The calculation follows from changing the reference concentration in two stages. First, each tissue phase is represented by its concentration relative to unbound plasma drug. Multiplying that ratio by the phase’s tissue volume gives its contribution to total tissue concentration relative to unbound plasma concentration. Adding the phase contributions gives the total tissue-to-unbound-plasma ratio. Second, total plasma concentration is related to unbound plasma concentration by \(C_{u,p}=f_{u,p}C_{p,\mathrm{total}}\). Substitution yields

\frac{C_{t,\mathrm{total}}}{C_{p,\mathrm{total}}} = \frac{C_{t,\mathrm{total}}}{C_{u,p}} \frac{C_{u,p}}{C_{p,\mathrm{total}}} = \left(\sum_j V_jQ_j\right)f_{u,p}.

The left side is \(K_p\). This also explains why applying \(f_{u,p}\) twice is an error: it is already the single conversion from an unbound-plasma reference to a total-plasma reference.

Check the inputs.
Confirm tissue composition, pH, pKa, lipophilicity, and plasma unbound fraction are defined for compatible conditions and concentration bases.

Calculate effective phase terms.
Use the Rodgers–Rowland parameterization for aqueous, neutral-lipid, and phospholipid contributions, including the ionization corrections required for the compound.

Weight and sum the phases.
Multiply each effective phase term by its tissue volume fraction and add the contributions.

Convert to total plasma reference.
Multiply the sum once by \(f_{u,p}\), then report \(K_p\) as a dimensionless ratio and retain adequate precision during intermediate calculations.

Diagnostics, Alternatives, and Limitations

A useful diagnostic is to inspect the contribution of each phase rather than only the final \(K_p\). If one term dominates, check the corresponding tissue volume and drug-property input first. An unexpectedly large lipid contribution may reflect a lipophilicity value on the wrong scale or an ionization correction applied inconsistently. An aqueous term far from the expected range can indicate a weak-acid versus weak-base sign error, incorrect pH assignment, or a pKa entered in the wrong units. Also verify that volume fractions refer to the same tissue-volume basis; adding a percentage to a fraction, for example, creates a hundredfold error.

The earlier material on Poulin–Theil prediction provides a useful comparison. Both approaches use tissue composition to estimate partitioning, but Rodgers–Rowland explicitly accommodates ionization and phospholipid-associated contributions in its treatment of ionizable drugs. The simpler neutral-compound framework is not a substitute when ionization materially affects phase distribution. Conversely, greater mechanistic detail does not guarantee a more accurate prediction if pKa, lipophilicity, pH, composition, or binding inputs are uncertain.

Rodgers–Rowland predictions are model-based estimates. They do not automatically capture active transport, tissue-specific metabolism, nonequilibrium distribution, saturable binding, or barriers that prevent the assumed equilibrium. A predicted \(K_p\) should therefore be assessed against available data and the intended PBPK use, rather than treated as a measured constant that must apply under every condition.

Wrong reasoning: “The tissue has a large lipid fraction, so its \(K_p\) must be large.” Correction: The drug’s effective phase partitioning also matters; a large phase volume does not guarantee strong drug association.
Wrong reasoning: “I used unbound plasma concentration in the denominator, so I still need to multiply by \(f_{u,p}\) twice to get \(K_p\).” Correction: Multiply the unbound-plasma-referenced tissue sum by \(f_{u,p}\) exactly once to obtain the total-plasma-referenced coefficient.
Wrong reasoning: “A predicted \(K_p\) is a probability or a direct measure of tissue exposure.” Correction: \(K_p\) is a dimensionless concentration ratio under specified assumptions; it does not by itself describe time course or establish a clinical tissue concentration.

Key takeaway: Rodgers–Rowland prediction combines tissue phase volumes with drug-specific effective partition terms, then applies \(f_{u,p}\) once to express tissue concentration relative to total plasma concentration. The result is only as reliable as its inputs and equilibrium assumptions.

Check Your Understanding

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

A tissue has phase contributions \(0.40\), \(0.10\), and \(0.20\), and \(f_{u,p}=0.50\). What is its predicted \(K_p\)?
For a weak acid, which pH belongs in the numerator of the tissue-to-plasma aqueous ratio?
If a calculation already gives the tissue-to-unbound-plasma ratio, what operation converts it to \(K_p\) relative to total plasma?
Name two inputs or assumptions to inspect if the predicted phospholipid contribution dominates unexpectedly.

Solutions: (1) \(K_p=0.50(0.40+0.10+0.20)=0.350\). (2) Tissue pH is in the numerator: \(Q_{\mathrm{aq}}=[1+10^{\mathrm{pH}_t-\mathrm{p}K_a}]/[1+10^{\mathrm{pH}_p-\mathrm{p}K_a}]\). (3) Multiply once by \(f_{u,p}\). (4) Check the phospholipid volume and the drug’s pKa, lipophilicity, and effective phospholipid parameterization, as well as the consistency of the input conditions.