Why Check Physicochemical Properties Together?
A PBPK parameter set can contain values that each look plausible in isolation but are difficult to reconcile as a group. For example, a reported pKa and logP imply a particular degree of ionization and, under a simple model, a pH-dependent logD. Solubility measured at that pH provides another check. If these values disagree substantially, the discrepancy may reflect a data-entry error, mismatched conditions, an inappropriate model assumption, or genuinely more complex chemistry.
These checks are screens, not replacements for measurements. Molecular weight (MW) can be checked against a chemical formula and the stated chemical form. Relationships among pKa, partitioning, and solubility are more conditional: they depend on the ionization model, pH, solid form, and assay method. A discrepancy is a reason to investigate, not a reason to silently alter a value.
State the Conditions Before Applying a Relationship
The relationships below are useful for a monoprotic weak acid or weak base at equilibrium. They assume that the measured pKa applies at the stated temperature and solution conditions; that the aqueous species are a neutral form and one ionized form; and that the neutral form partitions into the organic phase while the ionized form contributes negligibly to that phase. For the solubility equations, assume an equilibrium measurement in which the neutral intrinsic solubility is \(S_0\), the solid phase has not changed, and the solution is sufficiently dilute for ideal acid-base behavior. These assumptions can fail for compounds with multiple ionization sites, ion pairing, aggregation, complexation, changing solid form, or substantial non-ideal behavior.
For an acid \(HA\), the Henderson–Hasselbalch relationship gives the ratio of ionized to neutral aqueous forms. For a base \(B\), it gives the ratio of protonated to neutral forms. Combining these ratios with the assumptions above yields the following screening relationships:
Here \(S\) is total dissolved concentration at the stated pH and \(S_0\) is the intrinsic solubility of the neutral form. The equations are not universal laws for reported experimental values: they are consequences of the stated simplified model. In particular, they do not account for a salt dissolving with a different solid phase or for a kinetic solubility measurement that has not reached equilibrium.
Why the Screening Equations Have This Form
For the weak acid, let the neutral concentration be \([HA]\). The ionization ratio implies total dissolved concentration \(S=[HA]+[A^-]=[HA](1+10^{\mathrm{pH}-\mathrm{p}K_a})\). Under the ideal solubility assumptions, the neutral concentration at saturation is \(S_0\). Substitution gives the weak-acid solubility expression. For the weak base, total concentration is \([B]+[BH^+]\), and the corresponding ionization ratio gives the weak-base expression.
For distribution, let the organic-phase concentration be proportional to the neutral aqueous concentration, with ratio \(P\). If the ionized form contributes negligibly to the organic phase, total aqueous concentration is the neutral concentration multiplied by the same ionization factor. Therefore, the ratio of organic concentration to total aqueous concentration is \(P\) divided by that factor. Taking base-10 logarithms gives a practical check: the predicted \(\log D\) is \(\log P\) minus \(\log_{10}\) of the ionization factor.
Worked Checks
Worked Example: Estimating logD for a Weak Acid
Suppose a hypothetical monoprotic acid has \(\log P=3.00\), pKa \(=4.50\), and a measured \(\log D=0.150\) at pH 7.40. Assess whether the measurement is close to the simple model prediction.
The acid ionization factor is \(1+10^{\mathrm{pH}-\mathrm{p}K_a}\). The model predicts:
The measured value exceeds the prediction by \(0.150-0.09945=0.05055\), or about 0.051 log units. That difference corresponds to a ratio of \(10^{0.05055}\approx1.124\) in \(D\). This is a relatively small mismatch for an idealized screening equation, but its acceptability depends on measurement uncertainty and whether the values refer to comparable conditions. It should be recorded as a modest difference, not described as exact agreement.
Worked Example: A Larger logD Discrepancy
A hypothetical weak base is assigned \(\log P=2.00\) and pKa \(=8.00\). At pH 7.40, a dataset reports \(\log D=1.95\). Calculate the simple-model prediction and quantify the discrepancy.
For a base, the ionization factor is \(1+10^{\mathrm{p}K_a-\mathrm{pH}}\). Thus:
The reported value is \(1.95-1.303=0.647\) log units higher than the prediction. On the \(D\) scale, that is a factor of \(10^{0.647}\approx4.44\). The reported \(\log D\) remains below \(\log P=2.00\), so it does not violate the model’s upper bound, but the difference from the model prediction merits investigation. Check that pH, pKa, and logP refer to the same chemical form and suitable conditions; also examine assay methods, equilibration, and possible ionized-species partitioning. The calculation flags a discrepancy; it does not establish which input is wrong.
Worked Example: Checking Solubility and Converting Units
A hypothetical weak acid has pKa \(=4.50\) and intrinsic solubility \(S_0=0.0200\) mg/L. An equilibrium experiment at pH 7.40 reports a solubility of 80.0 mg/L. Under the ideal weak-acid model, is the reported value consistent with the supplied \(S_0\)?
The predicted total solubility is:
The reported-to-predicted ratio is \(80.0/15.91\approx5.03\). The measured value is therefore about five times the simple prediction. This is not, by itself, proof of an error: verify that the pH and temperature match, that \(S_0\) is an equilibrium intrinsic solubility for the same form, and that the measurement did not use a salt, cosolvent, or a different solid phase. If those conditions are confirmed, the size of the mismatch warrants checking the inputs and assay data.
Now suppose MW is 250 g/mol. The reported mass solubility can be converted to molar concentration, provided the reported mass refers to that same chemical form:
This conversion does not resolve the discrepancy, but it makes the concentration basis explicit. If a salt’s mass is reported while MW refers to the active moiety, the conversion must be adjusted to the stated chemical form.
Use Molecular Weight as a Cross-Check, Not a Solubility Law
MW is often the most direct property check: compare the stated value with the molecular formula, charge state, and chemical form used in the assay and model. A mismatch can arise because a record refers to a salt, hydrate, solvate, or parent moiety while another reports the mass of the whole material. Check units as well: g/mol and Da have the same numerical value for molecular mass, but mass concentration still needs consistent conversion before it is compared with molar concentration.
There is no general equation that determines aqueous solubility from MW alone. Nor does a high \(\log P\) impose a universal numerical upper limit on solubility. Lipophilicity, ionization, crystal packing, and experimental conditions all matter. A high pH-dependent solubility can be compatible with a lipophilic neutral form if ionization increases the total dissolved concentration. Conversely, a low measured solubility cannot be explained by logP alone. Treat a seemingly surprising combination as a prompt to inspect the full set of conditions and chemical identity.
Practical Audit and Failure Modes
Confirm that MW, pKa, lipophilicity, and solubility refer to the same compound form. Record whether solubility is a mass or molar concentration and note the units.
For logD and solubility, state pH and, where available, temperature, buffer, ionic strength, and whether the assay was an equilibrium or kinetic measurement.
Use the monoprotic equations only when the assumed ionization and partitioning behavior is plausible. Do not force them onto polyprotic or otherwise complex compounds without an appropriate model.
Calculate predicted values with unrounded intermediate results, compare on the original scale as well as the log scale where relevant, and trace each input to its measurement conditions before deciding whether a value needs correction.
A frequent mistake is to treat a screening equation as an exact identity: “The measured logD differs from the estimate, so the measurement must be wrong.” The correction is to state the model assumptions, quantify the mismatch, and check whether the chemical form and assay conditions satisfy them. The opposite mistake is to dismiss every inconsistency as assay noise. A large mismatch that persists after conditions and units are reconciled is important evidence that the property set needs review.
Compared with relying on a single property, a joint check can reveal inconsistencies across independent measurements. Compared with detailed mechanistic or empirical property models, these equations are transparent and quick but deliberately narrow. Their limitations include uncertainty in the pKa and \(S_0\), multiple ionization states, changes in solid form, non-ideal solutions, and non-negligible partitioning of charged species. They are best used to prioritize verification, not to generate a definitive replacement measurement.
Check Your Understanding
Answer each question using the assumptions stated in the tutorial. Brief solutions follow the questions.
- A weak base has \(\log P=1.50\) and pKa \(=7.00\). Using the simple model, calculate its predicted \(\log D\) at pH 7.00.
- A weak acid has \(S_0=0.010\) mg/L and pKa \(=5.00\). What is its predicted total solubility at pH 5.00?
- Why is a reported \(\log D\) value without a pH insufficient for a complete physicochemical comparison?
- A solubility value is reported in mg/L, but the model input is in mol/L. Which additional property is needed for conversion, and what identity issue must be checked?
- Under the stated distribution model, what should be checked if a measured \(\log D\) exceeds \(\log P\)?
Solutions. (1) The base ionization factor is \(1+10^{7.00-7.00}=2\), so \(\log D=1.50-\log_{10}(2)=1.50-0.3010=1.199\). (2) The acid solubility factor is \(1+10^{5.00-5.00}=2\), giving \(S=0.010(2)=0.020\) mg/L. (3) The ionization fraction, and therefore the distribution coefficient, depends on pH relative to pKa. (4) MW is needed to convert mass concentration to molar concentration; confirm that it refers to the same form whose mass is reported, such as parent compound versus salt. (5) The simple assumptions imply \(D\leq P\); check that both values use consistent conditions and definitions, then investigate measurement, identity, or model-assumption differences rather than treating the bound as universally exact.