Why Physicochemical Measurements May Disagree
A drug property is often reported as though it were a single, condition-free number. In practice, an assay measures a property under specified conditions, using a particular method and reporting convention. Two values can therefore differ because the underlying drug behaves differently under the tested conditions, because the methods do not measure exactly the same quantity, or because one result contains an experimental or reporting error.
For PBPK parameter sourcing, the first task is not to choose the value that appears most plausible. It is to determine whether the measurements refer to the same chemical species and comparable conditions. A pKa measured at one temperature and ionic strength is not automatically interchangeable with a value measured at another. A logD value is tied to pH as well as to the partitioning procedure. Solubility depends not only on pH but also on solid form, equilibration, and the concentration basis used to report the result.
The introductory material on pKa, ionization, logP and logD, intrinsic solubility, pH-dependent solubility, and salt forms provides the property definitions needed here. This tutorial focuses on reconciling measured values: identifying the conditions that matter, standardizing the reporting basis, and deciding what the evidence permits you to conclude.
A Common Basis for Comparison
For each reported result, record the analyte form, property definition, method, temperature, pH (where relevant), ionic strength or buffer composition, equilibration protocol, and reporting units. For solubility, also record the solid form and whether the measurement is equilibrium or kinetic. Missing metadata are not evidence that conditions were standard; they are an uncertainty that should remain visible in the assessment.
| Property | Conditions to check | Comparison question |
|---|---|---|
| pKa | Temperature, ionic strength, solvent composition, method, and species assignment | Does each value describe the same ionization step for the same chemical form? |
| LogD | pH, temperature, buffer and ionic strength, phase composition, and method | Were the distribution measurements made at the same pH and with comparable phase procedures? |
| Solubility | pH, temperature, ionic strength, solid form, equilibration time, and separation method | Do the values describe the same equilibrium state and concentration basis? |
| All three | Units, analyte identity, replicates, and reporting or rounding conventions | Could a basis conversion or reporting difference explain the apparent gap? |
Temperature and ionic strength deserve attention even when the stated pH is identical. They can affect equilibria and measured activities; the size and direction of a change are substance- and method-dependent. A pH value is also an operational measurement tied to the solution and procedure, not a guarantee that two buffers of different composition create identical chemical environments. Consequently, there is no universal correction factor that converts every literature value to a chosen reference condition.
Interpreting pKa and LogD Across Conditions
A pKa comparison begins by checking the ionization step and the chemical form. Polyprotic compounds may have multiple pKa values, and tautomeric or salt-form assignments can complicate which transition a method reports. Even if two values are labelled with the same nominal pKa, differences in temperature, ionic strength, solvent content, or analytical method can produce different measured results. A discrepancy should therefore be described alongside the measurement conditions, not reduced to a claim that one value must be wrong.
LogD is especially easy to miscompare because it depends on pH for an ionizable compound. It describes the distribution of all relevant forms between phases under a specified pH condition; it is not simply a second name for neutral-species logP. A simple derivation illustrates why pH must be matched. Assume a monoprotic base has neutral form \(B\) and ionized form \(BH^+\), only neutral \(B\) partitions appreciably into the organic phase, and the aqueous phase obeys the ideal Henderson–Hasselbalch relation. Then the neutral fraction in water is:
If \(P\) is the partition coefficient of the neutral form, then under these assumptions the distribution ratio is \(D=P f_B\). Taking base-10 logarithms gives:
This relation is a justification for the direction of the pH effect under the stated simplified model, not a universal conversion rule for experimental data. Ionized species may partition, activity effects may matter, and real assays may differ in phase composition or procedure. The practical lesson is narrower: values measured at different pH cannot be called equivalent simply because both are labelled logD.
Worked Example: Quantifying a pH-Driven Difference in LogD
Suppose a hypothetical monoprotic base has \(pKa=7.82\) and neutral-species \(\log P=2.00\). Estimate its logD at pH 7.40 and pH 7.96 using the simplified model above. Assume the ionized form has negligible organic-phase partitioning and the same neutral-species partition coefficient applies at both conditions.
At pH 7.40, the neutral fraction is:
Thus \(D=P f_B=10^{2.00}(0.2755)=100(0.2755)=27.55\), so \(\log D=\log_{10}(27.55)=1.440\), approximately.
At pH 7.96:
Then \(D=100(0.5799)=57.99\), giving \(\log D=\log_{10}(57.99)=1.763\). The predicted values differ by about \(1.763-1.440=0.323\) log units, even though the neutral-species \(\log P\) is unchanged. This is a model-based illustration, not evidence that an actual assay pair must differ by exactly this amount.
Solubility: Solid Form, Equilibration, and Reporting Basis
A solubility value is meaningful only with its experimental context. Different polymorphs, hydrates, solvates, amorphous material, or salt forms can yield different apparent or equilibrium results. A metastable form may dissolve to a higher concentration before converting to a more stable form; a salt may dissolve rapidly and then precipitate a less soluble free form. Thus an early high concentration and a later lower concentration can both be observations from the same experiment, but they do not represent the same state.
Equilibration time alone does not prove that equilibrium was reached. A measurement should be assessed for whether concentration stabilized, whether the solid phase remained or transformed, and whether the procedure separated undissolved material consistently. Filtration or centrifugation conditions can also affect what is counted as dissolved, particularly if fine particles or colloidal material remain in the sampled liquid. A kinetic solubility result from a short precipitation-sensitive procedure should not be silently treated as equilibrium solubility.
Units can create large apparent discrepancies. Mass concentration and molar concentration are related through molecular weight, but salt mass and free-drug-equivalent mass are not identical bases. For a substance with molecular weight \(MW\) in g/mol, a concentration \(C\) in mg/L converts to mol/L as \(C/(1000MW)\). If a salt contains one drug molecule per formula unit, its salt molar concentration corresponds to the same molar concentration of drug moiety, while the corresponding mass concentration of free drug is based on the drug moiety's molecular weight—not the salt's molecular weight.
Worked Example: Standardizing Solubility Units and Drug-Equivalent Basis
A hypothetical free drug has molecular weight \(300\) g/mol. One assay reports equilibrium solubility as \(0.60\) mg/mL of free drug. A second assay reports \(0.0020\) mol/L of dissolved drug moiety from a salt containing one drug molecule per formula unit. Convert both results to common units and compare them, assuming the values refer to the same pH, temperature, solid-state endpoint, and equilibrium basis.
First convert the mass concentration: \(0.60\) mg/mL equals \(600\) mg/L, or \(0.600\) g/L. Therefore:
The second result, \(0.0020\) mol/L, is also \(2.0\) mmol/L. On a free-drug-equivalent mass basis it is \(0.0020\ \mathrm{mol/L}\times300\ \mathrm{g/mol}=0.600\ \mathrm{g/L}=600\ \mathrm{mg/L}=0.600\ \mathrm{mg/mL}\). The converted values agree to the precision reported. This agreement does not establish that the assays are equivalent unless the assumed matching pH, temperature, solid-state endpoint, and equilibrium basis are supported by their records.
A Practical Reconciliation Workflow
A useful reconciliation is a sequence of checks rather than an average of the available numbers. First identify exactly what was measured; next align condition and reporting bases; then assess whether the remaining difference is credible relative to method and replicate variability. If key metadata are absent, retain that limitation rather than inventing a correction.
Check the molecular species, salt or free form, pKa transition, logD pH, and solubility endpoint. Exclude values that answer a different chemical or operational question.
Place temperature, pH, ionic strength or buffer, phase system, solid form, and equilibration protocol side by side. Mark unknown fields as unknown.
Convert mass and molar concentrations with the appropriate molecular weight. Distinguish salt mass from free-drug-equivalent mass and preserve the original reported value.
Review replicate results, calibration or analytical method, detection limits, and evidence for stable equilibration. Do not attribute a difference to a condition if the method could explain it just as plausibly.
Prefer a measurement that matches the intended property and relevant model conditions with well-documented methods. If no value is clearly preferred, retain multiple candidates or a qualified range for subsequent assessment rather than averaging incompatible results.
Worked Example: Reconciling Two Reported pKa Values
Two hypothetical records give a monoprotic base pKa of 7.82 and 7.96. The first was measured at 25 °C in a buffer with reported ionic strength of 0.10 mol/L. The second was measured at 37 °C in a different buffer; its ionic strength is not reported. Both records identify the same free-drug structure and assign the same ionization step. Determine what can be concluded, including the predicted protonated fractions at pH 7.40 under the ideal monoprotic-base expression.
For a base, the protonated fraction is \(f_{BH^+}=1/(1+10^{pH-pKa})\). Using the first pKa:
or 72.45%. Using the second pKa:
or 78.41%. The difference is \(78.41-72.45=5.96\) percentage points using the rounded displayed fractions; using unrounded fractions it is approximately 5.95 percentage points. The arithmetic describes what follows from each reported pKa under the ideal expression; it does not establish that either assay is more accurate.
The records match in assigned species and ionization step, but differ in temperature and buffer, and the second lacks ionic-strength information. Therefore, the values are not measurements under matched conditions. The appropriate conclusion is that they are condition-dependent candidate estimates whose difference needs to be evaluated with method details or new measurements at a common reference condition. Averaging them would obscure that limitation.
Diagnostics, Alternatives, and Limitations
Warning signs include a logD value with no pH, a pKa with no method or temperature, a solubility result without solid-form or equilibration information, and numbers that disagree by orders of magnitude until units are converted. A value reported to many decimal places is not necessarily more reliable than one reported to fewer places. Inspect the underlying assay conditions and replicate spread, not just the displayed precision.
When conditions differ, there are several reasonable alternatives to a direct comparison. A matched-condition repeat assay is the clearest way to assess whether the discrepancy persists. A validated thermodynamic or speciation model may support a conversion when its assumptions and required inputs are appropriate. If neither is feasible, report condition-specific values with their metadata and choose a value based on relevance and quality of evidence, stating the uncertainty. These options are preferable to pooling unmatched measurements as if they were replicate observations.
Reconciliation has limits. Some historical reports omit critical details; experimental methods may have systematic biases that cannot be recovered from summary values; and condition effects may interact rather than act independently. The simplified logD equation above, for example, assumes only the neutral species partitions and ideal aqueous speciation. It should not be used to “correct” a measured logD when those assumptions have not been checked. Similarly, unit conversion can resolve a basis mismatch but cannot make different solid forms or equilibration endpoints equivalent.
A common mistake is: “The two pKa values are close, so average them and use the mean as the drug’s pKa.” The correction is that numerical proximity does not show that conditions or methods are comparable. Preserve the original values and metadata, determine whether the difference is expected under the stated conditions, and only combine measurements when they are sufficiently comparable and their uncertainty supports doing so.
Check Your Understanding
Answer each question, then compare your reasoning with the solutions below.
- Why is a logD value reported at pH 6.5 not automatically interchangeable with one reported at pH 7.4 for an ionizable base?
- A pKa report omits temperature and ionic strength. What can you conclude, and what should remain uncertain?
- A hypothetical compound has molecular weight 250 g/mol and a solubility of 500 mg/L. Convert this concentration to mmol/L.
- Two solubility values differ, and one was measured after a short incubation while the other was measured after an extended equilibration with a different crystal form. Why is averaging them inappropriate?
- What is one strong alternative to reconciling values from unmatched assay conditions by calculation alone?
Solutions. 1. For an ionizable base, the neutral fraction and therefore distribution can vary with pH; the values describe different conditions and should not be treated as equivalent. 2. You can record the reported pKa and identity if available, but cannot establish its reference conditions or quantify the impact of the missing metadata; treat it as qualified rather than assuming standard conditions. 3. \(500\) mg/L is \(0.500\) g/L, and \(0.500\ \mathrm{g/L}/250\ \mathrm{g/mol}=0.00200\) mol/L, or \(2.00\) mmol/L. 4. The measurements may represent different equilibration states and solid-state properties, so their difference is not ordinary replicate variation; averaging would conceal the distinct endpoints. 5. Repeat the assay under matched, documented conditions, or use a validated model only when its assumptions and inputs are adequate.