Weak-base drug dissolution at the solid-liquid interface turns on the pH right at the particle surface, not the pH of the bulk medium. A mechanistic mass transport model published on 1 October 2026 found that bulk pH alone predicted dissolution with absolute average fold errors up to 4.70, while adding an explicit surface pH term pulled the same predictions to within 1.00 to 1.42 fold.
Why weak-base drug dissolution depends on surface pH
Poorly soluble weak bases are a standing problem in oral formulation. Their solubility swings steeply with pH, and their bioavailability falls when patients take acid-suppressing drugs alongside them. The usual answer is to add an acidifying excipient, which lowers the pH in the particle’s immediate surroundings and drags solubility back up.
Physiologically based biopharmaceutics modeling has gotten better at capturing that in the bulk medium, but most dissolution frameworks stop there. They treat the medium as uniform and ignore the pH gradient that functional excipients create inside the diffusion boundary layer. That omission is the gap this model addresses, and it is a wide one: when the acidifier sits millimeters from the drug particle, the pH that matters is not the one a probe reads in the vessel.
The model drug and the three formulations
The work used entrectinib, a tyrosine kinase inhibitor approved for ROS1-fusion positive non-small cell lung cancer and NTRK-fusion positive solid tumors. It is a BCS class II compound, meaning low aqueous solubility and high permeability, and it is a lipophilic base dosed as a 200 mg capsule.
| Formulation | Acidifying agent | Drug to acidifier molar ratio | Particle size D10, D50, D90 |
|---|---|---|---|
| F0 | none | not applicable | 8, 30, 121 micrometers |
| F1 | betaine hydrochloride, monoprotic | 1 to 1.5 | 4, 19, 49 micrometers |
| F2 | tartaric acid, diprotic | 1 to 1 | 4, 25, 65 micrometers |
Particle size matters twice over here. Smaller particles dissolve faster, and the boundary layer around each one is thinner, which sharpens the pH gradient the model is trying to capture. Both acidified formulations were milled finer than the unacidified one.
How surface pH was modeled
| Element | Detail |
|---|---|
| Modeling approach | mechanistic mass transport based on boundary layer theory |
| Extension of | the Mooney diffusion layer model |
| Governing physics | Fick diffusion laws with coupled diffusion-reaction kinetics |
| Acidifier types handled | monoprotic and diprotic |
| Media conditions covered | buffered and unbuffered |
| Solution method | quartic polynomial in surface hydrogen ion concentration, solved numerically |
| Diffusion coefficients | estimated from the Stokes-Einstein relation |
| Comparison platforms | SIVA V5 and DDDPlus V5 |
The model treats the acidifier’s own dissolution as non-rate-limiting, on the grounds that its aqueous solubility far exceeds the drug’s intrinsic solubility. It assumes every species is dissolved in the bulk at the start, that chemical equilibrium arrives instantly, and that the initial steady-state flux is constant.
Four equilibria govern what happens inside the boundary layer. The free base protonates to its singly charged form, that form protonates again to the doubly charged form, the acidifier dissociates, and water autoionizes. Solving them together produces a quartic polynomial in the surface hydrogen ion concentration.
Surface pH depends on the acidifier, not the medium
| Formulation | Bulk pH, acetate buffer | Predicted surface pH |
|---|---|---|
| F0, no acidifier | 4.5 | 4.59 |
| F1, monoprotic acidifier | 4.5 | 4.43 |
| F2, diprotic acidifier | 4.5 | 3.95 down to 2.93 |
The result reads oddly at first. In buffered acetate at pH 4.5, the unacidified formulation produced a surface pH slightly above bulk, 4.59, because the weak base buffers the diffusion layer itself. The monoprotic acidifier pulled surface pH down only marginally, to 4.43. The diprotic acidifier pulled it down hard, to somewhere between 3.95 and 2.93, more than 1.5 pH units below the bulk medium.
What drives the effect is the number of acidic protons the excipient can donate, not simply the presence of an acid. A formulator who adds a monoprotic acidifier expecting a large local pH shift will be disappointed. A diprotic acidifier at the same molar ratio moves the interface into a different solubility regime entirely.
Between pH 1 and 5 in unbuffered media, acid-base reactions pushed surface pH above bulk because of the drug’s own buffering capacity. In neutral and alkaline unbuffered media, the free base does not react and surface pH equals bulk pH. The model also reproduced a measurable bulk effect: an unacidified formulation in pH 4 medium raised bulk pH from 4 to 5.2 as it dissolved.
Dissolution conditions used in the study
| Medium | Volume | pH | Paddle speed |
|---|---|---|---|
| 0.1 M hydrochloric acid | 750 mL | 1.0 | 75 rpm |
| 0.0001 M hydrochloric acid | 500 mL | 4.0 | 75 rpm |
| Simulated gastric fluid, fasting state | 500 mL | 1.2 | 50 rpm |
| Acetate buffer | 500 mL | 4.5 | 50 rpm |
The four conditions span the range where a weak base either dissolves easily or does not. At pH 1 and at pH 1.2 in simulated gastric fluid, all three formulations dissolved completely within 15 minutes and the differences between them disappeared. The interesting behavior lived at pH 4 and 4.5, where solubility becomes limiting.
Prediction accuracy with and without surface pH
| Modeling basis | Average fold error range | Absolute average fold error range |
|---|---|---|
| Bulk pH only | 0.48 to 4.70 | 1.80 to 4.70 |
| Surface pH included | 0.72 to 1.03 | 1.00 to 1.42 |
An absolute average fold error of 1.00 means perfect average prediction. The bulk-pH framework hit 4.70 in the worst case, which is not a usable prediction. Adding surface pH compressed the entire error range into 1.00 to 1.42, the accuracy band formulation scientists generally want before trusting a model.
Where bulk pH failed and surface pH held
| Formulation | Medium | Bulk pH error | Surface pH error |
|---|---|---|---|
| F0 | 0.0001 M hydrochloric acid, pH 4 | 4.70 | 1.05 |
| F1 | 0.0001 M hydrochloric acid, pH 4 | 1.81 | 1.02 |
| F2 | 0.0001 M hydrochloric acid, pH 4 | 2.10 | 1.42 |
| F2 | acetate buffer, pH 4.5 | 1.80 | 1.39 |
| F0 | 0.1 M hydrochloric acid, pH 1 | 1.01 | 1.01 |
| F1 | simulated gastric fluid, pH 1.2 | 1.01 | 1.01 |
The pattern holds throughout: where the drug dissolves readily, both approaches agree, and where the drug dissolves poorly and the acidifier matters most, bulk pH breaks down. The worst case, an absolute average fold error of 4.70 for the unacidified formulation at pH 4, also carried bias in the wrong direction. Being off by nearly fivefold would corrupt any downstream exposure simulation built on it.
The diprotic formulation also disintegrated more slowly, with an estimated capsule disintegration time of around 14 minutes, which adds a second mechanism the dissolution model has to accommodate.
What this means for formulation development
The results of this weak-base drug dissolution work translate into a short list of design decisions.
- Stop treating the dissolution medium as uniform when the formulation contains a pH-modifying excipient. The relevant pH sits inside the diffusion boundary layer and can be more than 1.5 units from what a probe measures. Here, a diprotic acidifier drove surface pH from 4.5 down to between 3.95 and 2.93.
- Choose the acidifier by proton count, not by strength alone. A monoprotic acidifier moved surface pH by 0.07 units in buffered medium while a diprotic one moved it by more than 1.5. Matching the excipient’s chemistry to the local pH shift you need is the decision that matters.
- Expect bulk pH to drift during dissolution. The unacidified formulation raised bulk pH from 4 to 5.2 as it dissolved. A study that records only the starting pH will misdescribe the experiment.
- Check whether your model computes surface pH at all. Many commercial physiologically based and dissolution modeling platforms lack that calculation for weak bases formulated with pH-modifying excipients, which means an error range of 1.80 to 4.70 may be running silently underneath.
- Design the in vitro study so it can discriminate. At pH 1 and pH 1.2 all three formulations dissolved within 15 minutes, so those conditions cannot rank them. The discriminating conditions for a weak base are the ones where solubility limits release, here pH 4 and 4.5.
- Anticipate the acid-suppression interaction. Co-administration of a proton pump inhibitor cut entrectinib area under the curve by 25% and peak concentration by 23%. Formulations that build their own acidic microenvironment offset part of that loss, which is the clinical argument for the acidifier in the first place.
- Watch particle size alongside chemistry. The two acidified formulations were milled to a median of 19 and 25 micrometers against 30 for the unacidified one, and a thinner boundary layer sharpens the pH gradient the model captures.
Limitations
The model describes ion chemistry through concentrations rather than thermodynamic activities. Experimental ionic strength reached roughly 0.1 M in the 0.1 M hydrochloric acid medium and about 50 mM in the acetate buffer, conditions where mean ionic activity coefficients depart from unity. That is a known approximation rather than an error, but it bounds how far the predictions stretch to higher ionic strength.
The model assumes the acidifier dissolves essentially instantaneously because its solubility vastly exceeds the drug’s. For weakly soluble acidifiers, or formulations where the acidifier releases slowly from a matrix, that assumption would not hold. It also assumes instantaneous chemical equilibrium and a constant initial steady-state flux, which ignores the transient phase before the boundary layer establishes.
Everything here was validated on one model compound in capsules. Entrectinib is a lipophilic weak base of BCS class II, and while the physical chemistry transfers to other weak bases, the specific fold-error figures and acidifier ratios do not. Anyone applying this to a different base, a different salt form, or a different dosage form should re-derive the surface pH for their own weak-base drug dissolution problem rather than reuse these numbers.
The platform comparison ran on two commercial tools, and both produced comparable results, which strengthens confidence in the mechanism. It does not establish that either tool is validated for regulatory submission without further work.
Bottom line
Bulk pH is the wrong control variable for weak-base drug dissolution when the formulation contains a pH-modifying excipient. Modeling surface pH explicitly as a function of acidifier diffusion and acid-base equilibrium cut the absolute average fold error from a range of 1.80 to 4.70 down to 1.00 to 1.42. A diprotic acidifier pulled surface pH from a bulk 4.5 down to between 3.95 and 2.93 while a monoprotic one moved it only to 4.43. At pH 1 and pH 1.2 every formulation dissolved within 15 minutes and the differences vanished, so discriminating studies have to run where solubility limits release. Co-administration of a proton pump inhibitor lowered exposure by 25% in area under the curve and 23% in peak concentration, which is the clinical stake behind getting the interface chemistry right.
FAQ
What is surface pH in weak-base drug dissolution
Surface pH is the hydrogen ion concentration in the thin liquid layer immediately surrounding a dissolving particle, as distinct from the pH of the bulk medium. For a weak base it can differ substantially from the measured bulk value. In buffered acetate at pH 4.5, a diprotic acidifier drove surface pH down to between 3.95 and 2.93.
Why does bulk pH fail to predict dissolution of weak bases
Bulk pH ignores the pH gradient that functional excipients create inside the diffusion boundary layer. Modeling with bulk pH alone produced average fold errors from 1.80 to 4.70. Adding an explicit surface pH term, calculated from acidifier diffusion and acid-base equilibrium, brought the same predictions to 1.00 to 1.42.
Which acidifying excipients lower surface pH most
Diprotic acidifiers. Tartaric acid at a 1 to 1 drug-to-acidifier molar ratio pulled surface pH from a bulk 4.5 down to between 3.95 and 2.93. Betaine hydrochloride, a monoprotic acidifier at a 1 to 1.5 ratio, moved surface pH only to 4.43, so proton count drives the effect far more than acid strength does.
What is BCS class II and why does it matter here
BCS class II means low aqueous solubility with high permeability. Entrectinib, the model compound in this work, is a BCS class II lipophilic base dosed as a 200 mg capsule. Dissolution rate rather than membrane permeation limits its absorption, which is why the interface chemistry around the particle controls how much drug reaches the circulation.
How much does a proton pump inhibitor reduce entrectinib exposure
Co-administration of a proton pump inhibitor lowered entrectinib area under the curve by 25% and peak plasma concentration by 23%. That loss is the clinical reason acidifying excipients are added to weak-base formulations. The excipient creates a locally acidic environment that partially restores solubility even when gastric pH rises.
What dissolution media separate acidified weak-base formulations
Only the media where solubility limits release. At pH 1 in 0.1 M hydrochloric acid and at pH 1.2 in simulated gastric fluid, all three formulations dissolved completely within 15 minutes, so those conditions cannot rank them. Discrimination appeared at pH 4 in dilute hydrochloric acid and at pH 4.5 in acetate buffer.
What is average fold error in dissolution modeling
Average fold error compares predicted dissolution to observed dissolution across a profile, expressed as a ratio. An absolute average fold error of 1.00 means perfect average prediction, and larger values mean worse accuracy. This model reached 1.00 to 1.42 with surface pH, against 1.80 to 4.70 using bulk pH alone.
Does the acidifier in a formulation change the bulk pH
Yes. An unacidified formulation dissolving in pH 4 medium raised the bulk pH from 4 to 5.2, because the weak base consumes hydrogen ions as it dissolves. A dissolution study that records only the starting pH mischaracterizes the experiment. Tracking bulk pH through the run is necessary for interpreting the profile.
How is surface pH calculated in these models
Through coupled diffusion and reaction kinetics in the boundary layer. Four equilibria are solved together: free base protonating to its singly charged form, that form protonating again, the acidifier dissociating, and water autoionization. The result is a quartic polynomial in surface hydrogen ion concentration, solved numerically.
Do particle size and acidifier choice interact
They do. Smaller particles dissolve faster and have a thinner diffusion boundary layer, which sharpens the pH gradient. The two acidified formulations in this work were milled to a median of 19 and 25 micrometers against 30 for the unacidified one, so chemistry and particle size moved together and both need to be controlled.
What are the main assumptions behind surface pH modeling
The acidifier’s own dissolution is treated as non-rate-limiting because its solubility far exceeds the drug’s, all species start dissolved in the bulk, chemical equilibrium is instantaneous, and the initial steady-state flux is constant. Ion chemistry is described by concentrations rather than activities, which matters at the roughly 0.1 M ionic strength used here.
Which commercial tools handle surface pH for weak bases
The work compared two commercial platforms and found comparable results. Many physiologically based biopharmaceutics and dissolution modeling platforms lack a surface pH calculation for weak bases formulated with pH-modifying excipients, which means the bulk-pH error range of 1.80 to 4.70 can propagate silently into exposure predictions.

Comments are closed