How the release medium sets the dissolution rate of HPMCAS, and with it the release of drug from HPMCAS dispersions

Background and motivation

Amorphous solid dispersions (ASDs) are molecular mixtures of a poorly soluble drug and a polymer. They are thought to raise oral bioavailability by producing supersaturated solutions that drive membrane transport, so their release is best tested under nonsink conditions, where the extent and duration of supersaturation can be seen (Indulkar 2016; Sun 2016; Mosquera-Giraldo 2018) C1. Two features of ASD release make test design difficult. First, release depends on drug loading (DL): at higher loading it is controlled by the drug, at lower loading by the polymer. In the polymer-controlled regime the drug enters the medium at the rate the polymer dissolves, whatever the drug concentration already in solution, and forms colloidal drug-rich species, so classical Noyes–Whitney descriptions fail (Simonelli 1969 with polyvinylpyrrolidone; Yang 2023 and Saboo 2020 with copovidone) C2. Second, in biorelevant media, which contain micelles, the drug can exist as free drug, in micelles and as drug-rich colloids, and the relation between measured concentration and supersaturation is hard to establish (Elkhabaz 2018) C3. For these reasons the authors argue, following earlier work, that polymer release should be measured as well as drug release C4.

Simple buffered solutions, usually phosphate at pH 6.5 or 6.8, are often used instead of biorelevant media. pH matters for ionizable drugs, and it matters for polymers whose solubility depends on pH. For enteric coatings, studies of pH and buffer capacity have mostly asked when the coat ruptures; only a few early studies quantified polymer release, and several found that above a threshold pH it is the buffer capacity, not the pH, that dominates release (Shek 1978; Hens 2017; Sakamoto 2021) C5. Hydroxypropyl methylcellulose acetate succinate (HPMCAS) is an enteric polymer widely used in commercial ASDs, mainly because it delays crystallization from supersaturated solutions, and some studies have linked drug release from HPMCAS dispersions to the polymer's dissolution rate (Hiew 2022; Qian 2012; Wang 2018) C6. A recent study (Sakamoto 2021) found that the buffer system changes drug release from a dispersion of nifedipine in an ionizable polymer, but did not measure polymer release C6. There is a further difference from enteric coatings: in a coating the drug and the polymer sit in separate phases, whereas in an ASD they are mixed at the molecular level, so an ASD "may" respond differently to the medium C7.

The paper therefore asks how the buffer capacity, composition and pH of the medium change the release of HPMCAS, both as the neat polymer and from ASDs of two model hydrophobic drugs, indomethacin (IND) and indomethacin methyl ester (INDester), at low and high drug loading (Section 1). The two drugs differ only at the acid group (Figure 1): IND carries a free carboxylic acid and is mostly anionic at pH 6.8, while INDester is a methyl ester that stays un-ionized across the physiological pH range M1·M2. To remove differences in surface area, all release was measured from compacts with one exposed face (Wood's apparatus). Gel formation was imaged by confocal microscopy, and the apparent pH of the gel layer was estimated with a pH-sensitive dye built into the polymer P7·P9·P10.

Background concepts

Buffer capacity. Buffer capacity, written here in mM/ΔpH, is the amount of acid or base a solution absorbs per unit change in pH. For one buffer at one pH it rises with the buffer concentration: in this paper pH 6.8 phosphate of 5, 10, 25, 50 and 100 mM has buffer capacities of 3, 5, 13, 26 and 51 mM/ΔpH (SI Table S1) P14. The same concentration can give very different capacities at different pH values: 50 mM phosphate has 26 mM/ΔpH at pH 6.8 but only 18 at pH 7.5 and 2 at pH 9.0 P14. (Note by the AI writer, not discussed in the paper: phosphate buffers best near its second pKa, about 7.2, so its capacity falls away from that pH.) The paper cites intestinal buffer capacities of 2–8 mM/ΔpH in the fasted state, while the usual in vitro medium, 50 mM pH 6.8 phosphate, has 26 mM/ΔpH; FaSSIF, made in pH 6.5 phosphate, has 12 mM/ΔpH R24·M10.

How an enteric polymer dissolves. A glassy, entangled polymer that does not ionize dissolves by taking up solvent, swelling into a gel and freeing its chains once enough solvent has lowered the viscosity; the chains then diffuse away (refs 52–62) C32. A carboxylic acid polymer such as HPMCAS, insoluble at low pH, needs one more step: its acid groups must ionize before the chains can dissolve. Without a gel layer, the steps are (1) transport of water and base to the protonated polymer, (2) the acid–base reaction, (3) release of polymer from the solid, and (4) transport of polymer and of the released protons across the aqueous boundary layer to the bulk; a gel layer between the glassy core and the boundary layer adds transport through the gel and release of ionized polymer from it (Figure 9) C33. Whichever step is slowest limits dissolution. HPMCAS-MF has a reported pKa of 4.92 or 5.0, and its solubility rises steeply between pH 5.0 and 6.5 (Hiew 2022; Nguyen 2023) M3.

The Nguyen–Fogler picture. Nguyen and Fogler (2005) studied the methacrylic copolymers Eudragit L and S on a rotating disc R25. When dissolution is diffusion-limited, the rate is proportional to the polymer's solubility at the dissolving interface, and for an enteric polymer that solubility depends on the hydrogen ion concentration there. The interface is more acidic than the bulk because the polymer releases protons as it ionizes. Buffer species act as proton carriers: they remove protons from the surface, raise the surface pH and so increase the driving force for polymer diffusion. At high buffer concentration and fast rotation the rate stopped rising with rotation speed; Nguyen and Fogler read this plateau as a change from diffusion control to control by chain disentanglement from the gel R25. The same lower-than-bulk surface pH has been reported for dissolving weakly acidic drugs above pH 6, with larger differences at lower buffer capacity (Mooney 1981; Hens 2022) R26.

The rotating disc and the boundary layer. In a rotating disc apparatus the liquid next to the surface forms a boundary layer (also called the unstirred water layer) across which dissolved species diffuse. Under laminar flow its thickness falls with the square root of rotation speed, so a diffusion-controlled rate should rise linearly with the square root of rotation speed; a plateau would point to a slower step that does not depend on stirring, such as disentanglement E3·R25. The paper writes the flux and the boundary layer thickness as Equations 2 and 3, discussed in the next section.

Polymer entanglement. Chains entangle only above a critical molecular weight. Polymers above it form a gel in a good solvent because the solvent enters faster than the chains can disentangle; polymers near or below it "may" still swell into a gel through viscosity effects (Graessley 1980) C34. HPMCAS-MF has a reported molecular weight of about 18,000 Da, and its critical molecular weight for entanglement has not been reported M3.

Methods in detail

The study has no model of its own; it combines a set of release measurements, chosen to change one property of the medium at a time, with two measurements of the gel layer and one of water uptake, and reads the results through the diffusion equations of the rotating disc. The flow below summarizes the stages.

Contribution 1: release of drug and polymer measured separately, per unit area, across many media

① The problem. Most ASD release tests report only the drug, in one medium, from powders whose surface area changes as they dissolve. Such data cannot tell whether the polymer or the drug controls release, nor separate an effect of the medium from an effect of surface area C4.

② What they did. HPMCAS (AQOAT-MF, Shin-Etsu) and ASDs of IND or INDester with HPMCAS were made by rotary evaporation (IND from 1:2 v/v methanol/dichloromethane with the bath at 60 °C, INDester from tetrahydrofuran at 45 °C), dried under vacuum, cryomilled and sieved to 106–250 µm; PXRD and polarized light microscopy showed them to be amorphous M3·P5·R23. The results use 10 and 50 wt% DL P5. INDester was synthesized from IND and benzenesulfonic acid in methanol and was more than 98% pure by NMR (SI Figure S1) P13·R22. A pre-ionized triethylamine salt of HPMCAS (HPMCAS-TEA) was made by adding one equivalent of triethylamine relative to the polymer's carboxylic acid content (2.38 mmol per 2 g) P4. Neat amorphous IND was made by melt quenching P6.

For each run, 100 mg of powder was pressed at 1500 psi for 1 min into an 8 mm die, leaving one face of 0.5 cm² in contact with 100 mL of medium at 37 °C; the disc rotated at 100 rpm unless stated. Samples were taken every 5 min up to 30 min and every 10 min up to 1 h, with the volume replaced P7. Drug was assayed by HPLC-UV at 254 nm, and HPMCAS by HPLC with an evaporative light scattering detector (ELSD), with a separate calibration (1–250 µg/mL) and gradient for each medium (SI Table S2) P3·P8. Error bars in all release figures are SDs of three runs P7. The rate was calculated from the slope of the initial linear part of each release curve:

In plain terms: $R$ is the release rate of one component per unit of exposed area and per unit of that component's weight fraction in the compact; $k$ is the slope of the release curve, $V$ the medium volume (100 mL), $S$ the exposed area (0.5 cm²) and $x$ the weight fraction of drug or polymer E1. Dividing by $x$ puts drug and polymer from the same compact on one scale: if both leave together, as the compact erodes, their normalized rates are equal. Some samples crystallized during the runs, which is why only the initial linear part was used; the paper does not say which R21.

The media covered pH 6.8 phosphate at five buffer capacities (3–51 mM/ΔpH); FaSSIF (made with pH 6.5 phosphate, 12 mM/ΔpH) and the same buffer without sodium taurocholate and lecithin ("blank FaSSIF buffer"); FeSSIF-V2 in pH 5.8 maleate (25 mM/ΔpH); FeSSIF V1 at pH 5.0; 50 mM phosphate at pH 7.5 (18 mM/ΔpH) and at pH 9.0 (2 mM/ΔpH); and glycine–NaOH at pH 9.0 (25 and 51 mM/ΔpH) P14·M10·M11·M14. To separate buffer capacity from ionic strength, 50 mM phosphate was also run with enough NaCl to raise the ionic strength from 98 to 196 mM P12.

③ Why it works. A fixed area removes the effect of particle size and changing surface, so differences between media or formulations reflect the dissolution process itself. Measuring the polymer directly answers whether drug release follows it. ④ Difference from earlier work. Earlier enteric-coating studies focused on coat rupture, and the most relevant ASD study on buffer systems did not measure polymer release C5·C6.

Contribution 2: the gel layer, its thickness and its pH

① The problem. Whether HPMCAS forms a gel layer before dissolving was not well documented, although it is expected to matter for how the polymer dissolves C26. If the surface is more acidic than the bulk, as the Nguyen–Fogler picture suggests, that should be visible in the layer the polymer leaves behind.

② What they did. For thickness, HPMCAS containing about 0.01% w/w of the fluorescent dye prodan was compacted, partly dissolved for 60 min in pH 6.8 phosphate, air-dried and cut with a razor blade; the cross-section was imaged by confocal fluorescence microscopy P9. For pH, HPMCAS (or HPMCAS-TEA) was made with an indicator dye dispersed in it: chlorophenol red (pH range 4.5–6.8) for the pH 6.8 phosphate and pH 9.0 glycine media, bromothymol blue (6.0–7.5) for pH 7.5 phosphate. After exposure in the same apparatus, the gel was scraped off where possible and smeared on a glass slide, and its visible spectrum was compared with a calibration of the dye in solutions of known pH P10. The result is an average "apparent" pH of the removed gel, not a reading at the dissolving surface.

③ Why it works. The dye travels with the polymer, so it reports the pH where the polymer is. ④ Limits of the method. The exposure time before removal, the dye loading and whether a calibration in solution holds in a hydrated gel are not stated, and the TEA-salt value (6.7) lies at the top of chlorophenol red's stated range X6.

Contribution 3: rotation speed as a test of the rate-limiting step

① The problem. Release could be limited by diffusion of dissolved polymer away from the surface, or by chains freeing themselves from an entangled gel. The two predict different responses to stirring.

② What they did. HPMCAS was run at 100, 200, 300, 500 and 700 rpm (a laminar-flow range) in pH 6.8 phosphate of 5, 26, 51 and 77 mM/ΔpH and in a pH 9.0 medium of 51 mM/ΔpH, and the TEA salt in pH 6.8 phosphate; not every speed was run for every medium A4·R18. The rates were plotted against the square root of rotation speed, and also divided by the rate at 100 rpm (Figure 7) R18·R19. For diffusion-controlled flux from a rotating disc under laminar flow the paper gives:

In plain terms: the polymer flux $J_p$ equals the polymer's diffusion coefficient $D_p$ divided by the boundary-layer thickness $\delta$, times $C_p$, the polymer's solubility at the polymer–solution interface E2. A lower interfacial pH lowers $C_p$ and so the flux; this is how the authors connect buffer capacity to release C37.

In plain terms: the boundary layer thins as the disc turns faster, in proportion to the inverse square root of the rotation speed $\omega$; $\nu$ is the kinematic viscosity of the medium (Levich) E3. Combined with Equation 2, a diffusion-controlled rate at constant $C_p$ should be proportional to ${\omega }^{1/2}$, so that relative to 100 rpm it rises to $\sqrt{7}\approx 2.65$ at 700 rpm E3. A plateau at high speed would instead suggest disentanglement control, as in the Nguyen–Fogler work R25.

③ Why it works. Stirring changes only the boundary layer; if the slowest step lies in the gel, faster stirring stops helping. ④ Difference from earlier work. The same test was used by Nguyen and Fogler for methacrylic copolymers of high molecular weight; here it is applied to HPMCAS-MF and to its pre-ionized salt R25.

Contribution 4: water uptake of the dispersions

To test whether the drug changes how the polymer hydrates, 100 mg of neat HPMCAS and of the four dispersions (IND or INDester at 10:90 and 50:50) were held at 37 °C and 95% RH and weighed every 2 h for 12 h P11. The paper reports the uptake after 12 h and a water uptake rate in mg/h (Figure 8), without saying how the rate was taken R20·X9. Uptake from humid air is a proxy; the hydration of the gel in the release media was not measured X9.

Solubility was measured alongside: crystalline IND and INDester by shake flask in 100 mM pH 6.8 phosphate for 48 h at 37 °C with ultracentrifugation, and amorphous INDester by adding a methanolic stock to buffer containing 1000 µg/mL HPMCAS and pelleting the drug-rich phase P1·P2.

Experimental analysis

Solubility of the two drugs

In 100 mM pH 6.8 phosphate at 37 °C, crystalline IND dissolved to 459 ± 10 µg/mL and crystalline INDester to only 0.6 ± 0.1 µg/mL; amorphous INDester reached 3.8 ± 0.4 µg/mL in buffer containing 1000 µg/mL HPMCAS (Section 4.1) R1·R2·R3. The amorphous solubility of IND could not be measured: ionized IND reached concentrations high enough to crystallize. The authors estimate about 9.6 mg/mL by applying the amorphous/crystalline solubility ratio reported for un-ionized IND R4·C8. The two drugs thus span more than two orders of magnitude in solubility while being nearly identical in structure. Note by the AI writer: the two INDester values were measured in different media (with and without dissolved HPMCAS, and the buffer concentration for the amorphous value is not stated), so their ratio of about 6 is approximate X15.

Neat HPMCAS: buffer capacity dominates, micelles do not matter

Figure 2 is the central result. In pH 6.8 phosphate the neat polymer released at about 0.16, 0.26, 0.59, 0.89 and 1.01 mg min⁻¹ cm⁻² at 3, 5, 13, 26 and 51 mM/ΔpH R5. The rate rises steeply at low buffer capacity and flattens at high buffer capacity, where the authors say it "may be approaching a plateau value" C9. Raising the ionic strength of 50 mM phosphate from 98 to 196 mM did not change the rate within error (the values are not shown) R6·C10.

Table A. Release of neat HPMCAS from compacts at 100 rpm and 37 °C, ordered by buffer capacity; read from Figure 2 and SI Figure S2 (approximate; n = 3), buffer concentrations from SI Table S1 R5·R7·R8·P14.

Three comparisons in Table A carry the argument. First, micelles: FaSSIF (about 0.56) falls on the trend of pH 6.8 phosphate (about 0.59 at 13 mM/ΔpH), and SI Figure S2 shows that removing sodium taurocholate and lecithin from FaSSIF leaves the rate unchanged (about 0.57). The authors conclude that the micelles do not speed HPMCAS dissolution and that buffer capacity and pH matter more R5·R7·C11. Second, pH near the threshold: FeSSIF-V2 at pH 5.8 gave about 0.58, against about 0.89 in pH 6.8 phosphate of similar buffer capacity, and in FeSSIF V1 (pH 5.0) no polymer dissolved, "presumably" because the pH is too low for the polymer to reach the ionization it needs to dissolve R5·R8·C12·C14. The authors call the near-identical rates in FaSSIF and FeSSIF-V2 "likely" a coincidence C13. Third, species and pKa: 50 mM phosphate at pH 7.5 (18 mM/ΔpH) gave the highest neat-polymer rate measured at 100 rpm, about 1.09, while 50 mM phosphate at pH 9.0, which has a buffer capacity of only 2 mM/ΔpH, gave about 0.67; the authors attribute the lower value to that low buffer capacity R5·C15. A higher bulk pH alone, in other words, does not guarantee faster release.

The gel layer: thin, buffer-dependent and acidic

In confocal cross-sections (Figure 6) the dry compact shows grain boundaries between particles. After 60 min in pH 6.8 phosphate, no gel was visible at 5 mM/ΔpH; from 13 mM/ΔpH upward a lighter layer without grain boundaries appeared at the surface, and it thickened with buffer capacity, from 3 ± 1 µm at 13 to 19 ± 5 µm at 51 mM/ΔpH (labelled widths at 26 mM/ΔpH: 5.60 and 6.37 µm) R14·C27. Note by the AI writer: the widths labelled in Figure 6 (3.54–4.29 µm at 13 and 13.70–24.03 µm at 51 mM/ΔpH) average 4.0 and 17.2 µm, not exactly the text values; the number of readings behind the text values is not stated X7.

Table B. Gel layer and release of neat HPMCAS and its TEA salt. Gel pH and thickness from the text of Section 4.3 and Figure 6; release rates read from Figures 2 and 10 (approximate) R14·R15·R16·R5.

The scraped gel always read below the bulk pH (Section 4.3.2): 5.6, 6.0 and 6.2 in pH 6.8 phosphate of 13, 26 and 51 mM/ΔpH, 6.3 in pH 7.5 phosphate and 6.2 in pH 9.0 glycine–NaOH; in phosphate media that is 0.6–1.2 pH units below the bulk, and in the glycine buffer 2.8 units below R15·C28. The more buffer, the less acidic the gel. The TEA salt, whose acid groups are already ionized, gave a gel pH of 6.7 ± 0.1 in 50 mM pH 6.8 phosphate, which the authors describe as "extremely close" to the bulk R15·C41.

Figure 10 plots release against gel pH. For HPMCAS the rate rises from about 0.60 at gel pH 5.6 through about 0.90 and 1.01 to about 1.09 at gel pH 6.3, and the TEA salt sits apart at gel pH 6.7 and about 1.62 R16. The authors call the release rate "highly correlated" with gel pH and treat the salt as further support C51·C41. Extrapolating the data, they predict that release stops near a gel pH of 5, "consistent with" no release in FeSSIF V1 R17·C42. Note by the AI writer: a straight line through the four HPMCAS points reaches zero near gel pH 4.7, close to the authors' value; but the comparison is with the bulk pH of FeSSIF V1, whose gel pH was not measured and would by the same argument lie below 5 X13. The salt's medium is also uncertain: the text and Figure 10 give 50 mM pH 6.8 phosphate (26 mM/ΔpH), while both panels of Figure 7 label the same series "pH 6.8 51 mM/ΔpH", with the same 100 rpm rate X1.

Dispersions: the drug follows the polymer, and slows it

For the 10% IND dispersion (Figure 3), IND and HPMCAS left at the same normalized rate at every buffer capacity, from about 0.12 and 0.10 at 3 mM/ΔpH to about 0.80 and 0.78 at 26 mM/ΔpH, and also in FaSSIF (about 0.51 and 0.49) R9. The authors read this as drug release controlled by the polymer dissolution rate ("suggesting") C16. The drug rate rose about 7- to 8-fold across the buffer capacities (the text says "∼8"), following the polymer, and FaSSIF components did not change the release of either, "consistent with polymer-controlled release" R10·C18·C20. The 10% INDester dispersion behaved the same way (SI Figure S3): about 0.13 and 0.11 at 3, 0.57 and 0.56 at 13, 0.59 and 0.62 at 26 and 0.57 and 0.59 mM/ΔpH at 51, for drug and polymer R13. From this the authors write that any change in polymer release corresponds to a change in drug release C24.

Neat amorphous IND released at about 0.40 mg min⁻¹ cm⁻² in 50 mM phosphate, against about 0.80 for IND from the 10% dispersion, which the authors call "much slower" R9·C17. Note by the AI writer: this compares rates normalized by the drug's weight fraction (Equation 1). Per unit area, the dispersion released about 0.08 mg of IND per min per cm², about one fifth of the neat amorphous drug; the dispersion is faster only per unit of drug it contains X14.

Table C. Normalized release rates (mg min⁻¹ cm⁻², pH 6.8 phosphate, 100 rpm), all read from Figures 3, 4 and S3 (approximate; n = 3). Drug release was not measured for the 50% dispersions R9·R11·R13·X4.

Figure 4 compares polymer release from the dispersions with the neat polymer, and Table C collects the values. At 10% loading the dispersions released polymer almost as fast as neat HPMCAS up to 13 mM/ΔpH, but at 26 and 51 mM/ΔpH the neat polymer was clearly faster, even after normalizing for dilution by the drug R11·C21. The authors' reading is that at low buffer capacity the buffer "appears to dictate the rate-limiting step", while at higher buffer capacity the drug's effect on polymer release predominates C21. INDester had a "somewhat larger" effect than IND at 26 and 51 mM/ΔpH: at 51 mM/ΔpH about 0.58 against 0.76 R11·C22. At 50% loading polymer release was "clearly hindered" at every buffer capacity, and the INDester dispersion levelled off from 13 mM/ΔpH at about 0.44–0.45 R11·C23. Figure 5 plots the losses as differences from the neat polymer: they grow with buffer capacity, to about 0.25 and 0.43 for the 10% IND and INDester dispersions and about 0.43 and 0.55 for the 50% dispersions at 51 mM/ΔpH R12·C25.

Note by the AI writer: two readings of these data need care. In relative terms the drug's effect at 3 mM/ΔpH is not smaller than at 51: the dispersions released about 0.25–0.69 of the neat rate at 3 mM/ΔpH and about 0.45–0.75 at 51 mM/ΔpH, although at 5 and 13 mM/ΔpH the 10% dispersions were within about 12% of neat X3. The abstract's statement that INDester has the larger effect also holds only at 26 and 51 mM/ΔpH (and at 5 mM/ΔpH for 50% loading); at 13 mM/ΔpH the 50% INDester dispersion released polymer faster than the IND dispersion (about 0.44 against 0.40) X12.

Rotation speed: diffusion across the boundary layer

Polymer release rose with rotation speed in every medium (Figure 7a): at 5 mM/ΔpH from about 0.26 at 100 rpm to about 0.76 at 700 rpm, at 51 mM/ΔpH from about 1.01 to about 3.34, and for the TEA salt from about 1.63 to about 4.78 R18·C29. Divided by the rate at 100 rpm (Figure 7b), all systems fell on roughly one line against RPM^0.5 with no plateau R19. The authors read this as "consistent with a predominantly diffusion-controlled process", and in the Discussion as "in line with the supposition" that diffusion across the boundary layer, not disentanglement, limits HPMCAS dissolution under the conditions studied C30.

Table D. Release relative to 100 rpm, read from Figure 7b (approximate), against the proportionality to the square root of rotation speed that Equations 2 and 3 give at constant interfacial solubility; the last column was calculated by the AI writer R19·E3·X10.

The authors add a structural argument. HPMCAS-MF has a reported molecular weight of about 18,000 Da, below the critical molecular weights for entanglement estimated for pullulan, curdlan and xanthan (about 22, 24 and 34 kDa), so it is "unlikely" to form molecular entanglements, although it does form a thin gel that can be removed M3·R27·C45. The absence of a plateau even for the pre-ionized salt, where ionization and proton build-up are eliminated and interfacial solubility should be highest, supports diffusion control "certainly for the case of the neat polymer" C45. Note by the AI writer: Table D shows that the data are close to the square-root scaling up to 500 rpm but rise above it at 700 rpm, and that the spread at 300 rpm is wide; the trend is consistent with diffusion control, but no diffusion coefficient or interfacial solubility is given, so it is not a quantitative fit X10.

Water uptake

After 12 h at 95% RH and 37 °C, neat HPMCAS took up about 6.2% water, the IND dispersions about 5.1% (10%) and 3.9% (50%), and the INDester dispersions about 4.6% and 2.4%; the uptake rates were about 1.9, 1.3, 0.49, 1.1 and 0.38 mg/h in the same order (Figure 8) R20. Neat HPMCAS took up most water, fastest, and INDester lowered uptake more than IND C31. Note by the AI writer: a rate of about 1.9 mg/h kept for 12 h would give about 23 mg, far more than the 6.2% of a 100 mg sample, so the rate must describe an early period; the paper does not define it X9.

Discussion

The authors' chain of reasoning

The Discussion assembles the results into one mechanism, built on the Nguyen–Fogler framework, which the authors adopt for HPMCAS C36. HPMCAS must ionize to dissolve, and each ionizing acid group releases a proton at the surface C33. It is "probable" that, for a bulk pH of 6.8, these protons make the pH at the polymer–solution interface lower than the bulk pH, which lowers the polymer's solubility there and, through Equation 2, its flux C38. With a pKa near 5 and a solubility that rises steeply between pH 5.0 and 6.5, a drop of a few tenths of a pH unit matters; the steep initial rise of release with buffer capacity is "readily explained" this way C37·M3. A higher concentration of buffer species carries more protons to the bulk and lessens the drop; changing the bulk pH or buffer species changes the base and proton-carrier concentrations and so the interfacial pH. The authors describe the lower interfacial pH as "strongly supported" by the gel pH measurements and the relation between rate and gel pH as confirming the picture C39·C40. The flow below shows the chain and its two outcomes.

Rotation fits the same picture: faster stirring thins the boundary layer and speeds diffusion, and because the interfacial solubility differs between media, so does the slope of the stirring dependence C44. The drug adds a second effect. At low buffer capacity and low loading the dissolution rate "appears to be" controlled by polymer diffusion at low polymer solubility, so neat polymer and dispersion differ little; at higher buffer capacity the differences "can be attributed" to the drug reducing the amount and rate of water the polymer absorbs, since less water leaves a more viscous gel from which the polymer escapes into the boundary layer less easily C21·C46. At high loading the polymer is hindered at every buffer capacity, "indicating" that both hydration and polymer diffusion matter C47. In the Conclusions this becomes: the drug lowers polymer release, "most likely" as a result of reduced hydration rate and extent C46. The overall conclusion is that the polymer solubility at the polymer–solution interface "was elucidated as the rate-limiting step" for HPMCAS release, and that drug release from HPMCAS dispersions of the model acidic or neutral drugs is controlled by polymer release C52·C49.

What it means for release testing

The authors draw two practical consequences. Since intestinal buffer capacities are reported as 2–8 mM/ΔpH fasted and 3–15 mM/ΔpH fed, polymer release, and with it drug release, "might be" polymer-diffusion-controlled in vivo, so that boundary-layer thickness, which depends on particle size and hydrodynamics, is "expected" to have a large impact C43·R24. A lower drug release rate "is predicted" under physiological intestinal conditions than in the usual 50 mM phosphate C19. And any assessment of how drug loading affects release "needs to consider" physiologically relevant buffer capacity and hydrodynamics, because different relative outcomes may be seen in different media C48. The paper closes by saying the observations "can be translated" to predict in vivo outcomes and will help design more in vivo relevant release tests C53·C54.

Note by the AI writer (not discussed in the paper): the size of the effect is easy to underestimate. Neat HPMCAS released about 3.4 times faster in 50 mM phosphate (26 mM/ΔpH) than at 5 mM/ΔpH and about 5.6 times faster than at 3 mM/ΔpH, the two media closest to the cited fasted range R5·R24. A release test in standard buffer can therefore make an HPMCAS dispersion look several times faster than it would be in a weakly buffered intestine. The FaSSIF result also suggests that, for polymer-controlled HPMCAS dispersions, the buffer of a biorelevant medium matters more than its micelles R7·R9. Finally, the TEA salt shows that removing the proton source raises release 1.6- to 1.8-fold at 100 rpm, depending on which medium the salt was run in; salt forms of enteric polymers are a possible design lever, though TEA itself is a research tool here R18·X1.

What remains open

The authors note that the critical molecular weight for entanglement of HPMCAS-MF has not been reported; their argument that it is not entangled rests on comparison with other polysaccharides C45·R27. Beyond that, the mechanism has links that were not measured: the solubility of HPMCAS at the interfacial pH (the cited data stop at pH 6.5), the pH at the dissolving surface itself, the proton flux, and the water content and viscosity of the gel during dissolution X5·X9. Drug release was not measured for the 50% dispersions, where polymer release fell most X4. How the released INDester exists in the medium is also open: at 10% loading and 26 mM/ΔpH, about 0.03 mg/min from 0.5 cm² would give about 18 µg/mL in 100 mL after an hour, about five times its amorphous solubility, so drug-rich droplets or crystals are likely present, and the paper does not say how samples were treated before HPLC X11.

Limitations

Limits stated or implied by the authors. The rotation-speed argument is restricted to "the range of conditions studied herein", and diffusion control is claimed "certainly for the case of the neat polymer" C30·C45. The in vivo statements are hedged: release "might be" polymer-diffusion-controlled, a lower release rate "is predicted" C43·C19. Some samples crystallized during release, and only the initial linear part of each curve was used R21. The critical entanglement molecular weight of HPMCAS-MF is unknown C45.

The AI writer's own judgement (1): what is measured and what is inferred. The release trends are direct, in triplicate and consistent between figures; the buffer-capacity effect on neat HPMCAS is large and monotonic, and the FaSSIF control is clean R5·R7. The mechanism is inferred. The proposed rate-limiting quantity, polymer solubility at the interface, was not measured; the evidence is a correlation of four HPMCAS points and one salt point with an averaged "apparent" pH of gel scraped from the compact, plus cited solubility data that stop at pH 6.5 X5·X6. Media without a gel (5 mM/ΔpH phosphate, 50 mM glycine) have release rates but no gel pH, and the glycine gel pH at 51 mM/ΔpH is not plotted in Figure 10 X5. The rotation data match the square-root scaling only up to 500 rpm X10. The hydration explanation of the drug effect rests on water uptake from humid air with an undefined rate, not on hydration in the medium X9.

The AI writer's own judgement (2): scope and labelling. The conclusion that drug release is controlled by polymer release is shown for 10% loading only X4. The smaller effect of the drug at low buffer capacity holds in absolute terms, not in relative terms, and the INDester-over-IND ranking holds only at 26–51 mM/ΔpH X3·X12. The in vivo statements rest on phosphate, glycine and maleate buffers; intestinal bicarbonate, discussed in work the paper cites, was not used, and the paper calls FaSSIF (12 mM/ΔpH) "physiological intestinal conditions" although its own cited fasted range is 2–8 mM/ΔpH X8. Some labels conflict: the TEA salt's medium is 50 mM in the text and Figure 10 but "51 mM/ΔpH" in Figure 7, and Figure 7 uses a 77 mM/ΔpH medium and an unnamed pH 9.0 buffer of 51 mM/ΔpH that have no or no unambiguous recipe in SI Table S1 X1·X2. The comparison "neat amorphous IND is much slower" holds only per unit of drug X14. All conclusions concern compacts of one HPMCAS grade at 37 °C.

Conclusion

This paper turns a routine choice, the release medium, into a quantified variable for HPMCAS dispersions. In pH 6.8 phosphate, release of HPMCAS from compacts rose about six-fold from 3 to 51 mM/ΔpH, FaSSIF micelles had no effect, and pH near 5 or very low buffer capacity slowed release or stopped it R5·R7·R8. The gel layer was thin, formed only from 13 mM/ΔpH upward, and was more acidic than the medium, less so the more buffer was present; release rose with gel pH, and a pre-ionized salt with a near-neutral gel released fastest R14·R15·R16. At 10% loading both indomethacin and its methyl ester left at the polymer's pace, and both drugs slowed polymer release, most at high buffer capacity and high loading, while also lowering water uptake R9·R13·R11·R20. The authors' mechanism, in which the polymer's own protons lower interfacial pH and solubility and buffer species relieve it, fits all of these observations, but its central quantity, the interfacial solubility, was inferred rather than measured C52·X5. For formulation work the practical message is the authors' own: compare HPMCAS dispersions, and especially drug loadings, in media whose buffer capacity and hydrodynamics resemble the intestine, and measure the polymer as well as the drug C48·C4.