How the grade of HPMC-AS sets the free-drug ceiling of a supersaturated solution, and with it membrane transport

Background and motivation

Amorphous solid dispersions (ASDs) of a drug in a polymer are used to raise the oral absorption of poorly water-soluble drugs. When an ASD dissolves under non-sink conditions it usually produces a supersaturated solution, one in which the dissolved drug exceeds its crystalline solubility. Because both the amorphous drug and the supersaturated solution are metastable, the polymer must be chosen to stop crystallization, and hypromellose acetate succinate (HPMC-AS) has been studied extensively for this reason, including in earlier work by the first author on the role of its substituents (Section "Introduction", refs 1–19) C1.

Supersaturation helps permeation, but only up to a point. Above a certain concentration, the amorphous solubility, the solution separates into a drug-rich colloidal phase by liquid–liquid phase separation (LLPS) or, if that phase is glassy, glass–liquid phase separation (GLPS) (Ilevbare and Taylor 2013, ref 22; Mosquera-Giraldo and Taylor 2015, ref 23). The onset of LLPS marks the maximum thermodynamic activity of the drug and the highest free drug concentration the solvent can hold C2. The drug-rich phase is nanometre-sized at first and is not thought to cross a membrane, so in theory transport reaches a plateau at the amorphous solubility (Raina et al. 2014, ref 25) C3. The drug-rich phase is still useful: it can act as a reservoir that keeps the free concentration at its maximum while drug is absorbed (Indulkar et al. 2016, ref 27), and nanodroplets that do not coarsen may diffuse through the unstirred water layer next to the membrane and bring drug closer to the epithelium, a hypothesis supported by in vivo data (refs 29–31, including Wilson et al. 2018, ref 31) C4.

The formulation target that follows is an ASD that dissolves above the amorphous solubility and releases drug-rich nanodroplets; the right polymer type and amount can achieve this through polymer-controlled dissolution (Que et al. 2019, ref 32; Hate et al. 2019, ref 33) C5. What was not understood, the authors write, is how the polymer used in the ASD affects the absorption gain, and conventional dissolution tests give little mechanistic insight into it C6. The group had built a high-surface-area, flow-through apparatus with a hollow-fibre membrane that is very sensitive to solution phase behaviour and had shown extra mass flow from nanodroplets on top of supersaturation (Hate et al. 2017, ref 34; Hate et al. 2019, ref 33); the authors judge that it "appears to be a suitable tool" to evaluate polymer effects C7.

The discussion adds the background that sharpens the question. Certain polymers lower the amorphous solubility (Mosquera-Giraldo et al. 2018, ref 26; Ueda et al. 2019, ref 48) and change droplet size, colloidal stability and surface properties (Ueda et al. 2017, ref 28) C31. More hydrophobic HPMC derivatives distribute more into the drug-rich phase (refs 28, 42) C34, and a second component mixed into the drug-rich phase lowers the chemical potential of the drug there (Trasi and Taylor 2015, ref 50; Lindfors et al. 2006, ref 51) C36. If the polymer of an ASD enters the drug-rich phase, it could therefore lower the very ceiling that limits transport.

The paper sets out to test this with two grades of HPMC-AS that differ in substituent type: AS-LF, with less acetate and more succinate, and AS-HF, with more acetate and less succinate M5·M6. The plan has three steps: measure the amorphous solubility of several drugs with each grade; characterize the drug-rich phase formed above it (size, and polymer content by NMR); and measure the mass transport of solutions containing this colloidal drug, and of dissolving ASDs, in the flow-through apparatus A1·A3·A4·A5.

Background concepts

Amorphous solubility as a ceiling. When drug is added to water beyond its amorphous solubility, the excess does not stay dissolved; it forms a separate, drug-rich liquid (or glass). The dissolved, molecularly free drug in the water stays at the amorphous solubility, and the extra drug sits in the droplets (Taylor and Zhang 2016, ref 43) C23. The amorphous solubility therefore sets the maximum free drug concentration and the maximum thermodynamic activity a solution can reach C2. The paper measures it by the UV-extinction method: drug in methanol is added slowly to buffer, and the concentration at which the solution starts to scatter light at 450 nm is taken as the amorphous solubility (ref 22) P2·C9.

The solubility ratio. Dividing the amorphous solubility by the crystalline solubility gives the highest concentration-based supersaturation a solution can reach before a drug-rich phase forms. The paper defines it in words and tabulates it in Table 1 E1:

In plain terms: how many times more drug a solution can hold in the free state before LLPS than a solution saturated with crystals. Both solubilities are measured in the same buffer with the same dissolved polymer, so a polymer that solubilizes the drug raises the denominator, and a polymer that enters the drug-rich phase lowers the numerator E1·C16·C17.

Flux and supersaturation. The discussion states, citing earlier work, that membrane flux is directly proportional to the extent of supersaturation, which reaches its maximum at the amorphous solubility (refs 23, 25) E2·C29:

In plain terms: only free drug permeates, and its driving force is its supersaturation; once the free drug is capped at the amorphous solubility, a lower amorphous solubility means a lower maximum flux. Drug-rich droplets do not permeate the membrane themselves but can add to transport by dissolving near it as free drug is removed C24·C30.

Why a polymer in the droplets lowers the amorphous solubility. The amorphous solubility is the free drug concentration in equilibrium with the drug-rich phase. If that phase is pure drug, it is the solubility of the pure amorphous drug. If polymer mixes into it, the drug in the droplets is diluted, its chemical potential falls, and the water phase in equilibrium with it holds less free drug (refs 50, 51) C36. The paper does not compute this effect; it uses it to explain its data.

HPMC-AS grades. HPMC-AS is hydroxypropyl methylcellulose carrying acetyl and succinoyl groups (Figure 1e shows the repeat unit with R = -H, -CH3, -CH2CH(CH3)OH, -COCH3 or -COCH2CH2COOH) M5. The succinoyl groups are the ionizable ones. AS-LF has less acetate and more succinate than AS-HF; the authors describe AS-HF as the more hydrophobic grade for that reason M5·M6·C35. The exact substituent contents are given in Table S1 of the Supporting Information, which was not available for this article.

Reading polymer location by solution NMR. Solution 1H NMR quantifies the polymer that is still mobile in solution. Earlier work by the first author showed that the 1H NMR peaks of HPMC-AS diminish when the polymer distributes into drug-rich droplets, because its mobility is suppressed there (refs 28, 42). The paper uses this to read the loss of polymer signal, against a known dose, as polymer that has gone into the droplets C11.

The hollow-fibre absorptive dissolution apparatus. Donor and receiver solutions are pumped side by side (co-current) through a cellulosic hollow-fibre module with a large membrane area (100 cm²) and small pores (15 nm). Drug that crosses the membrane is read continuously in the receiver by a UV probe. The receiver concentration first rises while the module reaches steady state, passes a maximum (Cmax) and then falls slowly as the donor is depleted; Cmax and the area under the curve (AUC) are taken as measures of how much drug was transferred P7·C12. Because the pores are much smaller than the 100–300 nm droplets, only dissolved drug crosses C26.

Methods in detail

The study compares two polymer grades against each other, with and without drug-rich droplets, across four drugs and two transport scenarios. Every experiment ran at 37 °C. The drugs were atazanavir (ATZ, weak base, pKa 4.4) and nifedipine (NIF, neutral) in 50 mM phosphate buffer pH 6.8, and clozapine (CLZ, weak base, pKa 7.6) and ketoconazole (KET, weak base, pKa 6.8) in 100 mM phosphate buffer pH 8.0, chosen so that each drug is "predominantly un-ionized" M1–M4·M9·M10·C8. ATZ was bought as a metastable crystal form and converted to the stable form by stirring in water for 48 h at 37 °C before use M1.

Contribution 1: the amorphous solubility with each grade, across polymer concentrations

Problem. To know whether the polymer of an ASD moves the free-drug ceiling, the amorphous solubility must be measured with the polymer already dissolved at the levels an ASD releases, not only in plain buffer.

What they did. A methanolic stock (20 mg/mL) was pumped at 20–40 µL/min into 15 mL of buffer without polymer or with 100, 500, 1000 or 2700 µg/mL of AS-LF or AS-HF, at 37 °C; the final methanol content was about 0.2–1.2% (v/v). The concentration at which light scattering at 450 nm appeared was recorded as the amorphous solubility, in triplicate P2. The 2700 µg/mL level was used for ATZ only, because that is the polymer concentration a 1:9 ATZ:polymer ASD releases into the 50 mL donor of the transport test A2·C10. Crystalline solubility was measured by stirring excess crystals for 48 h at 37 °C in buffer with 0, 1000 or 2700 µg/mL polymer, ultracentrifuging at 50000 rpm (229600 ×g) for 30 min and analysing the supernatant by HPLC on a C18 column with UV detection at 278, 235, 297 and 230 nm for ATZ, NIF, CLZ and KET P3.

Why it works. The scattering onset detects the first drug-rich droplets C9. Measuring the crystalline solubility in the same polymer solution lets the solubility ratio separate two effects: solubilization by the polymer, which raises both solubilities, and polymer entering the drug-rich phase, which lowers only the amorphous one E1·C16.

Contribution 2: what the polymer does to the drug-rich phase

Problem. If the polymer lowers the amorphous solubility by entering the droplets, more polymer should leave the solution with the droplets for the grade that lowers it more. The droplets' size and stability also matter for the transport test.

What they did. For size, drug stock was added to 100–2700 µg/mL polymer to 300 µg/mL ATZ, 400 µg/mL NIF, 600 µg/mL CLZ or 300 µg/mL KET, about three times the neat amorphous solubility (for NIF, which crystallizes easily, about twice its amorphous solubility in AS-LF); the final methanol was 1.5–3.0%. The Z-average droplet diameter was read by dynamic light scattering (DLS, backscatter at 173°, 37 °C, n = 3) P4. ATZ droplets in 2700 µg/mL polymer and CLZ droplets in 1000 µg/mL polymer were also stirred at 300 rpm and measured after 30 min, 1, 2 and 3 h, to check that they stay stable for the length of a transport run P5. For polymer location, the same drug doses were made in 1000 µg/mL polymer in H2O/D2O (9/1) with the drug added from methanol-d4. Spectra were recorded at 800 MHz and 37 °C with WATERGATE water suppression and a 15 s repetition delay (stated as five times the T1 of HPMC-AS, 1.0–2.5 s). The polymer was quantified against an internal TSP standard from calibration curves of its hydroxypropyl methyl peak (about 1.2 ppm, used for NIF and KET) or acetyl peak (about 2.2 ppm, used for ATZ and CLZ), choosing the peak that does not overlap with the drug's peaks P6.

Why it works, and what it assumes. The read-out is the polymer that is still mobile, compared with the 1000 µg/mL dosed. The interpretation, taken from earlier work (refs 28, 42), is that the missing polymer is inside the droplets where its mobility is suppressed C11. The method does not measure the droplet composition directly.

Contribution 3: transport from supersaturated solutions and from dissolving ASDs

Problem. If AS-HF lowers the free-drug ceiling, transport should drop with AS-HF only when the drug is above the amorphous solubility; below it the free drug, and so the flux, should be the same for both grades.

What they did. The donor was 50 mL of buffer containing the polymer (2700 µg/mL for ATZ, 1000 µg/mL for CLZ) into which drug stock was added by solvent shift (final methanol 0.1–3.0%); the receiver was buffer with the same polymer. Both solutions were pumped co-currently at 4 mL/min through a cellulosic hollow-fibre module (100 cm², pore size 15 nm) at 37 °C, the receiver drug was read by a flow-through UV probe, and the apparatus was washed with 50:50 water:methanol for 30 min after each run (n = 3) P7. Donor concentrations covered three regions: below the amorphous solubility of both grades (ATZ 20, CLZ 50 µg/mL), at each grade's amorphous solubility (about 76 and 36 µg/mL ATZ, about 219 and 95 µg/mL CLZ, read from the axis of Figure 8) and above it (ATZ 150 and 300, CLZ 400 and 600 µg/mL) P7. For the ASD test, spray-dried 1:9 ATZ:polymer ASDs (from dichloromethane/methanol 1/1, 3% w/v solids, inlet 90 °C) were suspended in the donor at a dose that would give 300 µg/mL ATZ and 2700 µg/mL polymer on full dissolution, stirred at 300 rpm, with 10 µm filters keeping undissolved particles out of the module P1·P8. This 1:9 ratio had already been shown to dissolve above the amorphous solubility with nanodroplet formation (Hate et al. 2019, ref 33) R12.

Why it works. The test below the amorphous solubility is a built-in control: with the same free drug in both grades the authors expect no difference in transport, and a difference that appears only above the amorphous solubility can then be tied to the free-drug ceiling C22. Plotting Cmax against donor concentration separates the free-drug contribution (the level at the amorphous solubility) from the droplet contribution (the rise above it) C26·C27.

Experimental analysis

AS-HF lowers the amorphous solubility; AS-LF barely does

Table 1 gives the solubilities at the polymer levels used later in the transport study R1.

Table A. Crystalline and amorphous solubility and their ratio at 37 °C, mean (SD), n = 3 (from Table 1) R1·X1.

AS-LF lowered the amorphous solubility of ATZ, CLZ and KET by 5–8%, and raising AS-LF from 100 to 1000 µg/mL made no difference (Figure 2) R2·C13. AS-HF lowered it strongly for ATZ, NIF and CLZ and less for KET, and the effect grew with AS-HF concentration: ATZ fell to about 45, 40.5 and 37 µg/mL at 100, 500 and 1000 µg/mL AS-HF and to 35.5 µg/mL at 2700; CLZ to about 164, 127 and 95.0 µg/mL; NIF to about 165, 137 and 123 µg/mL; KET to about 63, 60 and 56.0 µg/mL R2. Without polymer NIF crystallized before its amorphous solubility could be read R3. The abstract calls the AS-HF effect a substantial reduction and the AS-LF effect "only a minor impact" C13.

Table B. Amorphous solubility (µg/mL) against dissolved polymer concentration (µg/mL), read from Figure 2 (approximate) R2.

The crystalline solubility moved the other way. It was unchanged for ATZ, rose about 1.25-fold for NIF and 1.4-fold for CLZ, and rose 2.4-fold (AS-LF) and 3.8-fold (AS-HF) for KET R1·X2. The authors call these increases "slight" and say they "may be due to" solubilization by HPMC-AS (ref 39) C14·C15. They note that solubilization which lowers the drug's activity would be expected to raise the amorphous solubility (refs 40, 41), yet AS-HF lowered it C16·C17. With the numerator falling and the denominator steady or rising, the solubility ratio fell in every polymer solution and most with AS-HF: ATZ from 71.5 to 63.2 and 33.2, KET from 54.3 to 21.8 and 10.6 R1·C17.

Note by the AI writer (from the card's extractor notes, not discussed in the paper): the printed ratio for CLZ with AS-LF, 5.9, does not follow from its own inputs (219.0/22.0 = 9.95), while the other eleven ratios reproduce; the authors' statement that the ratio fell most with AS-HF still holds with 9.95 X1. For KET, the large fall of the ratio is mostly a crystalline-solubility effect, since its amorphous solubility fell only from 75.0 to 56.0 µg/mL X2.

The drug-rich droplets: smaller and steadier with AS-HF

Without polymer the drug-rich droplets coarsened quickly and crystallized; with HPMC-AS they stayed at the nanometre scale, which the authors describe as effective inhibition of coarsening R4·C18. For every drug, the droplets became smaller as the polymer concentration rose, for example ATZ from about 385 to 232 nm with AS-LF and from about 377 to 196 nm with AS-HF between 100 and 2700 µg/mL (Figure 3) R5.

Table C. Fresh droplet Z-average and NMR-detected polymer (dose 1000 µg/mL) with drug-rich droplets present, both at 1000 µg/mL polymer; read from Figures 3 and 6 (approximate) R5·R8.

At 500 µg/mL polymer and above, AS-HF droplets were smaller than AS-LF droplets for all four drugs (Table C) R5. At 100 µg/mL, however, ATZ droplets measured about 377 versus 385 nm and CLZ droplets about 390 nm with either grade, so the authors' "regardless of the drug" holds for the higher polymer levels only R5·C18·X3. Under stirring at 300 rpm, AS-LF droplets grew in the first hour (ATZ about 231 to 264 nm, CLZ about 249 to 265 nm) and then stayed below 300 nm, while AS-HF droplets stayed at about 196–200 nm (ATZ) and 98–105 nm (CLZ) for 3 h (Figure 4) R6·C19. The droplets in the transport test are therefore 100–300 nm and stable over the run R6.

AS-HF is lost from the solution signal when droplets form

In the 1H NMR spectra (Figure 5), the AS-LF peaks barely changed when droplets were present and the drug's own peaks were visible; the AS-HF peaks fell clearly with NIF, CLZ and KET and slightly with ATZ, and no drug peaks stood out R7. Quantified (Figure 6, Table C), about 920–990 µg/mL of the 1000 µg/mL AS-LF stayed detectable, against about 415–880 µg/mL of AS-HF R8. The authors conclude that both grades distribute into the drug-rich droplets, AS-HF much more than AS-LF, and that this trend "was found to be independent of the drug" C20.

Note by the AI writer (not discussed in the paper): the size of the AS-HF loss does not follow the size of the amorphous-solubility drop across drugs. ATZ lost only about 12% of the AS-HF from the NMR signal but its amorphous solubility at 1000 µg/mL AS-HF was about 55% below the neat value; KET lost about 40% of the AS-HF but its amorphous solubility fell only about 25% X4. The drug doses also differ (300–600 µg/mL), so the amount of drug-rich phase that can take up polymer differs between drugs X4. The within-drug comparison of the two grades is consistent; a quantitative relation across drugs is not shown.

Transport: equal below the amorphous solubility, lower with AS-HF above it

Figure 7 shows the receiver profiles. Below the amorphous solubility of both grades, ATZ at 20 µg/mL and CLZ at 50 µg/mL gave identical profiles with AS-LF and AS-HF, with Cmax about 3.4 and 6.7 µg/mL R9. Above it, ATZ at 300 µg/mL reached about 18.0 µg/mL with AS-LF and about 13.3 µg/mL with AS-HF, and CLZ at 600 µg/mL about 73.5 and 55 µg/mL; the AS-LF profiles stayed higher for the whole hour (at 59 min about 14.6 versus 11.8 µg/mL for ATZ, 59.5 versus 42 µg/mL for CLZ) R9. The authors report Cmax and AUC as "significantly higher" with AS-LF; AUC values and a statistical test are not given C21·X6.

Table D. Receiver Cmax read from Figures 7, 8 and 9 (approximate); ATZ with 2700 µg/mL polymer, CLZ with 1000 µg/mL. Where Figures 7 and 8 give slightly different readings for the same run, both are shown R9·R10·R11.

The authors' reading is that above the amorphous solubility the free drug is fixed at the amorphous solubility, which is lower with AS-HF, so less of the permeating species is available; they write that the observation "can be readily attributed to the impact of the polymer on the drug amorphous solubility" C23·C24. As support for "this conjecture" they point out that, at the polymer levels of the transport study, the amorphous solubility with AS-HF is about half that with AS-LF, and so is Cmax when each grade is compared at its own amorphous solubility C25. The ratios are 35.5/75.8 = 0.47 for the ATZ amorphous solubility and about 4.6/10.1 = 0.46 for ATZ Cmax; for CLZ they are 95.0/219.0 = 0.43 and about 13.8/35.8 = 0.39, so the match is close for ATZ and approximate for CLZ R1·R10·X11.

Above the ceiling, the droplets add flux at a similar rate for both grades

Cmax kept rising with donor concentration above the amorphous solubility (Figure 8) R10. Because the droplets (Z-average 100–300 nm) are far larger than the 15 nm pores, the authors infer that they did not cross the membrane and that the "linear increase" of Cmax "indicated" that more drug-rich nanodroplets improve mass transport (ref 33) C26. Figure 8 prints a straight-line fit for each series E3–E6:

In plain terms: y is Cmax in µg/mL and x the initial donor concentration in µg/mL; the slope is the gain in peak receiver concentration per µg/mL of extra drug, which above the amorphous solubility goes into the drug-rich droplets C23. The fits have R² = 0.9969 and 0.9902 for ATZ and 0.9094 and 0.9918 for CLZ, each through three points E3–E6·X7. The authors call the slopes "approximately parallel" and conclude that the nanodroplet contribution to the flux improvement was similar for both grades despite their different effects on the amorphous solubility C27. In the discussion they add that CLZ droplets were about 100 nm with AS-HF and 250 nm with AS-LF, yet showed similar gains, so that "over this size range" transport depends more on the concentration of drug in nanodroplet form than on droplet size C40. For ATZ the slopes differ by under 2%; for CLZ by about 16%, and the CLZ AS-LF points (about 35.8, 64.2 and 73.8 µg/mL at about 219, 400 and 600 µg/mL) bend rather than lie on a line X7.

Dissolving ASDs reproduce the grade effect

Suspended 1:9 ATZ ASDs, at the same ATZ dose and polymer level as the solutions, gave receiver profiles close to those of the solvent-shift solutions (Figure 9) R11. The AS-HF ASD followed the AS-HF solution almost exactly (Cmax about 13.4 versus 13.3 µg/mL). The AS-LF ASD rose more slowly in the first minutes (about 12.1 versus 13.6 µg/mL at 3 min), peaked lower (about 16.0 versus 18.0 µg/mL) and joined the solution profile after about 35 min R11·X8. The authors call the profiles "very similar" and the transport from the AS-HF ASD significantly lower than from the AS-LF ASD C28. The grade gap at the peak is smaller for the ASDs (about 16.0 versus 13.4 µg/mL) than for the solutions (about 18.0 versus 13.3 µg/mL), but in the same direction R11·X8.

Discussion

The authors' chain of explanation

The discussion links the three sets of results into one chain. Earlier studies concluded that more hydrophobic HPMC derivatives distribute more into the drug-rich phase (refs 28, 42), and AS-HF is the more hydrophobic grade "as a result of the high extent of acetate substitution and a correspondingly lower amount of ionizable succinate groups" C34·C35. When drug-rich droplets form, AS-HF therefore distributes into them while "a majority of AS-LF remained in the aqueous phase", for all four drugs C33. Polymer mixed with drug in the drug-rich phase lowers the drug's chemical potential there, and because the amorphous solubility is set by that chemical potential, the amorphous solubility falls (refs 50, 51) C36. A lower amorphous solubility caps the free drug at a lower level, which lowers the driving force for membrane transport C24.

The same distribution is used to explain the droplet sizes. The smaller AS-HF droplets "can be attributed, in part" to the lower amorphous solubility, which slows Ostwald ripening (Lifshitz–Slyozov and Wagner, refs 52, 53), and "it can be speculated" that slow diffusion of polymer in the drug-rich phase hinders its redistribution into larger particles, more so for AS-HF with more polymer in the droplets C37·C38. Smaller droplets "would be expected" to contribute more to transport through faster diffusion in the aqueous boundary layer, where the laminar flow of the hollow fibre sets up a concentration gradient from the centre to the wall (refs 56, 57) and droplets entering the layer by Brownian motion dissolve where the local concentration falls below the amorphous solubility C39. The measured slopes did not show this size effect between about 100 and 250 nm C40.

Which effect matters more for absorption

The authors' overall judgement is that, although the drug-rich nanodroplets improve mass transport beyond a maximally supersaturated solution, the change in amorphous solubility caused by polymer entering the droplet phase "was found to have a greater overall impact" on in vitro absorption, both for solutions made by antisolvent addition and for ASD dissolution C41. They frame the formulation consequence as a trade-off. A high polymer loading is chosen to inhibit crystallization, but raising the polymer concentration, particularly of AS-HF, lowers the amorphous solubility through more polymer in the drug-rich phase C43. More polymer in the drug-rich phase helps inhibit crystallization in that phase (ref 48), and when the absorption window is long enough the longer-lasting supersaturation "may outweigh" the reduced flux C44. The improved crystallization inhibition therefore "should be balanced" against the decrease in amorphous solubility, and polymer selection should consider both the polymer's tendency to distribute into the drug-rich phase and the polymer concentration reached after the formulation dissolves C45.

What it suggests for formulation

Note by the AI writer (not discussed in the paper): three practical points follow from the data. First, within one polymer family the grade is a lever on the free-drug ceiling: for these four drugs AS-LF cost under a tenth of the amorphous solubility and AS-HF a quarter to about 60% R1·R2. Second, the penalty grows with the dissolved polymer concentration, and a 1:9 drug:polymer ASD releases nine times more polymer than drug (2700 against 300 µg/mL here), so a high polymer loading and a hydrophobic grade add up R2·P8. Third, the amorphous solubility measured in plain buffer can overstate what a polymer-rich ASD solution reaches; for ATZ the neat value (81.5 µg/mL) was more than twice the value with 2700 µg/mL AS-HF (35.5 µg/mL) R1. A screen of the amorphous solubility with the dissolved polymer present at the expected concentration would capture this before transport or in vivo tests.

Limitations

Limits the authors state or imply. The authors limit the size conclusion to the range studied ("over this size range") C40 and present the crystallization benefit of polymer in the drug-rich phase as a possibility that depends on the absorption window ("may outweigh") C44. The ripening mechanism for the smaller AS-HF droplets is offered "in part" and as speculation C37·C38. The in vivo relevance is stated as an implication ("highlights the importance of polymer selection"), not tested C47.

The AI writer's own judgement. The grade effects on the amorphous solubility and on transport are direct measurements, consistent across all drugs tested and backed by a built-in control below the amorphous solubility R1·R2·R9. The weaker parts are these:

• Where the lost polymer goes is inferred, not measured. The NMR read-out records polymer that loses mobility in the presence of droplets; it does not separate polymer mixed into the drug-rich phase, which the chemical-potential argument needs, from polymer adsorbed at droplet surfaces X10.
• Across drugs the polymer loss and the solubility drop do not match in size (ATZ about 12% AS-HF lost against about 55% solubility drop; KET about 40% against about 25%) X4.
• Polymer distribution was measured only at 1000 µg/mL, so the stated reason why more AS-HF lowers the amorphous solubility further is not measured at other concentrations; the ATZ transport runs used 2700 µg/mL X5.
• "Significantly" is used for Cmax, AUC and the ASD comparison without a stated test, and AUC values are not reported (n = 3) X6.
• Each Cmax line rests on three points; the CLZ slopes differ by about 16% and the CLZ AS-LF fit has R² = 0.9094 X7.
• CLZ and KET were tested in 100 mM phosphate buffer at pH 8.0 and ATZ and NIF in 50 mM at pH 6.8; with the cited pKa values CLZ is about 28% ionized at pH 8.0, and HPMC-AS ionization also differs between the two buffers, so effect sizes are best compared within a drug X9.
• One printed solubility ratio (CLZ with AS-LF, 5.9) does not match its inputs (9.95) X1.
• The Supporting Information, including the substituent contents of the two grades (Table S1) and the receiver profiles behind Figure 8 (Figure S2), was not available for this reading.

Conclusion

The paper shows, for four poorly soluble drugs, that the grade of HPMC-AS dissolved with a drug changes its amorphous solubility: AS-HF lowered it by a quarter to more than half, AS-LF by under a tenth, and more AS-HF lowered it further R1·R2·C13. Solution NMR showed far more AS-HF than AS-LF disappearing from solution when drug-rich droplets formed, which the authors read as polymer distributing into the drug-rich phase and lowering the drug's chemical potential there R8·C33·C36. In a hollow-fibre membrane apparatus the consequence was clear for ATZ and CLZ: equal transport below the amorphous solubility, lower transport with AS-HF above it, and Cmax at each grade's amorphous solubility roughly in proportion to that solubility R9·R10·C24·C25. The drug-rich nanodroplets added flux at a similar rate for both grades, and dissolving 1:9 ATZ ASDs kept the grade difference, so the authors conclude that the amorphous-solubility effect of the polymer had the "greater overall impact" R10·R11·C27·C41. The link through the drug-rich phase composition is an interpretation built on an NMR signal-loss read-out at one polymer level, and in vivo absorption was not tested X4·X5·X10·C47.