When a good dissolution curve hides a low flux: telaprevir ASDs with four polymers at three drug loadings

Background and motivation

An amorphous solid dispersion (ASD) is used to create a supersaturated solution of a poorly water-soluble drug. What happens next decides how much of that supersaturation is useful. The drug can crystallize, which lowers the concentration toward the crystalline solubility. If it does not crystallize and the concentration exceeds a threshold, the solution separates into a dilute aqueous phase and a drug-rich phase: liquid-liquid phase separation (LLPS) or, when the drug-rich phase is a glass, glass-liquid phase separation (GLPS). The threshold concentration usually agrees well with the amorphous solubility of the drug (Ilevbare 2013; Mosquera-Giraldo 2015; Almeida e Sousa 2015; Raina 2014; Trasi 2015). C3

Membrane transport experiments by the same group had shown what this means for absorption. The mass flow through a membrane rises with concentration only up to the amorphous solubility and then reaches a plateau; an apparent concentration above that does not raise the flux, and an amorphous precipitate gives a higher flux than a crystalline one and can serve as a "drug reservoir" (Mosquera-Giraldo 2015; Raina 2014; Indulkar 2016). C4 The flip side is that a dissolution test can mislead. Surfactants or drug-excipient complexes can raise the measured concentration while lowering the free drug that actually crosses a membrane; for resveratrol ASDs in HPMCAS, high concentrations came from soluble drug-polymer complexes and went with a lower mass flow (Wegiel 2015). C5 Diffusion measurements had also been used to find the rate-limiting barrier, detect complexation, relate in vitro results to in vivo data and measure supersaturation with surfactants and bile salts more rigorously than concentration ratios can. C6

Telaprevir (TPV) is a good test drug. Its commercial product is an ASD containing HPMCAS and SDS. R25 It has a dry glass transition temperature (Tg) of 103 °C and a wet amorphous precipitate Tg of 52 °C, so its drug-rich phase is a glass, and it undergoes GLPS above about 90-100 µg/mL, after which the mass flow across a cellulose membrane stays constant (Mosquera-Giraldo 2015). M1·C8 Earlier induction-time experiments by the group had shown that HPMCAS, HPMC and the new cellulose ester CA Sub inhibit TPV crystallization strongly and PVPVA less so. R24 The solid-state history matters too: TPV-PVPVA dispersions tend to phase separate, probably through water taken up during preparation and storage (Li 2016; Li 2017). C7·C54

The paper asks three questions: how polymer type and drug loading change the maximum mass flow; how complexation, crystallization in the ASD matrix and dissolution rate affect it; and how a surfactant changes dissolution and mass flow. Its approach is to compare formulations by drug release in the donor compartment and by mass flow into the receiver compartment of the same side-by-side diffusion cell. A1·A2·A3·P4

Background concepts

Apparent concentration, free drug and the flux ceiling. The donor concentration is measured by a UV fiber-optic probe. Above the GLPS concentration the solution contains drug-rich aggregates that scatter light, so the reading is an apparent concentration; the authors state that any value above the GLPS concentration does not represent molecularly dissolved drug. P4·C12 The receiver records the drug that has crossed the membrane. Only free (molecularly dissolved, unbound) drug crosses, so the receiver slope reports the free drug concentration in the donor. The paper expresses it as a mass flow rate, the slope of the receiver concentration multiplied by the receiver volume (written out in words in the Methods):

Here $J$ is the mass flow rate (µg/min), $C_{\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{e}\mathrm{i}\mathrm{v}\mathrm{e}\mathrm{r}}$ the receiver concentration (µg/mL) and $t$ time (min). In plain terms: the amount of drug arriving per minute; it grows with free drug in the donor until free drug is capped by the amorphous solubility. E2·P4 Only time points after the donor reached its maximum were used, and for PVPVA systems only those before crystallization; the windows ranged from 10-60 min to 800-1000 min (Table 1). P4

The drug-rich phase and what can enter it. Above the GLPS concentration the excess drug forms a dispersed drug-rich phase. If a second component dissolves in that phase, the drug's chemical potential there falls, and with it the free drug concentration in equilibrium with it. Earlier work had shown this for mixtures of two drugs that are miscible in the drug-rich phase (Trasi 2015; Alhalaweh 2016; Arca 2017). C33 For a polymer the effect had not been reported, "most likely" because the drug-rich phase usually contains too little polymer. C34 To measure the composition, the authors collected the drug-rich phase by filtration (to remove undissolved ASD) and centrifugation, and expressed the polymer share of the pellet as a mass fraction (Table 3 footnote):

Here $m_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}$ and $m_{\mathrm{d}\mathrm{r}\mathrm{u}\mathrm{g}}$ are the amounts (µg) of polymer and drug found in the drug-rich phase. In plain terms: the fraction of the separated phase that is polymer rather than drug. E1·C10

Two kinds of crystallization. The paper distinguishes crystallization inside the undissolved ASD particle (matrix crystallization) from crystallization out of the supersaturated solution (solution crystallization). Both stop the solution from reaching or keeping the GLPS concentration, but they have different causes: matrix crystallization follows slow release, solution crystallization follows weak crystallization inhibition in solution. C40·C52

Congruent and incongruent release. Release is congruent when drug and polymer leave the ASD in the same proportion as in the solid; it is incongruent when one leaves faster. Incongruent release had been reported for other systems (Corrigan 1985; Simonelli 1969; Purohit 2017; Bhardwaj 2018). C44

Methods in detail

The study has one measuring core and several probes of cause. The core is the diffusion cell, run with ASDs and with a drug-only reference. The probes ask, for each system that falls short of the reference: does the polymer bind the drug in solution, does it enter the drug-rich phase, does the membrane misbehave, does the film or particle release slowly, and does the drug crystallize?

The formulations

TPV-polymer ASDs were made at 1:9, 3:7 and 1:1 by weight by rotary evaporation (water bath 30 °C) from methanol (HPMCAS-MF and PVPVA), dichloromethane-methanol 2:1 (HPMC, Methocel E5) or THF (CA Sub), then dried under vacuum at room temperature for 24 h. P1 The 50% ASD was also made with 5% SDS (47.5% HPMCAS: 47.5% TPV: 5% SDS); its preparation is not described separately. P1·M6 The results report 10, 30 and 50% TPV for HPMCAS and PVPVA, 10 and 30% for HPMC, and 10% for CA Sub. A2 CA Sub is a cellulose acetate suberate (degree of substitution 1.82 for acetate and 0.9 for suberate, 21.2 kDa) made in-house. M5

The four polymers span a range of hydrophobicity. Their measured aqueous solubilities are 3.3 ± 1.3 (CA Sub), 20.5 ± 5.9 (HPMCAS-MF), 37.1 ± 0.5 (HPMC) and 45.9 ± 2.2 mg/mL (PVPVA), and their calculated solubility parameters 23.18, 21.47, 23.93 and 23.59 MPa^1/2. M2·M3·M4·M5 On this basis the authors class PVPVA and HPMC as more hydrophilic and HPMCAS and CA Sub as more hydrophobic ("can be considered"). C13 Note by the AI writer (not discussed in the paper): the two measures do not rank CA Sub and HPMCAS the same way; by solubility CA Sub is the most hydrophobic, by solubility parameter HPMCAS is. X6 The solubility method uses a 0.05 M potassium phosphate buffer, while the Table S2 footnote says 100 mM phosphate buffer. X5

Dissolution and mass flow in one cell

A PermeGear side-by-side cell with a 30 mm orifice held 32 mL of 100 mM sodium phosphate buffer pH 6.8 at 37 °C on each side of a regenerated cellulose membrane (MWCO 6-8 kDa). P4·M7·M9 The donor had a 2.0 cm cross-shaped stir bar and the receiver a 1.0 cm octagonal bar; the stirring speed is not printed. P4 UV fiber-optic probes (5 mm path in the donor, 10 mm in the receiver) read 270 nm with a baseline at 370 nm. P4 Most runs added enough ASD to give about 150 µg/mL TPV if fully dissolved; the 90% HPMCAS and 90% CA Sub ASDs were also added at 5 to 50 mg (theoretical 16 to 156 µg/mL). P4·A4 The reference curve came from adding a small volume of concentrated methanolic TPV to the donor at 5-130 µg/mL, without polymer. P5

Probing complexation and the drug-rich phase

To test for soluble drug-polymer complexes, the authors dissolved 1.05 mg/mL of each polymer in buffer, which matches a fully released 10% TPV ASD with 120 µg/mL drug. C9·P6 In these solutions they measured TPV crystalline solubility (excess crystals, 300 rpm, 37 °C, 72 h, centrifugation at 21,100 × g, triplicate), the amorphous solubility (GLPS concentration) by adding 120 µg/mL TPV from a methanol stock and centrifuging, the mass flow at 120 µg/mL in the diffusion cell, and the size and zeta potential of the aggregates by DLS. P7·P8·P6·P9 For the DLS, the solution viscosity was taken from measurements at 25 °C. P10·X9

To measure what enters the drug-rich phase, ASDs were dissolved at 150 µg/mL TPV in 15 mL buffer at 300 rpm and 37 °C for 20 min (90% PVPVA), 30 min (90% HPMCAS) or 1 h (70% HPMCAS, 90% CA Sub), filtered through 1.0 µm glass fibre and centrifuged. P12·P13·P14 The pellet was analysed for PVPVA by the iodine reaction (470 nm) or for cellulosic polymer by the phenol-sulfuric acid reaction, and for drug by Nile red fluorescence. P12·P13·P14 The authors assume that the filter passes the aggregates and that the pellet holds the GLPS particles with any polymer on or in them. C10 The methods contain inconsistencies here: the PVPVA standards are printed as standards "of TPV", the PVPVA run used "pH 6.5" buffer, and the volume to which the µg amounts refer is not stated. X4·X8

As a control for membrane effects, a hydrophobic PVDF membrane impregnated with lecithin in n-dodecane replaced the cellulose membrane for the 90% HPMCAS and 90% CA Sub ASDs (12 mg added). P20·M8

Films, particles and release of each component

Thin films were spin-coated from 50 mg/mL solutions (THF for CA Sub) in a glove box below 18% RH, to avoid water-induced phase separation before the test. P17·C11 Films on silicon were immersed in buffer for 1 s, 5 min or 20 min, dried and imaged by SEM; others were immersed for 15 s or 10 min and imaged by AFM with Lorentz contact resonance (AFM-LCR), which distinguishes materials by the resonance frequency of the cantilever in contact with the surface. P18·P19 Cryo-SEM imaged the aggregates in solution after 2 h; modulated DSC and XRPD characterized the starting ASDs; optical micrographs followed particles in buffer (the method for these is not described). P16·P2·P3·P22 For the 50% HPMCAS ASD with and without SDS, the release of TPV (UV) and HPMCAS (phenol-sulfuric acid) was followed in 75 mL buffer in a beaker at 300 rpm and 37 °C for 150 min (n = 3). P15

Experimental analysis

The ceiling: what 90% HPMCAS shows

Adding 5 to 50 mg of the 90% HPMCAS ASD to the donor gave apparent concentrations close to those expected from the amounts added, about 16 to 166 µg/mL against 16 to 156 µg/mL, reached within about 5-20 min (Figure 1a). R1 The authors read this as rapid and complete release ("indicate"). C14 The mass flow rose almost linearly to about 1.01 µg/min at about 78 µg/mL and stayed at about 1.05-1.18 µg/min at 94-156 µg/mL (Figure 1b). R1 The drug-only reference followed the same shape and levelled at about 0.97-1.07 µg/min from 90 µg/mL. R2 Above 90 µg/mL the solutions became turbid, and cryo-SEM showed discrete particles (Figure 1c); the text gives a diameter of about 300 nm from DLS and fluorescence data that are not shown. R3 The authors conclude that this ASD undergoes GLPS above 90 µg/mL, and they explain the agreement with the reference by fast release and maintained supersaturation. C16·C17

Note by the AI writer (not discussed in the paper): below 90 µg/mL the ASD points in Figure 1b lie about 0.08-0.17 µg/min above the reference points at matching concentrations, so "similar" holds for the plateau more than for the rising part. X11·C15 In SI Figure S1, cryo-SEM of the same ASD shows discrete particles at 90-130 µg/mL and cell-like networks of ridges at 150 and 200 µg/mL; the paper does not interpret the change. R4

Case 1: five ASDs reach the ceiling

With about 150 µg/mL TPV available, five ASDs passed the GLPS line and reached the reference ceiling (Figure 3). R5·C20 Table A lists how fast each got there and the flux it reached.

Table A. The five ASDs that undergo GLPS (read from Figure 3, approximate; mass flows from Figure 3f, where the reference ceiling is drawn at about 1.0 µg/min). R5·R2

The authors summarize the maximum for the 90% HPMCAS, HPMC and PVPVA ASDs as ~1.1 µg/min. C28 The 90% PVPVA solution, however, was less stable: it peaked at about 108 µg/mL, fell below 90 µg/mL from about 43 min and to about 20 µg/mL at 100 min, which the authors attribute to crystallization, so "the supersaturation advantage of this formulation was lost". R5·C21 With predissolved PVPVA the mass flow was 1.04 ± 0.01 µg/min before crystallization and 0.11 ± 0.01 µg/min after. R13

The films tell the same story about speed (Figure 4). After immersion, heterogeneous domains appeared and then disappeared, faster at 10% than at 30% drug; at 20 min the 90% HPMC film had fully dissolved and the other 10% films left only sparse residues, while the 70% films still covered the substrate. R6·C22 AFM-LCR after only 15 s showed the 90% HPMC and 90% HPMCAS films as a network of undissolved material around patches of bare substrate (contact resonance 138 kHz for substrate, 135 kHz for film), and the 90% PVPVA film as a continuous film with circular pits down to the substrate (140 and 137 kHz) (Figure 5). R7 The authors infer that part of each film dissolved within seconds ("it appears"). C23·C24 Two captions do not match the images: Figure 4 names a "70%HPMC: 10%TPV" film that the panels label 30% TPV, and Figure 5 names only two of its three films. X1·X2

Case 2: CA Sub, same apparent concentration, lower flux

The 90% CA Sub ASD released TPV more slowly: the 50 mg addition reached its plateau at about 30 min, against about 5 min for HPMCAS, and the 40 mg addition was still rising at 120 min (Figure 6a). R8·R1 Its films stayed homogeneous for 5 min in buffer and showed an unchanged surface by AFM after 10 min (Figure 7, SI Figure S2). R10·R11 The authors link the slow release to CA Sub being more hydrophobic than HPMCAS (aqueous solubility 3 against 21 mg/mL in the text). C25·C26 It still underwent GLPS, as cryo-SEM showed aggregates (Figure 6c). R9·C27 The text cites Figure S3 for the AFM spectra of this film; they are in Figure S2. X3

The apparent concentrations after 120 min were similar to those of the HPMCAS ASD, but the mass flow levelled at about 0.73-0.79 µg/min above the GLPS concentration (Figure 6b), against about 1.05-1.18 µg/min for 90% HPMCAS. R8·R1·C28 The authors put forward three possible explanations ("may be posited"): soluble TPV-CA Sub complexes lowering the free drug, CA Sub mixing into the drug-rich phase and lowering the chemical potential of TPV there, or CA Sub altering the membrane. C29 Each was then tested.

Table B. TPV with predissolved polymer (from Table 2; mass flow measured at 96-120 min, or 10-30 min before crystallization for PVPVA). R13·P6

Membrane. With the PVDF-lecithin membrane, the 90% HPMCAS ASD again gave more flux than 90% CA Sub (0.76 ± 0.04 against 0.64 ± 0.01 with 12 mg added), so the authors rule out a membrane artefact ("suggesting"); they add that no membrane-polymer interactions were found for HPMCAS-MF, HPMC and PVPVA either, although SI Table S3 shows data only for HPMCAS and CA Sub. R12·C30·X10

Complexation. With predissolved CA Sub, TPV's crystalline solubility rose to 25 µg/mL against 4.7-11 µg/mL with the other polymers, the amorphous solubility measured by centrifugation rose to 111 µg/mL against 96-102, and the mass flow at 120 µg/mL fell to 0.91 µg/min against 1.04-1.10 (Table B). R13 The authors find these two observations "consistent with" soluble complexes that reduce free TPV, as reported for resveratrol and HPMCAS; they treat the 111 µg/mL as free plus complexed drug, and note that very small insoluble complexes may also have escaped centrifugation ("It is also possible"). C31·C38·C32 CA Sub also gave the smallest and most negatively charged aggregates (76 nm, -23.6 mV). R13

Polymer in the drug-rich phase. The drug-rich pellet after dissolving the 90% CA Sub ASD held 9407 ± 1396 µg polymer and 1080 ± 6 µg drug, a polymer fraction of 0.90 (Table 3). R14 For 90% HPMCAS, 70% HPMCAS and 90% PVPVA the fractions were 0.21, 0.27 and 0.22. R14 The fractions follow from the printed amounts: for example, 415/(415 + 1593) = 0.21 for 90% HPMCAS. E1 The authors suggest that for CA Sub the dispersed phase "might be more correctly termed a polymer-rich phase", and that the amount of polymer in the drug-rich phase appears to depend on polymer solubility. C35·C36 Note by the AI writer (not discussed in the paper): the HPMCAS and PVPVA fractions are practically equal even though PVPVA is about twice as soluble, so the solubility link rests on CA Sub alone; and the absolute drug amounts are hard to reconcile with the stated conditions, since 1593 µg of the 2250 µg added in 15 mL would leave only about 44 µg/mL dissolved, below the GLPS concentration. X7·X8

The authors' summary: the CA Sub flux is lowered by soluble TPV-CA Sub complexes and by substantial mixing of CA Sub into the drug-rich phase, which lowers the chemical potential of TPV in the dispersed phase; "These two phenomena explain why the CA Sub system has a lower mass flow rate than HPMCAS dispersions". C37

Case 3: slow release at 50% HPMCAS, and the SDS rescue

Both the 50% HPMCAS: 50% TPV ASD and the 47.5% HPMCAS: 47.5% TPV: 5% SDS ASD were amorphous with a single Tg before the test (Figure 10). R15·C39 Their behaviour in buffer was very different (Table C).

Table C. The 50% HPMCAS ASD with and without SDS (read from Figures 8 and 9, approximate; n = 3 except the single representative runs of Figure 8d). R16·R17·R18·R19·P4

Without SDS the ASD released slowly and, according to the authors, did not undergo GLPS because the drug crystallized in the matrix (Figure 8e); its maximum mass flow was correspondingly low. R16·R18·C40 Readings above 90 µg/mL after about 600 min are ascribed to undissolved particles interfering with the measurement; the text locates this in "Fig. 9a", but the crossing is in Figure 8a. C41·X3 SEM of films at 50% drug left more material after 20 min than at 10% (SI Figure S3 against Figure 4). R20·C42

The authors' "plausible explanation" for the matrix crystallization is incongruent release. In the beaker HPMCAS left the particles ahead of TPV, about 102 against 59 µg/mL at 60 min, which the text calls "approximately twice as much polymer" as drug. R19·C43 The undissolved matrix therefore becomes richer in TPV, a TPV-rich layer forms at the particle surface, and with less polymer there to inhibit crystallization, the drug crystallizes. C43 Note by the AI writer (not discussed in the paper): the measured ratio is closer to 1.4-1.8 than to 2, and the HPMCAS values scatter (about 102, 81, 107 and 92 µg/mL from 60 to 150 min); the surface layer itself was not measured. X17·X16

With 5% SDS, the TPV dissolution rate increased and matrix crystallization was reduced: "the drug was released faster than matrix crystallization occurred", though not completely, and the solution passed the amorphous solubility and gave "the expected high mass flow rate". C45·R16 Release remained incongruent; in fact the gap between HPMCAS and TPV was larger and appeared earlier (about 164 against 94 µg/mL at 60 min). C47·R19 The final SDS concentration, about 15 µg/mL, is taken to be far below its critical micelle concentration, so solubilization of the drug is not expected. R21·C48 Note by the AI writer (not discussed in the paper): HPMCAS in the SDS run reached about 164 µg/mL, above the nominal 150 µg/mL that the 23.7 mg sample could release into 75 mL. X16

The rescue has a cost. At a higher dose (15 mg, about 230 µg/mL possible) the SDS ASD peaked at about 159 µg/mL and then fell from about 600 min to about 96 µg/mL at 1000 min, while the ASD without SDS kept rising to about 150 µg/mL (Figure 8d). R17 The authors conclude that SDS improves matrix release but "may promote solution crystallization", as seen for other drugs, and that for TPV supersaturation "is likely to be maintained" in the presence of the surfactant long enough for absorption in vivo. C49·C50

Case 4: PVPVA at 30 and 50% drug crystallizes from solution

The 70% PVPVA: 30% TPV and 50% PVPVA: 50% TPV ASDs peaked at about 83 and 72 µg/mL at about 50 min, below the 90 µg/mL line, then fell to about 20 and 25 µg/mL at 100 min and about 15 and 18 µg/mL at 200 min (Figure 11). R22 Their maximum mass flows, measured at 35-60 min before the fall, were about 0.89 and 0.70 µg/min. R22·P4 The authors write that both ASDs "dissolved quickly" but did not undergo GLPS because TPV crystallized rapidly from solution. C52 Note by the AI writer (not discussed in the paper): "quickly" holds against the 50% HPMCAS ASD, not against 90% PVPVA, which passed 90 µg/mL within about 4 min; and the crystals were not characterized. X15·X13

Both ASDs were amorphous by XRPD. By DSC the 50% PVPVA ASD showed one Tg and the 70% PVPVA ASD two overlapping Tgs, from which the authors infer that the latter "may have undergone partial phase separation" (Figure 12). R23·C53 They attribute the poor performance of the PVPVA ASDs to two factors ("can be attributed"): possible phase separation in the solid state, to which PVPVA ASDs are prone through water uptake, and weak inhibition of crystallization in solution because PVPVA's hydrophilicity favours interaction with water rather than with the hydrophobic drug. C55·C54 Note by the AI writer (not discussed in the paper): the 50% PVPVA ASD with a single Tg gave the lower flux, so solid-state phase separation cannot explain the difference between the two loadings. X14

Discussion

The paper's argument can be drawn as one chain that splits at two points: whether the drug reaches the GLPS concentration at all, and, if it does, how much free drug the drug-rich phase supports. C1

In this picture release speed and crystallization tendency set the first branch, and the polymer's affinity for the drug and for the drug-rich phase sets the second. The authors put it as: "We have demonstrated that drug loading, polymer selection, and surfactant addition critically influence the release profile of an ASD and the maximum achievable flux". C57 Most low-loading systems reached the maximum flux and "are expected" to enhance bioavailability, but high loadings are "normally preferred" for dosage size and cost, which is exactly where slow release and crystallization appear. C58·C59 Slow release with matrix crystallization "can be averted" by a surfactant that facilitates wetting, and polymer selection affects release, crystallization in matrix and solution, complexation and mixing in the drug-rich phase. C60·C61 An optimal ASD should therefore generate and maintain supersaturation and form drug-rich colloids that act as a reservoir. C56 Bile salts, which inhibited TPV crystallization from solution in other work, "may enable promising strategies" for high-loading ASDs, a point the authors leave for further study. C51

Note by the AI writer (not discussed in the paper): two practical lessons follow from the data. First, a polymer selected because it inhibits crystallization well can still cost flux: CA Sub was chosen for that reason, yet among the 10% ASDs it gave the lowest flux. R24·R8 Second, the failure mode at high loading depends on the polymer: the hydrophobic HPMCAS released too slowly and let the drug crystallize in the particle, the hydrophilic PVPVA released but let it crystallize in solution, and the surfactant that cured the first problem shortened supersaturation at a higher dose. R16·R22·R17 None of these differences is visible in the apparent concentrations alone for CA Sub, which is the authors' main methodological point: dissolution and diffusion should be measured together. C2·C28

Limitations

Stated by the authors. The authors qualify several of their own explanations. Complexation is "consistent with" the data, and very small insoluble complexes might have escaped centrifugation ("It is also possible"). C31·C32 Matrix crystallization from incongruent release is "a plausible explanation". C43 Phase separation of the 70% PVPVA ASD "may have" occurred. C53 Supersaturation with SDS "is likely" to last long enough in vivo, and bile salts as release promoters require "further study". C50·C51

The AI writer's own judgement (1): what was measured and what was inferred. The case 1 pattern and the low fluxes of CA Sub, 50% HPMCAS and high-loading PVPVA are measured directly, in one cell and against a drug-only reference. R5·R8·R16·R22·R2 The causes are partly inferred. Crystallization from solution is inferred from falling concentrations, with no solid-state analysis of the precipitates, and matrix crystallization rests on one optical micrograph. X13 For CA Sub, complexation and polymer mixing are proposed together, but neither the free drug concentration nor the chemical potential of TPV in the CA Sub-rich phase was measured, so their shares are unknown. X12 Below the GLPS concentration the CA Sub ASD follows the drug-only reference, and its deficit relative to HPMCAS there comes from HPMCAS values above that reference; the direct evidence for complexation is the predissolved-polymer experiment at 120 µg/mL. X11 The DLS diameters (76-156 nm) come from antisolvent addition with predissolved polymer and cannot be compared with the 300 nm stated for ASD dissolution. X9

The AI writer's own judgement (2): comparability and reporting. The maximum mass flows come from different time windows, from 10-60 min for 90% HPMCAS to 800-1000 min for 50% HPMCAS, and Table 1 lists a window for a 50% HPMC ASD whose results are not reported. X18 Several captions and cross-references are wrong (Figures 4, 5, 9a, 9f-g and S3), the PVPVA pellet experiment is described with a pH 6.5 buffer, and the basis of the Table 3 amounts is not stated. X1·X2·X3·X4·X8 The solubility link for polymer uptake rests on CA Sub alone, and the hydrophobicity ranking of CA Sub against HPMCAS depends on which measure is used. X7·X6 The work covers one drug, one membrane and one medium (pH 6.8 phosphate buffer at 37 °C), and n is not stated for several figures. P4

Conclusion

The paper combines dissolution and membrane mass flow in one experiment and uses it to sort telaprevir ASDs into four cases. C1·P4 Fast-releasing ASDs with HPMCAS, HPMC and PVPVA at 10% drug, and with HPMCAS and HPMC at 30%, passed the GLPS concentration and reached about 1.03-1.19 µg/min. R5 CA Sub reached similar apparent concentrations but about 0.73-0.79 µg/min, which the authors attribute to soluble complexes and to a drug-rich phase that is 90% polymer. R8·R14·C37 At 50% drug, HPMCAS released slowly and the drug crystallized in the particle (about 0.65 µg/min) until 5% SDS was added (about 1.07 µg/min), at the price of earlier crystallization at a higher dose; PVPVA at 30 and 50% drug lost the flux to crystallization from solution (about 0.89 and 0.70 µg/min). R16·R17·R22 The authors' conclusion is that drug loading, polymer and surfactant "critically influence" both release and the maximum flux, and that mass flow must be measured alongside dissolution when formulations are compared. C57·C2