Why some surface excipients rescue release from high-drug-loading amorphous particles: surface composition, wetting and exposed area

Background and motivation

Amorphous solid dispersions (ASDs) raise the oral bioavailability of poorly water-soluble drugs because the amorphous drug can supersaturate in the gut as long as crystallization is avoided. The polymer does several jobs: in the solid it raises the glass transition temperature and slows nucleation; in solution it stabilises the colloidal drug-rich phase formed above the amorphous solubility, maintains supersaturation and prevents matrix crystallization during dissolution (refs 2, 11–18 in the paper). The price is drug loading (DL). Much polymer is usually needed, and for high-dose drugs this means a large tablet or many tablets. The authors' example is the marketed elbasvir/grazoprevir product (Zepatier, 50 mg elbasvir plus 100 mg grazoprevir): the elbasvir ASD intermediate holds 20% drug with 70% HPMC and 10% TPGS, the grazoprevir intermediate 30% drug with 65% copovidone and 5% SDS, and these intermediates are only part of the final tablet. The authors attribute the difficulty of raising DL for these two drugs to their poor aqueous solubility.

Release usually gets worse as DL rises. With copovidone (PVPVA) this decline is abrupt above a certain DL (Indulkar 2019, ref 21), and recent work linked it to water-induced phase separation in the gel at the dissolving surface, which leaves a hydrophobic continuous phase that blocks further release of drug and polymer (Deac 2023, refs 22, 23). These observations had been made with copovidone. For the enteric polymer hypromellose acetate succinate (HPMCAS), other factors also matter: the pH and buffer capacity of the medium (Bapat 2023, ref 24) and, with basic drugs, ionic drug–polymer interactions (refs 20, 25).

The pH-dependent solubility of HPMCAS can, however, be turned into a manufacturing route. Drug and polymer are dissolved in a common organic solvent, and the solution is mixed rapidly with acidified water, in which HPMCAS is insoluble, in a rotor–stator or overhead mixer; the shear precipitates amorphous particles at once. The products are called co-precipitated amorphous dispersions (cPADs), and they have been reported to have larger specific surface areas than spray-dried or hot-melt-extruded ASDs (Jia 2022, ref 28; Schenck 2019, ref 31). The authors describe precipitation into acidified water as more environmentally friendly than manufacturing routes that need large volumes of organic solvent (refs 27–29).

In earlier work the same group (Hiew 2023, ref 29) added a water-soluble polymer, HPMC, to the acidified anti-solvent. The HPMC was deposited on the precipitated particles, producing what the authors call a hierarchical cPAD, and grazoprevir cPADs with DL as high as 70% then showed good in vitro release and good in vivo bioavailability; in vitro release predicted the in vivo exposure well. The explanation given was surface dilution: the anti-solvent held by capillarity in the filter cake carries the dissolved polymer, which coats the particle surfaces as the cake dries. This worked only when the polymer was soluble in the anti-solvent, and another hydrophilic polymer, hydroxypropyl cellulose, did not give the same benefit.

That last observation is the starting point of this paper: why are some secondary polymers effective and others not? The authors prepared hierarchical cPADs of grazoprevir and HPMCAS (75:25) with seven hydrophilic substances as the secondary excipient: five polymers (HPMC, PVP, PEG, gelatin, poloxamer 407) and two surfactants (SDS and TPGS). For comparison, they also took a high-DL ASD made by solvent evaporation (SE) and processed it under the same high-shear conditions in acidified water containing HPMC (and, in a further set, SDS). Release was measured in a two-stage biorelevant test, and the particles were characterised by scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS) and contact angle measurements. The authors state the aim as mechanistic understanding that can guide "the systematic selection of an appropriate polymer or other secondary excipient".

Background concepts

Hierarchical particles and secondary excipients. In a binary cPAD the particle consists of drug and the primary polymer (here HPMCAS) only. In a hierarchical cPAD a second, water-soluble component, the secondary excipient, is dissolved in the anti-solvent; after precipitation, filtration and drying it ends up mainly on the particle surface. The intended effect is to lower the fraction of drug at the surface (surface dilution) while the bulk DL stays high. The materials in this paper are named by a code that encodes the recipe (Table 1): for example, H-HPMC10-HCl-120m is a hierarchical cPAD precipitated into 0.001 N HCl containing 10 mg/mL HPMC, with the slurry mixed for 120 min; the suffix -L marks material made in a wet mill at larger scale, and -s (or -Ls) material sieved through 600 µm. SE-20, SE-30 and SE-75 are binary grazoprevir–HPMCAS ASDs made by rotary evaporation at 20, 30 and 75% DL, and SE75-… codes denote SE-75 particles treated afterwards in a high-shear slurry.

Surface composition by XPS. XPS measures the elements in roughly the top 10 nm of a sample. The authors fitted the carbon (C 1s) signal of each sample with the reference spectra of grazoprevir, HPMCAS and the secondary excipient, which gives the share of carbon atoms that belongs to each component. They then converted these atomic ratios to mole ratios using the number of carbon atoms per molecule, and to a weight ratio using the molecular weights, to report a "surface grazoprevir concentration" in percent. The paper describes this conversion only in words (Materials and methods, p.293). A surface grazoprevir concentration below the bulk DL indicates that the surface is enriched in excipient.

Wettability and contact angle. Before a solid can dissolve, the medium must displace the air at its surface and enter the pores of the powder. A common measure is the contact angle of a drop placed on the solid: a low angle means the liquid spreads (good wetting), a high angle means it beads up. The authors placed a 4 µL drop of SGF or FaSSIF on a layer of particles stuck to double-sided tape and followed the angle for 10 s by video. They classify materials with an angle at or below 90° as having good surface wettability and those above 90° as poorly wettable. On a powder bed the measured angle reflects both the chemistry of the particle surface and the roughness and porosity of the bed, so it is a comparative rather than an absolute measure.

Two-stage dissolution and the role of HPMCAS. The release test mimics the passage from stomach to intestine: the material is first stirred for 30 min in 50 mL simulated gastric fluid (SGF, pH 1.2), then 50 mL of double-strength FaSSIF V1 is added, giving fasted-state intestinal fluid at a final pH of 6.5, for a further 60 min. HPMCAS is insoluble at gastric pH and dissolves only in the intestinal stage, so almost no drug is released in the first 30 min, whatever the formulation. The dose was chosen so that complete release would give 150 µg/mL grazoprevir in the final 100 mL, which the paper states is the amorphous solubility of grazoprevir in FaSSIF V1 at 37 °C; the test is therefore non-sink. Concentrations were measured in line with UV probes using second-derivative spectra.

Accessible surface area and agglomeration. Release from a powder depends on how much particle surface the medium can reach. Primary particles can stick together into agglomerates, especially during drying, which hides surface. The authors probe this in two ways: by sieving dry material through a 600 µm sieve, which breaks large agglomerates, and by sonicating the material for 1 min in SGF before the dissolution test, which breaks agglomerates in the wet state. Neither changes the composition of the particles, so a change in release after sieving or sonication points to accessible surface area rather than to chemistry.

Methods in detail

The study has three strands that run in parallel: making high-DL particles with different surfaces, testing their release, and characterising the surfaces. The flow below summarises them.

Contribution 1: A screen of secondary excipients in hierarchical cPADs

① Problem. HPMC had worked as a secondary polymer and hydroxypropyl cellulose had not, but there was no explanation. If surface dilution alone were the mechanism, any water-soluble excipient deposited at a similar level should help.

② How. For the HPMC cPADs, 1.5 g grazoprevir and 0.5 g HPMCAS were dissolved in 10 mL tetrahydrofuran (THF). The anti-solvent, 100 mL of 0.001 N HCl containing 5 or 10 mg/mL HPMC, was chilled in an ice bath and stirred at 23,000 rpm with a rotor–stator homogenizer (IKA T 25 Ultra-Turrax), and the drug solution was injected from a syringe into the high-shear zone of the dispersion tool. A white precipitate formed immediately. The slurry was mixed at 8,000 rpm for 5 min, half of it was filtered off (the "-5m" material), and the rest was mixed for a further 115 min at below 10 °C ("-120m"). For the other six excipients the batch was halved: 0.75 g drug and 0.25 g HPMCAS in 10 mL THF were injected into 50 mL of 0.001 N HCl containing 0.25 g of excipient (5 mg/mL), and the slurry was mixed 5 min. A combined material used 5 mg/mL HPMC plus 0.5 mg/mL SDS in the anti-solvent. Two controls complete the set: the binary cPAD without any secondary excipient (B75-HCl-5m, made as in ref 29), and a cPAD precipitated into n-heptane in a dry-ice acetone bath, with 0.25 g of micronized HPMC added to the slurry afterwards (H-µHPMC5-heptane-5m), in which HPMC is dispersed rather than dissolved. All materials were isolated by vacuum filtration and dried with a nitrogen stream.

③ Scale and processing variables. To see whether the process transfers to larger equipment, 75 g grazoprevir and 25 g HPMCAS in 500 mL THF were pumped directly into a Quadro HV0 wet mill recirculating 5 L of 10 mg/mL HPMC in 0.001 N HCl at 1 °C (30 m/s tip speed during addition, 10 m/s during recirculation, slurry kept at or below 5 °C). Material was collected after 5 and 60 min of recirculation, dried overnight under nitrogen and then in a vacuum oven, and about 10 g of each was sieved through 30 mesh (600 µm) to break agglomerates. The authors also varied the slurry mixing time (5, 60, 120 min) because a long hold in the mother liquor might raise the risk of crystallization, while some time might be needed for the dissolved excipient to distribute uniformly.

④ How it differs from earlier work. The earlier paper (ref 29) compared two cellulose polymers. Here the excipient set spans polymers of different chemistry and molecular weight and two surfactants, and release is set against direct measurements of the surface (XPS, contact angle) rather than inferred from processing alone.

Contribution 2: Surface treatment of a conventional high-DL ASD

① Problem. If the surface, not the bulk, controls release at high DL, the same benefit should be obtainable by treating ready-made ASD particles, without co-precipitation.

② How. Binary grazoprevir–HPMCAS ASDs at 20, 30 and 75% DL were made by rotary evaporation from methanol:dichloromethane (1:1 v/v, 20% w/w solids, bath 50 °C), dried overnight under vacuum, cryomilled (three 1 min cycles in liquid nitrogen) and sieved below 250 µm. SE-75 particles were then stirred at 8,000 rpm for 5 min in 50 mL of 0.001 N HCl containing 10 mg/mL HPMC (0.75 g ASD) or 72.5 mg/mL HPMC (0.75 g ASD), or in 50 mL of deionized water containing 5 mg/mL SDS (0.5 g ASD), and filtered and dried. A fourth sample, SE75-MTBE-10s, was stirred for only 10 s in methyl tert-butyl ether and filtered at once; the paper does not state its purpose, and only its contact angle and its 90 min release (in Figure 11a) are reported. SE-20 and SE-30 serve as benchmarks: a treated 75% DL material that releases like SE-30 has, in effect, recovered the performance of a formulation with less than half its drug loading.

Contribution 3: Measuring surface composition, wetting and accessible area

① Problem. To separate the candidate explanations, each one has to be measured on the same materials: how much drug is at the surface, how well the surface wets, and how much surface is available.

② How. SEM (platinum-coated, 5 kV, 5000×) showed morphology before and after exposure to SGF. XPS (monochromatic Al Kα, five spots per sample) gave the surface grazoprevir concentration of five materials, fresh and after 30 min in SGF followed by overnight drying. Contact angles of FaSSIF were measured on seven materials over 10 s (n = 4). Accessible area was changed by sieving (above) and by 1 min sonication in SGF before the dissolution test; the paper does not describe the sonicator or its settings. Finally, to test whether HPMC is drawn into the precipitating particles, 2.5 or 5 mg of HPMC (as a 5 mg/mL solution in 0.001 N HCl) was added to HPMCAS dissolved in THF (5 mg/mL) at 37 °C, the precipitate was centrifuged off after 30 min, and the HPMC remaining in the supernatant was measured by the phenol–sulfuric acid method at 490 nm (n = 3).

③ Why it works. The three surface measurements respond to different things. XPS gives chemical composition but says nothing about whether water spreads. The contact angle gives wetting but not composition. Sieving and sonication change the exposed area without changing composition. A factor that tracks release across all materials while the others do not is the better candidate for the cause. Note by the AI writer (not discussed in the paper): the comparison is cleanest when the excipient level is the same for all materials; in this paper the HPMC cPADs used for the contact angle and XPS measurements were made with 10 mg/mL HPMC, the other excipients at 5 mg/mL.

Experimental analysis

Experimental setup. All release data come from the same non-sink two-stage test (37 °C, paddles at 100 rpm, n = 3, error bars one standard deviation). In every material less than about 1% of the drug dissolved during the SGF stage, so the comparisons below are about the FaSSIF stage. The paper gives release as curves (Figures 5–8); the values in the tables are read at 90 min from those figures and are approximate. Figures 5 and 7 carry dashed lines for the 90 min release of the unprocessed SE ASDs: about 63% for SE-20, about 43% for SE-30 and about 30% for SE-75.

Table A. Release of grazoprevir from 75:25 grazoprevir–HPMCAS hierarchical cPADs at 90 min of the two-stage test (read from Figures 5 and 8, approximate; PVP reads about 21% in Figure 5c and about 20% in Figure 8g).

Main result 1: the secondary excipient decides release. Table A shows a range of about 13% to about 47% at 90 min among cPADs that differ only in the excipient dissolved in the anti-solvent. Gelatin and poloxamer 407 lowered release below that of the binary cPAD and below SE-75, and the authors note that their addition "resulted in decreased drug release even when compared to the SE ASDs". PVP gave about 20%, PEG about 30% (the only polymer of the four that matched SE-75), TPGS about 31% and SDS about 37%; both surfactant cPADs stayed below 40%, "comparable only to the SE-75 ASD". The calculated DLs of the gelatin, PEG, PVP and poloxamer cPADs were similar to that of H-HPMC5-HCl-5m (73.9%, from Table S1 as quoted in the text), so the differences do not come from different amounts of excipient. SEM showed the same morphology for all cPADs: an intertwined network of rod-like structures that the authors describe as highly porous (Figure 4). Since morphology was alike and release was not, the authors conclude that "factors beyond particle morphology are also important".

The HPMC series adds a dose effect. With 5 mg/mL HPMC the cPADs released only about 23–24%, below SDS, TPGS and PEG at the same concentration, and the calculated DLs (73.9 and 73.7%) did not change with mixing time. With 10 mg/mL HPMC release rose to about 47% (5 min mixing) and about 52% (120 min), close to SE-30; the longer mixing lowered the calculated DL to 71.6% and gave slightly faster release just after the switch to FaSSIF. The authors call HPMC and SDS "the most promising" secondary excipients. Because the HPMC-based particles were seen not to wet well during the test while SDS is known to promote wetting, they combined the two: 5 mg/mL HPMC with 0.5 mg/mL SDS gave about 47%, comparable to SE-30, "suggesting that HPMC and SDS may have synergistic impact on dissolution enhancement".

Table B. Release from surface-treated 75% DL SE ASDs and the unprocessed references (read from Figure 6, approximate; 34 min is the second time point after the switch to FaSSIF at 30 min).

Main result 2: the same treatment lifts a conventional high-DL ASD. Table B shows that a 5 min high-shear treatment in HPMC solution raised the release of the 75% DL ASD from about 30% to about 43% (10 mg/mL HPMC, the SE-30 level) or about 57% (72.5 mg/mL HPMC, near SE-20, about 63%). SEM explains what was added: the cryomilled SE-75 particles were dense with smooth faces and sharp edges, and after HPMC treatment their surfaces were rough and irregular and smaller particles had agglomerated into larger ones (Figure 2), which the authors read as "suggesting that HPMC formed a layer around the particles". After 30 min in SGF the HPMC-treated particles looked smooth again, like milled particles (Figure 3). SDS treatment raised release to about 54% at 90 min, but release started more slowly than from SE-20 and SE-30 (about 12% against about 29% and about 18% at 34 min). The authors note that less SDS than HPMC was applied (0.25 g SDS for 0.5 g ASD against 0.5 g HPMC for 0.75 g ASD), and SEM still showed the smooth original faces with small deposits below 2 µm. The text cites Figure 2c for this, but Figure 2c is the HPMC72.5 sample; the SDS sample is Figure 2d.

Table C. Surface composition, wetting and release side by side (assembled by the AI writer from Figures 9, 10 and 11 and the text; values read from the figures are approximate; XPS n = 5, contact angle n = 4, release n = 3). The paper shows these quantities in separate figures.

Main result 3: surface drug content does not separate good from poor excipients. All five materials analysed by XPS had a surface grazoprevir concentration well below the 75% target loading before SGF exposure, from about 32% (SDS cPAD) to about 50% (HPMC cPAD), so every excipient diluted the surface (Figure 9). The poorly releasing PVP (about 33%) and poloxamer (about 45%) cPADs lie inside the same range as the SDS and HPMC cPADs. After 30 min in SGF the surface grazoprevir rose by about 19–24 points for the HPMC, SDS and PVP cPADs but hardly changed for the poloxamer cPAD and for SE75-HPMC72.5-HCl-5m. The authors conclude that "regardless of the surface grazoprevir concentration, the type of secondary excipient used ... was observed to impact drug release" (supported in the paper by Figure S1, not available here), and later that the polymers "diluted the particle surfaces to similar extents". Note by the AI writer (not discussed in the paper): "similar" covers differences of up to about 17 points (PVP about 33% against HPMC about 50% before SGF), and the HPMC sample is the 10 mg/mL, 120 min material, not a 5 mg/mL one.

Main result 4: wetting splits the materials into two groups. In the contact angle test (Figure 10) every drop started between about 93° and 137°. On the SDS cPAD the drop then spread completely (0° from the first second). On the two HPMC cPADs the angle fell over a few seconds: H-HPMC10-HCl-60m-s levelled at about 52°, and H-HPMC10-HCl-120m kept falling to about 30° at 10 s. On PVP, poloxamer, the heptane cPAD and SE75-MTBE-10s the angle stayed at about 119–137° for the whole 10 s. Contact angles in SGF were higher than in FaSSIF for the same materials (Figures S2, S3). Plotted against release (Figure 11a), the wettable group released about 37–52% at 90 min and the poorly wettable group about 8–20%, and the authors write that release "largely correlated with contact angle ..., suggesting a relationship between wettability and effective drug release". Within the wettable group the order does not follow the angle: the completely wetting SDS cPAD released less (about 37%) than the HPMC cPADs (about 52%). After 1 min of sonication the four materials tested spread further apart (Figure 11b: about 82%, 66%, 33% and 14% for SDS, HPMC, PVP and poloxamer), which the authors describe as "a clearer relationship", adding that release from the SDS cPAD depended on wetting and on de-aggregation by sonication.

Main result 5: exposed surface area helps only wettable particles. Sonication did not change release in SGF but raised release in FaSSIF for every cPAD except the poloxamer one (Figure 8, Table A). The gains were largest for the wettable materials: about 24 → 73% for H-HPMC5-HCl-120m (three-fold) and about 37 → 82% for H-SDS5-HCl-5m (two-fold), which the authors summarise as an absolute increase of 45–50%. PVP gained about 13 points, PEG about 11, TPGS about 18, gelatin about 16 (with a large standard deviation) and poloxamer about 1. The sonicated media were more turbid, which the authors take as a sign of broken-up agglomerates; SEM after sonication showed the same rod-like structure, only less agglomerated (Figure 4). Sieving gave the same message for dry material: the wet-mill cPADs released about 43% (5 min) and about 36% (60 min) unsieved and about 59% and about 52% after sieving through 600 µm (Figure 7), roughly 15 points more. The unsieved wet-mill material was slightly below the homogenizer material (about 43% against about 47%). The authors conclude that "the surface area accessible to the dissolution media enhances drug release only if the particles demonstrate good wettability", and that larger-scale cPADs "may be more susceptible to agglomeration", which sieving or milling can reverse.

Main result 6: HPMC must be dissolved in the anti-solvent. The heptane control, in which micronized HPMC was dispersed in the slurry after precipitation, released only 9.2% at 90 min (profile "data not shown") and had the highest contact angle measured, 137 ± 5°, against about 23% release for the cPAD with the same 5 mg/mL HPMC dissolved in 0.001 N HCl. The authors conclude that HPMC "needs to be dissolved rather than dispersed in the anti-solvent during precipitation". The authors link this to their colorimetric experiment ("This was further aided by polymer miscibility"): when acidified HPMC solution was mixed with HPMCAS in THF, a suspension formed at once, and of 2.5 mg and 5 mg HPMC added, 1.88 ± 0.05 mg and 3.87 ± 0.01 mg remained in solution, so 0.62 and 1.13 mg had gone with the HPMCAS (HPMCAS:HPMC 1:0.12 and 1:0.23 in the precipitate; Table 2). The authors read this as "some affinity of HPMC for the precipitated HPMCAS phase, even though HPMC is a highly water-soluble polymer".

Discussion

How the authors connect the results. The argument runs as a causal chain. A particle that is 75% drug has a very hydrophobic surface, so it wets poorly; the authors found this "hardly surprising". A secondary excipient that is dissolved in the anti-solvent stays on the particle surface after drying and dilutes the drug there. Whether this helps depends on what the excipient does to wetting:

The authors' summary of the chain is that "ensuring proper wetting of the particles is pivotal for effective drug release", that the accessible surface area "should be as high as possible", and that its benefit "can be negated" when the particle surfaces are hydrophobic. The decisive step in the chain, why one hydrophilic excipient makes the surface wettable and another does not, is not measured in the paper. The authors discuss three candidate properties:

Molecular weight. HPMC (20,000 Da), poloxamer 407 (12,600 Da), PVP (40,000 Da), gelatin (20,000–25,000 Da) and PEG (3350 Da) do not rank by molecular weight in release, so "it appears that molecular weight was not a reliable predictor" of their efficacy as wetting agents. Note by the AI writer (not discussed in the paper): the structure drawn for poloxamer 407 in Figure 1f (x ≈ 78, y ≈ 30, z ≈ 78) corresponds to about 8.6 kDa, not 12,600 Da; this does not affect the argument.

Gel strength. All these polymers can hydrate to gels. A strong gel "could create a barrier that hinders access of the dissolution media", whereas a weak gel erodes quickly; the low-viscosity HPMC used here erodes about as fast as lactose (ref 41). Poloxamer 407 and gelatin are used for sustained release because of their gels, so "polymers that form a strong gel may be less suitable excipients for surface dilution of hydrophobic particles". This fits the two lowest releasers in Table A, but gel strength was not measured.

Affinity for the drug or the primary polymer. HPMC distributed between the precipitating particles and the aqueous phase (Table 2) and made the high-DL surfaces hydrophilic (Figure 10), "which suggests that affinity with the drug and/or primary polymer may be another important attribute to consider when choosing a secondary excipient". The affinity of the other excipients for grazoprevir or HPMCAS was not measured.

The surfactants. SDS performed slightly better than TPGS. Earlier work by the group (refs 45, 46) showed that SDS raises release from high-DL ASDs, which was "postulated" to come from higher molecular mobility through plasticization, and SDS is known to help de-aggregate particles (ref 47). Here, the authors argue, SDS was not molecularly mixed with drug and polymer but sat "predominantly" on the particle surface, so it "likely promoted drug release by facilitating particle wetting". Its location in the particles was not measured. The authors add a caution: SDS is a plasticizer, plasticization can predispose amorphous formulations to phase separation and recrystallization (refs 48–50), and "further studies are warranted" on the stability of cPADs with SDS, to find a "sweet spot" between better wetting and stability risk.

Processing. The authors see two practical routes. High-DL particles with hydrophilic surfaces can be made in situ with rotor–stator high-shear devices, and existing high-DL ASDs can be surface-treated with the same devices. For the second route they argue that applying a secondary excipient onto ready-made particles avoids designing a three-component ASD, lets the formulator choose the primary polymer that best stabilises the amorphous drug, and "optimizes both the stability and dissolution properties" of the formulation; physical stability was, however, not tested in this paper. At larger scale, cPADs "could be more susceptible to agglomeration and consolidation especially during the drying process", but sieving or milling brought their release to the level of the small-scale material. The authors also regard co-precipitation as reproducible between facilities and scalable, because the mixing process can be modelled and controlled (ref 34).

What it suggests for formulation. Note by the AI writer (not discussed in the paper): three practical points follow from the data. First, a contact angle of the medium on a powder bed is a quick screen that separated the working from the failing excipients here; it did not rank materials within the wettable group. Second, the amount of excipient matters: at 5 mg/mL HPMC was no better than PEG or TPGS, and its advantage appeared at 10 mg/mL or in combination with 0.5 mg/mL SDS. Third, measuring the surface composition by XPS confirms that dilution took place but did not predict release; a surface can be diluted and still not wet.

Limitations

Limitations stated by the authors. The authors state few limitations explicitly. They flag that SDS is a plasticizer whose effect on the physical stability of cPADs needs study, and that larger-scale cPADs may agglomerate more, especially on drying. They note that the slightly longer wetting time of HPMC cPADs compared with SDS cPADs is "likely negligible" over the length of the release test, and that the relationship between contact angle and release became clearer only after sonication, with release of the SDS cPAD depending on de-aggregation as well as wetting.

The AI writer's own judgement (1): what the correlation can carry. The central relationship (Figure 11) rests on seven materials in two well-separated clusters, with no correlation statistic. It supports a threshold statement, wettable versus poorly wettable, rather than a graded relationship; inside the wettable group the completely wetting SDS cPAD released less than the HPMC cPADs. The 90° split also refers to the angles after the first second, since at 0 s every drop sat at about 93° or higher. The comparison of excipients is not at matched levels: the HPMC cPADs in Figures 9–11 were made with 10 mg/mL HPMC, the other excipients with 5 mg/mL, and at 5 mg/mL HPMC released less (about 23%) than SDS, TPGS or PEG. The statement that sonicated HPMC and SDS cPADs "released at least 70%" holds for H-HPMC5-HCl-120m (about 73%) and H-SDS5-HCl-5m (about 82%) but not for H-HPMC10-HCl-120m (about 66%). The highest values lie at the top of the FaSSIF calibration range (0–120 µg/mL, against 150 µg/mL for complete release), and the sonication conditions are not described.

The AI writer's own judgement (2): surface composition and controls. XPS was measured on five materials, and no untreated particle (SE-75 or the binary cPAD) was analysed, so surface dilution is judged against the bulk loading rather than against a measured untreated surface. The discussion states that XPS showed surface grazoprevir below 75% for the HPMC- and SDS-treated SE ASDs, but Figure 9 contains only the 72.5 mg/mL HPMC sample. The conversion of XPS carbon ratios to weight fractions is described as dividing mole ratios by molecular weight, whereas a mass ratio requires multiplying by it, and the molecular basis used for the polymers is not stated; the numbers are therefore best compared within this paper. The dissolved-versus-dispersed HPMC comparison rests on one heptane material that also differed in cooling, THF volume and the moment HPMC was added, and its release profile is not shown. The role of SE75-MTBE-10s is not explained, and its release (about 8%) is below that of untreated SE-75 without comment.

The AI writer's own judgement (3): textual errors. Several errors in the text do not change the conclusions but can mislead a reader. The discussion twice says that the SE ASDs were treated with SDS "in acidified water", whereas the methods and Table 1 give deionized water. The SDS-treated SE ASD is cited as Figure 2c instead of 2d. The surface-dilution mechanism of the earlier cPAD work is cited to ref 10, a clinical trial of ledipasvir/sofosbuvir, instead of ref 29, and the amorphous solubility of grazoprevir is cited to ref 1, whose title concerns anacetrapib nanoparticles. Figure 1c draws HPMC without methyl groups. One material appears under two names (H-HPMC10-HCl-60m-Ls in Table 1, H-HPMC10-HCl-60m-s in Figures 10 and 11). The supplementary tables and figures (calculated DLs, rank order with sonication, release against surface concentration, SGF contact angles) were not available for this reading.

Conclusion

The paper makes three contributions. First, it shows with one drug, one primary polymer and one loading that the choice of secondary excipient in hierarchical grazoprevir–HPMCAS cPADs changes 90 min release from about 13% (poloxamer 407) to about 47% (HPMC at 10 mg/mL, or HPMC with a little SDS), with all particles looking alike under SEM. Second, it measures the candidate explanations side by side: XPS shows that every excipient diluted the surface drug to a similar range, while the FaSSIF contact angle splits the materials into a wettable group that releases and a poorly wettable group that does not; breaking agglomerates by sonication or sieving adds most to the wettable materials. Third, it shows that the same surface treatment can be applied to a conventional 75% DL ASD with a rotor–stator mixer, raising its release to the level of ASDs with 20–30% drug. The authors' conclusion is that "the use of judiciously chosen secondary excipients ... is a promising strategy for the preparation of high DL formulations while ensuring good wettability for improved dissolution performance".

The work leaves the mechanism one step short: why HPMC and SDS make the surface wettable and PVP, poloxamer 407 and gelatin do not is attributed to gel strength and affinity for drug or polymer, but neither was measured, and the physical stability of the surface-treated particles was not tested. The next tests follow directly from the data: contact angles and XPS on all excipients at the same concentration and on untreated particles, a measure of gel strength or erosion of the surface layer, and storage stability of the treated materials, particularly those containing SDS.