Pretomanid amorphous solid dispersions with HPMCAS-HF: a stable formulation that works fasted and fails fed

Background and motivation

Pretomanid (PTM, also PA-824) is a nitroimidazooxazine that the FDA approved in 2019, in combination with bedaquiline and linezolid, for drug-resistant pulmonary tuberculosis (Section 1). It is poorly water soluble (10–20 µg/mL, log P 2.75), and it is better absorbed with food: in healthy adults, 200 mg taken with a high-calorie, high-fat meal raised Cmax by 76% and AUC0-inf by 88% (Winter 2013), and the label recommends dosing with food. M1·R28 Absolute oral bioavailability has never been measured in humans; in preclinical work it was 19% in monkeys given a carboxymethylcellulose suspension and 52% in rats given a cyclodextrin formulation. R29 The authors read this as room for a solubility-enabling formulation, which "may be useful" to raise the extent of absorption. C2

Earlier attempts to enable PTM were few: amorphous drug loaded into mesoporous silica at 28% w/w released more drug in pH 7.4 buffer than crystalline drug, and spray-dried porous particles were developed for inhalation, but inhaled exposure was low (6.7–13.5%) and an oral suspension reduced lung bacterial burden more in guinea pigs. Oral formulations with better bioavailability therefore remain of interest (Section 1).

The amorphous solid dispersion (ASD), in which a drug is mixed with a polymer at the molecular level, is the established route for such drugs: on release it creates a supersaturated solution and can raise absorption. Its main failure mode is crystallization, either in the hydrated ASD matrix or from the supersaturated solution after release; if the drug crystallizes quickly, the bioavailability gain "is likely to be reduced". C3 The polymer therefore has two jobs, to release the drug and to inhibit its nucleation and crystal growth, and commercial products use only a handful of polymers: neutral PVPVA and HPMC, and the weakly acidic (enteric) HPMCAS and HPMCP. C3

Two pieces of earlier work shape the study. The first concerns the drug: low-molecular-weight drugs tend to crystallize faster than heavy ones, from both the amorphous and the solution state (Alhalaweh 2014, 2019; Baird 2010). C5 With a molecular weight of 359.26 g/mol, PTM was expected to be a fast crystallizer for which "only a limited number of polymers" would work as inhibitors. C6 The second concerns enteric polymers. HPMCAS and HPMCP dissolve only when enough of their carboxylic acid groups are ionized, which happens above a threshold pH (Hiew 2021). C26 Immersion in gastric fluid, where these polymers are insoluble, has been shown to cause crystallization on the ASD surface for weakly basic drugs (Nguyen 2023, ref 30; Elkhabaz 2019), and the effect of food on ASDs made with enteric polymers has rarely been reported. C64·C79

The paper set out to measure how far and how long PTM can be supersaturated, find polymers that inhibit its crystallization, use them to design ASDs optimized for release and supersaturation in one- and two-stage dissolution tests that mimic fasted and fed conditions, and dose a subset to cynomolgus monkeys in both prandial states against the commercial tablet, which contains crystalline PTM. A2·A4·A8

Background concepts

Supersaturation and its ceiling. The driving force for crystal nucleation and growth from solution is the excess chemical potential of the solute over the saturated solution. The paper expresses it, as is usual, by the supersaturation ratio (Equation 1):

In plain terms: $S$ is the solution concentration $C$ divided by the crystalline (equilibrium) solubility $C_s$; at $S=1$ there is no driving force, and the higher $S$, the faster crystals tend to form. E1 For PTM in pH 6.5 phosphate buffer, $C_s$ is 16.2 µg/mL, so the screening concentrations of 80 and 160 µg/mL are supersaturations of about 5 and 10. R1.d·E1 Supersaturation cannot rise without limit. At the amorphous solubility the solution separates into a drug-rich and a drug-lean phase (liquid–liquid phase separation, LLPS), and further drug only adds to the drug-rich phase; the amorphous solubility is therefore the practical upper limit of supersaturation. C4 The ratio of amorphous to crystalline solubility (the A/C ratio) gives the largest supersaturation an ASD can reach in a given medium. E3

Induction time. The nucleation induction time is the time from creating a supersaturated solution until the first crystals are detected; here detection is the rise in light scattering at a wavelength where the drug does not absorb (440 nm). P3 A polymer that doubles or triples it gives modest protection; one that extends it from minutes to hours makes an ASD viable.

Glass stability and the reduced glass transition temperature. The tendency of a pure amorphous drug to crystallize is classed in the cited work (Baird 2010; Van Eerdenbrugh 2010), where class I marks rapid crystallizers. A related index is the reduced glass transition temperature, the ratio of the glass transition temperature to the melting temperature, both in kelvin:

In plain terms: $T_g$ is the glass transition temperature and $T_m$ the melting temperature, both in kelvin; the paper uses the ratio (stated in words, Section 4.1) as one index of how readily a drug crystallizes from the undercooled melt. PTM, with $T_g$ 10 °C and $T_m$ 147.3 °C, has $T_g/T_m$ = 0.67, close to its analog delamanid (0.64), and both are class I rapid crystallizers. E2·M1·R30·C59

Enteric polymers and their dissolution threshold. HPMCAS (hypromellose acetate succinate) carries acetyl and succinoyl groups. The succinoyl groups are carboxylic acids with a pKa of about 5, and the polymer dissolves only once a critical fraction of them is ionized, which requires the medium to reach a threshold pH. C26·C83 All grades have a pKa of about 5, but the succinoyl content sets the dissolution threshold. HF has the highest acetyl:succinoyl ratio and the lowest succinoyl content of the grades, which makes it the most hydrophobic grade, and it needs a higher pH than the others to start dissolving, reported as 6.5 or ">6.5" depending on the source cited. C15·C83·M3·X21 This hydrophobicity is what tends to make HF a good crystallization inhibitor, and its high threshold is what makes its release slow.

Ways to speed release from an enteric-polymer ASD. The paper tries three. A faster-dissolving HPMCAS grade can be blended in. A surfactant can be added; surfactants may improve wetting, dispersion and drug solubility but "may also promote" crystallization or phase separation. C19 Or a base can be co-formulated: it raises the microenvironmental pH at the dissolving surface and, "likely", forms a polymer salt that dissolves faster, as shown earlier for HPMCP (Qi and Taylor 2022). C26·C67 Bases here were added at a 1:1 molar ratio to the polymer's carboxylic acid groups; for every base the printed ASD composition (Table S2) corresponds to the same ≈0.79 mmol of base per gram of polymer. E7·P4

Biorelevant and two-stage dissolution. FaSSGF (pH 1.6) simulates the fasted stomach, FaSSIF V1 (pH 6.5) the fasted small intestine, and FeSSIF V2 the fed small intestine, with more bile salt and lipid and a lower pH (5.8 nominal; 5.0–6.0 here). In the two-stage or pH-shift test, the formulation spends 1 h in FaSSGF and the medium is then converted to FaSSIF. P11 For an enteric polymer, the gastric hour is a period in which the polymer cannot dissolve but the drug can still crystallize at the particle surface. C63·C64

Pharmacokinetic terms. Cmax and Tmax are the height and time of the plasma peak; AUC is the area under the plasma concentration–time curve (to 24 h, to the last sample, or extrapolated to infinity); MRT is the mean residence time. Relative bioavailability here is the dose-normalized AUC0-inf of an ASD as a percentage of that of the crystalline reference dosed fasted; food-effect relative bioavailability is a formulation's fed AUC0-inf as a percentage of its own fasted value. E4·E5

Methods in detail

The study is a staged formulation programme: screen polymers, build ASDs on the one that works, try to fix its release, then test the three best candidates as tablets in vitro and in monkeys. All in vitro work was at 37 °C. A1–A8

Stage 1: how fast does PTM crystallize, and what slows it?

The problem. Before choosing a polymer, the authors needed the size of the supersaturation window and the speed at which PTM leaves it.

What they did. Crystalline solubility was measured by shaking excess drug for 48 h at 37 °C in HCl pH 1.6, 50 mM phosphate buffer at pH 3.0, 5.0 and 6.5, 55 mM maleate buffer pH 5.8, FaSSIF V1 and FeSSIF V2, with ultracentrifugation and HPLC (n = 3). P1 Amorphous solubility was found by pumping a 50 mg/mL methanolic stock into the medium at 100 µL/min and recording the concentration at which the extinction at 400 nm rose sharply, the onset of LLPS (the paper does not state the temperature of this measurement). P2 Induction times were measured at 37 °C with a UV probe reading the scattering at 440 nm every 30 s. For the polymer screen, eight polymers were predissolved at 100 µg/mL in pH 6.5 phosphate buffer and PTM added to 80 µg/mL: Eudragit L100, CAP, HPMCP-50 and HPMCP-55, HPMCAS-LF, -MF and -HF, and the neutral HPMC. HPMCAS-HF was then retested at 160 µg/mL PTM with 1 mg/mL polymer, in buffer and FaSSIF, and at 100, 150 and 200 µg/mL with 500 µg/mL polymer. P3·A2·A3 Glassy PTM was made by quenching the melt in liquid nitrogen, then heated in the DSC or left at room temperature. P9·P19

A caveat on the screening conditions. Note by the AI writer (not discussed in the paper): the methods say induction time was measured "at a concentration corresponding to the amorphous solubility" in each medium, but Table 2 reports every induction time at 160 µg/mL. That is near the amorphous solubility in the buffers (141–166 µg/mL) but about half of it in FaSSIF (308 µg/mL), so the FaSSIF entry is at a supersaturation of about 3, not 10. X2·R1

Stage 2: screening ASD powders

The problem. HPMCAS-HF needs a high pH to dissolve, so its ASDs release slowly. The question was whether release could be sped up without losing crystallization inhibition. C65·C66

What they did. ASDs were made by rotary evaporation from dichloromethane:methanol 1:1 (10% w/v solids, 45 °C), dried overnight, cryo-milled and sieved to 106–250 µm. P4 The series were: HF alone at 10, 20 and 25% DL; MF alone at 10% DL; 50:50 (by weight) blends of HF with MF or LF at 10 and 20% DL; SLS or TPGS at 10% of the drug weight at 10 and 20% DL; TPGS at 5–50% of the drug weight at 20 and 25% DL; and eight amines at 20% DL. P4·A4·A5·A6 The amines spanned pKa 7.73 (triethanolamine) to 10.64 (triethylamine): triethylamine, 2-dimethylaminoethanol (DMEA), N-methyldiethanolamine (MDEA) and triethanolamine (tertiary), and tris, meglumine, ammediol and proline sodium (primary or secondary). M13–M20 Powders holding 10 mg PTM were tested in a USP II apparatus at 150 rpm, either in 50 mL FaSSIF V1 for 2 h or in a pH shift: 45 mL FaSSGF pH 1.6 for 1 h, then 5 mL of 10× concentrated FaSSIF (in 0.57 M phosphate, pH 7.3) to give 50 mL FaSSIF pH 6.5 for another hour. Full release would give 200 µg/mL, and the concentration was followed in situ by fiber-optic UV (n = 3). P11 Solid residues were examined by PXRD and SEM, Tg by DSC, and drug integrity by 19F NMR. P6·P8·P9·P10

Stage 3: tablets for the animal study

What they did. Three ASDs at 20% DL were spray-dried (inlet 80 °C, feed 6 mL/min): binary PTM-HF, PTM-HF-Tris and PTM-HF-TPGS (20% TPGS relative to drug). P5·P13 Each tablet held 150 mg ASD (30 mg PTM) with 30 mg croscarmellose sodium, 30 mg sodium starch glycolate and 90 mg microcrystalline cellulose, compressed to 300 mg, 3–4 kp, 11 mm diameter. The reference was 120 mg of crushed commercial PA-824 tablet (30 mg PTM; the product contains SLS) in a size 0 HPMC capsule. P13·M2 Tablets and capsules were tested in 150 mL (to keep the 200 µg/mL ceiling) at 75 or 150 rpm in FaSSIF V1, in FeSSIF V2 at pH 5.0, 5.5, 5.8 and 6.0, and in the pH shift. P12 After gastric immersion and after low-pH FeSSIF, the particles were examined by PXRD and SEM; how the solids were recovered and how long they stood in FeSSIF are not stated. P18·X19

Stage 4: the monkey study

Why monkeys. Dogs clear PTM rapidly and absorb it poorly, giving very low exposures with any formulation or route, so cynomolgus monkeys were the non-rodent species. C7

What they did. Four non-naive male cynomolgus monkeys (2.75–3.64 kg) each received 30 mg PTM as one tablet or capsule in a six-period crossover with washouts of at least 7 days (Table 1): the reference and the three ASDs fasted, and the reference and the binary PTM-HF ASD fed. Fasted animals had food withheld overnight until 4 h after dosing; fed animals were given food about 30 min before dosing and ate 44–73 g. Blood was sampled up to 48 h and PTM measured by LC-MS/MS (range 1–3000 ng/mL). P13·P14·P15 PK parameters came from noncompartmental analysis in WinNonlin with the linear-up/log-down trapezoidal rule. Relative and food-effect bioavailabilities were calculated from individual AUC0-inf values normalized to each animal's actual dose in mg/kg, and group differences were tested by one-way ANOVA with Duncan's method. P16·P17·E4·E5·E6

Note by the AI writer (not discussed in the paper): because of the dose normalization, the printed relative bioavailabilities do not equal ratios of the mean AUCs in Table 3 (for example 155% printed, 151% from the means for PTM-HF; 163% printed, 149% from the means for the fed reference). Both are legitimate, but they should not be mixed. X14

Experimental analysis

Result 1: a narrow window that closes in minutes

Table A. Solubility and induction time of PTM at 37 °C, mean ± SD, n = 3 (from Table 2). R1·X1

The crystalline solubility of PTM hardly changes with pH (15.2–17.8 µg/mL from pH 1.6 to 6.5), which the authors relate to its structure, and it rises about 3-fold in FaSSIF and 6-fold in FeSSIF through solubilization by bile components. R1·C8·C9 The amorphous solubility is 8–10 times higher in buffers, and SLS or TPGS at about 200 µg/mL raise it further in pH 6.5 buffer, to about 226 and 275 µg/mL against about 166 (Figure S1). R1·R2·C9 So PTM can in principle be supersaturated about tenfold. The problem is how long it stays there: without polymer, crystals appeared within 0.9–1.6 min at 160 µg/mL in every medium. R1·C10 The printed A/C ratio for FaSSIF (6.0) does not follow from the two solubilities in the same row (308/53.8 = 5.7); the other ratios reproduce. X1

The neat amorphous drug is no better. Melt-quenched PTM has a glass transition near 10 °C and, on heating, crystallizes just above it (exotherm peaking near 48 °C) and the crystals melt at 147.3 °C (onset; peak 151.67 °C, 47.80 J/g) (Figure S2A). A glass left at room temperature visibly crystallized within a few minutes (Figure S2B). R3·R4·C10 The DSC trace also shows an endotherm near 98 °C that the text does not mention; the form of the crystals produced on heating is not identified. X23

Polymer screen. At 80 µg/mL PTM (supersaturation about 5) with 100 µg/mL polymer in pH 6.5 buffer (Figure 1), most enteric polymers barely helped: induction times were about 1.6–3.1 min with Eudragit L100, CAP, HPMCP-50, HPMCP-55 and HPMCAS-LF, against about 1.7 min with no polymer. R5 HPMCAS-MF and the neutral HPMC extended it to about 21.5 and 23.6 min. With HPMCAS-HF no crystals appeared for up to 1000 min (the bar is cut at the top of the axis). R5·C11 The text describes the weak polymers as giving "2–5 min"; the bars read lower, and HPMCP-55 even falls slightly below the drug alone. X4

At 160 µg/mL, the supersaturation the authors call "approximately 10", 1 mg/mL HPMCAS-HF still delayed crystallization to 43.8 ± 1.3 min in buffer and 77.6 ± 13.6 min in FaSSIF (Table S1). R6·C12 With 500 µg/mL HPMCAS-HF, 100 µg/mL PTM stayed clear for more than 12 h, but 150 and 200 µg/mL crystallized almost immediately (Figure S3). R7 The authors read this as the polymer failing once the drug exceeds its amorphous solubility, in the presence of drug-rich droplets. C13 Note by the AI writer (not discussed in the paper): 150 µg/mL is just below the measured amorphous solubility in this buffer (166.1 µg/mL), the polymer concentration differs from Table S1, and no droplets were detected in these runs, so the data place the loss of inhibition between 100 and 150 µg/mL at 500 µg/mL polymer rather than tie it to the amorphous solubility. X3

The authors conclude that HPMCAS-HF was "the only realistic polymeric candidate" (Section 4.1). They contrast PTM with delamanid, a structural analog (Tg about 43 °C, also class I) whose supersaturated solutions were stabilized by HPMCAS, HPMCP and PVPVA alike, and attribute ("can be attributed") PTM's stronger crystallization tendency to its low Tg, low molecular weight and relatively linear structure. C14·C60·C61·R30

Result 2: grade blends help at 10% DL and hurt at 20% DL

At 10% DL in single-stage FaSSIF (Figure 2A), the binary HF ASD released more slowly and slightly less than the MF ASD (about 144 against 153 µg/mL at 120 min). R8.a–b·C16 In the pH-shift test (Figure 2B) all four ASDs released about 52–57 µg/mL in the gastric hour; after transfer, the MF ASD shot up to about 167 µg/mL and then fell to about 125 µg/mL, "consistent with crystallization", while HF and the two 50:50 blends kept rising, the blends to about 156–157 µg/mL against 143 for HF. R8.e–h·C16·C17 In FaSSIF the LF-HF blend reached about 164–173 µg/mL, but the MF-HF blend, though faster at first, ended below HF (about 136 against 144). R8.c–d·X5

At 20% DL the blends lost (Figure 2C, D): at 120 min, HF about 150 µg/mL, MF-HF about 133 and LF-HF about 87 in FaSSIF; about 116, 108 and 72 after the pH shift. R8.i–n Undissolved LF-HF particles clumped in the vessel (Figure S5A), and SEM of 10% DL blend particles after 1 h at pH 1.6 showed fibrous, needle-like surfaces where HF alone was largely smooth (Figure S4B). The authors attribute the loss "likely" to surface crystallization and agglomeration, and in the discussion add that HPMCAS-HF itself is what keeps the released drug in solution, so diluting it with other grades was "detrimental". R9·R10·C18·C69·C70 Note by the AI writer: the PXRD patterns in Figure S4A show only amorphous halos, and they come from 10% DL particles, so the crystallization explanation for the 20% DL loss rests on SEM texture. X6

Result 3: TPGS lifts release, SLS does not at higher loading

Table B. PTM concentration during release of HPMCAS-HF ASD powders with surfactant (% of drug weight), 10 mg PTM in 50 mL, 150 rpm; read from Figures 3 and 4 (approximate). The 20% DL TPGS 10% row combines the separate runs of Figures 3 and 4. R11·R12

At 10% DL either surfactant raised release, to about 176–180 µg/mL in FaSSIF and 159–168 µg/mL at the end of the pH shift (Figure 3A, B). R11.a–f The text puts this at "nearly 90%" in FaSSIF and "approximately 75%" in the pH shift; the figure's end values correspond to about 80–84% of the 200 µg/mL maximum. C20·X7 At 20% DL, SLS gave a flat, lower plateau (about 115 µg/mL), while TPGS still helped. R11.g–l·C21 Raising the drug loading from 20 to 25% cut the binary ASD's FaSSIF release from about 150 µg/mL at 60 min to under 70 µg/mL, and TPGS rescued it in proportion to its amount (Figure 4C). R12.a·R12.i–l·C23·C24 In the pH shift, more TPGS released more drug during the gastric hour, but all TPGS levels ended at nearly the same value after transfer (Figure 4B, D). R12.e–h·R12.m–p·C25 TPGS also plasticized the ASDs: Tg fell from about 79 to 63 °C at 20% DL (0 to 40% TPGS) and from about 75 to 53 °C at 25% DL (0 to 50% TPGS) (Figure S6). R13·C22 The 20% DL ASD with 20% TPGS was chosen for the tablets. C25

Result 4: bases speed release except the weakest

At 20% DL, seven of eight amines made release in FaSSIF much faster: peaks of about 158–180 µg/mL within about 15–30 min, against about 155 µg/mL reached only after about 65 min for HF alone (Figure 5A, C). R14.a–i·C27 Some then fell back, for example MDEA from about 158 to 131 µg/mL by 120 min. R14.d The exception was triethanolamine, the weakest base (pKa 7.73), which held release to about 70 µg/mL. R14.b·C27 The pH-shift test separated the bases more (Figure 5B, D): most raised gastric-hour release (about 69–81 µg/mL against 45), and final values ranged from about 54 (triethanolamine) to 167 µg/mL (proline sodium), with tris at about 144. R14.j–r·C28 The proline sodium ASD showed an extra small 19F NMR signal (Figure S7), which the authors read as reduced drug stability ("appeared to show"); note by the AI writer: its main PTM peak is not shifted more than in the other ASDs, so the concern rests on that minor peak. R15·C29·X22 Tris was chosen because it is already used as a counterion in approved products and has an established safety profile. C30

Result 5: tablets in vitro, and what stirring does

The reference capsule released PTM after a 5–6 min lag for the shell, then matched the original tablets (Figure 6, Figure S8). It reached about 31 µg/mL in FaSSGF, about twice the crystalline solubility at pH 1.6, which the authors attribute ("likely") to the SLS in the product; about 46 µg/mL in FaSSIF; and about 106 µg/mL in FeSSIF pH 5.8, close to the crystalline solubility there. In FeSSIF at pH 5.0–6.0 and 75 rpm it plateaued at about 103–107 µg/mL whatever the pH. R16·C31·C32·C33·C34 Note by the AI writer: the colour legend in the Figure S8 caption is swapped relative to the curve labels; the in-plot labels match Figure 6. X20

Table C. PTM concentration (µg/mL) from 20% DL ASD tablets, 30 mg in 150 mL; read from Figure 7 (approximate). R17

Every ASD tablet released faster and more than the reference capsule. C35 At 150 rpm in FaSSIF the three tablets peaked at about 160–175 µg/mL within 20 min and fell to about 124–128 µg/mL by 120 min (Figure 7A); the text calls the tris and TPGS differences "minor improvements". R17.a–c·C36·X9 In FeSSIF pH 5.8 at 150 rpm all three plateaued at about 160–171 µg/mL, which the authors find interesting given the polymer's reported dissolution threshold of pH 6.5 (Figure 7C). R17.g–i·C37

Slowing the paddle to 75 rpm separated the tablets (Figure 7B, D). The binary and TPGS tablets still peaked early, then fell to about 89 and 105 µg/mL in FaSSIF; the tris tablet rose slowly, to about 103 µg/mL at 20 min, but ended highest, at about 127 µg/mL. In FeSSIF pH 5.8 at 75 rpm the end values were about 128 (HF), 143 (TPGS) and 162 µg/mL (tris). R17.d–f·R17.j–l Photographs (Figure S9) show the tris tablet still intact on the paddle at 3 min and collapsed into one agglomerate at 5 min, while the binary tablet had dispersed by 3 min. R18·C39 The authors summarize that slower mixing "resulted in reduced drug release" and that the tris formulation's gelation was masked at high stirring. C38·C68·C74 Note by the AI writer: that summary holds for the early part of the tris curve; at 120 min the tris tablet was the highest at 75 rpm in both media. X10

Result 6: low pH in FeSSIF caps release and crystallizes PTM

Table D. Release of 20% DL ASD tablets in FeSSIF V2 adjusted to pH 5.0–6.0, 75 rpm; read from Figure 8 (approximate). Crystalline solubility in FeSSIF V2 pH 5.8 is 108.1 µg/mL. R19·R1.g

For the binary ASD, pH 5.8 and 6.0 gave supersaturated solutions that held, whereas at pH 5.0 and 5.5 the concentration only reached about the crystalline solubility (Figure 8A). R19.a–d·C40 Note by the AI writer: the crystalline solubility was measured only at FeSSIF pH 5.8, so "about the crystalline solubility" at pH 5.0–5.5 assumes it changes little with pH, as it does in buffers. R1 With tris the threshold for supersaturating release moved down to pH 5.5 (Figure 8B). R19.e–h·C41 After exposure to FeSSIF at pH 5.0 and 5.5, both ASDs showed sharp PXRD peaks at the positions of crystalline PTM, and SEM showed needle-like structures covering the particles (Figure S10). The authors write that this matrix crystallization "likely" accounts for the incomplete release, and in the abstract attribute the reduced release at lower pH to "reduced polymer solubility and an enhanced crystallization tendency of the drug". R20·C42·C56 Note by the AI writer: the tris ASD also showed crystalline peaks at pH 5.5, where it still released about 142 µg/mL, and the exposure time before the solid-state analysis is not given, so the PXRD shows that crystallization occurs at low pH but not how much it limits release. X16·X19

Result 7: the gastric hour crystallizes PTM in every ASD

In the pH-shift test at 150 rpm (Figure 9A), the ASD tablets released more than the reference during the gastric hour (about 75, 90 and 67 µg/mL at 60 min for HF, tris and TPGS against about 31) and more again after transfer, ending at about 128, 133 and 137 µg/mL against about 46. R21·R16.c·C43 But these final values stay below the single-stage FaSSIF peaks (about 160–175 µg/mL), which the authors read as "a deleterious change during immersion in the gastric stage, most likely due to drug crystallization". R21·R17.a–c·C44 PXRD after 1 h in gastric medium showed PTM peaks in all three ASDs, qualitatively HF-TPGS > HF > HF-Tris, and SEM showed needle-shaped crystals on the surfaces of the spray-dried particles (Figure 9B, C). R22·C45 Note by the AI writer: the comparison holds for maxima; by 120 min the single-stage curves had themselves fallen to about 124–128 µg/mL. X11

Result 8: fasted monkeys, every ASD beats the crystalline product

Table E. Mean (SD) PK parameters after a single 30 mg oral dose, n = 4 male cynomolgus monkeys (from Table 3; AUC0-last omitted). Footnotes: c, p < 0.001 versus PTM-HF fasted; d, p < 0.05 versus PTM-HF-Tris fasted; e, p < 0.01 versus PTM-HF-TPGS fasted; f, p < 0.05 versus reference fed. R23

Fasted, all three ASDs gave more exposure than the crystalline reference: relative bioavailability 155% for PTM-HF, 142% for tris and 141% for TPGS. R23.h·C46 Cmax values did not differ significantly, but the binary and TPGS ASDs were absorbed faster, with mean plasma concentrations at 1 h of about 1140 and 1530 ng/mL against about 520 for the reference (marked ** and *** in Figure 10A). R24·C47 The tris ASD lagged: about 215 ng/mL at 1 h and Tmax 4 h against 2–2.5 h for the others. R24·R23.b·C48 The binary ASD gave the highest AUC; the authors report its 155% as significant (p < 0.05) and call it the best fasted formulation. R23.f·C49 Note by the AI writer: Table 3 itself marks no significance for any fasted group, so this p value is stated only in the text. X12

The authors' reading is that adding TPGS or tris "did not translate to the in vivo situation", most clearly for tris, whose prolonged Tmax "can be attributed to the gelation tendency of this formulation, which was masked in the in vitro study when a high stirring rate was used"; "it appears that gelation of susceptible formulations may be more likely to occur in vivo than in a relatively well-stirred dissolution vessel". C73·C74·C76 They also concede that their initial in vitro conditions "were not optimized" to mimic physiology. C75

Result 9: food helps the crystalline product and hurts the binary ASD

With food, the crystalline reference showed the expected positive food effect (Figure 10B): absorption started later (about 70 against 520 ng/mL at 1 h; Tmax 4 against 2.5 h), but Cmax rose to 1702 ng/mL and the food-effect relative bioavailability was 163%. R23·R25·C50 In ratio-of-means terms Cmax rose by 23% and AUC0-inf by 49%, smaller than the 76% and 88% reported in humans, where the dose (200 mg) and the meal (high-calorie, high-fat) also differed. R33·R28·C78·X17 The authors link the positive effect to the higher solubility of PTM in fed-state medium (108.1 against 53.8 µg/mL in FaSSIF), in line with the expectation that high-log P, poorly soluble drugs gain from bile solubilization. C77·C78

The binary ASD went the other way. Fed, its Cmax fell to 765.8 ng/mL (p < 0.001 against fasted), Tmax rose to 19 h and MRT to 16.0 h, AUC0-24h and AUC0-last were significantly lower than fasted, and the food-effect relative bioavailability was 72.2%. R23.a–g·R23.i·C53 Mean plasma levels were still about 300 ng/mL at 24 h, when every other group was near zero. R25 The authors read the long Tmax and MRT as "slow and prolonged" release ("suggests"). C52 The individual profiles (Figure S11F) show that this mean hides two behaviours: monkey #3 absorbed as if fasted, peaking at about 1870 ng/mL at 4 h, while #1, #2 and #4 rose slowly to only about 380, 535 and 280 ng/mL at 24 h. R26 The authors note "a large variation between individual monkeys" and that group differences were "not always statistically significant". C54 Note by the AI writer: the negative food effect is therefore a three-of-four-animal result. X15

Discussion

How the authors explain the fed-state failure

The authors' explanation ties the polymer screen to the fed-state result through intestinal pH. The abstract presents it as a hypothesis ("was hypothesized"), the discussion as an attribution, with "likely" for the crystallization step. C55·C84

The causal chain as the authors give it (Sections 4.1 and 4.2). The fed branch was not measured in the animals. C61·C14·C65·C83·C49·C55·C84·R31

Each step has some support. Food delays gastric emptying, which the authors say "presumably" explains the later onset in both formulations. C51 Cynomolgus monkeys in the fed state are reported to have a longer and more variable gastric emptying time (up to 5 h) and a small-intestinal pH around 5–6, lower than humans or dogs. R31·C82 HPMCAS grades share a pKa of about 5, but HF, with the least succinoyl, needs the highest pH to start dissolving, "which may lead to an extended time lag and reduced dissolution rate". C83 In vitro, exactly this pH range (FeSSIF pH 5.0–5.5) capped binary-ASD release at about the crystalline solubility, and crystalline PTM appeared in the particles. R19·R20 The authors also cite albendazole ASDs with HPMC and HPMCP-55, whose bioavailability in rabbits fell from 100% at normal gastric acidity (pH about 1) to 62% at low acidity (pH about 5), as an example of pH sensitivity, and they suggest that pH variation between animals "might also contribute" to the large scatter in the fed group. R32·C81

The crystalline reference, by contrast, does not need a polymer to dissolve; its fed-state gain follows the higher solubility of PTM in bile-containing medium, as for many poorly soluble, high-log P drugs. C77·C78 So the sign of the food effect depends on the formulation, not only on the drug. The authors note that ASDs of posaconazole and ziprasidone were "not markedly affected by food" although their crystalline forms showed positive food effects, which makes the PTM result unusual among the few reports on enteric-polymer ASDs. C79

What the in vitro tests got right and wrong

The authors draw a general lesson: the findings "emphasize the limitations of in vitro assessment of ASD performance using standardized media conditions", and in vitro release in FeSSIF at its standard pH and in FaSSIF at high stirring did not predict the in vivo order. C57·C73 Three observations support this. In FeSSIF pH 5.8 at 150 rpm, the binary ASD reached about 160 µg/mL, a clearly supersaturated solution, yet fed exposure fell. R17.g·R23.i The ternary formulations, which improved powder release in vitro, gave no gain over the binary ASD in fasted monkeys. C66·C73·R23.h And the tris tablet, nearly equivalent at 150 rpm, was the slowest early at 75 rpm, matching its 4 h Tmax in vivo. R17.e·R23.b·C74 The authors conclude that HPMCAS-HF ASDs of PTM "may be highly sensitive to intestinal pH conditions" and call for in vitro methods that capture food effects for enteric-polymer ASDs. C85·C58

Note by the AI writer (not discussed in the paper): two of the in vitro tests did point in the right direction. The FeSSIF pH series showed the binary ASD losing its supersaturating release below pH 5.8, the pH range cited for the fed monkey intestine, and the low-stirring test exposed the slow early release of the tris tablet. For a formulation built on a high-threshold enteric polymer, these two variants, rather than the standard media, are the ones that bracket the risk. R19·R17·R31

Implications for formulation

The authors state that HPMCAS-HF was "the only viable choice from a crystallization inhibition perspective" for PTM, that ternary ASDs with TPGS or a base improved in vitro release, and that the binary ASD performed best in fasted monkeys. C14·C66·C49 They recommend further work on food effects and on in vitro methods that correlate with in vivo outcome for enteric-polymer ASDs. C58

Note by the AI writer (not discussed in the paper): the study shows a trade-off that is likely to recur for small, fast-crystallizing drugs. The property that makes HPMCAS-HF the best inhibitor, its hydrophobicity, goes with the highest dissolution threshold, which makes release depend on intestinal pH. The two fixes tested trade off differently: TPGS raised release but also gastric-stage crystallization (most PXRD peaks in Figure 9B), while tris lowered the pH needed for release to 5.5 but slowed tablet disintegration. Neither ternary ASD was dosed fed, so whether tris, which kept release high at FeSSIF pH 5.5, would have avoided the negative food effect remains open. R22·R19.e–h·R18·P13

Limitations

Limits stated by the authors

• Large variation between individual monkeys; differences between groups "not always statistically significant". C54
• The initial in vitro conditions "were not optimized to mimic the physiological conditions"; a lower stirring rate was needed to separate the formulations. C75
• Species differences in GI pH, transit time and motility can affect formulations with pH-dependent solubility; the fed cynomolgus monkey has a lower intestinal pH than humans. C82
• Further studies "are required" to understand food effects and to correlate in vitro observations with in vivo outcomes for enteric-polymer ASDs. C58

The AI writer's own judgement

Overall, the evidence is moderate. The polymer screen and the fasted crossover are direct and consistent, and the reversal of the food effect for the binary ASD is large. R5·R6·R23 The weaker points are these:

• Small study, uneven fed result. Four animals; the fed binary-ASD means combine one animal that absorbed as when fasted with three slow absorbers. X15
• Mechanism inferred, not measured. Intestinal pH, polymer dissolution and PTM crystallization in the animals were not measured; the support is the FeSSIF pH series, the PXRD and SEM after low-pH exposure, and cited monkey physiology. X16
• Statistics partly text-only. The significance of the binary ASD's 155%, of the other ASDs' gains and of the reference fed-versus-fasted difference is stated in the text but not marked in Table 3; relative bioavailabilities are dose-normalized and differ from ratios of the mean AUCs. X12·X14
• Text summaries stronger than the curves. "Minor improvements" for the tris and TPGS tablets, "reduced release" at low stirring, release percentages of the surfactant ASDs and the two-stage versus single-stage comparison read differently from the figures. X7·X9·X10·X11
• The polymer's failure above the amorphous solubility rests on three points at a different polymer concentration, one of them below the measured amorphous solubility. X3
• Small source inconsistencies: the FaSSIF A/C ratio (6.0 printed, 5.7 from the values), MRT called "median retention time" in the text and mean residence time in Table 3, swapped colours in the Figure S8 caption, and a reference to Table S2 for the blend compositions, which Table S2 does not contain. X1·X13·X20·X8
• Scope. One drug, one species, one dose (30 mg), male animals only, and only two formulations dosed fed. P13·P14

Conclusion

The paper takes a drug that crystallizes within minutes and finds that only one polymer, HPMCAS-HF, holds it supersaturated for hours. R5·R6·C14 ASDs built on it released far more PTM than the crystalline product in vitro and gave 141–155% relative bioavailability in fasted monkeys, with the plain binary ASD doing best; adding TPGS or tris improved release in vitro but not exposure in vivo. R17·R23.h·C49·C73 With food the binary ASD lost about a quarter of its fasted exposure while the crystalline product gained, and the authors attribute the loss ("was hypothesized", "likely") to slow dissolution of the enteric polymer at the lower fed-state intestinal pH and crystallization of the drug in the waiting matrix, an explanation supported by in vitro tests at pH 5.0–5.5 but not measured in the animals. R23.i·C55·C84·X16 Their wider conclusion is that HPMCAS-HF-based ASDs of PTM "may be highly sensitive to intestinal pH conditions" and that standardized in vitro media do not capture this. C85·C57