Crystals that form in the stomach: what happens to enteric posaconazole dispersions before and after the pH shift

Background and motivation

Many new drug candidates dissolve poorly, and the concentration a drug reaches in the gut lumen largely sets how much is absorbed (refs 1–4). Amorphous solid dispersions (ASDs), in which the drug is dispersed in a polymer in its amorphous form, are one of several ways to create a supersaturated solution: a solution that holds more dissolved drug than the crystal would allow. A supersaturated drug has a higher chemical potential than the crystal and can raise the flux across the intestinal wall, but the solution is metastable, and once the drug crystallizes the advantage is lost (refs 5–19). Polymers such as HPMC and hydroxypropyl methylcellulose acetate succinate (HPMC-AS) delay crystallization by acting on nucleation and crystal growth; inside the solid dispersion they also raise the glass transition temperature, slow molecular motion and can hydrogen-bond the drug (refs 20–23). The authors note that how polymers do this is "not fully understood". C1·C2

Inhibition inside the solid matters not only on storage but also before the dispersion has fully dissolved in the body. This is especially true for enteric polymers such as HPMC-AS and Eudragit L100. These are insoluble at gastric pH, so their dispersions do not dissolve in the stomach but sit there as a suspension until they reach the small intestine (refs 24–28). The authors point to a "limited understanding" of what this acid suspension does to later supersaturation and phase behaviour in the intestine. C3

Weakly basic drugs add a second twist. They are ionized and fairly soluble in the acidic stomach, but when the pH rises in the small intestine they turn neutral and their solubility can drop sharply, which creates supersaturation (refs 5, 29). The authors argue that in vitro tests for weak bases should therefore include this pH shift, and that biorelevant media, which mimic the change from gastric to intestinal pH and contain bile salts and lecithin, predict intraluminal behaviour more reliably. C4

Posaconazole is a case in point. It is a triazole antifungal, a BCS class II compound with a log P of 4.6, and a polyprotic weak base with pKa values of 3.6 (piperazine group) and 4.6 (triazole group) (ref 37). M1 Its oral bioavailability is variable, from 8 to 47% (refs 38–41): a non-fat or high-fat meal raised the AUC of the oral suspension 2.7- and 3.9-fold (ref 42), a proton pump inhibitor lowered absorption and an acidic drink raised it (ref 43). R21 When volunteers took acidified (pH 1.6) or neutral (pH 7.1) suspensions, Hens et al. found extensive precipitation in both cases, with some intestinal supersaturation (45 min) only for the acidified suspension (ref 44). R22 The marketed Noxafil delayed-release tablet is a hot-melt-extruded ASD of posaconazole in HPMC-AS (ref 45), designed to avoid extensive release in the stomach and some of this variability. After Noxafil, gastric aspirates showed little release in both fasted and fed volunteers, and fasted jejunal fluid stayed supersaturated for about 90 min, while precipitation seemed to dominate in the fed state (ref 46). C5·R22

The paper sets out to follow drug release, supersaturation and crystallization of posaconazole–HPMC-AS ASDs, and of Noxafil tablets, through an abrupt shift from fasted-state simulated gastric fluid (FaSSGF, pH 1.6) to fasted-state simulated intestinal fluid, version 2 (FaSSIF, pH 6.5). It adapts a two-stage dissolution set-up introduced for pH-dependent dissolution (Mathias 2013, ref 47; Mann 2017, ref 48), which needs no flow pump, keeps the pH consistent and allows flexible volumes, and it combines this with several methods that follow both the solution and the solid. C6·P6

Background concepts

Crystalline solubility, amorphous solubility and liquid–liquid phase separation. A crystal sets the lowest concentration at which dissolved drug is stable: the crystalline (equilibrium) solubility. An amorphous solid can dissolve to a higher concentration, the amorphous solubility. If more drug is added beyond the amorphous solubility, the solution does not simply become more concentrated: the excess separates as a second, drug-rich liquid phase of sub-micrometre droplets. This is liquid–liquid phase separation (LLPS), and the paper treats the LLPS onset concentration as the amorphous solubility. Above it, nanodroplets coexist with a solution saturated at the amorphous solubility and still supersaturated with respect to the crystal. The droplets can stay kinetically stable for a useful time, may help absorption (ref 65) and may agglomerate into larger amorphous precipitates. C7·C28

Supersaturation and desupersaturation. Supersaturation here means dissolved drug above the crystalline solubility. Its loss, usually by crystallization, is called desupersaturation. The paper distinguishes crystallization from the supersaturated solution, which can follow LLPS because supersaturation is then at its maximum, from crystallization inside the undissolved solid, called matrix crystallization. Crystals formed in the matrix provide seeds, existing crystal surfaces on which dissolved drug can grow without a new nucleation step (ref 64). Ideally the polymer suppresses both kinds of crystallization. C28

Enteric polymer and weak base: two pH-dependent components. In this system both components respond to pH, in opposite directions. Posaconazole has some solubility at gastric pH, while HPMC-AS is practically insoluble there and dissolves at intestinal pH (ref 66). The authors expect this to produce "a complex landscape of phase behaviors": the polymer holds the drug back in the stomach, and the drug loses solubility as the pH rises at the same moment as the polymer releases it. C29·M3

Second harmonic generation and two-photon ultraviolet fluorescence. Second harmonic generation (SHG) produces light at half the wavelength of the incident laser, but only in crystals without a centre of symmetry; posaconazole has four chiral centres (two R and two S), so its crystals are SHG-active while dissolved or amorphous drug is not (refs 52–56). Imaging SHG therefore shows crystals selectively. Two-photon excitation ultraviolet fluorescence (TPE-UVF) uses the drug's own fluorescence and shows posaconazole whatever its physical state. Read together, the two signals give the amount of crystalline drug and the amount of all undissolved drug in the same suspension. C10·M1

Environment-sensitive fluorescence. Posaconazole is auto-fluorescent, and its emission depends on how polar its surroundings are: in a less polar environment the quantum yield rises and the emission maximum moves to shorter wavelength, a blue (hypsochromic) shift (ref 59). When drug-rich droplets form, part of the drug sits in such a non-polar environment, so the emission spectrum changes at the LLPS onset; the same approach was used before for felodipine and indomethacin (refs 49, 50). C13

Methods in detail

The study has three strands that run in parallel: reference solubilities measured in the same media, concentration–time profiles under three contrasts (gastric stage or not, drug loading, dose), and tracking of the solid and colloidal phases during the test. The flow below summarizes them.

Materials and dispersions

Posaconazole (Form I) came from Chemshuttle, HPMC-AS MF grade from Shin-Etsu, and Noxafil delayed-release tablets (Merck) from a pharmacy; the media were prepared from commercial FaSSIF/FaSSGF and FaSSIF version 2 powders (Biorelevant). M1·M2·M3·M4 The ASDs were made at three drug loadings (DL), 10, 25 and 50 wt%, by dissolving drug and polymer in methanol at 10:90, 25:75 and 50:50 w/w, removing the solvent by rotary evaporation at 45 °C, drying under vacuum for 24 h and cryo-milling to a fine powder. Polarized light microscopy and PXRD found no detectable crystallinity before testing, and the ASDs were prepared fresh for each experiment. P1·R5 Each Noxafil tablet contains 100 mg posaconazole in a tablet of about 625 mg; the tablets were crushed, and the powder was used in every test to avoid confounding effects of tablet disintegration. M2·C8

Reference solubilities in the test media

Crystalline solubility. Excess crystalline drug was stirred at 300 rpm and 37 °C for 48 h in FaSSGF and in FaSSIF, ultracentrifuged at 35,000 rpm for 30 min at 37 °C, and the supernatant was diluted in methanol and assayed by HPLC (C18 column, water:acetonitrile 40:60, detection at 262 nm, standards 0.05–30 µg/mL, R² = 0.9993; n = 3). P2·P3

Amorphous solubility by UV extinction. A methanolic solution of posaconazole was pumped at 4 µg/min into 50 mL of stirred FaSSIF at 37 °C, while the extinction at 350 nm, a wavelength where the drug does not absorb, was recorded with a fibre-optic probe. The medium contained 20 µg/mL HPMC-AS to prevent crystallization during the titration. A jump in extinction marks the appearance of light-scattering droplets. P4

Amorphous solubility by fluorescence. Solutions of 0–26 µg/mL in FaSSIF with 20 µg/mL HPMC-AS were made by solvent shift from methanol at 37 °C and their emission recorded with excitation at 240 nm; a reference series was run in sodium phosphate buffer pH 6.5. P5 The methods state an emission range of 220–450 nm, while the published spectra run from 250 to 600 nm; the range actually recorded is therefore unclear, although the peaks used (368 nm in FaSSIF, about 372–385 nm in buffer) lie inside both. X2

The two-stage dissolution test and its controls

Problem. A single-medium test in FaSSIF tells nothing about what an hour in acid does to an enteric dispersion, and a flow-through transfer system is complex. What they did. The ASD or tablet powder was dispersed in 40 mL of FaSSGF (pH 1.6, medium 1) for 60 min, with samples at 15, 30, 45 and 60 min. Then 40 mL of a modified FaSSIF concentrate (medium 2), containing twice the bile salt and lecithin of standard FaSSIF at pH 7.5, was added; the resulting 80 mL (medium 3) matched FaSSIF-V2 in bile salt and lecithin and was brought to pH 6.5 with NaOH while the pH was monitored. Samples were taken at 5 to 300 min, micro-centrifuged at 14,800 rpm for 15 min, and the supernatant assayed by HPLC; all at 37 °C and 300 rpm, n = 3 (Fig. 2). P6

Two doses were used. At the high dose, 160, 64 and 32 mg of the 10%, 25% and 50% DL ASDs, or 100 mg of tablet powder, each supplied 16 mg posaconazole: 400 µg/mL in the gastric stage and 200 µg/mL after dilution, which the authors note is ten times the amorphous solubility. The 400 µg/mL corresponds to one 100 mg Noxafil tablet taken with 250 mL of fluid. At the low dose, one-fifth of these masses supplied 3.2 mg (80 and 40 µg/mL). P6·C31 As a control, the same materials were dispersed directly in FaSSIF "at the same conditions"; the paper does not state the volume or how this FaSSIF was prepared, so the control may differ from medium 3 in composition as well as in the missing acid stage. P7·X10

Following the colloidal and solid phases

Nanoparticle tracking analysis. One-millilitre samples taken 5 or 30 min after the shift were micro-centrifuged as above, which according to the authors pellets the large undissolved particles and most of the amorphous precipitate but leaves nanodroplets in the supernatant; the supernatant was analysed by NTA at 37 °C (532 nm laser, 30 s, triplicate) for droplet size and number. P8·C9

SHG and TPE-UVF imaging. For imaging, 8 mg of drug (80, 32, 16 and 50 mg of the three ASDs and the tablet powder) was dispersed in 20 mL FaSSGF and shifted after 60 min with 20 mL of the concentrate, so the drug concentrations matched the high-dose dissolution test. Aliquots taken at 30 and 60 min in FaSSGF and at 30, 60, 120 and 180 min in FaSSIF were placed in a nylon washer on a glass slide and imaged on a SONICC instrument: SHG at 1064 nm excitation, TPE-UVF with excitation at 532 nm, over fields of 1925 µm × 1925 µm and several focal planes 100 µm apart. P10 The integrated SHG intensity was converted into a volume fraction of crystalline drug, and the TPE-UVF intensity into a volume fraction of all undissolved drug. The authors define an estimate of matrix crystallization from the two, stated in words with the axis of Fig. 12:

In plain terms: of the drug that has not yet dissolved, the share that is already crystalline. The authors apply it after 30 min in FaSSGF. Reading Fig. 11 at 30 min reproduces the bars of Fig. 12 (for example about 0.062 divided by about 0.09 gives about 69% for the 50% DL ASD, against about 68% plotted), so the ratio is internally consistent. E1·R18·R19·R20 Neither signal is calibrated in the paper, and the fields of view of the two methods were not congruent, a point the authors raise themselves. C26·X6

Crystal form. To obtain enough precipitate for PXRD, 5 mg of drug dissolved in 0.5 mL methanol was added to the vessel after 180 min in FaSSIF, and the pellet from ultracentrifugation was dried and scanned from 5 to 50° 2θ. P9 Because the precipitate was spiked with dissolved drug, and the protocol arm and volume are not stated, the patterns identify the crystal form but say little about how much crystallized in the test itself. X9

Experimental analysis

The reference lines: crystalline and amorphous solubility

Crystalline posaconazole dissolved to 115.1 ± 9.4 µg/mL in FaSSGF but only 1.69 ± 0.06 µg/mL in FaSSIF after 48 h at 37 °C (Table I, n = 3). The authors attribute the difference to nearly complete ionization of the drug in the acid, in agreement with earlier reports (ref 58). R1·C11 In FaSSIF with 20 µg/mL HPMC-AS, the extinction at 350 nm stayed flat at about 0.005–0.008 up to about 17 µg/mL, began to rise at about 18 µg/mL and reached about 0.3 at 37 µg/mL; tangents drawn to the two parts of the curve cross at about 20 µg/mL, which the authors take as the LLPS onset (Fig. 3). R2·C12 Fluorescence gave the same answer: the emission at 368 nm stayed at about 380–395 a.u. from 16 to 19 µg/mL and then rose step by step to about 900 a.u. at 24 µg/mL (Fig. 4), which the authors assign to the formation of a posaconazole-rich amorphous phase. R3·C13 In plain sodium phosphate buffer pH 6.5 the onset was much lower, 8 µg/mL; above it the emission maximum moved from 385 nm to shorter wavelength (dashed lines at about 385 and 372 nm in Fig. 5) while its intensity rose, and this buffer peak lies at the wavelength of the peak that grows in FaSSIF, which the authors take as "helping confirm" that the FaSSIF peak marks drug-rich aggregates. R4·C14 The roughly 20 µg/mL in FaSSIF is the green "amorphous solubility" line in all later concentration plots, and the 1.69 µg/mL the grey crystalline line. R2·R1

Note by the AI writer (not discussed in the paper): the buffer spectra in Fig. 5 also show a narrow band near 480 nm, twice the 240 nm excitation wavelength, which grows from about 70 to about 430 a.u. above 8 µg/mL; the paper does not comment on it. R4·X2

Result 1: in acid, little drug dissolves and the solution stays subsaturated

During the hour in FaSSGF the dissolved drug rose steadily. After 60 min the ASDs had released around 7.5% (10% DL), approximately 10% (25% DL) and approximately 11.5% (50% DL) of the 400 µg/mL dose, and the tablet powder 8.5%; in the figure this is about 28, 40, 45.5 and 32 µg/mL (Fig. 6, first four points of each panel). R10·R6.d·R7.d·R8.d·R9.d Every value is far below the 115.1 µg/mL crystalline solubility in FaSSGF, so the gastric solution was subsaturated in all cases. C16·C33·R1.a Release rose with drug loading, the same trend Li and Taylor reported for a non-ionizing drug in poorly soluble polymers (ref 69), and the authors conclude that dispersion in HPMC-AS "substantially reduces" gastric release. C29·C30 This reading compares release with the crystalline solubility; no neat amorphous or crystalline drug was tested in FaSSGF. X7 At the low dose the 10% DL ASD reached about 11 µg/mL after 60 min (Fig. 13). R12.d

Result 2: crystals form inside the undissolved particles during the acid stage

No crystals were detected in the ASDs before testing, by polarized light microscopy, PXRD or SHG. R5 After only 30 min in FaSSGF, SHG showed crystals in every system: a few for the 10% and 25% DL ASDs and the tablet powder, and many large ones for the 50% DL ASD (Fig. 9, left column). In FaSSIF the SHG-active domains became more numerous between 30 and 180 min, and the 50% DL ASD had larger and more crystals than the others at every time point. R16·C22 The TPE-UVF images (Fig. 10) show large drug-containing ASD particles in FaSSGF for all systems; after the shift the signal fell sharply for the 10% and 25% DL ASDs and the tablet powder, "consistent with drug release as the polymer dissolved", but stayed high for the 50% DL ASD, "consistent with extensive crystallinity in the ASD matrix". R17·C25

Table A. Crystalline and total undissolved drug from imaging, high-dose two-stage test. Times are total minutes (FaSSGF to 60 min, FaSSIF after). Read from Fig. 12 and Fig. 11 (approximate; the paper omits error bars in Fig. 11 and does not define those in Fig. 12). R20·R18·R19

Table A puts numbers on the images. The SHG crystallinity of the 50% DL ASD was two to three orders of magnitude above the others throughout, and its final value was, in the authors' words, roughly 250 times that of the 25% DL ASD; the ratio at 30 min puts the 50% DL ASD (about 68%) far above the 25% DL ASD (about 0.050%), the tablet powder (about 0.047%) and the 10% DL ASD (about 0.026%). R18·R20·C27·C38 The authors describe crystallinity as appearing to increase with time "for all systems until it reached a maximum at 240 min"; in the plotted values the 10% DL, 25% DL and tablet series first dip after 30–60 min and rise only later, so the increase is net rather than steady. C27·X5 For the 50% DL ASD the SHG fraction exceeds the TPE-UVF fraction from 120 min onwards (about 0.157 against 0.105, and about 0.187 against 0.02 at 240 min), which cannot be true if both measure the same drug on one scale; the volume fractions are best read as relative indices, with the 30-min ratio as the paper's quantitative estimate. X6 The number of imaged samples is not stated, and at the low dose the SHG signal was too weak for reliable observations. X13·C24

The authors' reading of this result is the core of the paper. Because the gastric solution was subsaturated, "Solution-phase crystallization cannot occur"; the crystals must have formed inside the suspended ASD particles, "perhaps somewhat counterintuitively", in a medium where crystals should dissolve. They conclude that crystals formed in the HPMC-AS matrix were "apparently 'protected'" from dissolution and so remained available for transfer to the intestine, that crystallization was initiated by the gastric suspension ("strongly suggesting"), and that HPMC-AS prevents crystallization during ASD preparation but not on suspension in acid. C34·C35·C23·C37 The much greater crystal burden at 50% DL "raises questions" about how well polymers with pH-dependent solubility protect against matrix crystallization, beyond controlling release. C36

Result 3: after the shift, supersaturation depends on how many crystals came along

Table B. Posaconazole concentration (µg/mL) at the high dose, all values about, read from Fig. 6 (n = 3). FaSSIF times count from the pH shift for the two-stage test and from the start for the FaSSIF-only control. "Hidden" marks a two-stage point covered by the FaSSIF-only marker. Reference lines: amorphous solubility about 20 µg/mL, crystalline solubility 1.69 µg/mL. R6·R7·R8·R9

Table B shows four distinct behaviours after the shift in the two-stage test. The 10% DL ASD reached about 22–24 µg/mL within minutes, near the amorphous solubility, and held it for about two hours; it was still about 19 µg/mL at 150 min and fell to about 6–8 µg/mL by 180 min. The 25% DL ASD held about 21–22 µg/mL for half an hour, was about 18 µg/mL at 60 min and about 6–8 µg/mL from 90 min. The 50% DL ASD, which had released the most drug in acid, fell from about 17 µg/mL at 5 min to about 6 µg/mL by 30 min and never reached the amorphous solubility. The tablet powder reached about 20–21 µg/mL for the first 15 min, dropped to about 14 µg/mL at 30 min and settled near 11 µg/mL, the highest residual of all. R6·R7·R8·R9·C17 The discussion summarizes the first two as supersaturation lasting 150 and 60 min; the results text quotes the same events on the total-time axis of Fig. 6 ("After 210 min" and "about 120 min"), so the two sets of numbers describe the same curves. C46·C17·X1

The authors explain the pattern through two sources of supersaturation and one sink. On the shift, drug dissolved in the acid turns neutral and becomes supersaturated (mechanism 1), and the dissolving polymer releases the remaining amorphous drug rapidly (mechanism 2); since little drug dissolved in acid, mechanism 2 "seemingly predominated" for the low-loading ASDs, whose particles were still largely amorphous. C40·C42·C45 In theory the 50% DL ASD should have supersaturated most, having released the most drug in acid. C41 Its collapse "can be attributed to the presence of extensive matrix crystallization during the gastric stage": the crystals freed when HPMC-AS dissolves "are expected to grow rapidly" on the high initial supersaturation, and since desupersaturation depends on crystal surface area, initial supersaturation and further primary or secondary nucleation, the authors find the rapid loss "unsurprising". C43

Result 4: without the acid stage, supersaturation lasts longer

Put directly into FaSSIF, the 10% DL ASD stayed at about 21–25 µg/mL for 180 min (about 18 µg/mL at 240 min), the 25% DL ASD stayed near 19–23 µg/mL for about 90 min, and the tablet powder held about 20 µg/mL from 30 to 60 min; the 50% DL ASD still fell to about 6 µg/mL within 30 min. R6·R7·R9·R8·C15 The authors summarize that the FaSSIF-only profiles show longer supersaturation "in all instances, but most notably for the lower drug loading ASDs, presumably due to the initial absence of seeds". C51 They add that the poor result of the 50% DL ASD even without an acid stage highlights its "susceptibility" to competitive crystallization during release, and that HPMC-AS "is potentially an effective solution-crystallization inhibitor" for posaconazole when matrix crystallization is absent and the polymer is soluble. C44·C52 Two points limit this comparison: crystallinity was not imaged in the FaSSIF-only runs, and the control medium is not described in enough detail to know whether it matched medium 3. X8·X10 A few text values also differ slightly from the figure: the text gives the tablet powder peak as "approximately 21 µg/mL after 15 min" and the 50% DL plateau as about 5 µg/mL, while the figure shows about 18 µg/mL at 15 min with the peak of about 20.5 µg/mL at 30 min, and a plateau of about 6 µg/mL. X12

Result 5: nanodroplets form at the lower loadings

Table C. Mean nanodroplet size (nm) by NTA in the supernatant after micro-centrifugation, two-stage test, read from Fig. 7 (approximate, n = 3). No result is shown for the 50% DL ASD. R13

NTA found nanodroplets of about 300 nm for the 10% and 25% DL ASDs and smaller ones, about 220 nm, for the tablet powder (Table C). R13·C19 The particle counts (Fig. S2, not available here) were lower at the low dose, and the number of 10% DL nanodroplets fell between 5 and 30 min while their mean size grew from about 304 to 339 nm. R14·R13 The authors conclude that the 10% and 25% DL dispersions underwent LLPS, and the abstract describes this as "transient formation of amorphous nanodroplets prior to crystallization". C46 Two caveats come from the methods. The 50% DL ASD has no NTA result in the figure, and the text does not say whether it was measured, so the scheme's "No observable nanodroplet reservoir" for high loading is not a measured null. X11·C59 And because the HPLC samples went through the same centrifugation that leaves nanodroplets in the supernatant, the measured concentration may include some colloidal drug; the plateau of about 24 µg/mL for the 10% DL ASD, above the 20 µg/mL LLPS onset, fits this. X3

Result 6: a smaller dose reaches the same peak but desupersaturates sooner

At one-fifth of the dose (80 µg/mL in acid, 40 µg/mL after the shift) the 10% DL ASD reached about 21 µg/mL in FaSSIF, stayed at about 17–20 µg/mL to 90 min and was down to about 8 µg/mL at 120 min, whereas at the high dose the fall came between 150 and 180 min (Fig. 13). R12·R6 According to the text, the 25% DL ASD also desupersaturated faster at the low dose, while the 50% DL ASD and the tablet powder did not change notably, and the FaSSIF-only profiles were similar at both doses (Fig. S1). R11·C18 Since the amorphous solubility was reached at both doses, the authors argue that the extent of supersaturation "clearly cannot explain" the difference. C47 With some crystals present at the start of the FaSSIF stage in every ASD, they propose that the longer supersaturation at the higher dose "can be most likely be explained by a reservoir effect": as crystal growth removes dissolved drug, undissolved ASD or amorphous nanodroplets dissolve and replace it, and at the low dose this reservoir is used up sooner. C48

Crystal form of the precipitates

The precipitates of all three ASDs, collected after 180 min in FaSSIF, showed the peaks of posaconazole Form I at 7.5, 10, 18, 20, 22 and 26° 2θ; those of the 50% DL ASD were the most intense, which the authors read as a higher extent of crystallization (Fig. 8). R15·C20 The tablet-powder precipitate gave broader, less clear Form I peaks, which "could suggest" smaller or more defective crystals, "possibly" because of tablet excipients, plus titanium oxide peaks at 25, 38 and 48°; sodium chloride from the medium appeared at 32 and 46° in all samples. R15·C21 Since each precipitate was spiked with 5 mg of dissolved drug before collection, the intensities partly reflect the spike. X9

Discussion

How the authors connect the findings

The authors put the results together in Fig. 14. The chain starts in the stomach and splits at the moment of transfer, depending on how much of the undissolved drug has crystallized by then. C59

In the conclusions the authors state the central link without a hedge: "These seed crystals were transferred to the intestinal compartment and promoted further crystallization", and the supersaturation generated in the intestinal medium was "impacted by" the gastric release and the extent of matrix crystallization. C56·C55 The intermediate steps carry hedges: the crystals were "apparently" protected, the 50% DL collapse "can be attributed" to them, mechanism 2 "seemingly predominated", and the dose effect is "most likely" a reservoir. C35·C43·C45·C48

The onset of desupersaturation is not the onset of crystallization

One argument in the discussion reaches beyond posaconazole. In this system crystals are present from the start of the intestinal stage, yet the 10% DL ASD stays supersaturated for about 150 min. R16·R6 The authors therefore stress that "the onset of desupersaturation for this complex system does not represent the onset of crystallization, as is commonly assumed"; it marks the time when crystal growth outpaces the dissolution of undissolved amorphous material, or when that material runs out. C49 Following from this, they call it "imperative" to choose the ASD dose in dissolution tests deliberately, both for in vivo relevance and for the insight gained by varying the dose number. C50

What it suggests for formulation

Stated by the authors. Characterization in a two-compartment biorelevant setting was "essential" to understand the effect of the pH shift on release, supersaturation and precipitation, and the work "emphasizes the significance of assessing the impact of pH change on dissolution". C54 The crystallization-inhibiting ability of ASD polymers "should be assessed in different environments", varying the ionization of drug and polymer by pH and looking at both matrix and solution crystallization. C53 The results "provide insights" for formulation prediction of poorly soluble weak bases in delayed-release ASDs and for "potential origins" of intraluminal absorption variability. C58

Note by the AI writer (not discussed in the paper). Three practical readings follow from the data, beyond what the authors state. First, for posaconazole in HPMC-AS MF, the crystal burden after 30 min in acid jumps by about three orders of magnitude between 25% and 50% DL, so a formulation meant to supersaturate after gastric passage should stay well below 50% DL; where exactly the limit lies cannot be read from three loadings. R20 Second, enteric protection limits gastric release but not gastric crystallization: drug kept in an insoluble, wetted polymer matrix crystallized while the medium was subsaturated. Since gastric residence varies between people (refs 70, 71) and only 60 min was tested here, residence time is a variable worth testing. C34·P6 Third, the marketed tablet held the amorphous solubility only briefly after transfer but kept the highest residual concentration; together with its broadened PXRD peaks this hints that tablet excipients change crystal growth, which the authors suggest only for the crystal size. R9·C21

Limitations

Limitations stated by the authors. SHG at the low ASD dose showed limited crystallinity but could not give reliable observations because the signal was weak. C24 The fields of view of SHG and TPE-UVF were not exactly congruent, so overlays "cannot be trivially generated". C26 The authors also frame the FaSSIF-only explanation with "presumably" and the dose effect with "most likely", and they present the questions about pH-dependent polymers as open ("raises questions"). C51·C48·C36

The AI writer's own judgement (1): the causal chain is inferred. The paper shows two things directly: crystals form in the undissolved ASD during the acid stage, far more at 50% DL, and supersaturation after the shift is shorter after an acid stage and at high loading. R16·R20·R6·R7·R8 That the gastric crystals act as seeds, and that a reservoir of undissolved amorphous drug prolongs supersaturation, are interpretations: there is no seeding experiment (for example crystals added to a FaSSIF-only run), no growth-rate or crystal-surface measurement, no measurement of undissolved amorphous drug over time, and no crystal imaging in the FaSSIF-only runs. X8 The 50% DL ASD also crystallized within 30 min in FaSSIF alone, so its collapse after the acid stage cannot be assigned to gastric seeds alone; the authors themselves call it susceptible to competitive crystallization during release. C44·R8 Finally, the FaSSIF-only control is described only as "at the same conditions", so the two arms may differ in medium as well as in the acid stage. X10

The AI writer's own judgement (2): what the numbers can and cannot carry. The imaging volume fractions are uncalibrated, and for the 50% DL ASD the crystalline fraction exceeds the total undissolved fraction after 120 min, so only the 30-min ratio of Fig. 12, read as a relative index, should be quoted; within it, the three low values (about 0.026–0.050%) overlap within their bars, and only the 50% DL ASD stands apart. X6·X13 Durations of supersaturation appear on two time axes in the text, and a few text values differ from the figure, so durations are best read from Fig. 6 in FaSSIF time. X1·X12 HPLC supernatants may contain nanodroplets, and the amorphous-solubility line comes from FaSSIF with 20 µg/mL HPMC-AS rather than from medium 3, which by simple arithmetic contained up to about 1.8 mg/mL dissolved HPMC-AS at the 10% DL high dose. X3·X4 The PXRD precipitates were spiked with dissolved drug, and no NTA result is shown for the 50% DL ASD. X9·X11 The supporting figures with the low-dose profiles and particle counts were not available for this reading. R11·R14

Conclusion

The paper makes three contributions. First, it shows with SHG imaging that posaconazole held in undissolved HPMC-AS particles crystallizes within 30 min in fasted gastric fluid, although the surrounding solution stays below the crystalline solubility; at 50% drug loading about two-thirds of the undissolved drug was estimated to be crystalline, at 10% and 25% loading less than 0.1%. R16·R20·R10·C34 Second, it links this crystal burden to what happens after the pH shift: the low-loading dispersions reached the amorphous solubility of about 20 µg/mL, formed nanodroplets of about 300 nm and stayed supersaturated for about 150 and 60 min, whereas the 50% loading desupersaturated at once, and every system stayed supersaturated as long or longer without the acid stage. R2·R13·R6·R7·R8·C46·C51 Third, by varying the dose it shows that the duration of supersaturation depends on the amount of drug present, not only on its peak, which the authors explain "most likely" by a reservoir of undissolved amorphous drug, and from which they argue that desupersaturation onset is not crystallization onset. R12·C48·C49 In the authors' words, the seed crystals "were transferred to the intestinal compartment and promoted further crystallization". C56