Keeping the acid away: an enteric coat for tablets of a delamanid–HPMCP amorphous solid dispersion

Background and motivation

Delamanid is an anti-tuberculosis drug used against multidrug-resistant tuberculosis (MDR-TB). It is a weak base with a pKa of 4.3, poorly soluble and, in the authors' words, "rapidly crystallizing" (Section 4). M1 Above pH 5.5 its solubility in buffer is below 0.02 µg/mL, a value from the group's earlier work. R18 The commercial tablet, Deltyba (Otsuka), is absorbed much better with food: in the regulatory assessment the authors cite, a 200 mg dose with a standard meal raised Cmax 3.4-fold and AUC0-inf 2.9-fold against fasting, with high variability between subjects. R24

An amorphous solid dispersion (ASD) disperses the drug at the molecular level in a polymer so that it dissolves faster and to a supersaturated concentration. The benefit lasts only as long as the drug does not crystallize, and polymers that keep a solid ASD amorphous on storage can fail once the ASD meets water. Crystals that form on the ASD surface and grow fast were the proposed cause of poor release from bicalutamide–copovidone ASDs, through a crystalline drug layer at the surface (Moseson et al., 2022c). C7·C14 Weakly acidic polymers such as HPMCAS and hypromellose phthalate (HPMCP) are widely used because, where they dissolve, they delay nucleation and slow crystal growth. C6 But these are enteric polymers: they do not dissolve at gastric pH. Several studies have shown that weakly basic drugs still leak out of such ASDs in acid, more at pH 1 than at pH 4.5 in one study, because the drug is molecularly dispersed in the insoluble polymer (Monschke and Wagner, 2019; Elkhabaz et al., 2019; Wang et al., 2021). R26·C8·C35 Posaconazole–HPMCAS ASDs crystallized in fasted simulated gastric fluid at pH 1.6 at high drug loading (Elkhabaz et al., 2019). R20

The direct predecessor of this paper is the group's own study of DLM ASDs with HPMCP and HPMCAS (Nguyen et al., 2023b). There, surface crystallization at gastric pH was strongest at pH 3.0 and lowered release after transfer to intestinal pH. R19 Forming a salt of DLM inside the ASD during spray drying (chloride or edisylate) had improved solid-state stability and release at intestinal pH (Van Duong et al., 2022a), but every formulation still released differently depending on the pH of the medium. C11

Enteric coatings are an old answer to acid: a continuous film of a polymer that dissolves only above a threshold pH keeps gastric fluid away from the tablet core. The threshold depends on the polymer chemistry; the methacrylic acid–ethyl acrylate copolymer Eudragit L 100-55 dissolves above pH 5.5 and is sold as a water-dispersible coating powder, Acryl-EZE II. C9·M11·M12 Enteric coats had been used on ASDs of ezetimibe–lovastatin with Soluplus, on darunavir–HPMC nanoparticles and on niclosamide–copovidone tablets, where the coat avoided crystallization in acid (Jara et al., 2022). C10·R21 None had been used on an ASD made with an enteric polymer, perhaps, the authors suggest, because such ASDs are assumed to be protected already. C10·C34

The authors therefore hypothesized that an enteric coat would protect an HPMCP ASD of DLM at the pH values where the polymer cannot prevent crystallization, keep the dissolution benefit at higher pH, and so reduce the variability of release across gastric pH values that reflect fasted and fed states. C12

Background concepts

Enteric polymer versus enteric coat. An enteric polymer carries acidic groups: phthalate in HPMCP, methacrylic acid in Eudragit. At low pH these groups are not ionized and the polymer does not dissolve; above a threshold pH they ionize and the polymer dissolves. In an ASD, the enteric polymer is the matrix in which the drug is dispersed. In an enteric coat, the same kind of polymer forms a continuous shell around the tablet. The paper's argument rests on the difference: an insoluble ASD matrix still lets dispersed drug escape and crystallize at its surface, while a continuous coat keeps the medium out. C35 The text gives HPMCP as "not soluble until pH 5" and soluble at pH 5.8 or above; Eudragit L 100-55 dissolves above pH 5.5. M3·M11

pH-dependent solubility of a weak base. The solubility of a weak base rises steeply as the pH falls below its pKa, because the protonated form is much more soluble than the neutral one. The paper compares its measured buffer solubilities of DLM with a curve "predicted by the Henderson-Hasselbalch equation using reported pKa of 4.3" (Figure 6). The equation itself is not printed. E1·R13 For DLM the measured buffer values follow this curve: about 10 µg/mL at pH 1.6, about 0.6 µg/mL at pH 3.0 and practically zero at pH 4.5 and above (read from Figure 6). R13 At gastric pH 1.6 the drug is therefore partly soluble and leaches; at pH 3.0 it is barely soluble while HPMCP is also insoluble, the combination the authors call the worst case. C41

In situ salts. The chloride and edisylate ASDs were made by adding hydrochloric acid or 1,2-ethanedisulfonic acid at a 1:1 molar ratio to the spray-drying solution, so the ASD contains protonated DLM with its counterion. P1·M4·M5

Two-stage (pH-shift) dissolution. A single-stage test runs at one pH. A two-stage test first holds the dosage form at a gastric pH and then raises the pH to an intestinal value, to mimic passage from stomach to small intestine. Gastric pH is not fixed: it is low in the fasted state and higher and variable after a meal; the authors quote a gastric emptying time of 15–60 min fasted and 2–5 h fed, and a gastric pH of about 6–6.5 thirty minutes after a meal. R22·C19 Testing several gastric pH values (here 1.6, 3.0 and 5.0) therefore asks how robust release is to the stomach a tablet happens to meet. A3

Biorelevant media. FaSSIF V1 and FeSSIF V2 are simulated fasted and fed intestinal fluids containing bile salts and lipids; FeSSIF V2 carries about 7.8 mM lipid against 0.75 mM in FaSSIF V1. P16 FEDGAS is a high-fat simulated fed gastric medium: a gel of fat (63.8 g), bile salts (0.34 g), stabilizers and carbohydrates (67.8 g) per 170 g, dispersed with citrate salts at pH 3.0, 4.5 or 6.0 into a milky liquid (Table 3). P15 Such media can raise the solubility of a lipophilic drug by orders of magnitude and change how tablets disintegrate.

Surface crystallization and seeding. Molecules at the surface of an amorphous solid are more mobile than those inside, so crystals often form there first. Once formed, crystals can act as seeds: when the surrounding polymer later dissolves, they enter a supersaturated solution and pull drug out of it by growth. The authors invoke both ideas to explain why a short stay in acid can spoil release at intestinal pH. C37·C38

Methods in detail

The study has three parts: it reproduces the failure and its suspected cause on thin ASD films, measures how long the tablet coat survives in different gastric media, and then tests coated tablets in single-stage, two-stage and fed-state dissolution. All formulations share one ASD composition and one tablet core. The flow below summarizes the work.

Common materials. DLM free base, chloride and edisylate ASDs were spray dried at 25 wt% drug loading (DL) with HPMCP grade P-50 (Buchi B-290, inlet 75 °C, feed 4 mL/min, nitrogen 700 L/h). The free base and chloride were sprayed from dichloromethane:methanol 1:1; for the edisylate, methanol was replaced with acetone to avoid esterification of the sulfonic acid with the alcohol. The particles were below 20 µm and were dried overnight under vacuum. P1·C13 Each 500 mg tablet contained ASD equivalent to 50 mg DLM, 40 mg sodium starch glycolate, 40 mg croscarmellose sodium, 6 mg colloidal silica, 6 mg magnesium stearate and microcrystalline cellulose to weight, compressed in an 11 mm die (Table 1). P2 Tablets were pan coated with an aqueous Acryl-EZE II dispersion (100 g in 1000 mL water; inlet 70 °C, bed 55 °C) together with placebo tablets, reaching 14–16% weight gain; Table 1 lists 75 mg of Acryl-EZE II per tablet. Deltyba tablets were coated in the same way. P3

Contribution 1: showing the failure and its suspected cause on films

Problem. The earlier study had linked poor release after a gastric stage to surface crystallization. Release data for uncoated tablets were needed as a reference, and the crystals needed to be seen directly, with and without an enteric layer. R19·C1

What they did. Release from uncoated tablets was measured in single-stage tests at pH 5.8 (maleate) and 6.5 (phosphate) and in pH-shift tests: 60 min in 450 mL HCl pH 1.6, 470 mL phosphate pH 3.0 or 480 mL phosphate pH 5.0, after which 50, 30 or 20 mL of 0.57 M phosphate buffer pH 7.3 was added in the same vessel to give 500 mL at pH 6.5, sampled 30 min later. P10·P11·P14 One tablet in 500 mL corresponds to 100 µg/mL at complete release. Part of these data (the free base and edisylate) was taken from the earlier study, and the rest was measured here. R1·R2·X10 In parallel, ASD films about 8 µm thick were spin coated on cover slips from the same solutions, and part of each film was covered by spin coating a 10% w/v Eudragit L 100-55 solution, giving an enteric layer about 2 µm thick (thicknesses by confocal microscopy, Figure S1). The films were immersed for 1 h in phosphate buffer pH 3.0 and examined by polarized light microscopy and scanning electron microscopy (Figures 2 and 3). P4·P5·R5

Why it works. A thin film puts the whole ASD surface under the microscope, so crystals can be seen within an hour, and a partly coated film carries its own control: coated and uncoated areas of the same film experience the same medium. pH 3.0 is the pH at which the earlier study found the most surface crystallization. R19 The film test shows the step that the release data cannot show, but it is a model of the tablet, not the tablet itself. X7

Contribution 2: measuring how long the coat survives

Problem. An enteric coat helps only if it stays intact as long as the tablet stays in the stomach, and the stomach after a meal is both less acidic and slower to empty than the fasted stomach: the authors cite 2–5 h fed residence and pH 6–6.5 shortly after a meal. Food also changes viscosity and water transport. R22·C19

What they did. Coated tablets of the edisylate ASD were stirred overnight in 500 mL of gastric medium at 37 °C and 150 rpm (USP apparatus II) and photographed: HCl pH 1.6; phosphate buffer pH 3.0 and 5.0; acetate buffer pH 3.0, 4.5 and 6.0; and FEDGAS pH 3.0, 4.5 and 6.0 (Figure 4). P6 The text refers the composition of the acetate buffers to Table 3, but Table 3 lists the FEDGAS gel with sodium citrate, citric acid and sodium chloride, so the acetate recipe itself is not printed. X1 A USP acid-stage tolerance test at pH 4.5–5 for 2 h was also run, with release values in Table S3. P7

Why it works. Covering pH 1.6 to 6.0, with and without fat, brackets the range from a fasted stomach to a fed one and around the pH 5.5 dissolution threshold of the coating polymer. A5·M11 The end point is visual, and only one formulation (edisylate) was tested. X12

Contribution 3: coated tablets in two-stage and fed-state dissolution

Problem. If the coat works, coated tablets should release nothing in the gastric stage and then release at intestinal pH as well as in a single-stage test, for every gastric pH and every form of the drug. C12

What they did. Coated tablets of the three ASDs were tested in single stage at pH 5.8 and 6.5 and in two stages: 60 min at pH 1.6, 3.0 or 5.0, after which the gastric medium was discarded and the tablet moved to phosphate buffer pH 6.5 for another 60 min (Figure 5). P10·P12 Concentrations were read in situ from the second-derivative UV spectrum (330–350 nm) with fibre-optic probes. P8 For the fed state, coated edisylate tablets were tested in FaSSIF V1 and FeSSIF V2, and after 17 h in FEDGAS pH 3.0 or 4.5 followed by transfer to FeSSIF V2 (Figure 8). The equilibrium solubility of DLM was reported in buffers and biorelevant media (Figure 6), and Deltyba release was followed in FaSSIF V1, FeSSIF V2 and FEDGAS, where samples were filtered through 0.2 µm nylon and analysed by HPLC instead of in situ UV (Figure 7). P9·P13·R13·R14·R16·R17 How the solubilities were measured (time, temperature, solid form, replicates) is not described in the main text. X8

How it differs from the reference tests. The coated tablets were not tested with the same protocol as the uncoated ones: they were transferred to fresh pH 6.5 medium and followed for 60 min, while the uncoated tablets were shifted in the same vessel and followed for 30 min. In the transfer protocol any drug released in the gastric stage is discarded, which for coated tablets is less than 0.5%. P11·P12·X6 Comparisons of variability between Figure 1B and Figure 5 are therefore fair in kind but not exact in value.

Experimental analysis

Main result 1: without a coat, release depends on what happened in the stomach. In single-stage tests at pH 5.8 and 6.5, where HPMCP dissolves, all three uncoated ASD tablets released about 82–96 µg/mL within 60 min (Figure 1A). R1 The authors read this as rapid release into a solution in which the polymer maintains supersaturation. C40 A gastric stage changed the picture (Figure 1B and Table A). At pH 1.6 the uncoated tablets already released about 30–41 µg/mL into the acid (where DLM solubility is about 10 µg/mL, Figure 6) and reached about 62–93 µg/mL after the shift. At pH 3.0 almost nothing dissolved in the acid (about 1–2 µg/mL), and after the shift the tablets reached only about 33.5–54 µg/mL. At pH 5.0 the forms diverged: the free base and the chloride stayed at about 7 and 0.4 µg/mL in the acid and at about 6.5 and 0.3 µg/mL after the shift, while the edisylate released about 83 µg/mL already in the pH 5.0 stage and kept about 82 after it. R2·R13 The authors summarize this as good single-stage release but "diminished release for two-stage dissolution tests", worst after an initial pH 3.0, where both polymer and drug have low solubility. C15·C41 In their own figure the pH 3.0 worst case holds for the edisylate; for the free base and the chloride the lowest values follow pH 5.0. X5 The difference between the forms at pH 5.0, close to where HPMCP begins to dissolve, is not discussed in the paper. X11

Table A. Drug concentration (µg/mL; complete release = 100) for uncoated and coated tablets of 25% DL ASDs, read from Figures 1B and 5 (approximate). The free base and edisylate uncoated values come from the earlier study; the two sets were obtained with different two-stage protocols. R2·R10·X6·X10

Main result 2: the films crystallize in acid, but not under the coat. After 1 h in phosphate buffer pH 3.0, polarized light showed birefringent needle- and star-shaped crystals over the films of all three ASDs, and SEM showed the same star-like crystals on the surface (Figures 2A and 3). R3·R4 The authors call the crystallization "rapid" and note that "salt formation did not affect the crystallization tendency of DLM at this pH", in agreement with their earlier study. C16·C17 On films partly covered by the roughly 2 µm Eudragit L 100-55 layer, the covered areas stayed clear while the uncovered areas showed crystals (Figure 2B), and the text reports no crystals under the coat even after overnight incubation (Figure S2). R5·R6·C18 Note by the AI writer (not discussed in the paper): in two of the three partly coated films (panels iv and v), the uncovered side shows small dark dots rather than the star-shaped crystals of Figure 2A; only panel vi shows star-shaped crystals, so the identification of the dots as crystals rests on the authors' reading. X15

Main result 3: the coat holds in acid, fails at pH 4.5–5.0 in buffer, and lasts longer with fat. Table B collects the soak tests. At pH 1.6 and 3.0 the coated tablets stayed intact for 15 h to more than 17 h with minimal release. R7·C20 At pH 4.5–5 they passed the USP acid-stage tolerance requirements, releasing less than 0.2% in 2 h, but in buffer the coat failed soon after: at around 130 min in acetate pH 4.5 and about 150 min in phosphate pH 5.0. R7·R8·C21 Once the coat failed, drug leached into the medium (below 5 µg/mL) and/or crystallized, and release after transfer to a higher pH was "notably reduced" (Table S3, Figure S3). R9·C22 At pH 6.0 the coat dissolved within 10–15 min in acetate buffer, as expected above the pH 5.5 threshold of Eudragit L 100-55. R7·C23 FEDGAS delayed failure: at pH 6.0 the tablet was intact at 10 min and had disintegrated by 30 min (about 20–30 min against less than 15 min in buffer), and at pH 4.5 it lasted more than 17 h against about 2 h in buffer. The authors relate this to the glycerides, carbohydrates and bile salts of FEDGAS changing viscosity, osmolarity and surface tension and reducing water diffusivity; none of these properties was measured in this work. R7·C24

Table B. Integrity of enteric-coated DLM edisylate ASD tablets (25% DL, 15% coating weight gain) in 500 mL gastric media at 37 °C and 150 rpm, from Figure 4 and the text of Section 3.2. R7

Main result 4: coated tablets release well after every gastric pH. In single-stage tests the coated tablets released nothing for about 11–20 min, the time the coat needs to dissolve and the tablet to disintegrate, and then reached about 81.5–92 µg/mL at 60 min at pH 5.8 and 6.5; acetate pH 6.0 (Figure S4) also gave more than 80 µg/mL. The edisylate started first (about 11–12 min) and the free base last (about 16–20 min) (Figure 5). R10·R11·C25 In the two-stage tests, the gastric stage produced a flat line near zero at pH 1.6, 3.0 and 5.0, which the authors give as less than 0.5% release. After transfer, release began within about 11–16 min and reached, at 60 min after transfer, the values in Table A: about 76–84 µg/mL for the free base, 78–84 for the chloride and 82–86 for the edisylate. R10·C26 Within one form the values after the three gastric pH values differ by about 4–8 µg/mL, against about 52–86 µg/mL for the uncoated tablets (computed by the AI writer from the plot readings in Table A). R2·R10 The authors describe the variation as "much lower" and the release as "remarkably improved and less variable relative to that observed for uncoated tablets". C26·C43

Two statements in the text go a little further than the figure. The text gives final concentrations above 80 µg/mL for both salts and the free base, but the free base after pH 5.0 stays at about 75–77 µg/mL and the chloride after pH 5.0 ends at about 78. X3 The discussion states that release after transfer was "the same" as in single-stage tests when the coat stayed intact; the two-stage end values are similar but up to about 15 µg/mL lower, most clearly for the free base (about 76–84 against about 91–92). C46·X4 Neither affects the main comparison, which is about spread rather than absolute level. For the commercial tablet the coat also reduced the differences between pH conditions, but the overall release of coated Deltyba was much lower than for uncoated Deltyba or the in-house ASD tablets (Figures S5 and S6, not available here). R12·C27

Main result 5: food raises solubility and commercial-tablet release, but hardly changes coated-tablet release. Table C collects the biorelevant data. In buffers, DLM solubility followed the Henderson–Hasselbalch curve for pKa 4.3, from about 10.3 µg/mL at pH 1.6 to practically zero at pH 4.5 and above. Food components raised it strongly: FeSSIF V2 gave 2.30 ± 0.31 µg/mL, 4.4-fold FaSSIF V1 (0.52 ± 0.12), and FEDGAS gave 57.97 ± 0.64 µg/mL at pH 3.0 and 13.10 ± 2.32 µg/mL at pH 4.5. The authors put FEDGAS pH 4.5 and the fed intestinal medium at approximately 650- and 100-fold the corresponding buffers. R13·R18·C28 Deltyba, uncoated, released rapidly and then gradually in FeSSIF V2, to about 77 µg/mL at 120 min, against a near plateau of about 31–37 µg/mL in FaSSIF V1 (Figure 7); in FEDGAS it reached about 34.5 µg/mL at pH 3.0 and about 20 at pH 4.5 by 60 min, much more than in buffer of the same pH (Figure S7). R14·R15·C30 The text gives the FeSSIF and FaSSIF values as "up to 80" and "a maximum concentration of 35" µg/mL, a few µg/mL from the plotted curves. X2 After a buffer stage at fed gastric pH and transfer to fed-state media, Deltyba crystallized (Figure S7). R15·C31 The authors say the higher solubility "likely accounts for" the better release of Deltyba in these media and that "the impact of any crystallization is mitigated by the higher solubility". C29

Table C. Solubility and release in biorelevant media. Solubilities as printed in Section 3.4 (mean ± spread, number of replicates not stated); release read from Figures 5, 7 and 8 (approximate); for the post-FEDGAS tests, times count from the transfer to FeSSIF V2. R13·R14·R16·R17·R18

The coated edisylate tablets behaved almost the same in fasted and fed intestinal media: after a lag of about 9–11 min they released about 96 µg/mL by 60 min in both, and about 100 µg/mL by 120 min in FeSSIF V2 (Figure 8A). R16 After 17 h in FEDGAS pH 3.0 or 4.5 and transfer to FeSSIF V2, release began after about 9–10 min and reached about 87 and 86 µg/mL at 120 min (Figure 8B); the text gives "approximately 90 µg/mL". R17·X2 The authors summarize that simulated fed-state media had "only a minor impact" on release from coated tablets. C32

Discussion

The authors' account of the mechanism. The discussion strings the results into one chain. At gastric pH, HPMCP is unionized and insoluble, yet some drug leaves the ASD because it is molecularly dispersed in the polymer; the extent rises with drug loading, whereas a continuous enteric coat keeps gastric release below 0.5 µg/mL. C35·C26 The rapid surface crystallization at low pH "can be attributed to the enteric polymer being a poor crystallization inhibitor when it is unionized, and/or the drug having a higher tendency to crystallize when exposed to certain pH conditions", with drug at the surface the most vulnerable because of its higher molecular mobility. C37 When the pH rises and the polymer dissolves, those crystals are released from the matrix and, "presumably", seed crystallization of the supersaturated drug, which accounts for the large variations in Figure 1B. C38 The coat interrupts the chain at its first step. The diagram below shows the chain as the authors give it; the crystals were observed on films, and the later steps for tablets are inferred from release. X7

A third branch applies when the coat fails: at pH 4.5–5.0 in buffer this happened after about 130–150 min, water and buffer species reached the core, drug leached and/or crystallized, and release after transfer fell. C22·R7·R9 What decides between the branches is therefore the integrity of the coat, which depends on the pH relative to the 5.5 threshold of Eudragit L 100-55, on how long the tablet stays in the stomach and on food components. C23·C24·C45

Food. For uncoated tablets, the authors see release variability as "a complex interplay of the impact of pH and food components on drug (and polymer) solubility, as well as crystallization tendency". C39 Food components raise DLM solubility, so even if some drug crystallizes, more of it stays in solution; this "likely" explains the better release of Deltyba in fed media, and the authors suggest that food-dependent solubility "likely" contributes to the positive food effect of poorly soluble drugs such as posaconazole, ziprasidone and pretomanid, and to the improved fed-state exposure of Deltyba. C29·C49·R24 For coated tablets, food mattered less: release in FeSSIF V2 matched that in FaSSIF V1, and a 17 h stay in FEDGAS left the tablet closed and able to release about 86–87 µg/mL afterwards. R16·R17·C32 The authors connect this to patients: many people with MDR-TB may not have access to the high-fat meals that help Deltyba, so a formulation that releases well without solubilizing food components and less dependently on gastric pH "may be of practical importance". C51·C50

Where the approach applies. The evidence covers one drug, one polymer (HPMCP P-50), one drug loading (25%), one tablet core and one coating level (14–16% weight gain), in vitro only. A1·A2·P3 The coat-life data come from edisylate tablets alone. X12 Within this scope the authors present the coat as "a promising remediation strategy" that "may" reduce in vivo absorption variability linked to prandial state, and they call more generally for closer examination of gastric crystallization from ASDs whose polymers are insoluble in acid, since in vitro tests rarely cover the range of gastric pH. C5·C4·C42·C47

Note by the AI writer (not discussed in the paper): what this means for formulation. An enteric polymer inside an ASD and an enteric coat around a tablet do different jobs. The polymer keeps the drug amorphous and supersaturated where it dissolves; in acid it neither dissolves nor stops dispersed drug from leaking or crystallizing at the surface. For a fast-crystallizing weak base this argues for testing two-stage release at several gastric pH values, including a weakly acidic one such as pH 3.0, before choosing an enteric-polymer ASD. The coat removed the gastric penalty but did not raise release above single-stage values, and it added a 10–20 min lag. Its weak window is a fed stomach at a pH between about 4.5 and 5.5 for more than about 2 h, where it failed in buffer; FEDGAS prolonged it, but real meals vary in fat and pH. Salt formation in the ASD did not stop surface crystallization at pH 3.0, so a salt chosen for solid-state stability still needs its own gastric-stage test.

Limitations

Limitations stated by the authors. The authors are explicit about the coat's limits: it protected the tablet core at pH 5.0 or below "for at least 2 h in buffer", and "once any cracks in the coating occurred, water and buffer species could penetrate into the core, resulting in drug crystallization and decreased drug release". C45·C22 They note that pH-shift experiments could not be run in FEDGAS because of its fat and carbohydrate content. R15 Their broader conclusions are hedged: the coat "appears to provide a promising remediation strategy" and "may mitigate" variability; the observations "may be of importance for in vivo behavior"; and the strategy "may represent an approach to reduce in vivo absorption variability". C5·C42

The AI writer's own judgement (1): what is observed and what is inferred. Crystals were imaged only on spin-coated films about 8 µm thick, after 1 h at pH 3.0. Tablets or ASD particles were not examined after a gastric stage, at pH 1.6 or 5.0, after coat failure or in the intestinal stage. The causal step "surface crystals cause the lost release" and the seeding step are therefore inferred from release data, and the conclusion that the coat prevents crystallization in tablets rests on the films plus the release curves. X7·C52·C38·C43 The pH 5.0 results also need an explanation the paper does not give: uncoated edisylate tablets released about 83 µg/mL during the pH 5.0 stage, the free base and chloride hardly any, although the text says HPMCP is "not soluble until pH 5". X11

The AI writer's own judgement (2): comparability and reporting. The coated and uncoated tablets were tested with different two-stage protocols (transfer to fresh medium for 60 min against an in-vessel shift for 30 min), and the free base and edisylate uncoated data come from the earlier study, so the comparison of spread is qualitative. X6·X10 Some summary statements do not hold for every arm: the worst gastric pH is 3.0 only for the edisylate, two arms stay below the stated 80 µg/mL, and two-stage release is up to about 15 µg/mL below single-stage release. X5·X3·X4 Replicate numbers and the meaning of error bars are not given for any test, coat integrity was judged visually on one formulation, the solubility method is not described, the acetate buffer recipe is missing (the text points to Table 3, which lists citrate salts), and a few text values differ from the plotted curves by about 3–4 µg/mL. X13·X12·X8·X1·X2 The tablet hardness is printed as 14–15 kN, a value that suggests a misprinted unit. X14 The Supplementary material, which holds the tolerance-test values, the coat-failure data, the Deltyba two-stage data and the overnight film images, was not available for this reading (card section H).

Conclusion

The paper makes three points. First, uncoated tablets of 25% DL DLM–HPMCP ASDs release well in a single stage at intestinal pH but poorly and variably after a gastric stage, and spin-coated films of all three forms crystallize on the surface within 1 h at pH 3.0, which the authors take as the cause ("attributed to surface drug crystallization"). R1·R2·R3·R4·C52 Second, an Acryl-EZE II coat at 14–16% weight gain keeps the tablet closed for more than 15–17 h at pH 1.6–3.0 and for about 2–2.5 h at pH 4.5–5.0 in buffer, longer in a high-fat medium, and dissolves within minutes at pH 6.0. R7·R8·C45 Third, coated tablets of all three forms release less than 0.5% in the gastric stage and about 76–86 µg/mL after transfer to pH 6.5, in fasted and fed media alike; in the authors' words, release was "remarkably improved and less variable". R10·R16·R17·C43

Under the conditions tested (one drug, one polymer, one loading and one coat, in vitro), the coat turns a release that depended on gastric pH into one that does not. The next steps the data point to are direct: look for crystals inside uncoated tablets after a gastric stage, test the coat at fed gastric pH between 4.5 and 5.5 with real meals and longer residence, and check whether the in vitro gain carries over to absorption. X7·C5