When a weakly basic drug locks up an enteric polymer: bedaquiline, HPMCAS and the release medium

Background and motivation

An amorphous solid dispersion (ASD) dissolves a poorly soluble drug at the molecular level in a polymer, so that on release it can form a supersaturated solution and raise membrane transport (Monschke and Wagner 2019, ref 1). The polymer also slows crystallization in the solid and in solution. Among the polymers in marketed ASDs, hydroxypropyl methylcellulose acetate succinate (HPMCAS) stands out for its crystallization inhibition (Butreddy 2022, ref 9). C4 It is an enteric polymer: its succinate carboxylic acid groups must be ionized before it dissolves, which in practice means above about pH 5.0–5.5 (refs 9, 10). C1

Release from HPMCAS ASDs is therefore more complicated than release from non-ionic polymers such as PVPVA or HPMC. The introduction lists the known complications. Release appears to be governed by dissolution of the polymer, particularly at low drug loading (DL); as the polymer ionizes it releases protons, so the unstirred layer at the dissolving surface can be more acidic than the bulk, which slows the polymer further (Nguyen and Fogler 2005, ref 15); and dispersed drug reduces water uptake into the matrix (Bapat 2023, ref 14). C1·C2 Earlier work from the same group measured gel-layer pH values of 5.6, 6.0 and 6.2 for HPMCAS-MF in pH 6.8 phosphate buffer at buffer capacities of 13, 26 and 51 mM/ΔpH. R22

The complication that matters most here is electrostatic. A weak base that is protonated at intestinal pH can pair with the ionized carboxylates of HPMCAS. For lumefantrine (pKa 8.7) with HPMCAS, poor release at pH 6.8 was attributed to such ionic interactions, which were thought to hinder hydration of the carboxylate groups (Hiew 2021, ref 16); later work "supported the conjecture" that lipophilic cationic drugs form a hydrophobic electrostatic complex with HPMCAS at the gel-layer interface and in solution, so that release is poor even at 5% DL (Bapat 2024, ref 11). C3 The same group has also shown that the release medium, its buffer capacity and its solubilizing micelles, can change how HPMCAS ASDs perform (ref 14). C1

Bedaquiline (BDQ), a diarylquinoline anti-tubercular drug, is a natural test case. It is a weak base with a pKa of 8.5 (measured by the authors, in the SI) and a log P of 7.13, practically insoluble in water (<1 µg/mL), and it is recommended to be taken with food. M1 The authors consider it a good ASD candidate because its absorption is limited by both solubility and dissolution rate. C6

The paper sets out two hypotheses. First, complexation between BDQ and HPMCAS should fall at higher pH, where BDQ is less ionized, and in simulated intestinal media, whose mixed micelles can solubilize the drug. Second, release should be low where much complex forms and high where little forms. Nucleation induction times were added to decide whether crystallization could be the cause of any poor release. C5

Background concepts

Ionization of an acid and a base. HPMCAS behaves as a weak acid with a pKa of about 5 (the paper gives 4.97 in the results and "~5" or 5.0 in the discussion), and BDQ as a weak base with a pKa of 8.5. M1·M2·X18 The authors calculate the percent ionization of each from the Henderson–Hasselbalch relation, which the paper names but does not print (Figure 10). E2·R21 Two consequences follow. Between about pH 5 and 8.5 both molecules carry charge, the polymer negative and the drug positive; near pH 6.7 both are about 98% ionized, the point of maximum mutual ionization. R21·C30 At pH 6.0 the polymer is 91.5% ionized and at pH 5.8 88.1% ionized, while BDQ is more than 99% protonated. R21 (The 88.1% value is not reproduced from either pKa the paper gives, which yield 87.1% or 86.3%; the difference is about one percentage point. X18)

How much ionization HPMCAS needs. The authors estimate that HPMCAS must be about 77% ionized before it starts to dissolve, taking the ionization at a dissolution threshold pH of 5.5 with a pKa of 5.0. C28 This threshold carries much of their later argument: a polymer that is 88–92% ionized has only a modest margin above it.

Polyelectrolyte complexes. Each HPMCAS chain carries many carboxyl groups, so it is a polyelectrolyte. Polyelectrolytes form insoluble complexes with oppositely charged polymers, surfactants and small molecules; precipitation is typically greatest when the charges are neutralized, and hydrophobic interactions add a contribution whose balance with electrostatics changes with pH (refs 28–32). C29 For a very lipophilic cation such as protonated BDQ, the pair is hydrophobic enough to fall out of solution; the paper calls it a hydrophobic ion pair. C56

Polymer-controlled release. If drug leaves an ASD only as fast as the polymer dissolves, the polymer's dissolution sets the release rate. The paper writes the polymer dissolution rate as a flux across the aqueous boundary layer (Eq. 1): E1·C41

In plain terms: the polymer flux $J_p$ equals its diffusion coefficient $D_p$ divided by the thickness of the boundary layer $\delta$, times its solubility at the polymer–solution interface $C_p$. Anything that lowers $C_p$, such as part of the polymer being tied up in an insoluble complex, lowers the polymer flux and, under polymer control, the drug release. E1·C41·C42

Simulated intestinal fluids. FaSSIF-V1 (fasted state, pH 6.5), FeSSIF-V1 (fed state, pH 5.0) and FeSSIF-V2 (fed state, pH 5.8) contain solubilizing mixed micelles (the paper does not print their composition; Table SI was not available). M5·M6·M7 The "blank" versions are the same buffers without the micellar components: blank FaSSIF-V1 is a pH 6.5 buffer and blank FeSSIF-V2 a pH 5.8 buffer. A2 Comparing a medium with its blank isolates the effect of the micelles at constant pH.

Crystalline and amorphous solubility. The crystalline solubility is the equilibrium solubility of the crystal; the amorphous solubility, higher, is the concentration at which an amorphous drug-rich phase separates from solution, measured here by adding a DMSO stock until a precipitate forms and ultracentrifuging at once (ref 21). P4·P6 Their ratio sets how much supersaturation an ASD can generate.

Methods in detail

The study has four strands: a solution test for the complex, release tests of the ASDs in many media, checks for crystallization, and X-ray photoelectron spectroscopy (XPS) of the complex and the ASD. The method flow below shows how they connect. P7·P11·P12·P13·P15

The ASDs

BDQ and HPMCAS (LF or MF grade) were dissolved in acetone:dichloromethane 1:1 v/v at 5% w/v total solids and rotary evaporated at 50 °C over 20 min, at drug:polymer ratios of 5:95, 10:90, 20:80 and 30:70. P1 The products were dried under vacuum for a further 48 h, cryomilled (four 1 min cycles at 10 cps after 2 min pre-cooling in liquid nitrogen) and sieved; the fraction below 250 µm was used. P1 All ASDs were X-ray amorphous and showed a single glass transition (the traces are in the SI). R1 The two grades differ in acid content: 1.54 mM (LF) and 1.19 mM (MF) of COOH groups per gram of polymer, taken from earlier work of the group. M2·M3

Contribution 1: a solution test for the drug–polymer complex

The problem. A complex inside a dissolving ASD particle is hard to see. The authors therefore built a simpler test in which drug and polymer meet in the medium itself, and asked how much of the polymer leaves solution. P7

What they did. Stock solutions in tetrahydrofuran held HPMCAS at a fixed 25 mg/mL and BDQ at six levels (Table 1), chosen so that the BDQ:COOH molar ratio matched that of ASDs at about 5, 10, 20, 30, 40 and 50% DL: 0.0625:1 to 1.2:1 for LF and 0.08:1 to 1.5:1 for MF. P7 A 100 µL aliquot went into 5 mL of medium stirred at 990 rpm, giving 0.5 mg/mL of polymer; after 30 min the mixture was ultracentrifuged (35,000 rpm, 30 min, 25 °C) and the polymer left in the supernatant was measured. P7 A stock with polymer only served as the control, and percent complexation is the polymer lost relative to that control. P7 Polymer was measured by phenol–sulfuric acid colorimetry at 490 nm in plain buffers and by HPLC with evaporative light scattering detection (ELSD) in the simulated intestinal fluids, each with calibrations from 1 µg/mL upward. P8·P9

The media were pH 5.8 buffer (blank FeSSIF-V2), FeSSIF-V2, pH 6.0 buffer, pH 6.5 buffer (blank FaSSIF-V1), FaSSIF-V1 and pH 7.0, 8.5 and 10.5 buffers. P7 Only media in which HPMCAS alone dissolves were used: the polymer-only control gave a clear solution at 0.5 mg/mL from pH 5.8 upward, whereas at pH 5.0 (blank FeSSIF-V1) HPMCAS dissolves only to about 250 (LF) and 110 µg/mL (MF), according to earlier work. R3·R4·C11 The blank buffers held their pH during the test; the pH 6.0, 7.0, 8.5 and 10.5 buffers fell by 0.2–0.5 units as the polymer released protons. R9 The pellets were dried and analysed by ATR-FTIR. P10

Why it works, and its limit. The test isolates one question, whether drug and polymer precipitate together in a given medium, and answers it at the pH and with the micelles of that medium. It does not reproduce the gel layer of a dissolving particle, where the polymer is far more concentrated and the pH is set by the polymer's own protons; that limit returns in the discussion. X14

Contribution 2: release tests in plain and biorelevant media

ASD powder equal to 5 mg of BDQ (100, 50, 25 and 16.66 mg for 5, 10, 20 and 30% DL) was added to 50 mL of medium at 37 °C with paddles at 150 rpm, so 100 µg/mL means complete release; concentrations were followed by in situ fiber optic UV every minute for 120 min, in triplicate. P11·P12 In the two-stage test the first 60 min were in pH 1.6, 3.0 or 5.0 buffer, to mimic pH variation with food intake or acid-reducing drugs, and then concentrated phosphate buffer raised the pH to 6.5. P11 Single-stage tests used pH 6.5 buffer (5 and 10% DL), FaSSIF-V1 (10 and 20% DL), FeSSIF-V1 (20% DL) and FeSSIF-V2 (20 and 30% DL). A5 The text describes the pH-shift test for 5 or 10% DL, but only 5% DL results are shown. X3

Contribution 3: ruling crystallization in or out

Two kinds of evidence address crystallization. The undissolved residue at 120 min was photographed, examined under polarized light and analysed by powder X-ray diffraction (PXRD). P14 Separately, BDQ was added from a 20 mg/mL DMSO stock to 5 mL of medium at 37 °C to give 100 µg/mL, without polymer or with 400 µg/mL of LF or MF already dissolved (the polymer level of a fully dissolved 20% DL ASD at this dose), and light scattering was followed every 10 min for 24 h; the end point was checked by polarized light microscopy and the particle size by dynamic light scattering. P13·R18

Contribution 4: XPS of protonation and surface composition

XPS reads the binding energy of core electrons in the top ~10 nm of a solid; the protonated nitrogen of BDQ gives an N 1s peak at about 402 eV, the neutral nitrogen one at about 399 eV. P15·R19 The authors fitted N 1s spectra with a protonated and an unprotonated component taken from BDQ maleate, which they treat as 100% ionized because BDQ forms only a monoprotic salt (ref 24). P15·C24 They expressed the result as percent protonation of the ionizable nitrogen, and fitted the C 1s region with reference spectra of BDQ and HPMCAS to estimate the surface drug content. P15 The samples were BDQ free base, BDQ maleate, a 40% DL BDQ–MF ASD, the insoluble complex from MF at 1:1 BDQ:COOH, and the 40% DL ASD after 1 h in pH 6.5 buffer, measured on 3–5 spots each. P15·A8 The 40% DL ASD is not among the loadings listed in the ASD methods, and its preparation is not described. X9

Experimental analysis

Solubility: very low at intestinal pH, very high in micelles

Table 2 shows the expected behaviour of a base with a pKa of 8.5: the crystalline solubility of BDQ is 67.1 µg/mL at pH 1.6 and falls to 0.52 µg/mL at pH 5.0, below 0.1 µg/mL at pH 5.8 and below detection at pH 6.5. R2·C7 The simulated intestinal fluids reverse this: 7.8 µg/mL in FaSSIF-V1, 258.8 µg/mL in FeSSIF-V2 and 628 µg/mL in FeSSIF-V1. R2 The authors attribute this order to both pH and the amount of solubilizing species, the latter "likely predominating". C8 The amorphous solubility could be measured only in the three weakly acidic buffers, where it was 13.4, 1.9 and 0.63 µg/mL at pH 5.0, 5.8 and 6.5; at pH 5.0 that is about 26 times the crystalline value, and the authors expect a more than 20-fold advantage elsewhere too ("can be anticipated"). R2·C9

Table A. BDQ solubility at 37 °C, mean ± SD, N = 3 (from Table 2). R2

Note by the AI writer (not discussed in the paper): several of these values lie below the limits the methods give for the HPLC assay (LOD 1 ± 0.5 µg/mL, LOQ 3.5 ± 0.2 µg/mL), so the 0.52, <0.1, 1.9 and 0.63 µg/mL entries, and the 20-fold ratio built on 0.52 µg/mL, are best read as approximate. X11·P5

Complexation depends on pH: greatest at pH 5.8–6.0

In plain buffers, BDQ removed HPMCAS from solution in proportion to how much drug was present and how strongly both were ionized (Figures 1A and 2A). R5·R7 At pH 6.0 complexation rose steeply with the BDQ:COOH ratio and levelled off at about 90–96% from a ratio equivalent to 20% DL, so that "virtually no polymer remained soluble"; at pH 5.8 it reached about 80–90%. R5.a·R5.b·R7.a·R7.b·C12 At pH 6.5 the plateau was lower, about 58.5% (LF) and 65.5% (MF) at the highest ratio, at pH 7.0 about 50 and 57%, at pH 8.5 about 18 and 22%, and at pH 10.5 only about 7 and 5.5%. R5·R7 The authors rank the pH values 6.0 > 5.8 > 6.5 > 7.0 > 8.5 > 10.5 for LF and call the MF trend similar. C12 The ranking holds well at ratios of about 0.5:1 and above; at the two lowest ratios the pH 6.5 to 10.5 values crowd together at about 3–18.5%, and for MF at 0.16:1 the pH 7.0 and 8.5 values exceed the pH 6.5 value. X6·R7.c·R7.d·R7.e

Table B. Percent of HPMCAS removed from solution as insoluble complex, 0.5 mg/mL polymer, 30 min (read from Figures 1 and 2, approximate). The lowest and highest ratios correspond to ASDs of about 5 and 50% DL. R5·R6·R7·R8

The text rounds a few of these values: complexation at pH 10.5 is given as "<5%", whereas the plotted values reach about 7–7.5% at some ratios; and blank FeSSIF-V2 is given as "~85%" at LF ratios of 0.5:1 to 1.2:1, whereas LF reads about 89–90% there (MF reads about 84–84.5%). C12·C13·X5 The differences are a few percentage points and do not change the pattern.

Micelles largely prevent the complex

Figures 1B and 2B set each simulated intestinal fluid against its blank buffer at the same pH. At pH 6.5, LF complexation at the highest ratio fell from about 58.5% in the blank to about 25% in FaSSIF-V1; at pH 5.8, from about 90% in the blank to about 17% in FeSSIF-V2. R5.a·R5.c·R6 For MF the corresponding values were about 65.5 against 26.5% and 84.5 against 29%. R7.a·R7.c·R8 The authors summarize this as "~20% or less" for LF and "~30% or less" for MF, with similar ratio dependence in the two fluids; LF in FaSSIF-V1 actually reaches about 24–25% at the two highest ratios. C13·X5 Because the same fluids raise the crystalline solubility of BDQ by orders of magnitude, the authors infer "a high affinity of the drug for the surfactant micelles". C36·R2

What the complex is made of

ATR-FTIR spectra of the dried pellets (Figure 3) show, for complexes formed at pH 6.0, 6.5 and 7.0, the carbonyl band of HPMCAS (1720 cm⁻¹ in the text; BDQ has no carbonyl) together with BDQ bands at 1210–1285, 1380–1490 and 1530–1650 cm⁻¹ that are much broader than in crystalline BDQ. R11 The authors read this as polymer plus disordered drug, "as would be expected for a drug-polymer ionic complex". C14 Pellets from pH 8.5 and 10.5 instead show sharp bands of crystalline BDQ, and at pH 10.5 the polymer bands are essentially absent, "consistent with" the very low complexation there. R11·C15 The residue composition of the MF complex at 1:1 BDQ:COOH was estimated at about 52% drug and 48% polymer (Table SV). R10

Release in plain buffer stays low

In the two-stage test with 5% DL ASDs (Figure 4A, C), less than 10% of the drug dissolved in the acid stage: about 9% at pH 1.6, 4–5% at pH 3.0 and 2–3% at pH 5.0 after 60 min. R12 The highest value at pH 1.6, where BDQ is most soluble, is "presumably" due to that solubility, because HPMCAS does not dissolve at low pH. C16 After the switch to pH 6.5 the concentration jumped, but only to about 23–25 (from pH 1.6), 19.5–20 (from pH 3.0) and 20–22 µg/mL (from pH 5.0) at 120 min, for both grades. R12 In pH 6.5 buffer alone, 5% DL ASDs reached about 30 (LF) and 28 µg/mL (MF), and 10% DL ASDs about 17 and 19.5 µg/mL, at 120 min (Figure 4B, D). R13 The text summarizes these results as "approximately 20 %" for both loadings and grades. C17 Note by the AI writer: the plotted 5% DL single-stage curves end nearer 30% than 20%, and pH-shift data are shown only for 5% DL. X2·X3

The vessels at 120 min held undissolved material: settled particles for LF and a turbid suspension for MF (Figure 5). R14 Polarized light showed no birefringence, and PXRD of all residues showed amorphous halos, which the authors take as evidence against crystallization. R14·C18 Note by the AI writer: the PXRD patterns of the three pH-shift residues also carry one sharp peak near 32° 2θ, absent from the pH 6.5 single-stage residue; the text does not discuss it. X16

Release in simulated intestinal fluids

Figures 6–8 show the same ASDs in the biorelevant media. In FaSSIF-V1, 10% DL ASDs reached about 97 (LF) and 87.5 µg/mL (MF) at 120 min, and 20% DL ASDs about 80 and 80.5 µg/mL; the text gives 80–90% for 10% DL, with some undissolved material left, and about 80% for 20% DL. R15·C19·X4 In FeSSIF-V2 (pH 5.8), 20 and 30% DL ASDs released quickly, passing about 66% within 20 min, and ended at about 87.5–95 µg/mL ("approximately 90 %"). R17·C21 In FeSSIF-V1 (pH 5.0), 20% DL ASDs crept to only about 25 (LF) and 22.5 µg/mL (MF), which the authors put down to the low solubility of HPMCAS at pH 5.0. R16·C20

Table C. BDQ concentration (µg/mL) at the end of the release test, 5 mg BDQ in 50 mL, so the values equal percent released; 120 min, or the last point at about 117 min in FeSSIF-V1 and FeSSIF-V2 (read from Figures 4 and 6–8, approximate). R12·R13·R15·R16·R17

The comparison the paper rests on is clearest at 10% DL, the only loading tested both in pH 6.5 buffer and in FaSSIF-V1: about 17–19.5% against about 87.5–97%. R13·R15·X15 At the other loadings the media differ in drug loading as well as in composition. X15

The drug stays amorphous

Table 3 tests whether supersaturated BDQ crystallizes. At 100 µg/mL, far above the amorphous solubility of about 2 and 0.6 µg/mL at pH 5.8 and 6.5, BDQ without polymer precipitated from the pH 5.8 and 6.5 buffers within 1 min as particles larger than 5000 nm; these stayed amorphous for up to 2 h and showed birefringence by 3 h. R18·R2·C22 With 400 µg/mL of LF or MF already in solution, BDQ again precipitated at once, but as particles of 413–967 nm in the buffers, which showed no crystallization for 12–24 h. R18·C22 In FeSSIF-V2 and FaSSIF-V1, too, no crystallization was seen for 12–24 h with polymer. R18

Table D. Nucleation induction time and particle size of BDQ (100 µg/mL) with and without HPMCAS, 37 °C; size mean ± SD, N = 3 (from Table 3). The times are lower bounds: how long no crystallization was seen. R18·X17

On this evidence the authors conclude that BDQ "remained amorphous in the presence of HPMCAS for extended time periods, hence crystallization was not considered a failure mechanism". C23

XPS: the complex is ionic, and so is part of the dry ASD

Figure 9 and Table 4 give the XPS results. The free base showed a single N 1s peak at about 399.5 eV and 4 ± 4% protonation; the maleate salt showed peaks at 399 and 402 eV and fitted to 101 ± 5%. R19·R20.a·R20.b The insoluble MF complex was 55 ± 4% protonated and the dry 40% DL ASD 52 ± 7%. R20.c·R20.d Since BDQ is nearly insoluble from pH 5.0 upward, one might expect the precipitate to be free base; instead about half of the ionizable nitrogen is protonated, which the authors say "confirmed" drug–polymer ionic interactions. C35 The similar value in the dry ASD shows, in their reading, that these interactions form "not only during precipitation, but also during ASD manufacture from organic solvents", as reported before for other basic drugs with HPMCAS. C25·C35

Table E. XPS percent protonation (relative to the ionizable nitrogen) and surface drug concentration in the top ~10 nm (from Table 4). R20

The surface data add a second observation. The dry 40% DL ASD had 32 ± 2 wt% drug at its surface, below its bulk loading, which the authors call "some polymer enrichment"; after 1 h in pH 6.5 buffer the surface held 55 ± 2 wt% drug and only 15 ± 3% protonation, "indicating preferential loss of polymer from the surface", with about 45% polymer remaining. R20.d·R20.e·C26 The complex, about 52% drug in bulk, had only 26 ± 1 wt% drug at its surface, so its surface is polymer-rich. R10·R20.c·C26 The text describes the post-immersion spectrum as showing reduced intensity "at 399 eV"; the table and the spectrum show that it is the protonated component at about 402 eV that is lost. X7 The panel labels of Figure 9 (BDQ-MF, CR 40/60, DR 40/60) also do not match the caption, so which of the two middle spectra is the complex and which the dry ASD is not certain from the figure; Table 4 is unambiguous. X8

Discussion

Why complexation peaks at pH 5.8–6.0 rather than at maximum mutual ionization

With both species about 98% ionized near pH 6.7, the most complex might be expected there. R21·C32 Instead it peaks at pH 5.8–6.0, and the authors offer two reasons. At pH 6.5 the highly ionized polymer is more soluble, which "may" shift the balance from polymer in the complex toward dissolved polymer. C32 At pH 5.8 and 6.0 the polymer is 88–92% ionized while BDQ is more than 99% protonated, so drug cations are in slight excess over carboxylates; since HPMCAS needs about 77% ionization to stay dissolved, pairing only a small share of its carboxylates pushes it below that threshold, and "the excess BDQ may explain why complexation is highest at these pH conditions". C33·C28 At pH 10.5, where BDQ is mostly neutral, the small residual complexation "can be attributed to hydrophobic interactions". C31 The authors add that the buffers contained salts, including NaCl, and that ionic strength can screen electrostatic interactions. C34 None of these mechanisms was tested directly; they are offered to explain the shape of Figure 1A. C32·C33·C31

How the complex is thought to stop release

The authors start from the "premise" that release from these ASDs is controlled by polymer dissolution. C40 They justify it by the pH dependence: release is low at pH 1.6, where BDQ is most soluble but HPMCAS is not, and rises only when the polymer ionizes; drug-controlled release would show the opposite trend. C40·R12·R2 Under polymer control, the polymer flux of Eq. 1 sets release, and anything that lowers polymer solubility at the interface, $C_p$, lowers it. C41·E1 The chain they propose runs as follows. C42·C43·C46·C49·C56

Each link carries a hedge. For the plain-buffer case, the authors take complexation at pH 6.0 as the relevant value because the gel-layer pH is lower than the bulk; at the ratios of 5 and 10% DL ASDs they read "~55 and ~75%" from Figure 1A, and conclude that "a relatively low extent of complexation between drug and polymer can be extremely detrimental". C44·C45 Note by the AI writer: the discussion assigns the ratios 0.08:1 and 0.16:1 to HPMCAS-LF and Figure 1A, but Table 1 gives them for MF; the values of about 55 and 75% are the MF values of Figure 2A, while the LF values at 0.0625:1 and 0.125:1 are about 43 and 77%. X1·R5.b·R7.b The conclusion is unchanged, since both grades lose roughly half to three quarters of their polymer at these ratios. The authors then trace the effect to the 77% ionization threshold: it is "not the ionization per se, but rather the hydration resulting from ionization" that dissolves the chains, and the complex "is expected to impede hydration and hence polymer plasticization, dilution of polymer chains and ultimately release". C46 A second factor is the drug-rich surface seen by XPS after 1 h at pH 6.5, "expected to be highly unfavorable to release" because BDQ is nearly insoluble at that pH. C43·R20.e

For the simulated intestinal fluids, the authors argue that the mixed micelles compete for BDQ. BDQ is predominantly ionized at the pH of FaSSIF-V1 and FeSSIF-V2, and ionization usually reduces micellar solubilization (ref 38), yet the large solubility gains show that BDQ cations are "clearly still able to interact" with the micelles; "the positive charge on BDQ may allow for favorable interactions with micelles". C38·C37 This affinity "explains the observed reduction in complexation". C39 The improved release is then "partially explained" by the disrupted complex, and this is "likely the predominant mechanism" in FaSSIF-V1. C49 In FeSSIF-V2 a second factor applies: with a crystalline solubility of 258.8 µg/mL against a maximum of 100 µg/mL, the test is "no longer conducted under non-sink conditions" and the driving force for release is very high. C50·R2.h FeSSIF-V1 fits the polymer-control picture from the other side: it has the highest BDQ solubility of all media (628 µg/mL) but sits at pH 5.0, where HPMCAS barely dissolves, and release stays at about 22.5–25%. C51·R2.g·R16

What it means for testing and formulation

The authors draw a practical conclusion. Judged by plain buffers, "BDQ is not a good candidate for an ASD formulation with HPMCAS"; judged by FaSSIF-V1 or FeSSIF-V2, the same ASDs release well even at higher loadings, "now suggesting that an ASD with BDQ should have good bioperformance". C47 With an amorphous solubility more than 20-fold the crystalline one, an ASD that releases readily under non-sink conditions "can" maintain high supersaturation while the drug is absorbed. C48 They contrast BDQ with delamanid, a weak base of low pKa that is not extensively ionized where HPMCAS is anionic, for which biorelevant media made no difference in earlier work; neat HPMCAS also released at the same rate with or without the micellar components (refs 8, 14). R23·C52 Hence "it is likely that the impact of media properties on release from ASDs varies between drugs", depending on drug–polymer interaction tendency and on drug solubilization by the medium, and "clearly caution should be applied when performing screening studies to identify high performing formulations if simple buffers are used". C53·C54

Note by the AI writer (not discussed in the paper): two properties together flag the risk shown here. The drug must be fully protonated in the pH range where HPMCAS dissolves, which requires a basic pKa well above about 7, and the drug cation must be lipophilic enough that its pair with the polymer precipitates; BDQ, with a pKa of 8.5 and a log P of about 7, has both. M1 For such a drug, a low release result in plain buffer says more about the test than about the formulation, and a release test in a blank buffer and in its micellar counterpart at the same drug loading separates the two. X15

Limitations

Limits the authors state. The authors note that buffer salts and NaCl can screen the electrostatic interactions, so complexation may depend on ionic strength. C34 They describe the improvement in simulated intestinal fluids as only "partially explained" by complex disruption, and give FeSSIF-V2 a second explanation, the sink condition. C49·C50 They expect media effects to differ from drug to drug. C53

The AI writer's own judgement. The evidence is strongest where it is direct: complexation as a function of pH and medium, protonation by XPS, release in each medium, and the absence of crystallization are all measured, with replicates. R5·R6·R7·R8·R20·R13·R15·R18 The causal link from complex to release is weaker, for four reasons.

• The complex was measured in a solution test (a THF stock injected into the medium at 0.5 mg/mL polymer, 30 min, ultracentrifuged at 25 °C), not in the gel layer of a dissolving ASD; polymer release from the ASDs and polymer solubility at their surface were not measured. The dry 40% DL ASD does show ion pairing by XPS. X14·R20.d
• Buffer and simulated intestinal fluid were compared at the same drug loading only once (10% DL, pH 6.5 buffer against FaSSIF-V1); the other loadings differ by medium, and FeSSIF-V2 has no matching blank-buffer release test, although the introduction says release was measured at pH 5.8. X15
• In FaSSIF-V1, micellar solubilization alone could raise release: the crystalline solubility rises from undetected to 7.8 µg/mL, and the authors' own more-than-20-fold factor would put the amorphous solubility above the 100 µg/mL maximum. The paper does not separate this from complex disruption, and amorphous solubility in the biorelevant media was not measured. X13·X12
• The surface-enrichment argument for the 5 and 10% DL ASDs uses XPS of a 40% DL ASD whose preparation is not described and which was not release-tested. X9·X10

Several printed values also differ from the figures: about 20% release for the 5% DL ASD in pH 6.5 buffer against about 28–30% plotted, 80–90% in FaSSIF-V1 for 10% DL against about 97% plotted for LF, the LF/MF mix-up in the discussion, and the XPS peak named for the post-immersion change. X2·X4·X1·X7 Several low solubility values lie below the stated HPLC quantification limit. X11 None of these reverses a conclusion, but they matter to anyone reusing the numbers.

Conclusion

For bedaquiline with HPMCAS LF and MF, the study shows directly that the drug removes the polymer from solution as an insoluble, ionic complex whenever both are substantially ionized, most at pH 5.8–6.0, and that the mixed micelles of FaSSIF-V1 and FeSSIF-V2 largely prevent it. R5·R7·R20·R6·R8·C12·C35·C36 The same ASDs release about 17–30% of their drug in plain pH 6.5 buffer and about 80–97% in FaSSIF-V1 and FeSSIF-V2, without crystallizing in either. R13·R15·R17·R14·R18 The authors attribute the poor buffer release to the complex lowering polymer solubility at the dissolving surface, and the good release in intestinal fluids to its disruption by micelles, with the sink condition contributing in FeSSIF-V2. C56·C42·C49·C50 The link between the complex measured in solution and the release of the particle is inferred rather than measured, and micellar solubilization is not separated from complex disruption. X14·X13 The practical message the authors draw is that ASDs of lipophilic cationic drugs with enteric polymers should not be judged on simple buffers alone. C54·C53