Bile salts as crystallization inhibitors for supersaturated telaprevir: monomers, micelles and drug-rich droplets

Background and motivation

Many new drug candidates are poorly soluble in water, and a common answer is to deliver them in a higher-energy form, such as an amorphous solid dispersion (ASD), that dissolves to give a supersaturated solution. The extra concentration only helps absorption if the solution does not crystallize first, so the ability to maintain supersaturation over the time the drug spends in the small intestine decides whether such a formulation works (refs 1–4). C1 Polymers are the usual crystallization inhibitors, and a wide range of them stabilize both the amorphous solid and the supersaturated solution (refs 5–10). C2

Bile salts are a second, endogenous class of additive that the formulation scientist cannot avoid: they are present in the upper intestine and in biorelevant dissolution media. They are surfactants with an unusual shape. A rigid steroid nucleus carries its hydroxyl groups on one face (facial polarity), so bile salts aggregate stepwise rather than in one sharp step and have broad critical micelle concentration (CMC) ranges. The six most abundant in human intestinal fluid are sodium taurocholate (STC), sodium taurodeoxycholate (STDC), sodium taurochenodeoxycholate (STCDC), sodium glycocholate (SGC), sodium glycodeoxycholate (SGDC) and sodium glycochenodeoxycholate (SGCDC), with reported CMCs of 2 to 12 mM; the trihydroxy bile salts (STC, SGC) are more water soluble and have higher CMCs than the dihydroxy ones, and the total amount and mix vary greatly between people (refs 14–19). C3·M2

Earlier work from the same group showed that bile salts can act as crystallization inhibitors. Chen et al. (ref 12) found that STC below its CMC slowed solution crystallization of structurally diverse drugs, while sodium dodecyl sulfate (SDS) promoted it, and that SGC and SGDC inhibited celecoxib and nevirapine crystallization as monomers. Li et al. (ref 13) compared 13 bile salts below their CMC with celecoxib, nevirapine and flibanserin and found that most slowed crystallization, but to different extents, so that bile salts are "not interchangeable". C4 Chen et al. (ref 11) also saw STC raise the celecoxib induction time more than ten-fold both below and above its CMC, whereas chain-like surfactants (SDS, Polysorbate 80, Poloxamer 188) enhanced crystallization. C30

The authors identify three gaps. First, bile salts in intestinal fluid sit in mixed micelles, yet nearly all earlier crystallization data were for monomers, so little was known about how the aggregation level changes inhibition. Second, nobody had tested bile salts in the more complex solutions that dissolving ASDs actually produce: above the amorphous solubility a sub-micron amorphous drug-rich phase forms by liquid–liquid phase separation (LLPS), or by glass–liquid phase separation (GLPS) when the water-saturated drug is glassy at the test temperature (refs 20–22). C5 Third, the mechanism of bile salt inhibition, and whether it complements polymers, was not known.

A further, methodological problem sits underneath. A surfactant that solubilizes the drug raises its crystalline solubility, so at a fixed concentration it lowers the true driving force; comparing additives at equal concentration then confuses solubilization with genuine inhibition (ref 23). C6 In a companion study (ref 24), the authors measured how the six bile salts, as monomers and as micelles, change the thermodynamics of supersaturated telaprevir, developed a way to calibrate the supersaturation, and found that the trihydroxy bile salts change telaprevir thermodynamics less than the dihydroxy ones. C6 This paper uses that calibration to compare all six bile salts, below and above their CMC, at the same activity-based supersaturation, and in solutions just below and just above the amorphous solubility of telaprevir. A1·A2·A3

Background concepts

Supersaturation as an activity ratio. Crystallization from solution is favoured whenever the chemical potential of the dissolved drug exceeds that of the crystal. The paper defines the driving force, the supersaturation $\delta$, as an activity ratio (Eq. 1, PDF p.8): E1

In plain terms: $\mu$ is the chemical potential of the dissolved drug, $a$ its activity, $R$ the gas constant and $T$ the temperature; the reference state (°) is a solution in equilibrium with the stable crystal, so $\delta =1$ at crystalline solubility and $\delta =11.5$ means the drug is 11.5 times as "active" as in a saturated solution. E1

Activity cannot be read from concentration once micelles hold part of the drug. The authors therefore measured it from the flux of drug across a regenerated cellulose membrane (molecular weight cutoff 6–8 kDa) in a side-by-side diffusion cell; only free drug crosses the membrane, and the mass flow rate $F$ is proportional to activity (Eq. 2, PDF p.8): E2·M9·P1

In plain terms: the ratio of the flux from the test solution to the flux from a crystal-saturated solution is the supersaturation itself, whatever fraction of the drug sits in micelles. Figure 2 turns the flux data of the earlier study into curves of $\delta$ against telaprevir concentration for buffer and for each bile salt at 1.86 and 12 mM. E2·R1

Amorphous solubility and the drug-rich phase. Above a certain concentration, the amorphous solubility, a supersaturated solution separates into a drug-lean continuous phase and a dispersed drug-rich phase. For telaprevir this phase is glassy (GLPS), and earlier work put the droplet size at about 300 nm (ref 42). R12·C5 The two supersaturations chosen here bracket this limit: $\delta =11.5$ is a homogeneous solution in which telaprevir is molecularly dissolved, and $\delta =28.8$ just exceeds the amorphous solubility, so a glassy second phase is present (ref 24). C8·P3

Monomers and micelles. Below the CMC a surfactant is present as single molecules (monomers); above it, extra surfactant forms micelles while the monomer concentration stays roughly at the CMC. The paper uses 1.86 mM bile salt as the "monomeric" level and 12 mM as the "micellar" level, and also 18 mM for STC and SGC and a series of STC concentrations below its CMC. A2·P3·P4 Which level is which rests on CMCs measured here for one trihydroxy and one dihydroxy bile salt only (STC and SGDC). R5·C14

Induction time. The induction time $t_{ind}$ is the time from creating a supersaturated solution to the first detectable crystallization. It contains two parts (Eq. 3, PDF p.9): E3

In plain terms: $t_n$ is the time needed to form critical nuclei and $t_g$ the time for those nuclei to grow to a detectable size; reading $t_{ind}$ as a measure of nucleation assumes that $t_n$ dominates, which the paper later checks for two bile salts. E3·C9·C22

Zeta potential. The zeta potential is the electric potential at the slipping plane around a particle. Telaprevir is taken to be neutral at pH 6.5 and the bile salts are anions, so a more negative zeta potential of the drug-rich droplets after adding bile salt is read as bile salt sitting on the droplet surface. C10 The authors stress that this is qualitative evidence, since zeta potential includes physically and chemically adsorbed ions within the slip plane and does not measure how much is adsorbed. C40

Methods in detail

All experiments used telaprevir in 50 mM sodium phosphate buffer, pH 6.5, at 37 °C. M1·M10·P2 The bile salts were used as received (STC practical grade; the others 97–99% pure), and HPMCAS-MF served as a polymer comparator. M2·M3·M4·M5·M6·M7·M8 The work has three parts: fixing the driving force, timing crystallization under many conditions, and three side experiments that probe the mechanism.

Contribution 1: comparing additives at a constant driving force

The problem. If a bile salt solubilizes telaprevir, a solution of fixed concentration has a lower driving force in its presence, and a longer induction time could simply reflect that. C25 What they did. Using the flux data of the earlier study and Eq. 2, the authors plotted $\delta$ against telaprevir concentration for buffer and for each bile salt at 1.86 and 12 mM (Figure 2) and read off the concentration that gives $\delta =11.5$ and $\delta =28.8$ in each medium. P1·R1 The figure marks these concentrations with arrows but prints no numbers. Read from it, buffer, all 1.86 mM media except SGCDC, and 12 mM STC and SGC need about 60 µg/mL for $\delta$ 11.5 and about 150 µg/mL for $\delta$ 28.8; SGCDC needs about 90 µg/mL for $\delta$ 11.5; 12 mM STDC, SGDC and STCDC need about 140 µg/mL for $\delta$ 11.5, and 12 mM STDC and SGDC about 300 and 330 µg/mL for $\delta$ 28.8. R1.a–f With STCDC and SGCDC the activity levels off at $\delta$ about 21–24 and never reaches 28.8, which the authors attribute, citing the earlier study, to partitioning of these bile salts into the telaprevir-rich phase. R1.b·R1.c·R1.f·C13 Why it works. Because every crystallization test then runs at the same activity, the authors can rule out solubilization as the cause of longer induction times and attribute differences to direct effects on crystallization. C26

Note by the AI writer (not discussed in the paper): Figure 2 draws one arrow for the three 12 mM dihydroxy curves at $\delta$ 11.5 and one line for SGCDC at both concentrations, so the exact concentration used in each run is not recoverable from the paper. X12

Contribution 2: induction times across aggregation states and phase states

What they did. A methanolic telaprevir stock (12 or 20 mg/mL) was titrated into 50 mL of buffer, with or without predissolved bile salt or polymer, stirred at 300 rpm and 37 °C. A UV/vis spectrometer with a fibre-optic dip probe (1 cm or 0.5 cm path) followed the absorbance at 270 nm, the absorption maximum, and the extinction at 370 nm, where telaprevir does not absorb. P2 When crystals form, the 270-nm signal drops as dissolved drug is consumed and the 370-nm signal rises from scattering at the same moment; that moment is the induction time (Figure 3). P2·R2 In the example of Figure 3 the apparent concentration held near 140–150 µg/mL and then fell to about 30 µg/mL at an induction time of about 180 min, while the extinction jumped from about 0.02 to about 0.85; the paper does not say which solution this was. R2 Runs were done in triplicate, with standard deviations as error bars, and ended at 960 min (16 h); bars drawn at 960 min with the label ">960" mean that no crystallization was detected. P2·R3

The design crosses bile salt (six, plus buffer), bile salt concentration (1.86 mM and 12 mM, plus 18 mM for STC and SGC) and supersaturation ($\delta$ 11.5 and 28.8). A1·A2·A3 A separate series varied STC from 0.19 to 1.86 mM at 60 µg/mL telaprevir ($\delta$ 11.5), with runs ending at 720 min. P4·R6 For comparison, HPMCAS-MF was tested at 5 µg/mL and 1 mg/mL at both supersaturations. P3·A4 Solutions with 1.86 mM bile salt were also inspected under cross-polarized light. P9

Why it differs from earlier work. The earlier bile salt studies (refs 12, 13) used monomeric bile salts and homogeneous solutions; this design adds the micellar level, the two-phase solutions that ASDs produce, and the activity calibration. C4·C5·C6

Contribution 3: three probes of the mechanism

CMC by surface tension. To place the monomer and micelle levels, the surface tension of STC and SGDC solutions was measured by the Wilhelmy plate method (Krüss K12, 37 ± 0.5 °C, accuracy within ±0.05 mN/m), without and with 60 µg/mL telaprevir. P5 STC was chosen as the representative trihydroxy and SGDC as the representative dihydroxy bile salt; the other four were assumed to behave like their group. C14

Zeta potential of the drug-rich droplets. Telaprevir was titrated into 10 mL of buffer containing 1.86 mM of each bile salt, at 300 rpm and 37 °C, until the solution turned turbid, giving 180 µg/mL; no crystals were visible under cross-polarized light. The zeta potential of the droplets was measured with a Malvern Nano ZS and compared with droplets formed without bile salt. P6

Separating nucleation from growth. A measured induction time mixes nucleation and growth (Eq. 3). Following earlier work (refs 32, 33), the authors combined unseeded induction times with seeded growth rates measured under similar conditions to estimate the relative nucleation rate (Eq. 4, PDF p.14): E4·P8

In plain terms: $J$ and $J_0$ are the nucleation rates with and without bile salt, $R_0/R$ is the mass growth rate without bile salt divided by that with bile salt, and $t_{u,0}$ and $t_u$ are the unseeded induction times without and with bile salt. A bile salt that lengthens the induction time a lot but barely slows growth must be slowing nucleation strongly. E4 The growth rates came from seeded solutions in buffer, 1.86 mM STC and 1.86 mM SGC, as the slope of concentration against time over the first 10 min (Figure 11). P7·R9 The paper gives no description of the seeded experiment itself: seed form and amount, volume, temperature and number of replicates are not stated. X3

Experimental analysis

Homogeneous solutions: monomers hold, three dihydroxy micelles do not

Without bile salt, the homogeneous telaprevir solution ($\delta$ 11.5) crystallized within about 30 min. R3.a With 1.86 mM of any of the six bile salts, no crystallization was detected within 960 min, and no crystals were visible under cross-polarized light (Figure 4). R3.b–g At 12 mM the picture splits: STC, SGC and SGCDC still held for more than 960 min, but STDC, SGDC and STCDC crystallized after about 140, 125 and 180 min, which the text rounds to "about 2 hours". R3·C11 Table A collects these values together with the two-phase results discussed below.

Table A. Induction times (min, mean ± SD, n = 3) of telaprevir in pH 6.5 buffer at 37 °C, read from Figures 4 and 5 (approximate); ">960" means no crystallization within 960 min. STCDC and SGCDC solutions cannot reach δ 28.8. R3·R4

The CMC measurements (Figure 6) support the authors' labels for the two bile salts tested. Surface tension fell with the logarithm of concentration to a minimum near 4 mM for STC (about 46.3 mN/m) and near 2 mM for SGDC (about 45.6 mN/m), which the figure labels as "CMC~4mM" and "CMC~2mM". R5 With 60 µg/mL telaprevir the surface tension was lower at low bile salt concentration (for SGDC about 51.0 instead of about 55.6 mN/m at a log concentration of −4), but one CMC label covers both curves, and the authors take the effect of telaprevir on the CMC to be negligible. R5.a–d·C14 On this basis they write that "it can be reasonably assumed" that most bile salt at 1.86 mM is monomeric and that micelles are "most likely" dominant at 12 mM. C14

Two further observations sharpen the monomer–micelle contrast. At 18 mM STC or SGC, all $\delta$ 11.5 solutions crystallized within 16 h, whereas none did at 12 mM; no induction times are given. R7 And for the trihydroxy STC, the effect builds up with concentration below the CMC (Figure 7): about 29 min in buffer, about 127 ± 21 min at 0.19 mM, about 225 ± 47 min at 0.47 mM, about 360 ± 42 min at 0.70 mM, and no crystallization within 720 min at 0.93 and 1.86 mM. R6 The text describes 0.47 mM STC (less than 0.025% w/w) as moving the induction time from within 30 min to "approximately 3 hours"; the bar in Figure 7 is nearer 3.75 h. R6.c·X4 The authors relate this delay to a small-intestinal transit time of about 3–4 h (ref 29) and call it a sign of "the potential of STC as a crystallization inhibitor for amorphous formulations". R13·C16

Two-phase solutions: faster crystallization and a different ranking

At $\delta$ 28.8 the solutions contain glassy telaprevir-rich droplets. Without bile salt they crystallized immediately. R4.a With 1.86 mM bile salt the four testable bile salts spread widely: SGDC held for more than 960 min and STDC for about 550 min, whereas STC (about 60 min) and SGC (about 30 min) gave little protection (Figure 5, replotted in Figure 8). R4.b–e In the presence of the drug-rich phase, the monomeric dihydroxy bile salts were therefore the better inhibitors, SGDC the best. C18 At 12 mM the pattern flipped for each class: STC (about 390 min) and SGC (about 175 min) did better than at 1.86 mM, while STDC (about 205 min) and SGDC (about 650 min) did worse. R4·C12

Figure 9 sets the 12 mM data at the two supersaturations side by side. STC and SGC held the homogeneous solutions for more than 960 min but lost much of their effect with droplets present, as one would expect from a larger driving force. STDC and SGDC, in contrast, gave longer induction times at the higher supersaturation (about 140 to 205 min and about 125 to 650 min). R3·R4·C19 The authors call this "different from previous cases where crystallization induction times decreased as crystallization driving force increased" and say that "further mechanistic studies are warranted". C19·C20

Note by the AI writer (not discussed in the paper): at $\delta$ 28.8 the 12 mM STDC and SGDC solutions held about 300–330 µg/mL telaprevir, about twice the 150 µg/mL of the other media, because these micellar solutions solubilize part of the drug (Figure 2). Equal activity therefore does not mean equal amounts of drug-rich phase or of micelle-bound drug, which may bear on the unusual 12 mM dihydroxy results. R1.d–e·X18

Bile salts on the droplet surface

The zeta potential of telaprevir-rich droplets without bile salt was −9.2 mV. R8.a With 1.86 mM bile salt it was about −45 mV (STC), −40 mV (SGC), −47 mV (STDC), −48 mV (SGDC), −53 mV (STCDC) and −50.5 mV (SGCDC) (Figure 10, Table B). R8.b–g The text summarizes this as a "~4-5 fold increase in the magnitude" to "around -50mV"; the figure values correspond to about 4.4- to 5.8-fold. R8·X5 The authors' reading is that this "implies adsorption of bile salts at the telaprevir-rich droplet-water interface", that the small negative charge of the bare droplets means a high tendency to aggregate, and that at about −50 mV the droplets "can be considered to be electrostatically stabilized". C21·C38

Table B. Zeta potential of telaprevir-rich droplets at 180 µg/mL telaprevir, read from Figure 10 (approximate) except the printed −9.2 mV; the ratio column was computed by the AI writer from these values. R8·X5

Note by the AI writer (not discussed in the paper): the zeta samples were made at 180 µg/mL telaprevir, whereas the $\delta$ 28.8 induction-time solutions with 1.86 mM STC, SGC, STDC and SGDC held about 150 µg/mL, and STCDC and SGCDC, which have zeta values, were never tested in two-phase induction-time experiments. The zeta data describe the droplets in general, not the exact solutions whose induction times were measured. X6

Nucleation or growth?

In seeded solutions starting at about 70 µg/mL, telaprevir fell to about 13 µg/mL in 60 min in buffer, to about 22 µg/mL with SGC, and only to about 62 µg/mL with STC (Figure 11). R9 The initial 10-min slopes gave mass growth rates of 89.2 µg/min in buffer, 6.4 µg/min with STC and 60.5 µg/min with SGC, that is, growth slowed 14-fold and 1.5-fold. R9 With an unseeded induction time of about 30 min in buffer and 960 min ("more than 16 hours") with either bile salt, Eq. 4 gave $J/J_0=2.5\times {10}^{-3}$ for STC and $2.9\times {10}^{-6}$ for SGC: nucleation was slowed "by at least 400 fold and 340,000 fold". R10 The authors conclude: "It is evident that the impact of bile salts on telaprevir nucleation rate is greater than the impact on telaprevir crystal growth rate", so that taking nucleation as the dominant part of the induction time "may be applied in this study". C22 They add that the seeded data "reveal the potential of bile salts as crystal growth inhibitors", an area for future study. C23

Table C. Inputs and outputs of Eq. 4 as printed on PDF p.15, with a recomputation by the AI writer. R9·R10·E4·X1

Two points qualify this result. The recomputation from the printed inputs gives slightly different values (2.58 × 10⁻³ and 3.06 × 10⁻⁶, that is about 390-fold and 330,000-fold), so the printed fold changes are best read as orders of magnitude. X1 More importantly, the bile salt induction times were not measured, only bounded (no crystallization within 960 min), the seeded runs started near 70 µg/mL rather than at the 60 µg/mL of the unseeded runs, and only STC and SGC were tested, all at $\delta$ 11.5. X2

Bile salts against a polymer

HPMCAS-MF had been one of the most effective telaprevir inhibitors among many polymers in earlier work (ref 37). C46 In homogeneous solution, HPMCAS-MF at 5 µg/mL and 1 mg/mL and the 1.86 mM bile salts all held for more than 960 min (Figure 13). R11·R3 With the drug-rich phase present, 5 µg/mL HPMCAS-MF gave about 110 min and 1 mg/mL about 435 ± 60 min: better than STC and SGC, close to STDC (about 550 min) and below SGDC (more than 960 min). R11·R4 The authors summarize that HPMCAS-MF "was better than the least effective bile salt, but not as effective as the best bile salt", and that the bile salt monomers showed "comparable" inhibition to the polymer, as celecoxib studies also suggest (refs 10, 13, 50). C47 The paper does not state how the supersaturation was set in the polymer solutions; Figure 2 has no HPMCAS-MF curve. X13

Discussion

How the authors explain the pattern

The authors' argument starts from the calibration: since all runs had the same driving force, solubilization is excluded, and because bile salts solubilize less than conventional surfactants anyway, the delay of crystallization is "more likely to be a result of intermolecular interactions". C25·C26 The seeded experiments then let them read induction times as nucleation inhibition. C22 All monomers inhibit, and below the CMC more STC means a longer delay, so the monomer is, in their words, "the effective crystallization inhibitory species". C28·C32·R6 The difference at 12 mM is then explained through the CMC, as summarized in their schematic, Figure 12: C32·C34·C41

For the trihydroxy branch, the higher monomer concentration at 12 mM is described as "a possible contributing factor" to their "persistent, or even improved" inhibition. C32 For the dihydroxy branch, the authors write "we can speculate that bile salt micelles promote crystallization by providing interfaces that act as heterogeneous nucleation sites", so that the 12 mM dihydroxy solutions "may have both inhibitory and promoting species present". C34 SGCDC, the one dihydroxy bile salt that still held at 12 mM, is explained with a molecular-simulation result from the literature: amphiphilic additives that interact strongly with the solute can disturb the packing of emerging nuclei (ref 38), and SGCDC had earlier been found to interact more strongly with telaprevir (ref 24); this explanation "might be relevant". C33 The 18 mM STC and SGC tests, which crystallized within 16 h, are offered as support for "our contention that micelles oppose the inhibitory effect of monomers". C35·R7 The summary statement is that micelles of dihydroxy bile salts other than SGCDC "appear to negate the inhibitory impact of the monomers". C41

For the two-phase solutions, the authors expect the drug-rich phase to accelerate crystallization because it adds interfaces on which heterogeneous nucleation can occur (ref 25). C36 That bile salts still delay crystallization here is taken as "suggesting that bile salts interact with the drug-rich phase", supported by the zeta potentials; bile salt at the particle–water interface "can be expected to modify heterogeneous nucleation". C37·C39 With both micelles and droplets present, the pattern is "complex, unpredictable", and the authors judge these to be "probably the most relevant systems", since micelles are present in vivo and many formulations form a drug-rich phase on dissolution. C42

What the paper means for testing and formulation

The authors' practical conclusions concern dissolution testing. Bile salts are "not interchangeable" for crystallization inhibition, as earlier found for solution thermodynamics (ref 24) and crystallization (ref 13), and STC, the only bile salt in common simulated intestinal fluids (ref 34), "may not be a good surrogate for the mixture of bile salts found in vivo". C43·C44·R14 They suggest this "possibly contributes" to some of the in vivo–in vitro discrepancies reported in the literature (ref 35), and conclude that further studies of bile salt effects are "essential" for biopredictive dissolution media. C24·C50 On the formulation side, they note that bile salt monomers inhibit about as well as HPMCAS-MF and that low STC levels delay crystallization on the time scale of intestinal transit. C47·C16

Note by the AI writer (not discussed in the paper): the data carry a practical warning that the authors only imply. In homogeneous solution all six bile salts looked equally protective at 1.86 mM; the differences appeared once drug-rich droplets were present, which is the state an ASD typically creates. There, at 1.86 mM, STC was among the weakest bile salts (about 60 min) and SGDC the strongest (more than 960 min). R3·R4 A crystallization test in an STC-only medium, run above the amorphous solubility, could therefore misjudge the risk for a formulation in either direction, at least for telaprevir. A second point is methodological: the 12 mM dihydroxy media needed about twice as much telaprevir for the same supersaturation, so a comparison at equal concentration would have mixed solubilization and inhibition, which is exactly what the activity calibration avoids. R1

Limitations

Limitations stated by the authors. The authors state that the zeta-potential measurements "can only provide qualitative evidence" of interfacial telaprevir–bile salt interactions and are "not a quantitative measure" of adsorption. C40 They write that "further mechanistic studies are warranted" for the improved 12 mM dihydroxy inhibition at higher supersaturation, that additional studies are needed to relate inhibition among bile salts in two-phase solutions, that the micelle-plus-droplet case is "complex, unpredictable", and that more detailed experimental and molecular-modelling studies are required to understand how bile salt–drug interactions produce the differences. C20·C42·C48 They present growth inhibition by bile salts as preliminary. C23 The assignment of monomer and micelle states rests explicitly on the assumption that CMCs vary little within each group and that telaprevir does not change them. C14

The AI writer's own judgement (1): what the induction-time data can and cannot show. The core result, more than 960 min against about 30 min for every bile salt at 1.86 mM and $\delta$ 11.5, is large, triplicate and obtained at a matched driving force, and it is the strongest part of the paper. R3·R1 But most values are bar-chart readings and the best ones are censored at 960 min (720 min in Figure 7), so the most effective bile salts cannot be ranked against each other. X16 The monomer and micelle labels rest on CMCs measured for STC and SGDC only; 1.86 mM is close to the SGDC CMC of about 2 mM, and the cited CMC range of 2 to 12 mM means 12 mM may be near the CMC of some bile salts. X7 The 18 mM evidence for micelles opposing monomers is given in one sentence without times, and Figure 2 contains no 18 mM calibration. X8 Even the dihydroxy bile salts at 12 mM extended the induction time four- to six-fold over buffer, so "negate" overstates what the data show. X17

The AI writer's own judgement (2): the mechanism and the summary statements. The nucleation-dominance argument rests on two bile salts at one supersaturation, with induction times that are lower bounds, growth rates from an undescribed seeded method at a slightly different concentration, and a small recomputation mismatch. X1·X2·X3 It supports the reading for STC and SGC at 1.86 mM; for the other conditions it is an assumption. Several summary sentences go beyond the data. The abstract's "trihydroxy bile salts showed better inhibition compared to dihydroxy bile salts" at higher concentrations holds at $\delta$ 11.5 except for SGCDC, but not at $\delta$ 28.8, where 12 mM SGDC (about 650 min) beat STC (about 390 min) and SGC (about 175 min). X9 The graphical-abstract statement that monomers are "more effective" than the corresponding micelles fails for STC and SGC in the two-phase solutions. X10 The abstract's statement that all six bile salts inhibited in solutions containing a second phase covers only four, since STCDC and SGCDC could not be tested. X11 Smaller points: the text gives "approximately 3 hours" where Figure 7 shows about 3.75 h, the zeta fold change is wider than "~4-5", and the caption of Figure 13 misprints HPMCAS-MF as "HMPCAS-MF". X4·X5·X14

Conclusion

The paper's main achievement is methodological as much as factual: by holding the activity-based supersaturation constant, it separates the inhibitory effect of bile salts from their solubilizing effect, and then shows that this effect depends strongly on aggregation state and on the presence of a drug-rich phase. C26·C27 For telaprevir at pH 6.5 and 37 °C, all six major human bile salts are effective nucleation inhibitors as monomers in homogeneous solution; the micelles of STDC, SGDC and STCDC sharply reduce that effect; and once glassy drug-rich droplets are present, crystallization is faster, the bile salts adsorb on the droplets and the ranking among bile salts changes. C28·C41·C12·C21 The explanations offered, monomer share set by the CMC, micelles as nucleation sites and modified nucleation at the droplet surface, are hedged by the authors and not tested directly. C32·C34·C39 For formulation and testing, the message the authors draw is that bile salts are not interchangeable and that a medium with STC alone may not represent intestinal bile when crystallization kinetics matter. C43·C44