Surfactants inside a ritonavir–copovidone amorphous solid dispersion: what they do to release above the limit of congruency

Background and motivation

An amorphous solid dispersion (ASD) disperses a poorly water-soluble drug in a polymer so that the drug dissolves from a high-energy amorphous state. The higher thermodynamic activity of the amorphous drug gives a supersaturated solution and a higher diffusive flux across a membrane, which can raise oral bioavailability compared with formulations that do not supersaturate (refs 4–10). C1 The polymer has at least two jobs: it inhibits crystallisation, in the solid by lowering mobility and drug activity and in solution by adsorbing to crystal surfaces, and it speeds up release (refs 11–20). C2

Release from an ASD depends strongly on the drug loading (DL). At low loading the polymer dissolves quickly and carries the drug with it, so the solution concentration can exceed the drug's amorphous solubility. The excess drug then separates by liquid–liquid phase separation (LLPS) into drug-rich nanodroplets, which redissolve quickly, act as a reservoir that keeps membrane flux at its maximum, and have been linked to higher oral bioavailability in vivo (Ilevbare 2013; Indulkar 2016; Wilson 2018; Stewart 2017; refs 26–29). C3·R26 Purohit and Taylor (2017, ref 23) showed for ritonavir ASDs that dispersions which underwent water-induced drug–polymer demixing, called amorphous–amorphous phase separation (AAPS), released slowly and formed no nanodroplets, whereas dispersions that stayed miscible dissolved quickly and formed nanodroplets. R27

The present paper builds directly on an earlier study of ritonavir (RTV) and copovidone (PVPVA) by the same authors (Indulkar 2019, ref 30). There, drug and polymer were released together ("congruently") about 300-fold faster than amorphous drug alone, with nanodroplets, up to about 25% DL; this threshold is the limit of congruency (LoC). R2 Above 25% DL the ASDs underwent AAPS, and at 30% DL and above drug release became incongruent and very slow. R2 The 30:70 binary ASD shows the problem in Figure 2 (reprinted from ref 30): almost nothing dissolves for about 15 min, drug and polymer then dissolve together for about 25 min, after which PVPVA runs ahead and dissolves completely while ritonavir release stops near 50% (about 48.5% at 120 min, read from the figure). R1 The authors "posited" that fast release with nanodroplets requires the drug to stay well dispersed in the polymer throughout hydration and dissolution, so that the polymer controls the rate and carries the drug into solution. C4

Low LoCs are common with PVPVA: 5% for ledipasvir, 10% for nilvadipine, 15% for cilnidipine and 20% for itraconazole (refs 31–33). R22 Since release declines sharply above the LoC, an ASD should be formulated below it; for a drug with a low LoC, that means a large tablet or a high pill burden, so the authors call strategies to raise the LoC "essential" for all but the most potent compounds. C5 Surfactants are an obvious candidate. They are already used in marketed ASDs, for example Span 20 in Kaletra, Tween 80 in Venclexta and SDS in Incivek (refs 46–48), but how they act on the phase behaviour and dissolution of ASDs was, in the authors' words, "poorly understood". R23·C29

The study therefore takes the 30% DL RTV–PVPVA ASD, which releases incongruently and incompletely, as a model, adds a hydrophilic non-surfactant additive (PEG 3350) or one of five surfactants spanning a wide range of hydrophilic–lipophilic balance (HLB), and asks how each changes release, the size and solid state of the precipitate formed in solution, and the surface of the partially dissolved tablet. C5·C30·A1 The 30% loading was chosen because the binary ASD at that loading shows all three phases of interest: a lag, a congruent period and then incongruent, incomplete release; it undergoes AAPS, but the phase-separated ritonavir does not cover the whole surface, so PVPVA still dissolves completely. C30

Background concepts

Limit of congruency and congruent release. Release is congruent when drug and polymer leave the ASD at the same fractional rate, so that their % release curves overlap. The LoC is the highest drug loading at which this happens; for RTV–PVPVA it is about 25% DL. R2 Above it, release is incongruent: the polymer dissolves but a fraction of the drug stays behind.

Polymer-controlled and drug-controlled release. The authors describe two limiting regimes with two equations (Discussion). At low loading, with non-interacting components, the polymer dissolves fast and brings the drug into solution at the same relative rate (refs 21, 22, 30): C20·E1

In plain terms: the drug's dissolution rate ${(dC/dt)}_{\mathrm{d}\mathrm{r}\mathrm{u}\mathrm{g}}$ equals the polymer's dissolution rate ${(dC/dt)}_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}$ multiplied by the mass ratio of drug ($m_{\mathrm{d}\mathrm{r}\mathrm{u}\mathrm{g}}$) to polymer ($m_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}$) in the ASD (Eq. 1). Because the polymer dissolves much faster than the drug alone, the drug concentration can rise above the amorphous solubility, which leads to LLPS and drug-rich nanodroplets. E1·C20 At high loading the drug controls dissolution, and the rate can be approximated by the Nernst–Brunner equation (Eq. 2): E2·C21

In plain terms: the drug dissolves at a rate set by its diffusion coefficient $D$, the dissolving area $A$, the medium volume $V$ and the diffusion-layer thickness $h$, driven by the difference between its amorphous solubility $C_s$ and the bulk concentration $C$ at time $t$; once $C$ reaches $C_s$, release stops. E2 In ref 30, a 50% DL ASD placed in a medium already saturated at the amorphous solubility released nothing, its surface was enriched in drug by AAPS, less than 10% of the components dissolved, no LLPS occurred, and the profile followed Eq. 2; a 10% DL ASD in the same medium dissolved completely and formed droplets. R3 Neither equation is fitted to data in the present paper; they define the two regimes between which the 30% DL ASD sits. E1·E2·C22

Amorphous–amorphous phase separation (AAPS). When water enters an ASD of a hydrophobic drug in a hydrophilic polymer, the polymer prefers to interact with water, the drug–polymer contacts are disrupted and the matrix separates into a drug-rich and a polymer-rich amorphous phase. The authors call this "inevitable" for hydrophobic drugs in PVPVA, especially at high loading (refs 35–38). C23 Large phase-separated drug domains dissolve slowly and may also crystallise (refs 23, 30, 31, 39). C23

Release as a race. Dissolution, AAPS and crystallisation are kinetic processes that run at the same time, so their effect depends on their relative rates. If the drug is released before extensive AAPS or crystallisation can begin, these transformations do not matter; if AAPS is faster than release, incomplete release is "likely", especially when it forms contiguous drug-rich domains at the dissolving surface. C24 Hence, in the authors' formulation, a good ASD should either release faster than AAPS and crystallisation or resist them. C25 This framing is the key to reading the results.

Droplet growth: coalescence and Ostwald ripening. Nanodroplets formed by LLPS feed absorption by keeping the dissolved drug at its amorphous solubility until they are used up, but their small size makes them prone to growth. They can grow by coalescence (droplets colliding and merging; for ritonavir droplets, Sun et al. reported singlets forming doublets and then larger droplets, ref 43) or by Ostwald ripening (in a polydisperse system, particles below about 200 nm are more soluble because of their curvature, the Kelvin effect, so they dissolve and the larger particles grow, with particle volume rising linearly in time; refs 40–42). C26·C27 Size matters in vivo: for anacetrapib in dogs, formulations that formed larger particles gave a lower AUC and Cmax (Kesisoglou 2019, ref 44). R24 The authors therefore ask that an optimised ASD not only release completely but also keep the colloids small after LLPS. C28

Surfactants and crystallisation. Surfactants can promote crystallisation: they lower the interfacial energy of a new crystal surface, which lowers the energy cost of nucleation (refs 52–54); they can change the rate of solute attachment and molecular ordering or act as templates (refs 55–57); and, as plasticisers, they can lower the Tg of the ASD and raise drug mobility (refs 58, 59). C40 The surfactants studied are summarised in Table 1: SDS (HLB 40, MW 288.37, anionic), Tween 80 (HLB 15, MW 1310), TPGS (HLB 13.2, MW about 1513), Span 20 (HLB 8.6, MW 346.46) and Span 85 (HLB 1.8, MW 957.52); the HLB values are cited from a handbook (ref 34). M4·M5·M6·M7·M8

Methods in detail

The study is a single experimental comparison: one drug, one polymer, one drug:polymer ratio (30:70), six additives at one level (5 parts), and a set of read-outs chosen to separate four possible effects of a surfactant, namely on release rate, AAPS, crystallisation and droplet size. A1·C51 The flow below summarises it.

Making the dispersions

Ritonavir (AbbVie), PVPVA (BASF) and one additive were dissolved together at 30:70:5 by weight in dichloromethane:methanol 1:1 at 20% w/v solids and dried on a rotary evaporator with a 50 °C water bath, then dried again overnight in a vacuum oven at room temperature, milled and sieved to 100–150 µm. P1·M1·M2 The additives were PEG 3350 (a hydrophilic, non-surface-active polymer) and the surfactants SDS, Tween 80, TPGS, Span 20 and Span 85. M3·M4·M5·M6·M7·M8 Because the additive is added on top of the 30:70 drug:polymer ratio, the ternary ASDs contain 30/105, or 28.6 wt%, ritonavir, although the paper refers to them as 30% DL. P1·X13 The binary comparators were not prepared here: the 30:70 release profile (Figure 2) and the early PVPVA release of a 25:75 ASD at the LoC (Figure 9) are taken from ref 30. P2

Release from a tablet of constant area

The problem with powder dissolution is that surface area changes as the particles dissolve. The authors used Wood's intrinsic dissolution apparatus: 100 mg of powder was compressed in an 8 mm die at 1600 lbs, the die was mounted on an overhead stirrer and turned at 100 rpm in 100 mL of 50 mM phosphate buffer, pH 6.8, at 37 °C, so that one face of constant area dissolved. P3·C6 One millilitre was withdrawn at each time point and replaced with fresh medium. The samples were deliberately not filtered or centrifuged: release came from an intact tablet by erosion rather than disintegration, and the authors wanted to count drug held in colloids. P3·C7 Ritonavir was assayed by reverse-phase HPLC (C18 column, 0.1% trifluoroacetic acid in water and acetonitrile 55:45, UV 210 nm, linear from 0.1 to 10 µg/mL) and PVPVA by size exclusion chromatography (Viscotek A2500 column, phosphate buffered saline pH 7.4 and methanol 70:30, UV 205 nm, linear from 5 to 200 µg/mL). P5·P6 Measuring both components is what lets the authors judge congruency: if the % release curves of drug and polymer overlap, release is congruent.

A control separated a surface-wetting effect from an effect of the surfactant inside the ASD: a binary 30% DL tablet was dissolved in medium that already contained 50 µg/mL SDS, the concentration that the SDS in one tablet would give on complete dissolution. P4·C8 (100 mg of a 30:70:5 ASD contains about 4.8 mg SDS, which in 100 mL gives about 48 µg/mL, so the stated equivalence holds; this check is the AI writer's.) P3·C8

Following the droplets and the precipitate

Droplet size in the dissolution medium was measured by dynamic light scattering (Malvern Zetasizer Nano ZS, backscatter at 173°), starting from the time point at which the ritonavir concentration passed the amorphous solubility, at times chosen from the release profiles. P7 Precipitates from the Tween 80 and Span 20 ASDs, the two that released poorly, were examined under crossed polarisers at 20× magnification. P8 The number of DLS runs is not stated, and Figure 4 shows single values without error bars. X12

Looking at the tablet surface

Three techniques probed what happens on the dissolving surface. ATR-IR spectra (64 scans, 4 cm⁻¹ resolution) were taken on tablets ejected at chosen times and blotted dry, and compared with crystalline ritonavir. P9 SEM images were taken of ejected, vacuum-dried and platinum-coated tablets. P11 DSC (5 °C/min from −60 to 150 °C) measured the Tg of all dry ASDs and, for the two poorly releasing surfactant ASDs (Tween 80, Span 20), of material scraped off the exposed surface after dissolution; these samples were run in open pans after a 5 h nitrogen purge to remove water. P10 The logic is that a single Tg near that of the mixture means a homogeneous ASD, whereas a Tg near that of pure ritonavir after dissolution points to a drug-rich phase left behind by AAPS. C18·C19

Experimental analysis

Release: every surfactant removes the lag, but the end points differ widely

Figure 3 shows ritonavir and PVPVA release from the six 30:70:5 ASDs, to be read against the binary ASD of Figure 2. Table A collects the values at 20 and 90 min, read from the figures. R1·R4·R5·R6·R7·R8·R9

Table A. Release from 100 mg tablets in 100 mL pH 6.8 buffer at 37 °C, read from Figs. 2 and 3 (approximate); the binary ASD is from ref 30. R1·R4·R5·R6·R7·R8·R9

Three groups emerge. Span 85 and TPGS released drug and polymer together throughout, to about 87–89% ritonavir and 94–95% PVPVA at 90 min, which the authors call "complete and congruent release". R8·R9·C58 Span 85 reached its plateau at about 60 min, TPGS more slowly and was still rising at the last time point. R8·R9·C10·C11 SDS released fastest: about 60% of the ritonavir at 10 min and 75% at 15 min, with both components then levelling off at about 80%. R5 The text describes this as near complete release "in 20 min" in the Results and as about 80% "in ~15 min" in the Discussion, after which "the dissolution completely ceased". R5·C12·C37·X15 Tween 80 and Span 20 started at once but levelled off after 30–40 min, below 50%: Span 20 at about 41% ritonavir with wide error bars, Tween 80 at about 25% ritonavir and 41% PVPVA. R6·R7·C43 The text gives the Tween 80 plateaus as "30 to 40%" for drug and polymer, slightly above the plotted ritonavir values of about 25–28%. R6

Two observations hold across all five surfactants: none showed the lag phase of the binary ASD, and all gave congruent release for at least the first 20 min. C31 PEG 3350, a hydrophilic polymer without surface activity, shortened the lag only slightly and left the profile close to the binary ASD (about 46% ritonavir at 90 min), which the authors read as "suggesting" that hydrophilicity alone does not improve release. R4·C9 SDS dissolved in the medium at the concentration a tablet would give did not change the release of the binary ASD (data not shown), so the authors consider poor wetting "unlikely" to explain the surfactant effects. R10·C13

Note by the AI writer (not discussed in the paper): the abstract states that all surfactant-containing ASDs improved on the binary ASD, but in the Discussion the authors write that the Tween 80 and Span 20 ASDs "do not fare any better than the binary ASD at 30% drug loading". Table A supports the second statement for the amount released: at 90 min the Tween 80 ASD had released about 25% ritonavir, less than the binary ASD at the same time (about 45%), and Span 20 about 41%. The "improvement" for these two is the loss of the lag phase. C61·C45·X3 Likewise, "complete" in the paper means about 87–95% at 90 min. X14

Early polymer release: only SDS is faster than the binary ASD

To separate a faster dissolution rate from other effects, Figure 9 compares PVPVA release in the first 20 min with that of a binary 25:75 ASD at the LoC, which releases congruently (from ref 30). R20·C32 Table B gives the values read from the figure.

Table B. PVPVA released in the first 20 min, read from Fig. 9 (approximate), in the axis unit printed in the figure ("mg/mL"; see the note below). R20

The SDS ASD released PVPVA "approximately twice" as fast as the binary ASD at the LoC; Table B shows ratios of about 2.5 at 10 min and 1.9 at 15 min, falling to about 1.2 at 20 min when the SDS ASD had levelled off. C32·R20·X6 The Span and TPGS ASDs followed the binary curve closely, and Tween 80 was slower than all others, about 6.6 against 14.0 at 10 min. R20·C34 The authors read this as two different routes to better release: SDS "perhaps" works by making dissolution faster than AAPS, while Span 85 and TPGS give complete release without a faster rate and so "can potentially" act by changing the kinetics of AAPS or the composition and morphology of the separated regions. C33·C35·C36

Note by the AI writer (not discussed in the paper): the axis of Figure 9 reads "Concentration (mg/mL)", but a 100 mg tablet of the 30:70:5 ASD contains only 66.7 mg PVPVA, at most about 0.67 mg/mL in 100 mL. The plotted values match milligrams of PVPVA released per vessel: 35.4 at 10 min for SDS is 53% of 66.7 mg, the same as the PVPVA release read from Figure 3B at 10 min (about 53%), and the Span 85 and Tween 80 values at 20 min (about 26.8 and 12.1) correspond to 40% and 18%, again as in Figure 3. The comparison with the 25:75 binary ASD is also not normalised to its larger polymer content (75 mg per tablet). The direction of the effect is not in doubt, but "twice as fast" is a ratio of unnormalised amounts at 8–15 min. X5·X6

SDS: ritonavir crystallises on the dissolving surface

Why does the fastest-releasing ASD stop at about 80%? Figure 6 shows the surface of the SDS tablet by SEM: smooth and homogeneous after 2 min of dissolution, covered with needle-shaped crystals after 15 min. R15 Figure 7 shows the matching ATR-IR spectra: the 2 min spectrum is similar to that of the dry ASD, while the 15 min spectrum differs and carries peaks of crystalline ritonavir. R16·C17 The text lists these peaks at about 1280, 1610 and 1700 cm⁻¹; the arrows in Figure 7 sit at about 1705, 1610 and 1535 cm⁻¹, so one of the three listed positions does not match the figure. R16·X7 SEM and IR of the tablets with the other surfactants showed no surface crystals (data not shown). R17

The authors conclude that ritonavir crystallisation on the undissolved matrix "likely served as a barrier to further drug release, halting dissolution", and call SDS "a double-edged sword": its fast dissolution ensures nearly complete release, but it promotes ritonavir crystallisation before the tablet has dissolved, which leads to incomplete release under the non-sink conditions used. C38·C39 They add that, from these data, it is "difficult to ascertain" whether AAPS preceded the crystallisation, and that surfactants "can" make a crystallisation-inhibiting polymer partly or fully ineffective even in a homogeneous ASD, particularly when surfactant and polymer interact strongly. C41·C42 The two SEM panels are at about 9,700× and 40,000× magnification, so their textures should be compared qualitatively. X17

Tween 80 and Span 20: a lower Tg after dissolution

Before dissolution, every ASD showed a single glass transition (Figure 8A), which the authors read as "suggesting miscibility". R18·C18 Table C lists the printed values; four of the five surfactants lowered the Tg relative to the binary ASD, while SDS raised the midpoint slightly. R18

Table C. Glass transition temperatures printed in Fig. 8 (DSC, 5 °C/min). The dry onsets of the Span 20 and Tween 80 ASDs differ slightly between Fig. 8A and the "before dissolution" traces of Figs. 8B–C. R18·R19·X16

After dissolution, the surface material of the Span 20 ASD showed a Tg onset of 42.18 °C against 67.85 °C before, and that of the Tween 80 ASD 40.25 °C against 66.01 °C. R19 The authors note that this value is "very close" to the reported Tg of ritonavir (ref 35; the value is not printed) and conclude that "phase separation of ritonavir has likely occurred in these ASDs upon contact with water, with the subsequent loss of PVPVA and enrichment of RTV on the surface". C19 The time and concentration at which these two ASDs stopped releasing "correlates well" with the binary ASD. C43 Their reading is that Tween 80 and Span 20 "may" improve initial wetting and dissolution, which removes the lag, but that they do not raise the dissolution rate above that of the polymer, so dissolution "may" be slower than AAPS and the build-up of a drug-rich surface phase; release from these ASDs was compromised by AAPS. C44·C45 Span 20 also produced needle-shaped crystals in solution (Figure 5B), which the authors read as solution crystallisation of ritonavir induced by Span 20. R14·C16

Note by the AI writer (not discussed in the paper): the AAPS assignment rests on one lowered Tg per sample. No second, polymer-rich Tg is reported; the after-dissolution traces show large unassigned thermal events above about 100 °C; and the water content of the scraped material is not given, although residual water in a still-mixed matrix would also lower the Tg. The suggestion that these surfactants act through wetting also sits alongside the SDS-in-medium control, which the authors used to argue that wetting is unlikely to explain the release behaviour. X10·X8·C13

Droplet size: the Spans hold it near 1 µm

Figure 4 follows the size of the colloids formed after LLPS for the binary ASD and four surfactants (the Tween 80 droplets were too large for DLS). Table D gives the readings over time. R11·R13

Table D. Droplet size (nm) by DLS during release, read from Fig. 4A (approximate; single values, — = no point). Arrows in the figure mark the release plateau. Tween 80: spherical droplets of about 10 µm by microscopy (Fig. 5A). R11·R13

In general the droplets grew as release went on and the total ritonavir concentration rose. C14 Without surfactant they reached about 2000 nm at the release plateau (near 140 µg/mL ritonavir; about 2200 nm at 90 min in the figure). R11 With SDS they were about 1100 nm at the maximum concentration of about 240 µg/mL and grew to about 2400–2500 nm; with TPGS they were large from the start and reached about 4000 nm by the end (3–4 µm in the Discussion); with Span 20 they grew from about 1000 to 1500 nm after a maximum of about 160 µg/mL, and with Span 85 they stayed near 1000 nm. R11·R12·C15 Plotted against concentration (Figure 4B), the Span droplets stayed at about 320–940 nm up to more than 160 µg/mL, the TPGS droplets at about 2080–3340 nm. R12 The Tween 80 ASD gave spherical droplets of about 10 µm at a final ritonavir concentration of about 90 µg/mL. R13

The authors group the surfactants into those that let the droplets grow beyond 2000 nm and those that keep them below; the Span-type (hydrophobic) surfactants maintained droplet size, whereas TPGS and Tween 80 increased it. C47·C53 Table 2 of the paper condenses all results: SDS, no AAPS, crystals on the solid, near complete release (~80%), 2 µm droplets with coalescence; Tween 80, AAPS, no crystals, <50% release, >10 µm with coalescence; Span 20, AAPS, crystals in solution, <50% release, ~1 µm stable; TPGS, no AAPS or crystals, complete release, ~4 µm with coalescence; Span 85, no AAPS or crystals, complete release, ~1 µm stable. R21

Note by the AI writer (not discussed in the paper): the abstract names "Span 20 and TPGS" as forming the smallest and most size-stable droplets; Figure 4, the text and Table 2 all give this role to Span 20 and Span 85, so the abstract is most likely a misprint. X1·C63 For Span 20, the maximum ritonavir concentration given in the text (about 160 µg/mL) and in Figure 4B is higher than the roughly 41–43% release of Figure 3D would give for 28.6 mg ritonavir in 100 mL (about 120 µg/mL), so concentrations and percentages from different figures should not be converted into each other for this system. X9

Discussion

How the authors connect the results

The authors' explanation follows from the "race" framing: at 30% DL water drives AAPS, and dissolution, AAPS and crystallisation compete; a surfactant changes the outcome through one or more of four attributes, namely release rate, AAPS, crystallisation and droplet size. C24·C51 The chain below lays out the cases they describe; the arrows are their proposed causes, which were not measured directly.

For droplet size, the authors offer two mechanisms. Tween 80 and TPGS "could possibly" partition into the drug-rich droplets and lower their Tg, which "may" make the droplets more prone to coalescence than droplets of neat drug. C48 The two Spans, being poorly water soluble, "probably" mix into the droplet phase and lower the chemical potential of ritonavir there; as small droplets dissolve, their surfactant fraction rises relative to larger droplets until the chemical potentials balance across sizes, which stops Ostwald ripening, as reported for amorphous felodipine particles containing the poorly soluble additive Miglyol (ref 61). C49·R25 "Alternatively", the Spans "may" prevent coalescence by steric repulsion. C50 None of these mechanisms was tested: neither the surfactant content nor the Tg of the droplets was measured. C48·C49

The authors' overall statement is that "the impact on the release profile was found to be a balance between dissolution rate and the rate/extent of phase transformation (AAPS/crystallization)", and that no correlation between the physicochemical properties of the surfactants and their performance was observed. C59·C64 The HLB values bear this out: SDS (HLB 40) and Span 85 (HLB 1.8), at opposite ends of the scale, both released fast or fully, while Tween 80 (HLB 15) and Span 20 (HLB 8.6), in the middle, both stalled. M4·M5·M7·M8·R21 In the Conclusions they generalise further: ionic surfactants such as SDS "may" act by interacting with the polymer and raising the release rate, whereas nonionic surfactants "may" modify drug–polymer phase separation. C56 Note by the AI writer (not discussed in the paper): only one ionic surfactant was tested, and two of the four nonionic surfactants (Tween 80, Span 20) did not prevent AAPS, so the second half of that statement holds for Span 85 and TPGS only. X11

The best surfactant, and what it means for formulation

Ranking the surfactants on all four attributes, the authors conclude that Span 85 "fulfils the criteria of an ideal surfactant (resists AAPS and crystallization, and stabilizes the nanodroplets)": it gave complete release, no crystallisation and droplets near 1 µm. C54 TPGS also released completely but produced droplets larger than the binary ASD, and Tween 80 the largest droplets of all. C53 SDS gave the best initial dissolution rate, but its release was compromised later by crystallisation; Tween 80 and Span 20 were compromised by AAPS. C52·C38·C45

The formulation messages the authors state are general: an ASD should release faster than AAPS and crystallisation, or resist them; an optimised ASD should both release completely and keep its droplets small; and drug release is "extremely sensitive to the addition of a small amount of a surface-active excipient", so surfactants "need to be carefully considered and thoroughly examined". C25·C28·C60 They also note that, although not observed here, surfactants "may" induce crystallisation of the solid ASD during manufacture or storage, especially at high drug loading (refs 16, 60). C46

Note by the AI writer (not discussed in the paper): the practical reading for a formulator is that a surfactant cannot be selected on HLB or charge, and that removing the lag phase is not evidence of a rescue; the Tween 80 and Span 20 ASDs both removed the lag and still released no more ritonavir than the binary ASD. A candidate surfactant has to be checked separately for the extent of release, for crystals on the tablet and in solution, and for droplet size. The paper also does not show whether any surfactant raises the LoC itself, since only one loading and one surfactant level were tested. A1·R6·R7·X3

Limitations

Limits stated by the authors. For the SDS ASD, "it is difficult to ascertain if AAPS in the ASD preceded crystallization or not". C41 The mechanisms proposed for the surfactant effects are hedged throughout: SDS "perhaps" outruns AAPS; Span 85 and TPGS "can potentially" modify AAPS; Tween 80 and Span 20 "may" improve wetting; the Span effect on droplets "may be due to" inhibition of Ostwald ripening, with steric repulsion as an alternative. C33·C35·C44·C49·C50 Solid-state crystallisation induced by surfactants during manufacture or storage was not studied. C46

The AI writer's own judgement (1): what was and was not measured. The release profiles of drug and polymer, the surface crystals with SDS (seen by both SEM and IR at 15 min) and the droplet sizes are direct data, and the differences between surfactants are large. R4·R5·R6·R7·R8·R9·R15·R16·R11 Phase separation, however, was tested only on the Tween 80 and Span 20 ASDs, and only through one lowered Tg onset on material scraped from the wet surface; Table 2 nevertheless lists "No" AAPS for SDS, TPGS and Span 85, for which no phase-separation measurement was made. X4·X10 The binary comparators are reprinted from ref 30, the ternary ASDs contain 28.6 wt% rather than 30 wt% ritonavir, the number of replicates is not stated, and Figures 4 and 9 carry no error bars. X13·X12

The AI writer's own judgement (2): inconsistencies in the text. The abstract names TPGS rather than Span 85 as giving small, stable droplets; the Conclusions list Tween 80 among the surfactants that promoted crystallisation, while Table 2 says "No" and the Tween 80 precipitate was droplets; the abstract claims improvement for every surfactant, while the Discussion says that the Tween 80 and Span 20 ASDs do not fare better than the binary ASD. X1·X2·X3 Figure 9 has an axis unit that does not fit its values, and the IR peak list in the text does not match the arrows in Figure 7. X5·X7 None of these changes the main findings, but readers should take the figures and Table 2, not the abstract, as the record of what was observed.

Conclusion

This paper tests surfactants as a way to rescue release from a ritonavir–copovidone ASD above its limit of congruency. At 30:70:5, all five surfactants removed the slow start of the binary ASD, and that is where their common effect ends. C31 Span 85 and TPGS gave near-complete, congruent release; SDS released polymer about twice as fast as a binary ASD at the LoC but let ritonavir crystallise on the tablet, which "likely" stopped release near 80%; Tween 80 and Span 20, judged from a lower Tg after dissolution, did not prevent AAPS and stalled below 50%. R8·R9·C32·C38·C19 The surfactant also set the droplet size: about 1 µm with the Spans, several micrometres with TPGS and about 10 µm with Tween 80. R11·R13 The authors read the outcome as a balance between dissolution rate and phase transformation, unrelated to HLB, and name Span 85 the best of the five. C59·C64·C54 The mechanisms behind these effects, from SDS–polymer complexes to surfactant partitioning into droplets, remain proposals to be tested.