How a few percent of surfactant lets a high-Tg drug release from copovidone at higher loading

Background and motivation

An amorphous solid dispersion (ASD) disperses a poorly soluble drug in amorphous form within a polymer. The amorphous form is typically 5 to 100 times more soluble than the stable crystal, and some ASDs raise the apparent drug concentration far beyond that, up to 10 000-fold, and the dissolution rate up to 1 700-fold (Section 1). More than 24 ASD products have been approved. The authors note that this performance depends on how the ASD dissolves, not only on the amorphous solubility.

For copovidone (poly(1-vinylpyrrolidone-co-vinyl acetate), PVPVA, VA64), earlier work from the Taylor group established a characteristic pattern. At low drug loading (DL) the drug and the polymer leave the tablet surface together, at about the rate of the neat polymer; this is congruent release, and the drug usually appears in solution as drug-rich nanodroplets, which may help absorption. Above a critical loading the drug release "falls off a cliff". The highest loading that still gives congruent release is the limit of congruency (LoC), named for the point where the normalized drug and polymer release rates diverge (Section 4.1). The LoC depends on the drug and is often low, which means large tablets, a high pill burden, or, if the loading is pushed, no nanodroplets and poorer absorption (Section 1).

Two earlier observations frame this paper. First, the drug's glass transition temperature (Tg) seems to matter: lowering the dissolution temperature below the wet Tg of an indomethacin–PVPVA ASD halved its LoC (Saboo et al., 2020), and ASDs of high-Tg drugs (danazol, paclitaxel, ivacaftor, elbasvir, ledipasvir) released poorly unless the loading was very low (Yang et al., 2021; Que et al., 2019). Ledipasvir, with a Tg of 161 °C, released poorly above 5% DL in PVPVA. Second, surfactants can help: adding SDS raised the ledipasvir–PVPVA LoC from 5 to 30% DL (Que et al., 2019), and surfactants have long been used in ASDs as wetting agents and as plasticizers for hot melt extrusion. Yet why surfactants improve release was not understood, and their selection remained "mostly empirically driven" (Section 1).

The paper therefore asks how surfactants improve release from ASDs at higher drug loadings, using a drug whose high Tg gives it a low LoC. The model drug is atazanavir (ATZ), a weakly basic HIV protease inhibitor (pKa 4.5) with a dry Tg of 105 °C and a wet (water-saturated) Tg of 55 °C, a crystalline solubility of 1.1 µg/ml and an amorphous solubility of about 65 µg/ml at pH 6.8, cLogP 4.5, BCS class II (values cited from Indulkar et al., 2015 and 2017). The additives span anionic, cationic and nonionic surfactants and one non-surfactant plasticizer, and the ASDs were characterized by dissolution, DSC, water sorption, confocal fluorescence microscopy and 1H NMR.

Background concepts

Surface-normalized release and the LoC. Release is measured from a compact of fixed surface area that rotates in the medium (a rotating-die, intrinsic-dissolution style test). Both components are assayed: ATZ by reversed-phase HPLC and PVPVA by size-exclusion HPLC. The cumulative milligrams released are plotted against time, and the slope $k$ is normalized by the exposed surface and by how much of that component the ASD contains (Eq. 1):

Here $R$ is the normalized release rate (mg/min·cm²), $k$ the slope of cumulative mg released against time, $S$ the tablet surface area (0.5 cm² for the 8 mm die) and $x$ the fraction of the component in the ASD. In plain terms: if drug and polymer leave the surface in the same proportion as they are in the solid, their normalized rates are equal; this is congruent release, and the loading at which the two rates part is the LoC. The paper reports both normalized rates and simple percent released at 90 min. It does not state over which time window $k$ was fitted.

Drug loading in a ternary ASD. The paper expresses loading as a drug:polymer weight ratio. "10% DL" means 10 g ATZ and 90 g PVPVA in a binary ASD, and 10 g ATZ, 90 g PVPVA and 5 g additive in a ternary one (Section 2.2.2). The additive therefore sits on top of the drug and polymer, at about 4.8 wt% of the solid, and a ternary tablet holds slightly less ATZ by mass (about 9.5 wt%) than the binary one.

Glass transition, plasticization and mobility. Below its Tg an amorphous solid is a glass with little molecular motion; above it, it becomes rubbery and mobile. Water is a plasticizer: it lowers the Tg, which is why ATZ's wet Tg (55 °C) lies 50 °C below its dry Tg. Any small molecule with a low Tg that mixes into the ASD does the same. The paper uses the empirical rule $T_g/T_m=2/3$ (Kanno, 1981) to estimate the Tg of SDS (46 °C) and VP (−82 °C) where no clear Tg could be measured. Because ATZ's Tg (105 °C) is close to that of PVPVA (108 °C), the dry Tg of the binary ASD hardly changes with loading: 108.9 °C at 10% DL and 109 °C at 20% DL (Table 1, Section 4.1).

How a glassy polymer dissolves. The discussion borrows from polymer science (Devotta et al., 1995; Miller-Chou and Koenig, 2003; Papanu et al., 1989). Solvent first diffuses into the glassy polymer, which swells, plasticizes and forms a gel-like layer. In that layer the chains become mobile and disentangle by reptation over a characteristic time, and the free chains then diffuse into the bulk solution. For a high-molecular-weight neutral polymer such as PVPVA under stirring, either solvent ingress or chain disentanglement is usually the slow step. Solvent enters glassy matrices more slowly than rubbery ones, and a matrix with a higher initial Tg needs more solvent before its chains can disentangle. Small added molecules, whether good solvents or non-solvents for the polymer, can speed dissolution; Cooper et al. (1986) found that smaller plasticizers work better and that a non-solvent additive gives a maximum in the dissolution rate, beyond which it slows dissolution again.

Water-induced phase separation seen by fluorescence. A hydrated ASD can split into a drug-rich and a polymer-rich amorphous phase (amorphous–amorphous phase separation). The paper detects it in thin spin-coated films exposed to 100% relative humidity (RH), using two probes: Nile red, a hydrophobic dye that stains the drug-rich phase (imaged at 561 nm), and Alexa 488, a hydrophilic dye expected in the polymer-rich phase (imaged at 488 nm). A uniform image means no separation within the microscope's resolution; heterogeneous domains mean the film has separated. A continuous drug-rich phase has been linked in earlier work to poor dissolution (Que et al., 2019; Purohit and Taylor, 2017). Here the authors use the speed of separation mainly as a sign of molecular mobility, since phases cannot separate in a matrix that cannot move (Barnes et al., 1999).

Nanodroplets, partitioning and amorphous solubility. When dissolved ATZ exceeds its amorphous solubility (about 65 µg/ml at pH 6.8), the excess forms drug-rich nanodroplets. A surfactant or polymer in the medium may stay in water, adsorb on the droplets or dissolve into them. If an additive mixes into the drug-rich phase, it lowers the drug's thermodynamic activity there and so lowers the measured amorphous solubility (Ueda and Taylor, 2021). The paper uses 1H NMR to measure how much ATZ and additive remain dissolved, and thus which additives have an affinity for the drug-rich phase.

HLB and CMC. The hydrophilic–lipophilic balance (HLB) ranks surfactants from lipophilic (low) to hydrophilic (high); the critical micelle concentration (CMC) is the concentration above which a surfactant forms micelles. Table 3 lists both for each additive alongside its molecular weight.

Methods in detail

The study has one main measurement, release, and four probes of mechanism. The flow below shows how they fit together.

Making the dispersions

ATZ, PVPVA and the additive were dissolved in methanol (ATZ below 15 mg/ml) or, for the Span ASDs, in dichloromethane because Spans dissolve poorly in methanol. The solutions were sonicated for 30 min, the solvent removed by rotary evaporation at 55 °C and 150 rpm, and the solids dried for 30 min in the evaporator and overnight under vacuum, then milled and sieved below 250 µm. PXRD and DSC confirmed that the products were amorphous (data not shown). The additive level, 5 g per 100 g of drug and polymer, was taken from earlier studies in which this amount was enough to change release (Harmon et al., 2016; Saboo et al., 2021; Que et al., 2019). The eight additives were sodium dodecyl sulfate (SDS, anionic), cetrimonium bromide (CTAB, cationic), D-α-tocopherol polyethylene glycol 1000 succinate (TPGS), polysorbate 80 (Tween 80), the sorbitan esters Span 20, Span 80 and Span 85 (all nonionic), and 1-vinyl-2-pyrrolidone (VP), chosen as a plasticizer without surfactant character.

Table A. The additives, ordered by molecular weight (Table 3 and Table 1 of the paper; physical state from Section 3.2). The paper gives no source for the CMC and HLB values or for the Tg of Span 80, Span 85 and Tween 80.Card check: Table 3 prints the same CMC (14) and HLB (15) for Span 80 as for Tween 80, while the text calls Span 80 "more lipophilic" and "hydrophobic"; the Span 80 HLB should not be used without another source (card note X8).

Release testing

For each test, 100 mg of ASD was compressed in an 8 mm die at 2 000–2 500 psi for 1 min. The die rotated at 100 rpm in 100 ml of 50 mM phosphate buffer, pH 6.8, held at 37 °C (or 55 °C for the temperature test). Samples were taken every 5 min to 20 min, every 10 min to 60 min, and at 75, 90 and 120 min, with the volume replaced. Each condition was run in triplicate. ATZ was assayed on a C18 column at 210 nm and PVPVA on a size-exclusion column at 205 nm. The binary ASDs at 5 and 10% DL are shown as full release curves (Figure 1). The ternary ASDs are reported as percent released at 90 min (Figures 2, 4, 6) and as normalized rates (Figures 3, 7). The release enhancement factor in Table 3 and Figure 12 compares ATZ released at 90 min from a ternary 10% DL ASD with that from the binary 10% DL ASD. The Figure 12 caption gives it in milligrams released, the Table 3 heading in percent released; written as an equation from the caption's words:

In plain terms: how many times more drug a 10% DL ASD with the additive releases in 90 min than the same ASD without it. Note by the AI writer (not discussed in the paper): because a ternary tablet holds 10/105 of its mass as ATZ against 10/100 for the binary, a factor computed in milligrams is about 5% smaller than one computed in percent released; the paper does not show the calculation.

The temperature test

If poor release is due to low molecular mobility, then running the same test closer to the Tg should help. The authors therefore dissolved the binary ASD at 10, 15 and 20% DL, the TPGS ASD at 10% DL and the SDS ASDs at 15 and 20% DL at 55 °C, the wet Tg of ATZ, and checked for crystals by polarized light microscopy (Figures 5–7). The binary 15 and 20% DL ASDs appear only at 55 °C in the figures.

Tg by DSC

Modulated DSC (2 °C/min heating, modulation 1 °C every 60 s, Tg from the midpoint of the second heating) gave the Tg of the 10% DL ASDs with and without additive. To see whether each surfactant mixes with the drug, 2:1 ATZ:surfactant blends were made by rotary evaporation and examined in the same way: one Tg means a miscible blend, two Tgs a phase-separated one.

Wet films

Thin films of drug, polymer, additive and the two probes were spin-coated from 30 µL of methanolic solution (300 rpm for 12 s, then 3 000 rpm for 50 s), dried overnight under vacuum, and imaged on a Nikon A1R confocal microscope before and after exposure to 100% RH at 37 °C, for up to 3 h (binary) or 10 min (ternaries). The binary film was also exposed at 50 °C for 10 min (Figure 11).

Water sorption

Dynamic vapor sorption at 37 °C (dried 2 h at 0% RH, then 5 to 95% RH in 10% steps) compared the binary 10% DL ASD with the SDS and Span 80 ternaries (Figure S4, not available here).

Partitioning into ATZ nanodroplets by 1H NMR

Each additive was dissolved in buffer (pH 6.8, with 10% D2O; WATERGATE water suppression) at 500 µg/ml (PVPVA, VP) or 200 µg/ml (SDS, TPGS). Spans, which dissolve poorly, were used at 20 µg/ml in D2O with a single-pulse sequence. After warming to 37 °C, ATZ in methanol-d4 was spiked in under stirring to a total of 200 µg/ml, well above its amorphous solubility, so that nanodroplets formed. NMR peak areas against an internal reference (TSP, 50 µg/ml) gave the ATZ still in solution (the nanodroplet ATZ follows by difference from 200 µg/ml) and the additive still in solution. An additive that leaves the solution has gone into, or onto, the drug-rich droplets.

Experimental analysis

The baseline: release collapses between 5 and 10% DL

At 37 °C the binary 5% DL ASD released drug and polymer in step: the ATZ and PVPVA curves in Figure 1 overlap at every time point, reaching about 51% at 30 min and about 100% by 75–90 min. At 10% DL release almost stopped. The text gives less than 3% ATZ, and PVPVA release also fell. The authors therefore place the LoC of ATZ–PVPVA at 5% DL, the last loading tested that released fully. The normalized rates of the 10% DL ASD were about 0.07 mg/min·cm² for ATZ and about 0.24 for PVPVA (Figure 3), against about 2.7 for a compact of neat PVPVA. So the polymer, too, was held back, not only the drug.Card check: in Figure 1 the 10% DL points at 90 min sit at about 4.7% ATZ and 11% PVPVA, whereas Figures 2, 4 and 6 show about 2.4–2.8% ATZ and 7.6–7.8% PVPVA for the same ASD at 90 min and the text says less than 3%; the paper does not say whether the figures come from different runs (card note X1).

The additive screen at 10% DL

Every additive raised the release of both components from the 10% DL ASD (Figures 2 and 3), but by very different amounts.

Table B. Release from 10% DL ASDs at 37 °C (percent released read from Figure 2 and normalized rates read from Figure 3, approximate, n = 3; enhancement factors printed in Table 3).Card check: Table 3 gives CTAB a factor of 32, equal to SDS, but Figure 12 plots CTAB at about 36 and Figure 2 shows CTAB releasing more ATZ (about 106%) than SDS (about 93%) with the same ATZ content; the CTAB value is unresolved between 32 and 36 (card note X2).

Three readings follow. First, in every ternary the drug and the polymer stayed close to each other in both extent and rate, so the additives restored release that was close to congruent rather than freeing the drug alone. Second, the effect does not follow the additive class: two ionic surfactants of opposite charge (SDS, CTAB) and a nonionic one (Span 80) gave the largest gains, while two other nonionics (Span 85, TPGS) gave small ones. Third, VP, a small molecule with no surfactant character, raised total release about 30-fold and the release rate about 50-fold, "comparable to the most effective surfactants" (Section 3.1). The SDS and CTAB ASDs released so much drug that the medium held colloidal species after 90 min, detected by DLS (Table S1), because the concentration exceeded the amorphous solubility; Tween 80 and TPGS did not release enough to form them.

Higher loadings: only Span 80 holds up

The ternaries that released almost completely at 10% DL were made at 15% and 20% DL (Figure 4).

Table C. ATZ released by 90 min at 37 °C, % (PVPVA in brackets), read from Figures 2 and 4 (approximate, n = 3).

From these data the abstract states that SDS or CTAB "doubled the LoC (from 5 to 10% DL)" with a "more than 30-fold increase in total release", and that Span 80 "increased the LoC to 15% DL". The three systems fail differently above their LoC. CTAB collapsed at 15% DL. SDS declined gradually, about 60% and 39% of the drug at 15% and 20% DL, with drug and polymer still in step. Span 80 released fully at 15% DL but at 20% DL its PVPVA still reached about 42% while ATZ fell to about 4%: polymer left without the drug. VP, which matched the best surfactants at 10% DL, released only about 6% of ATZ at 15% DL; the text does not discuss this bar of Figure 4.

Note by the AI writer (not discussed in the paper): each LoC here is the last loading that released fully, tested in 5% steps, so the true threshold lies somewhere below the next step. The LoC as defined by diverging normalized rates would need the rates for the 15% and 20% DL ternaries, which are in Figure S1 and were not available for this reading.

Temperature instead of an additive

At 55 °C the binary 10% DL ASD released about 63% ATZ and 84% PVPVA by 90 min (Figure 6), against about 3% and 8% at 37 °C; its normalized ATZ rate rose from about 0.07 to about 1.6 mg/min·cm² (Figure 7). The release curve (Figure 5) shows ATZ trailing PVPVA throughout, about 43.5% against 58.5% at 50 min, and levelling off near 64–67% from about 70 min while PVPVA reached about 100% at 120 min. Polarized light showed crystals at 90 min, which is "likely the reason underlying the cessation of ATZ release". The binary 15% and 20% DL ASDs released about 51% and 22% ATZ at 55 °C (74% and 52% PVPVA). The TPGS 10% DL ASD rose from about 11% to 54% ATZ but crystallized after 1 h. The SDS ternaries gained no drug release: at 15% DL ATZ stayed at about 59–60% while PVPVA rose from about 66% to 99%, and at 20% DL ATZ fell from about 39% to 28%. Neat PVPVA itself dissolved only somewhat faster at 55 °C (about 2.7 to 3.7 mg/min·cm²).

The authors read the binary result as evidence that "the poor release of the ASD components at 37 °C may be linked to the Tg of the drug and/or the ASD" (Section 3.1), and later describe release at 55 °C as "once again rapid and congruent with polymer release" (Section 4.1).Card check: ATZ trails PVPVA at every time point at 55 °C and stops at about 64–67% while PVPVA reaches about 100%, so release at 55 °C is strongly improved but not fully congruent (card note X9). The paper shows no 37 °C value for the binary 15% DL ASD, although the text says heating raised its release (card note X10).

Tg and wet-film phase separation

Table D. Tg values from Table 1 and Table 3; film descriptions from Section 3.2 and Figures 8–10 (qualitative; the images were not quantified).Card check: the text says the ATZ:TPGS and ATZ:Tween 80 blends gave no single Tg, whereas Table 1 lists a single Tg of 98 °C for ATZ:Tween 80 and two Tgs for ATZ:SDS (card note X5). CTAB has no row in Table 1 although Table 3 gives it a 9 °C depression (card note X17).

Every additive lowered the ASD Tg, by 7 to 16 °C (Section 3.3). The neat additives have low Tg values, measured, cited or calculated, from −82 °C (VP) to 46 °C (SDS), and the authors attribute the plasticization to this. The 2:1 blends show that the Spans mix with the drug (single Tgs of 32, 36 and 68 °C), whereas SDS does not (two Tgs).

In the wet films the contrast was stark. The binary 10% DL film showed no phase separation within the microscope's resolution after at least 3 h at 100% RH and 37 °C (Figure 8), but every ternary film separated within 10 min (Figures 9, 10). The morphology split by the physical state of the surfactant. With the solid surfactants (SDS, CTAB, TPGS) the red domains stood out only weakly from a red background, which the authors read as domains that "appear to each contain mixtures of drug and polymer"; the effect was strongest for SDS. With the liquid surfactants (Tween 80 and the Spans) the domains were sharp: red drug-rich domains on a black background, and black pits in the green polymer channel where the Nile-red-rich regions lacked Alexa 488 signal, "consistent with the formation of drug-rich and polymer-rich domains". The binary film also separated within 10 min when exposed at 50 °C instead of 37 °C (Figure 11). The authors take these observations together as suggesting that "surfactants increase mobility, facilitating phase separation", and the contrast pattern as a sign that the surfactants "may impact the miscibility between the drug and polymer in the presence of water".

Water sorption pointed the same way, if weakly: up to 95% RH the binary ASD took up 44% water and the SDS and Span 80 ternaries 49% each (Section 3.4), which the authors describe as a "slightly" higher affinity for water, "consistent with plasticization and an increase in free volume".

Affinity for the drug versus additive size

Table E. 1H NMR of ATZ nanodroplet suspensions, 200 µg/ml total ATZ, 37 °C (Table 2 of the paper).Card check: for Span 80, 5 of 20 µg/ml is a 75% reduction, not the printed 68%; the other rows reproduce (card note X3).

Within detection limits PVPVA and VP did not enter the droplets. SDS lost less than 10% from solution, which "can likely largely be accounted for by the SDS adsorbed on the nanodroplet surface", and TPGS was described as "only minimally affected" although its loss (13%) exceeds that of SDS. The three Spans clearly left the solution, by 65%, 68% and 25%, and at the same time the ATZ remaining in solution, a measure of the amorphous solubility, fell by about 28% (from 64 to 46 µg/ml). The authors attribute this, as for other miscible systems, to lower drug activity once the additive mixes into the drug-rich phase (Ueda and Taylor, 2021). They estimate droplet compositions of 29:3, 62:5 and 60:4 (ATZ:Span by weight) for Span 20, Span 80 and Span 85.Card check: these ratios do not follow from Table 2; ATZ in the droplets divided by the Span lost from solution gives about 11:1, 10:1 and 30:1, and the calculation is not shown (card note X4).

Placing these results next to release (Table B) gives the paper's key negative result: SDS barely interacts with the drug and Span 80 partitions strongly into it, yet both release fully, so there is "little correlation between release enhancement and the affinity of the surfactant for the drug in the presence of water" (Section 4.2). What does order the gains is size. In Figure 12 the five additives below about 430 g/mol (VP, SDS, Span 20, CTAB, Span 80) give enhancement factors of about 23.5–36, and the three large ones lower values: Span 85 (957.5 g/mol) about 13, Tween 80 (1310 g/mol) about 19.5 and TPGS (1513 g/mol) about 4. Among the small additives, charge and hydrophilicity made little difference ("does not appear to have substantial impact"); among the larger ones, chemical structure "appears to have a greater impact" (Tween 80 above the smaller Span 85).

Note by the AI writer (not discussed in the paper): placing the two Table 3 columns side by side shows that the size of the Tg drop does not order the gains either. Span 85 (7 °C) and CTAB (9 °C) lower the Tg least yet differ 2.5-fold in effect (13 versus 32), and TPGS lowers it by 14 °C but helps least (4). The plasticization that the authors propose is therefore not captured by the dry Tg alone; size, which the authors link to how fast an additive helps water in, does better.

Discussion

The authors' mechanism

The discussion (Sections 4.1 and 4.2) builds a single chain from the polymer-dissolution literature. ATZ and PVPVA have similar Tg values, so the ASD stays glassy whatever the loading. A glassy matrix takes up water slowly and needs more water before its chains can disentangle; for the polymer, and with it the drug, to leave the surface, the matrix must become mobile. The authors contrast ATZ with ritonavir, a structurally related drug with a Tg of 50 °C, whose 20% DL PVPVA ASD has a Tg of about 85 °C and whose LoC is about 25% (Indulkar et al., 2019): water "is expected" to penetrate the ritonavir ASD faster, and "these observations suggest that the low LoC for the ATZ ASDs may be linked to the high Tg of the drug".

For the additives the authors write: "Our supposition is that the improved release performance is due, at least in part, to plasticization by the surfactants, with a resultant increase in the molecular mobility of the polymer chains, which leads to an improved polymer release rate in the presence of the drug, and consequently an improved drug release" (Section 4.2). The abstract puts it as "We speculate that surfactants act as plasticizers", the conclusions as "it is postulated", and they state that the mechanism "is still not completely clear". The support offered is indirect: all additives lower the Tg; all ternary films phase separate quickly while the binary film does not, and "mobility is required for phase separation to occur"; heat has the same effect as an additive on the binary ASD; and the SDS and Span 80 ASDs take up slightly more water. Hydrophilic surfactants "would reasonably be anticipated" to add a thermodynamic pull for water as well.

The size trend is fitted into the same picture. In Cooper's experiments on poly(methyl methacrylate), smaller plasticizers sped solvent ingress more, while the chemistry of an additive mattered through its thermodynamic compatibility with the polymer. The ATZ data show the same split: among small additives chemistry hardly matters, among large ones it does, so "kinetic factors appear to be very important in enhancing ATZ release from PVPVA ASDs". The point that both a hydrophilic, PVPVA-interacting surfactant (SDS) and poorly water-soluble ones (the Spans, "likely a non-solvent for the polymer" in the case of Span 20) help is consistent with the literature finding that solvents and non-solvents can both speed polymer dissolution. SDS also nearly doubled the release rate of neat PVPVA when added at 10%, whereas the Spans did not change it (Figure S2).

What is new relative to earlier work

Earlier papers had shown that surfactants can raise release at higher loadings, notably SDS with ledipasvir (LoC 5 to 30% DL; Que et al., 2019), and that internally added SDS speeds dissolution of PVP ASDs at concentrations below its CMC, so bulk solubilization is not the cause (Dave et al., 2012, 2013). This paper adds a systematic screen of eight additives on one high-Tg drug, a non-surfactant plasticizer control, a temperature control, and two measurements that rule out simple explanations: release gains do not follow charge or HLB, and they do not follow the additive's affinity for the drug-rich phase. What remains standing is additive size, which the authors connect to the kinetics of water ingress and chain mobility.

What it suggests for formulation

The authors are cautious. They point out that the same plasticization that helps release "can increase the crystallization tendency of the drug on long term storage", that several marketed ASDs contain SDS, TPGS or Tween 80, so balancing the two "appears possible" for some drugs, but that for rapidly crystallizing compounds "it may not be possible to incorporate surfactants". In vivo effects (membrane permeability, metabolism, clearance) were outside the study but "need to be included when selecting a surfactant" (Section 4.3).

Note by the AI writer (not discussed in the paper): for a high-Tg drug whose LoC in PVPVA is too low, a small, low-Tg additive at about 5% is a lever worth testing; here it doubled or tripled the usable loading. The window was narrow, though (Span 80 failed at 20% DL, SDS declined gradually above 10%), and the dry Tg drop did not predict which additive would work, so a release test, not a Tg measurement, is the screen. The additives that helped release also made wet films separate within minutes, and both the binary and the TPGS ASDs crystallized when release was forced at 55 °C; the physical stability of these ternary ASDs was not tested and should be checked before an additive is chosen.

Limitations

Limitations stated by the authors. The mechanism of improved release "is still not completely clear" (Section 5), and plasticization is put forward as a supposition. Surfactant plasticization may raise crystallization on storage, and in vivo effects of surfactants were outside the study (Section 4.3). At 55 °C the binary ASD and the TPGS ASD crystallized, which ended ATZ release (Sections 3.1, 4.1).

The AI writer's own judgement (1): the mechanism rests on indirect evidence. Molecular mobility, water ingress and chain disentanglement in the hydrating compact, the links of the proposed chain, were not measured. The evidence is a dry Tg, the speed of phase separation in thin films under vapour (a different geometry and hydration route from a dissolving compact), and a 44 versus 49% difference in water uptake measured for two of the eight additives. The dry Tg drop does not order the release gains (Span 85 and CTAB, TPGS), and VP, the plasticizer control, failed at 15% DL where Span 80 succeeded, so plasticization as measured by Tg cannot be the whole story. The temperature test also changes more than mobility: neat PVPVA dissolved faster at 55 °C and ATZ crystallized. The high-Tg explanation of the low LoC rests partly on a comparison with ritonavir taken from other studies, with no drug of intermediate Tg tested here.

The AI writer's own judgement (2): coarse loadings and internal inconsistencies. The LoCs come from percent released at 90 min in 5% DL steps; the normalized-rate data for the 15% and 20% DL ternaries, which would apply the LoC definition directly, are in supplementary material that was not available. Several values disagree within the paper: the binary 10% DL release at 90 min in Figure 1 (about 4.7% ATZ) against Figures 2, 4 and 6 and the text (below 3%); the CTAB enhancement factor (32 in Table 3, about 36 in Figure 12); the Span 80 row of Table 2 (68% printed, 75% from its own numbers); the droplet ATZ:Span ratios, which do not follow from Table 2; the miscibility of the ATZ:Tween 80 blend (single Tg in Table 1, not single in the text); the film temperature (50 °C in Figure 11, 55 °C in Section 4.1); the HLB of Span 80 (printed as 15, as for Tween 80); and cross-references to a non-existent "Table 4" and to "Table 3" for the NMR data, which are in Table 2. None of these changes the main release results, but they argue for reading individual numbers from the figures with care. Finally, the NMR partitioning was run at additive levels (20–500 µg/ml) and drug:additive ratios unlike those in the ASD, and in a different medium for the Spans, so it ranks affinity rather than reproducing the dissolving surface.

Conclusion

The paper shows three things clearly. Atazanavir–PVPVA ASDs release drug and polymer together only up to 5% DL at 37 °C, and release at 10% DL recovers at 55 °C until the drug crystallizes. Adding 5 g of SDS or CTAB per 100 g raises the LoC to 10% DL and Span 80 to 15% DL, with the drug and polymer still released together, and the plain plasticizer VP works as well as the best surfactants at 10% DL. Across eight additives the gain tracks molecular size rather than charge, hydrophilicity or affinity for the drug-rich nanodroplets. The authors' explanation, that the additives plasticize the glassy matrix so that water enters and the polymer chains can disentangle and dissolve, fits these observations but was not measured directly; in their own words, "We speculate that surfactants act as plasticizers, which may facilitate polymer release from ASDs containing a high Tg drug". For formulation, the work points to small additives as a way to raise the usable loading of high-Tg drugs in PVPVA, to be weighed against the higher crystallization risk that the same plasticization may bring.