Surfactants in clopidogrel–copovidone dispersions: how far they raise drug loading and how they keep nanodroplets small

Background and motivation

Many new drug candidates dissolve poorly, and their absorption is limited by how much drug is dissolved in the gut. An amorphous solid dispersion (ASD), in which the drug is mixed at the molecular level into a water-soluble polymer, can push the dissolved concentration above the crystalline solubility. When enough drug is released to exceed the amorphous solubility, the excess separates as a second, drug-rich liquid phase: tiny drug-rich droplets, called nanodroplets, typically 50–1000 nm across. The introduction collects the reasons why these droplets are thought to help. They may act as a reservoir that keeps the dissolved drug at its maximum level while drug is absorbed (Indulkar 2016, ref 3), and they may drift into the mucus layer near the gut wall (refs 11, 21, 22). An enzalutamide–PVPVA ASD that formed nanodroplets gave a higher exposure in rats than one that did not (Wilson 2018, ref 19); anacetrapib droplets of 100 nm or less did better than droplets of 200 nm or more (Kesisoglou 2019, ref 11); and itraconazole–HPMCAS colloids of 150–400 nm gave 10–50% higher bioavailability (Stewart 2017, ref 21). C1

The droplets are not stable. A supersaturated solution that holds a drug-rich phase is metastable, so the droplets can grow by creaming, Ostwald ripening, flocculation and coalescence, and the drug can crystallize, which removes the supersaturation altogether (refs 23–27). C2 Surfactants are the obvious tool to stop droplets from growing, since they stabilize emulsions and suspensions, and they are already used in amorphous products to aid dissolution and hot-melt extrusion. They cut both ways, however: micellar surfactant raises solubility but can lower the chemical potential of the drug and so its absorption, too little surfactant lets particles coalesce while too much promotes Ostwald ripening, and surfactants have been reported to induce crystallization in solids and in solution (refs 4, 15, 28–37). C3

The second strand of background is the release behaviour of PVPVA dispersions, much of it from the same group. PVPVA dissolves quickly, and at low drug loading (DL) drug and polymer leave the dissolving compact at the same normalized rate, called congruent release; this "polymer-controlled" regime generates nanodroplets and is considered the optimum. C44 Above a certain DL the two rates diverge, drug release becomes slow and incomplete, and release is said to "fall off the cliff". The highest DL that still releases congruently is the limit of congruency (LoC). Reported LoCs for PVPVA ASDs run from below 5% to about 30% DL: 5% for ledipasvir, about 10% for nilvadipine (printed "nivaldipine"), about 15% for cilnidipine and 25% for ritonavir (Que 2019, ref 12; Saboo 2019, ref 81; Indulkar 2019, ref 16). C45·R28 A high drug glass transition temperature ($T_g$) and strong drug–polymer hydrogen bonding both lower the LoC, although in the drugs studied so far the two went together and are hard to separate (Saboo 2020, ref 13; Yang 2021, ref 84). C49·R28 Surfactants have raised the LoC before: SDS lifted the ledipasvir–PVPVA LoC from 5 to 30% DL, TPGS improved felodipine release from 10 to 35% DL, and TPGS greatly improved anacetrapib release at 20% DL (Que 2019, ref 12; Saboo 2021, ref 68; Harmon 2016, ref 20). R29

Against this background the paper sets two aims. The first is to find the lowest surfactant concentration that keeps drug-rich nanodroplets from growing for at least 4 h, which the authors take as the biologically relevant time for gastric and small-intestinal transit. C4 The second is to test how an anionic (SDS), a cationic (cetyltrimethylammonium bromide, CTAB) and two non-ionic surfactants (D-α-tocopherol polyethylene glycol 1000 succinate, TPGS, and polysorbate 80, Tween 80 or TWN) change drug release from PVPVA dispersions, with solution 1H NMR used to see how each surfactant meets the drug. The model drug is clopidogrel free base (CPD), chosen because it crystallizes slowly, so that nanodroplets can be studied over long times without crystals interfering. C5

Background concepts

Amorphous solubility and liquid–liquid phase separation. If drug is added to water beyond its amorphous solubility, the solution splits into a water-rich phase holding dissolved drug and a drug-rich phase dispersed as droplets. This is liquid–liquid phase separation (LLPS), and the concentration at which it starts (the LLPS concentration) matches the amorphous solubility (Ilevbare 2013, ref 7). If the drug's $T_g$ lies above the test temperature, the drug-rich phase is glassy and the process is called glass–liquid phase separation (Mosquera-Giraldo 2015, ref 9). Here the onset is detected as a rise in light extinction at 450 nm, a wavelength the drug does not absorb, while drug is pumped in from a methanol stock (the solvent-shift method). P1

Congruent release and the LoC. In a surface-normalized dissolution test, a compact of fixed area is rotated in buffer and both the drug and the polymer are assayed in the medium. Comparing their release rates per unit area, each normalized to the component's fraction in the solid, shows whether they leave together. The paper's normalized rate is

In plain terms: $k$ is the slope of the cumulative mass released (mg) against time (min), $S$ is the exposed compact area (0.5 cm²) and $x$ is the fraction of that component in the dispersion, so $R$ (mg/min·cm²) is the rate at which a component leaves per unit area, scaled as if the compact were made of that component alone. Equal $R$ for drug and polymer means congruent release. E1·P14

Critical micelle concentration and HLB. Below its critical micelle concentration (CMC) a surfactant is present as single molecules; above it, micelles with an oily core form and can take up a lipophilic drug. The CMC is found here from the fluorescence of pyrene, whose ratio of two emission peaks changes when it moves into a micelle core, and in some cases from surface tension. P4·P5 The hydrophilic–lipophilic balance (HLB) is a literature number describing how water-loving a surfactant is; the higher it is, the more hydrophilic. R1

Zeta potential. The zeta potential is the electrical potential at the surface of a particle as it moves in solution. Charged surfactants that adsorb on a droplet change it, and a large magnitude (strongly positive or negative) makes droplets repel each other. The authors use it as a readout of whether a surfactant sits on the droplet surface. C6·P7

Seeing two populations of drug by NMR. Solution NMR detects only molecules that tumble fast. Drug dissolved in water gives sharp peaks; drug inside a viscous droplet gives broad, often undetectable peaks. A change in peak position (chemical shift) reports a change of surroundings: a shift to lower ppm ("up-field") is read as a less polar environment, such as a micelle core. When a surfactant is taken up into droplets, its peaks can also disappear from the spectrum, and the drop in its dissolved concentration (depletion) measures how much went in. C14·C16·P9

Water-induced amorphous–amorphous phase separation. A drug–polymer glass that takes up water can split into a drug-rich and a polymer-rich amorphous phase (AAPS), because the polymer prefers water and the drug does not. The authors image this in thin films containing two fluorescent dyes: Nile red, which favours the hydrophobic, drug-rich phase, and Alexa 488, which favours the hydrophilic, polymer-rich phase. A uniform film shows even fluorescence; a separated film shows domains, and the shape of the domains (droplets, a bicontinuous network, or one continuous phase around islands of the other) is the morphology. C7·P17

Methods in detail

The study has two halves that run in parallel and meet in the discussion: a screen of the four surfactants in solution, with CPD nanodroplets made by solvent shift, and a test of the same surfactants built into rotary-evaporated CPD–PVPVA dispersions. All work is in 50 mM sodium phosphate buffer at pH 6.8 and 37 °C unless stated. M9

Contribution 1: a solution screen for nanodroplet stability

The problem. To know whether a surfactant can protect the nanodroplets that an ASD releases, one first needs the drug's amorphous solubility with and without surfactant, the surfactant's CMC in the same medium, and a measure of how fast the droplets grow.

What they did. CMCs of SDS, CTAB and TPGS were measured in buffer at 37 °C from the ratio of pyrene's emission at 371 and 382 nm (excitation 332 nm, 0.2 μg/ml pyrene), and in some cases by Wilhelmy surface tension; Tween 80 gave no clear break by fluorescence and was measured by surface tension only. P4·P5 The LLPS concentration was found by pumping a methanolic CPD stock into 15 ml buffer (methanol at most 2%, run under 10 min, 300 rpm) while recording extinction at 450 nm every 15 s, at least in triplicate; for buffer and some surfactant media it was checked by centrifuging a supersaturated dispersion (21,100 g for 45 min, or 184,048 g for 40 min when small particles did not pellet) and assaying the supernatant by HPLC, and by a diffusion-cell flux measurement. P1·P2·P3 Nanodroplets were made by adding a 10 mg/ml methanolic CPD stock to buffer with or without surfactant and held at 37 °C and 200 rpm for 4 h; Z-average size and polydispersity index (PDI) were measured by dynamic light scattering (DLS) every hour, with nanoparticle tracking analysis (NTA) as a check, in triplicate. P6 Zeta potential was measured at 70 μg/ml CPD, of which about 50 μg/ml is in droplets, without surfactant and with SDS 50, CTAB 50, TPGS 4 and 50, and Tween 80 5 and 50 μg/ml. P7

How it differs. The point of the screen is to find the smallest effective level rather than to solubilize the drug: the levels shown in Fig. 2 (CTAB and SDS at 50 μg/ml, TPGS at 4 μg/ml, Tween 80 at 5 μg/ml) are near or below the CMCs. P6·R1 The authors then estimate how much surfactant would cover the droplet surface with one layer, from the droplet radius (DLS, NTA) and literature molecular areas, and compare it with the effective level (Table IV); the calculation itself is in the SI. E2·R24

Contribution 2: NMR of drug–surfactant interactions, including drug inside the droplets

The problem. A surfactant can act on the dissolved drug (micellar solubilization), sit on the droplet surface, or move into the droplets. These have different consequences, but they are hard to tell apart.

What they did. Spectra were taken on an 800 MHz spectrometer with a cryoprobe, in buffer made up in H2O/D2O 9:1 with trimethylsilyl propionate (TSP, 50 μg/ml) as reference, with the water signal suppressed (WATERGATE); 64 scans were used for SDS and CTAB and 128 for TPGS and Tween 80. CPD was spiked in from methanol-d4 stocks (below 2% v/v) at 37 or 50 °C. P8 For quantification of TPGS remaining in solution, single-pulse spectra in D2O buffer were used (relaxation delay 15 s), because peak areas from this sequence are less distorted. P9 Three kinds of experiment were run: CPD at 70 μg/ml at 37 and 50 °C without surfactant (Fig. 4); CPD at 10–170 μg/ml with 50 μg/ml TPGS at 37 °C (Fig. 5) and at 70 μg/ml with TPGS at 50 °C (Fig. 6); and CPD at 70 μg/ml with each surfactant at 50 μg/ml (Fig. 7). R8·R11·R13·R14

Why it works. CPD has a low $T_g$ (−9 °C), so at 50 °C the drug inside the droplets is mobile enough to give its own, broad set of peaks, which lets the authors watch the droplet phase directly rather than infer it. M1·C15·C35

Contribution 3: surfactants inside the dispersion and surface-normalized release

What they made. Binary CPD–PVPVA and ternary CPD–PVPVA–surfactant ASDs were made by dissolving the solids in about 50 ml methanol, sonicating 20 min, removing the solvent on a rotary evaporator (150 rpm, 50 °C bath), drying under vacuum for at least 24 h, grinding and sieving through 425 μm. The surfactant was fixed at 5 g per 100 g of drug plus polymer, and compositions are given as weight ratios: "CPD 40 DL" is 40 g CPD with 60 g PVPVA, plus 5 g surfactant in the ternary ASD. ASDs from 20 to 60 DL were prepared; release results are shown for 40, 50 and 60 DL. P10

How release was measured. 100 mg of powder was compressed in an 8 mm die (surface 0.5 cm²) at 1500 psi for 60 s immediately before testing, and the die was rotated at 100 rpm in 100 ml degassed buffer at 37 °C (Wood's apparatus). These are non-sink conditions with respect to the amorphous solubility. Samples were taken every 5 min to 20 min and then every 10 min to 120 min, with medium replaced. CPD was assayed by reversed-phase HPLC (UV 217 nm) and PVPVA by size-exclusion HPLC (UV 205 nm); results are means of three runs. P11·P12·P13 The normalized initial release rate was obtained with Eq. 1 from the initial linear part of the cumulative release, and differences from the binary ASD were tested by ANOVA with Tukey's test (P < 0.05). E1·P14·P15 Droplet size in the medium was measured by DLS at 90 min for the 40 and 50% DL ASDs. P16

Contribution 4: what water does to the solid

Thin ASD films containing Nile red and Alexa 488 were spin-coated (30 rpm for 12 s, then 2000 rpm for 50 s) and heated at 70 °C, then imaged by confocal microscopy (excitation 561 nm for Nile red, 488 nm for Alexa 488) freshly made and after 5 min at 100% relative humidity (RH). Binary films and films with SDS, TPGS and Tween 80 were imaged at 40, 50 and 60% DL; the authors give the detection limit as about 500 nm. P17·R20·R21·R22·R23 Note by the AI writer (not discussed in the paper): no CTAB films are shown, and the films see water vapour for 5 min, whereas the compacts sit in buffer for up to 2 h. X13

Experimental analysis

Surfactant CMCs and the amorphous solubility of clopidogrel

The CMCs in buffer at 37 °C span more than an order of magnitude: SDS 487 ± 34 μg/ml, CTAB 50 μg/ml, TPGS 40 μg/ml and Tween 80 about 14 μg/ml (Table I). R1 The authors compare them with the literature: the SDS value agrees with 500 and 433 μg/ml reported in similar media, the CTAB value is below the 328 μg/ml reported in water, the Tween 80 value falls inside the reported 13–34 μg/ml, and the TPGS value is close to 30 μg/ml reported in water. R2·C8 Elsewhere in the text the TPGS CMC in buffer is given as "~50 μg/ml"; Table I gives 40 μg/ml. X1

CPD separates into a drug-rich phase above 23 ± 3 μg/ml in buffer, a value that ultracentrifugation (18.3 ± 1.0 μg/ml) and a diffusion cell (20–25 μg/ml) roughly bracket (Table II). R3.a Surfactant raised this concentration only at the higher level: TPGS and Tween 80 at 5 μg/ml gave 20 ± 1 and 25 ± 2 μg/ml, at 50 μg/ml 29 ± 2 and 30 ± 2 μg/ml. R3·C9 SDS at 50 μg/ml, still below its CMC, gave 20 ± 1 μg/ml, and in a concentration series SDS raised the LLPS concentration only once its CMC was exceeded, as seen earlier for atazanavir (Indulkar 2017, ref 5). R3.b·R4·C9 CTAB gave 20 ± 1 and 24 ± 2 μg/ml at 5 and 50 μg/ml. R3.c·R3.d Two further media show how sensitive the value is to the solvent: 17 ± 1 μg/ml in D2O buffer, which the authors attribute to the higher polarity of D2O ("can be attributed"), and 49 μg/ml in unbuffered water. R3.i·R3.j·C10

Table A. Surfactant CMCs and the LLPS concentration of CPD at 37 °C in 50 mM phosphate buffer pH 6.8 (from Tables I and II; mean ± SD, n = 3 for LLPS). R1·R3

Nanodroplet size: a little surfactant goes a long way, except CTAB

Without surfactant, CPD nanodroplets in buffer grew quickly: relative to time zero, the Z-average was about 1.5 times larger after 1 h and about 12 times larger after 2 h, and the PDI rose about 4.2- and 4.6-fold (Fig. 2, values read from the plot). R6.a·R6.f Adding surfactant gave smaller droplets from the start, for example about 520 nm in buffer against 280 nm with 4 μg/ml TPGS. R5·C11 SDS at 50 μg/ml, TPGS at 4 μg/ml and Tween 80 at 5 μg/ml held the Z-average ratio at about 0.81–1.01 and the PDI ratio at about 0.9–1.25 for 4 h. R6.c–e·R6.h–j CTAB at 50 μg/ml did the opposite: the Z-average was about 12 times larger within 1 h, and large agglomerates floated on the surface. R6.b·C12 No bars are plotted for buffer at 4 h or for CTAB at 2 and 4 h, and the figure does not say why. X10

Table B. Z-average of CPD nanodroplets relative to time zero (read from Fig. 2a, approximate; n = 3), with the minimum effective concentrations and calculated monolayer coverages of Table IV. R6·R24

For the non-ionic surfactants the authors find "reasonable agreement" between the effective level and the amount needed to cover the droplets with a monolayer. C41 Strictly, TPGS worked at 4 μg/ml, below its calculated coverage of 10.4 μg/ml, and Tween 80 at 5 against 3.1 μg/ml, so the agreement is within a factor of about 2.6. X15 SDS needed 50 μg/ml, more than its calculated range of 0.3–17.8 μg/ml (the range comes from a wide range of literature molecular areas, 6.7–425 Ų), which the authors explain as "most likely" reflecting its high HLB and weaker tendency to sit at an interface, consistent with its higher CMC. R24.a·C42 CTAB failed even though 50 μg/ml is well above its calculated 2.4 μg/ml, "by an unknown mechanism". R24.b·C38

Zeta potential: where the charged surfactants sit

CPD nanodroplets in buffer carry a zeta potential of about −16 mV. SDS at 50 μg/ml made it about −69 mV, CTAB at 50 μg/ml reversed it to about +29 mV, and TPGS and Tween 80 brought it close to zero (about −4 and −1 mV for TPGS at 4 and 50 μg/ml, about −5 and −1.5 mV for Tween 80 at 5 and 50 μg/ml; Fig. 3, read from the plot). R7 The authors read the larger magnitude with the ionic surfactants and the smaller one with the non-ionics as evidence that all four "support adsorption of the surfactant at the nanodroplet surface". C13 For SDS they add that little SDS was lost from the aqueous phase while the charge became more negative, which they take as the typical case of stabilization by an adsorbed layer; the depletion data are not shown in the main text. R25·C39

NMR: dissolved drug, drug in droplets and the four surfactants

At 37 °C the spectrum of CPD at 70 μg/ml showed only one set of aromatic peaks, from dissolved drug, although droplets were present; at 50 °C a second set appeared, broad and up-field by roughly 0.5–0.7 ppm (Fig. 4; for example the H8 proton at about 6.77 ppm dissolved and about 6.25 ppm in the droplets, positions read from the plot). R8 These extra peaks were absent below the LLPS concentration and grew with the amount of droplets (SI). R10 At 50 °C the solubility rose to about 30 μg/ml, leaving about 40 μg/ml of the 70 μg/ml in the droplets. R9 The authors assign the broad peaks to drug inside the droplets, in a less polar and less mobile environment, and explain their appearance at 50 °C by the higher mobility of the droplet phase given CPD's $T_g$ of −9 °C; the same was seen for ibuprofen ($T_g$ −45 °C) at 37 °C (Ueda 2019, ref 25). C14·C15·R31

TPGS gives the clearest picture. With 50 μg/ml TPGS at 37 °C, every CPD peak moved up-field at every CPD level from 10 to 170 μg/ml (for example H8 from about 6.76 to about 6.68–6.70 ppm), including levels below the LLPS concentration; the authors read this as solubilization in micelles ("This suggests a solution phase interaction"). R11·C16 The TPGS peaks moved up-field too, which "suggests that the chemical environment around TPGS was changed". C17 At 50 °C the droplet peaks of CPD moved down-field and became sharper and more intense with TPGS, which the authors read as TPGS partitioning into the droplets and making them more polar and more mobile ("suggests"). R13·C18 Quantitative single-pulse NMR then showed dissolved TPGS falling from 50 to 33 ± 5.4 μg/ml in the presence of 70 μg/ml CPD as droplets, with no visible signal from TPGS inside the viscous droplets; the authors derive a droplet composition of 50:17 w/w CPD:TPGS and report that HPLC after pelleting confirmed the uptake (SI). R12·C19

The other three surfactants behaved differently (Fig. 7). With SDS the CPD peaks broadened but stayed at their buffer positions; with CTAB the H3/H7 multiplet split into two peaks while the other peaks stayed put; with Tween 80 all but one broad peak (about 7.57 ppm) disappeared. R14 The authors summarize these in Table III: SDS shows no interaction with dissolved drug or droplets; CTAB takes dissolved CPD into micelles but does not interact with the droplets; TPGS takes dissolved CPD into micelles and partitions into the droplets; Tween 80 interacts "strongly" with dissolved CPD and partitions slightly into the droplets. C20·C21·C22·C23

Note by the AI writer (not discussed in the paper): two points in this summary rest on thin main-text evidence. The droplet composition of 50:17 assumes about 20 μg/ml of CPD dissolved, whereas with 50 μg/ml TPGS the LLPS concentration is 29 ± 2 μg/ml, which would leave about 41 μg/ml in the droplets; the temperature of the depletion measurement is not stated. X8 And the CTAB entry "incorporation into surfactant micelles" sits beside an LLPS concentration at 50 μg/ml CTAB (24 ± 2 μg/ml) that is within the scatter of buffer (23 ± 3 μg/ml), at a CTAB level equal to its CMC; the supporting spectra are in the SI. X14

Release from binary dispersions: congruent up to 40% DL

In every test the drug passed its LLPS concentration within the first 10 min, so nanodroplets formed in all media. R16 At 40% DL the binary ASD released CPD and PVPVA nearly together (about 57 vs 74% at 20 min, about 76 vs 97% at 60 min; Fig. 8a, read from the plot), with the polymer slightly ahead and the drug levelling off near 76%. R15.a·R15.b·C24 At 50% DL the polymer clearly ran ahead (about 37 vs 68% at 20 min), and drug release levelled off near 55% while the polymer dissolved completely (Fig. 8f). R15.k·R15.l·C27 Note by the AI writer: the binary panels of Fig. 8 stop at 60 min, while all other panels run to 120 min. X2

Fig. 9 condenses these into a total percent release and a normalized initial rate for 40, 50 and 60% DL. R17·R18 For the binary ASDs the drug released fell from about 77% to about 52% to about 7%, while PVPVA came out at about 100, 103 and 108%; the drop is steeper from 50 to 60% than from 40 to 50%. R17.a·R17.f·R17.k·C28 The normalized rates tell the same story: about 4.5 vs 5.2 mg/min·cm² for CPD vs PVPVA at 40% DL, 3.6 vs 6.7 at 50% and 0.2 vs 4.1 at 60%. R18.a·R18.g·R18.m The authors place the LoC of the binary system at "~40%", which they call high compared with other PVPVA dispersions, and read the slightly faster polymer release at 40% DL as the reason only about 80% of the drug comes out: the compact left behind becomes richer in drug, and once release is drug-controlled there is no driving force above the amorphous solubility ("likely"). C46·C25

Table C. Binary CPD–PVPVA ASDs (read from Fig. 9, approximate; mean of n = 3). R17·R18

Two features of these numbers need care. First, release was tested only at 40, 50 and 60% DL, so the data place the LoC between 40 and 50% DL rather than at a point. X6 Second, the text says Fig. 9a–c shows release after 60 min, but several bars match the 120 min points of Fig. 8 rather than the 60 min points (for example 40% DL with TPGS: about 81% in Fig. 9a, about 48% at 60 min and about 82% at 120 min in Fig. 8c), so the time basis of the percent-release bars is unclear. X2 The authors' wording also runs slightly ahead of the bars in places: they call drug and polymer rates "similar" at 40 and 50% DL, though the binary 50% DL ASD gives about 3.6 vs 6.7 mg/min·cm², and they say PVPVA is released "up to 10 times faster" than CPD at 60% DL, though the binary bars give about 20-fold. C29·X3·X4

The most surprising result is the polymer rate itself. A neat PVPVA compact dissolved at about 2.6 mg/min·cm², but in the binary ASDs PVPVA dissolved at about 5.2, 6.7 and 4.1 mg/min·cm² at 40, 50 and 60% DL: adding a lipophilic drug made the polymer dissolve about twice as fast ("Surprisingly", "approximately" twofold). R18.f·R18.a·R18.g·R18.m·C30 The authors connect this to the low $T_g$ of CPD, which lowers the $T_g$ of the ASD (data in the SI), and to work in which 20 wt% poly(ethylene oxide) doubled the dissolution rate of poly(methyl methacrylate) by plasticization; in the conclusions the faster polymer release "is thought to result from a plasticization effect of the drug". R26·R30·C48

Release with surfactants: only SDS helps

At 40% DL the surfactants changed the extent of drug release little, apart from CTAB (about 77–81% CPD for none, Tween 80, TPGS and SDS; about 54% with CTAB; Fig. 9a), and none of these differences was marked significant. R17.a–e The rates moved more: SDS (about 4.8 vs 5.4 mg/min·cm² for CPD vs PVPVA) and CTAB (about 3.9 vs 5.2) kept the fast release of the binary ASD, whereas Tween 80 (about 2.5 vs 3.0) and TPGS (about 1.9 vs 2.1) slowed both drug and polymer significantly. R18.a–e·C30·C56 With TPGS and Tween 80 release simply went on longer, so by 120 min CPD reached about 82 and 80% (Fig. 8b, c). R15.c·R15.e

At 50% DL SDS stood out. With SDS the ASD released about 70% CPD against about 52% without surfactant, and drug and polymer left at almost the same rate (about 5.1 vs 5.3 mg/min·cm²) instead of 3.6 vs 6.7; both changes were significant. R17.f·R17.i·R18.g·R18.j The profile shows it from the start: about 57% CPD and 65% PVPVA at 20 min, levelling off near 72% and 81% (Fig. 8i). R15.q·R15.r Note that SDS lowered the PVPVA release at 50% DL (about 81% against 103%), so the closer match partly comes from a slower polymer. R17.i·R17.f TPGS and Tween 80 released about 60 and 59% CPD at 50% DL, but at much lower rates (about 1.5 and 1.3 mg/min·cm²), and CTAB about 44%. R17.g·R17.h·R17.j·R18.h·R18.i The authors conclude that only SDS improved release, extending the LoC "from 40 to 50% DL", and that SDS "appeared to be the best surfactant choice for CPD"; the slower release with the non-ionic surfactants is "likely unfavorable" unless a slower release is wanted. C55·C57

Table D. ASDs at 50% DL with 5 g surfactant per 100 g of drug plus polymer; rates in mg/min·cm² (read from Fig. 9b and 9e, approximate; n = 3; significance at P < 0.05, ANOVA and Tukey). R17·R18

CTAB was harmful in a characteristic way. With CTAB, dissolved CPD rose to a maximum (about 65% at 40–50 min at 40% DL, about 45% at 50 min at 50% DL) and then fell, to about 25% at 120 min at 50% DL (Fig. 8e, j). R15.i·R15.s Polarized light microscopy showed no crystals, and the authors attribute the loss to CTAB agglomerating the drug-rich droplets ("can be attributed"), in line with the solution screen. R16·C26 At 60% DL no surfactant helped: every ASD released only about 7–11% CPD, although TPGS, SDS and CTAB raised the very low CPD rate significantly (about 0.45–0.65 against 0.2 mg/min·cm²). R17.k–o·R18.m–q

The droplets in the release medium followed the solution screen. After 90 min, the 40% DL ASDs gave Z-averages of about 790 nm (binary), 510 nm (Tween 80), 580 nm (TPGS), 690 nm (SDS) and more than 2000 nm (CTAB), and the 50% DL ASDs about 1960, 780, 890, 1260 and more than 2000 nm; PDI was about 0.22–0.44 with SDS, TPGS and Tween 80, against about 0.48 and 0.75 for the binary ASDs and about 0.61 and 0.84 with CTAB (Fig. 10, read from the plot). R19 The authors summarize this as "somewhat smaller" droplets with a narrower distribution with surfactant, especially at 50% DL, and coarsening with CTAB. C31

Films under humidity: separation from 50% DL, and SDS

All films were uniform before exposure. After 5 min at 100% RH the binary 40% DL film was unchanged, the 50% DL film had separated into a bicontinuous structure of Nile red-rich and Alexa 488-rich domains, and in the 60% DL film the Nile red (drug-rich) phase was continuous around discrete polymer-rich domains (Fig. 11). R20 The authors find this "consistent with" the release results and with earlier reports in which a continuous drug-rich phase went with poor release (Saboo 2021, ref 68). C32

Table E. Film morphology read from the confocal images (Figs. 11–14) and the text. R20·R21·R22·R23

With SDS the picture changed at 50% DL. The results section describes the 50% DL SDS film as showing "a small extent of heterogeneity in fluorescence intensity", the discussion as remaining "homogeneous (within the detection limit of the measurement which is ~500 nm)"; at 60% DL the SDS film separated, with a continuous drug-rich phase and smaller polymer domains than without SDS (Fig. 12). R21·X12 TPGS films were bicontinuous at 50% DL and coarsened into large irregular structures at 60% DL ("suggesting more extensive coarsening"), and Tween 80 films separated at 50 and 60% DL (the text says "above 50%"). R22·R23·C33·X16

Discussion

Why clopidogrel reaches a high LoC

An LoC of about 40% DL is high for PVPVA, where the drugs studied before reached 5–25%. C46·R28 The authors propose two drug properties as the reason ("likely"). C47 The first is the low $T_g$ of CPD (−9 °C). Polymer dissolution depends on how far the matrix $T_g$ lies above the test temperature, and earlier work points the same way: the very high-$T_g$ drug ledipasvir had an LoC below 5% DL, the indomethacin LoC fell when the temperature dropped below the estimated $T_g$ of the water-saturated ASD, and the loratadine–PVPVA LoC fell from 30% to below 10% when the temperature was lowered from 37 to 5 °C (refs 12, 13, 84). R28·C48 CPD lowers the ASD $T_g$ (SI), which the authors take as a plasticization that also explains the doubled PVPVA rate. R26·C48 The second property is the absence of specific interactions: neither CPD nor PVPVA has a hydrogen-bond donor, and drugs that hydrogen-bond strongly to PVPVA "tend to have lower LoCs", as for indomethacin against its methyl ester (Saboo 2020, ref 13). M1·M2·C49 Centrifugation found less than 5% PVPVA in the CPD droplets, in contrast to ibuprofen, which pulls in much PVPVA (Ueda 2020, ref 60). R27·R31·C50 The authors are careful that the two effects are linked in the earlier data: the strongly hydrogen-bonding drugs studied before also had high $T_g$ values, so "it is somewhat difficult to deconvolute these two effects". C49

Why release fails above the LoC

A low $T_g$ helps release but also speeds unwanted changes, including AAPS. C51 PVPVA likes water far more than CPD (cLogP 3.8), so water drives the two apart, and the films show this from 50% DL. C52·M1 The authors link this to release through earlier micro-CT and surface studies, which found a porous drug-rich layer at the surface of high-DL PVPVA dispersions (refs 12, 77, 81): when enough drug-rich phase forms to make a continuous or semi-continuous barrier, the polymer dissolves out and leaves drug behind. C53 For CPD they say the results "support a link" between extensive phase separation at 50% DL and above and poor release, and the full sequence is offered as a proposal: "at high DLs, AAPS is proposed to occur rapidly", with a continuous drug-rich phase, fast polymer loss and surface enrichment of drug. C53

What SDS does, and what the non-ionic surfactants do

The authors offer a "plausible mechanism" for SDS: since the 50% DL film with SDS stays (nearly) uniform while the binary film separates extensively, "SDS may prevent, delay, or change the domain size of phase separation at this drug loading, leading to improved release". C58 SDS does not stop visible separation at 60% DL, which matches the poor release there, and TPGS and Tween 80 "clearly do not prevent phase separation", which matches their lack of benefit. C58 TPGS "appears to enhance the rate of ripening" of the separated domains, which "can possibly be explained" by TPGS partitioning into the CPD-rich phase, as seen by NMR, changing the kinetics and thermodynamics. C59 For the solution behaviour the authors propose that surfactants stabilize droplets by adsorbing at the surface with steric and/or electrostatic stabilization ("consistent with" the low effective levels), and that the non-ionic surfactants' uptake into the droplets "does not appear to be related to nanodroplet stabilization" and "likely reflects" their lower polarity. C39·C40 In the conclusions this becomes a general statement: non-ionic surfactants "tended to partition into the drug-rich phase, while ionic surfactants interacted at the nanodroplet-water interface". C60

Note by the AI writer (not discussed in the paper): the ionic/non-ionic split rests mainly on TPGS. By the authors' own NMR summary, Tween 80 partitions only slightly into the droplets, and neither SDS nor CTAB shows any interaction with the droplets by NMR; the placement of the ionic surfactants at the interface comes from the zeta potential and from a statement that SDS was barely depleted. X7

What it suggests for formulation

The authors draw three practical points. Surfactants worked at very low concentrations, which they see as favourable given toxicity and design limits on surfactant content. C38 Stabilizing the dispersed drug phase is "promising" for keeping a large droplet surface in equilibrium with dissolved drug during absorption; surfactants that coarsen droplets, like CTAB, "would be a poor choice", and this kind of solution screen "may be useful when selecting surfactants". C43 And for CPD, SDS "appeared to be the best surfactant choice", while the slower release with non-ionic surfactants is "likely unfavorable" unless slow release is wanted. C57 They also note that several marketed ASD products contain SDS, TPGS or Tween 80, although it is not known whether the surfactant sits inside the ASD intermediate. R32·C54

Note by the AI writer (not discussed in the paper): the solution screen and the release test rank the surfactants differently. TPGS and Tween 80 held droplet size at 4–5 μg/ml but slowed release from the ASD, while SDS needed 50 μg/ml in solution yet was the only surfactant that raised the usable loading. R24·R18·R17 A 100 mg compact with 5 g surfactant per 100 g carries about 4.8 mg of surfactant, about 48 μg/ml in 100 ml if all of it dissolves (AI arithmetic), which is near the SDS level that stabilized droplets and far above the non-ionic levels. P10·P11·R24 For selecting a surfactant to raise drug loading, the release test, not the droplet screen, decided the outcome here.

Limitations

Stated by the authors. The surfactant-dependent kinetics and morphology of water-induced phase separation were observed, but "the significance of such differences to ASD release performance, if any, is currently unknown". C61 The abstract concludes that surfactants "may lead to some improvements in formulation performance, but predictive capabilities and mechanisms of surfactant effect still require further studies". C62 Why CTAB coarsens the droplets is left open ("by an unknown mechanism"), and the separate roles of $T_g$ and hydrogen bonding in setting the LoC are "somewhat difficult to deconvolute". C38·C49

The AI writer's own judgement.

• The LoCs are brackets. Release was tested at 40, 50 and 60% DL only, so the binary LoC lies between 40 and 50% DL and the LoC with SDS between 50 and 60% DL; ASDs from 20% DL were prepared but no lower-loading release is shown. X6
• The time basis of the percent-release bars in Fig. 9 is unclear: some match the 120 min points of Fig. 8 rather than the stated 60 min. Compare ASDs at matched times from Fig. 8. X2
• Some summary wording goes beyond the bars: "similar" rates at 50% DL (binary about 3.6 vs 6.7 mg/min·cm²), "up to 10 times" faster polymer release at 60% DL (binary about 20-fold), and PVPVA released "almost completely" regardless of loading (about 79–85% in three surfactant ASDs, and over 100% in others without comment). X3·X4·X5
• The morphology–release link is a correlation across loadings. Films were imaged after 5 min of water vapour, the dissolving compacts were not imaged, and CTAB films were not shown. The 50% DL SDS film is described both as slightly heterogeneous and as homogeneous. X13·X12
• The mechanistic explanations (low $T_g$ and no hydrogen bonding, plasticization, surface drug enrichment, SDS delaying separation, TPGS speeding ripening) are proposals; the wet-ASD $T_g$ and the surface composition were not measured in the main text. C47·C48·C53·C58·C59
• Several supporting data sets are in the SI, which was not available for this reading: the SDS concentration series for the LLPS concentration, the size data at lower surfactant levels, the NMR details for each surfactant, the ASD $T_g$ values, the PVPVA content of the droplets and the monolayer calculation. R4·R5·R10·R26·R27·E2
• Some figure references in the text point to the wrong panels of Fig. 11, and the TPGS CMC is given as 40 μg/ml in Table I and "~50 μg/ml" in the text. X11·X1

Overall, the release ranking (only SDS helps at 50% DL; TPGS and Tween 80 slow release; CTAB loses drug) and the solution screen rest on triplicate data with statistics and are solid within this system; the mechanisms that connect them are plausible but untested. R17·R18·R6·X6·X13

Conclusion

This paper asks whether surfactants can raise the drug loading of a PVPVA dispersion and keep the nanodroplets it releases small, using clopidogrel, a slowly crystallizing, low-$T_g$ drug without hydrogen-bond donors. Without surfactant, CPD–PVPVA releases drug and polymer together up to about 40% DL, a high LoC that the authors attribute to the drug's low $T_g$ and lack of interaction with the polymer, and CPD makes PVPVA dissolve about twice as fast as the neat polymer. C46·C47·R18 Of four surfactants built in at 5 g per 100 g, only SDS improved release, extending the LoC to 50% DL, possibly by preventing or delaying water-induced phase separation; TPGS and Tween 80 slowed release, and CTAB agglomerated the droplets and lost drug from solution. C55·C58·C56·C26 In buffer, 50 μg/ml SDS and 4–5 μg/ml TPGS or Tween 80 kept droplet size steady for 4 h, CTAB did not, and NMR showed TPGS solubilizing the drug and moving into the droplets. R6·R24·R12 The authors end cautiously: surfactants "may lead to some improvements", but the mechanisms and a way to predict them still need work. C62