Plasticizers let PVPVA dispersions of high-Tg drugs release well at higher drug loading

Background and motivation

Amorphous solid dispersions (ASDs) raise the apparent solubility of poorly soluble drugs, but they work best only within a narrow composition window. The amorphous drug itself dissolves slowly, much more slowly than a hydrophilic polymer such as copovidone (polyvinylpyrrolidone/vinyl acetate 64, PVPVA). When the drug:polymer ratio is low, the polymer controls release: drug and polymer leave the dispersion together and quickly, the drug exceeds its amorphous solubility and forms drug-rich nanodroplets. When the ratio rises, release slows down, the drug concentration usually stays below the amorphous solubility and no nanodroplets appear. The highest drug loading (DL) at which near-complete, congruent release of drug and polymer is still obtained under non-sink conditions is called the limit of congruency (LoC). (Craig 2002, ref 7; Taylor 2016, ref 9; Indulkar 2019, ref 12; Saboo 2019, ref 14; Saboo 2020, ref 17) C2

A low LoC is a practical problem. If good release requires, say, 90% polymer, a high-dose drug ends up in large or multiple tablets; if the loading is pushed above the LoC instead, the formulation underperforms. The authors note that polymer-controlled release has been "widely reported" only at low drug:polymer ratios, "in particular for high Tg compounds". (He 2015, ref 6; Que 2019, ref 16) C3 For PVPVA dispersions a link between a high drug glass transition temperature ($T_g$) and a low LoC has emerged from the group's earlier release studies: indomethacin-PVPVA has an LoC of 10% DL, whereas its ester analogue with a $T_g$ of 1.7 °C reaches a 2.5-fold higher LoC; atazanavir and ledipasvir have LoCs at or below 10% DL; the lower-$T_g$ drugs ritonavir and miconazole reach 25% DL, and clopidogrel and loratadine up to 40% DL. (Saboo 2020, ref 17; Que 2019, ref 16; Indulkar 2019, ref 12; Yang 2021, ref 18) C4·R15

Several routes to higher drug loading have been tried: polymer combinations, drug ionization and surfactants (refs 19–29). Plasticizers, by contrast, are added to ASDs for processing. They help spray drying produce smooth particles and lower the temperature and melt viscosity of hot-melt extrusion (HME); triacetin, citrate esters, tocopheryl derivatives and surfactants are used, and low-$T_g$ drugs can themselves plasticize polymers in melt-extruded products. (Wan 1992, ref 30; Mendonsa 2020, ref 33; Zhu 2002, ref 35) C5 Plasticizer efficiency is commonly judged by how far a given amount lowers the polymer's $T_g$, $T_g$ being a key indicator of chain mobility. (Immergut 1965, ref 39) C7 To the authors' knowledge, the effect of plasticizers on drug release, and on release as a function of drug loading, had not been studied. C1

The starting point was the group's own earlier work on atazanavir (ATZ). Surfactants raised the LoC of an ASD containing a high-$T_g$ drug, and this was "suggested" to come from their plasticizing effect; a non-surfactant small-molecule additive also improved release at higher loading. (Correa-Soto 2022, ref 37) C6 The same study established that a binary ATZ-PVPVA dispersion releases rapidly and completely at 5% DL but hardly at all at 10% DL, giving an LoC of 5% DL. R2·R1.a Surfactants have drawbacks: they can promote drug crystallization and migrate to interfaces, so other additives that allow higher drug loading are of interest. (Mosquera-Giraldo 2014, ref 46; Indulkar 2020, ref 48; Yu 2022, ref 49) C27

The paper therefore tests one hypothesis: that non-surfactant plasticizers raise the LoC of ASDs containing high-$T_g$ drugs. C8 It uses two high-$T_g$ drugs, atazanavir (ATZ, $T_g$ 104 °C) and ledipasvir (LED, $T_g$ 161 °C), and the lower-$T_g$ ritonavir (RTV, $T_g$ 50 °C) as a control, all in PVPVA. M1·M2·M3 The plasticizers come from two families: the glycerol series (glycerol, glycerol diacetate GDA, glycerol triacetate GTA, glycerol tributyrate GTB) and the citric acid series (citric acid, triethyl citrate TEC, triethyl o-acetylcitrate TEAC, tri-n-butyl citrate TBC, tributyl o-acetylcitrate TBAC). Citrates were chosen because they are widely used, nontoxic and approved for medical plastics, personal care and food contact; the glycerol derivatives were added to widen the chemical range. M5–M13·C9

Background concepts

Limit of congruency and congruent release. The LoC is not a thermodynamic property but an operational threshold from a dissolution test. In this paper a formulation is below its LoC if more than 80% of the drug is released within 120 min at a normalized release rate similar to that of the polymer; "complete release" means more than 80%. P6 Because DL is defined on the drug + polymer basis only (see Methods), a "25% DL" ternary dispersion contains 25 g drug, 75 g PVPVA and an extra 5 g plasticizer. P2

Surface-normalized release rate. Release is measured from a compact of fixed face area that rotates in the medium, so that rates of drug and polymer can be compared directly. The paper defines (Eq. 1):

Here $k$ is the slope of the cumulative mass released (mg) against time (min), $S$ is the compact face area (0.5 cm²) and $x$ is the mass fraction of the component (drug or polymer) in the dispersion. In plain terms: dividing by $x$ puts drug and polymer on the same footing, so that if both leave the compact together, as an eroding block, their normalized rates are equal. Equal drug and polymer rates are the signature of congruent release. E1·P6

Plasticization, free volume and the glass transition. A plasticizer is a small molecule with a low $T_g$ that, in the usual free volume picture, penetrates a polymer, separates its chains and weakens the interchain forces. This raises the free volume (the unoccupied space between molecules), lowers the $T_g$ and raises molecular mobility. (Immergut 1965, ref 39; Marcilla 2017, ref 53) C37 The paper measures plasticization as the drop in $T_g$ of an amorphous drug:plasticizer mixture relative to the neat amorphous drug. P8

How a glassy polymer dissolves. Polymer dissolution is slower and more complex than that of a small molecule. It runs through three successive steps: water (solvent) ingress, which lowers the local $T_g$ and turns the glass into a gel; reptation and disengagement of the chains at the gel–liquid interface; and diffusion of the freed chains away from the interface. Any of the three can limit the rate. For a glassy, high-molecular-weight polymer such as PVPVA under stirring, the authors take disengagement as "most likely" rate-limiting, while noting that it is hard to separate from water ingress, since both depend on water content. (Devotta 1995, ref 51) C33 Below the LoC the drug leaves with the polymer, so whatever speeds polymer dissolution should also speed drug release. C41

Liquid–liquid phase separation and nanodroplets. When dissolved drug exceeds its amorphous solubility, the solution separates into a drug-rich and a water-rich phase; the drug-rich phase appears as nanodroplets. Their formation is a marker of good, polymer-controlled release from an ASD. C2

Methods in detail

The study is a screening design built around one read-out, the LoC from surface-normalized dissolution, followed by three probes of why plasticizers help: $T_g$ depression measured by DSC, water vapour sorption measured gravimetrically, and the dissolution of drug-free PVPVA with plasticizer. A1·P3·P8·P9·R13

Making the dispersions

Drug (ATZ, RTV or LED), PVPVA and plasticizer were dissolved together in methanol, sonicated to a clear solution and rotary-evaporated with the water bath at 55 °C and 170 rpm; the glassy solid was dried under vacuum at room temperature for at least 24 h, ground and passed through a 250 µm sieve. PXRD showed no sharp peaks for any dispersion. P1·R24 ATZ and RTV dispersions received 5 g of plasticizer per 100 g of drug + PVPVA ("5% w/w"), LED dispersions 10 g per 100 g ("10% w/w"); the drug loading is the drug share of drug + polymer only. ATZ 10% DL thus means 10 g ATZ, 90 g PVPVA and, in the ternary dispersion, 5 g plasticizer. P2 Note by the AI writer (not discussed in the paper): on a total-solid basis "5% w/w" is about 4.8 wt% plasticizer and "10% w/w" about 9.1 wt%, and the drug's share of the total solid is slightly below the nominal DL; comparisons with papers that state composition on a total-solid basis need this conversion. X8

Surface-normalized release and the LoC

Each test used a compact made just before the run from 100 mg of dispersion in an 8 mm die at 2500 psi for 60 s, giving a 50 mm² face. It rotated at 100 rpm in 100 mL of 50 mM phosphate buffer, pH 6.8, degassed and held at 37 °C. Samples of 100–200 µL were taken every 5 min to 20 min, every 10 min to 60 min and at 75, 90 and 120 min, with medium replaced each time. P3·M14 ATZ and LED were measured by reversed-phase HPLC at 210 nm, RTV at 241 nm, and PVPVA by size exclusion chromatography at 205 nm, each with a linear calibration (r² 0.999). P4 Release rates were computed with Eq. 1; when a release profile showed a lag, the slope was taken from the linear part after the lag, between 30 and 60 min. Rates are means of three runs, shown with standard deviations. E1·P6·R3 Nanodroplet size was measured by dynamic light scattering (backscatter, 173°, 37 °C) after 90 min. P5

The LoC of a formulation is the highest DL among those tested that meets the release criterion (more than 80% of the drug in 120 min at a rate similar to the polymer's). P6 Note by the AI writer (not discussed in the paper): the paper names the next loading above the LoC only for RTV (35% DL) and, implicitly, for TEC (25% DL); for the four plasticizers that reached 25% DL with ATZ and for GTB with LED (20% DL) it does not say whether higher loadings were tried, so these LoCs are at least the reported values. The binary LoCs of all three drugs are cited from earlier studies rather than re-measured. X6·X7

$T_g$ of drug:plasticizer glasses

To measure how strongly each plasticizer depresses the drug's $T_g$ and whether the two mix, amorphous drug:plasticizer mixtures at 8:1 were prepared by rotary evaporation. They were dried under vacuum for only 1 h to avoid crystallization, then held at 20 °C under nitrogen in the DSC for at least 20 min, cooled to −60 °C and heated at 2 °C/min to 210 °C in modulated mode (±1 °C every 60 s); $T_g$ is the midpoint of the heat-capacity step, mean ± SD of n = 3. P8·R17 The $T_g$ of dispersions was measured in the same instrument with a heat–cool–heat cycle between −20 and 220 °C. P7 Note by the AI writer: the 8:1 mixtures contain no polymer and about 11 wt% plasticizer, more than the ternary dispersions, so their $T_g$ values show the plasticizer's effect on the drug, not the $T_g$ of the dispersions tested for release; the dispersion $T_g$ values are in Table S2, which was not available. X8·R23

Water vapour sorption

About 5 mg of powder was dried at 0% RH and 37 °C for 1 h and then exposed to 95% RH at 37 °C for at least 8 h in a DVS-Advantage instrument; water content and its time derivative (dm/dT) were recorded. P9 The figures show the first 200 min (Figure 10) and 300 min (Figure 11), so the water contents quoted below are uptakes at those times, not equilibrium values. X19

Experimental analysis

Atazanavir: which plasticizers raise the LoC

With 5% w/w plasticizer, every additive raised the ATZ LoC above the binary 5% DL, but by very different amounts (Figures 2 and 4). In the glycerol series the LoC rose to 7.5% DL with glycerol and with GDA, to 10% with GTA and to 25% with GTB. R1.b–e·C10 In the citric acid series citric acid itself raised it "slightly", to 7.5% DL, TEC to 20%, and TEAC, TBC and TBAC to 25%, an increase "up to fivefold". R1.f–j·C12 Table A sets these values beside two properties the paper reports for each plasticizer: its calculated Log P (Figure 1) and how far it lowers the $T_g$ of amorphous ATZ (Table II).

Table A. ATZ-PVPVA dispersions with 5% w/w plasticizer: LoC (Figures 2 and 4), calculated Log P (Figure 1) and $T_g$ reduction of the 8:1 drug:plasticizer glass relative to neat amorphous ATZ at 105.5 °C (Table II, mean of n = 3). Columns from different figures are set side by side by the AI writer. R1·M5–M13·R17

Two patterns stand out. Within the glycerol series, each added ester group raised both Log P and the LoC, and the jump came with the butyrate ester GTB (Log P 2.58). R1.b–e·M5–M8 The citric acid series did not follow Log P as neatly: TEAC, with a Log P of only 0.35, reached the same 25% DL as TBC and TBAC (Log P 2.83 and 3.06). R1.f–j·M10–M13·X15 The paper's own summary is that the lipophilic plasticizers "appear more effective at improving release at higher drug loadings" than the hydrophilic ones. C44 The $T_g$ drop, on the other hand, does not order the LoCs at all: GTA lowered the ATZ $T_g$ most (43.3 °C) but reached only 10% DL, while TEAC lowered it least (30.4 °C) and reached 25% DL. R17·R1·X14 The authors themselves list "the lack of a precise correlation between the extent of plasticization of the ASD and the polymer release rates" among the outstanding questions. C45

Release rates: congruent, and faster than PVPVA above the old limit

Raising the LoC is one thing; the paper also asks how fast drug and polymer leave. Table B collects the normalized release rates read from Figures 3 and 5 (approximate; the paper prints no values).

Table B. Normalized release rates (mg/min·cm²) of ATZ and PVPVA, read from Figures 3, 5A and 5B (approximate; means of n = 3). R3·R6·R7

In every dispersion the drug and polymer rates are close to each other, which is what congruent, polymer-controlled release looks like. R3·R6·R7 With GTA at 10% DL and GTB at 25% DL both components left faster than neat PVPVA, and the authors note that these two plasticizers raised "not only" the LoC but also the release rate. R3·C11 At 20% DL the three lipophilic citrate esters gave ATZ rates of about 4.6–5.1 mg/min·cm², well above neat PVPVA (about 2.7), while TEC, whose LoC is 20%, did not exceed it. R6·C13 At each formulation's own LoC the rates fell back: the authors describe them as "generally similar" to neat polymer, and for TEAC and TBAC at 25% DL they are at or below it (about 2.0 and 2.4 for ATZ). R7·C13 For TEC at 20% DL and TEAC and TBAC at 25% DL, release began only after a lag of about 20 min, followed by almost complete release of drug and polymer; TEAC at 20% DL showed no lag. R4 In all cases the drug exceeded its LLPS concentration and formed drug-rich nanodroplets of 500 to 1500 nm; the paper does not specify which dispersions were sized. R5·X18

Note by the AI writer (not discussed in the paper): the conclusion section states that "in all cases" the dissolution rate was higher than that of the neat polymer for drug loadings below the LoC. The figures do not show rates below the LoC for most formulations, and at the LoC the TEC, TEAC and TBAC dispersions are at or below the neat polymer. The statement holds for the lipophilic esters at 20% DL, for GTA and GTB, and for the LED dispersions, but not for every bar in Figure 5B. No statistical test is reported, and the neat PVPVA standard deviation (about 0.5) is as large as several of the differences. C47·X5·X20

Ledipasvir: twice the plasticizer

LED has a dry $T_g$ of about 160 °C and a binary LoC of 5% DL. M2·R11.a If low mobility is what limits release, a plasticizer "is expected" to raise its LoC. C14 The first screen, at 10% DL with 5% w/w of the citrate plasticizers or GTB, gave minimal release except with citric acid, which released completely (data not shown). R8 Because the LED $T_g$ is about 55 °C above that of ATZ, the authors "rationalized" that more plasticizer was needed and raised the level to 10% w/w. C15 Table I of the paper shows what this did for TEC and TEAC (Table C).

Table C. Glass transition and extent of release of 10% DL LED dispersions with 5 or 10% w/w TEC or TEAC (from Table I; the plain dispersion's value is from ref 16). R9

Doubling the plasticizer lowered the dispersion $T_g$ by 10.7 °C with TEC and 13.3 °C with TEAC; the text rounds these to 10 and 14 °C. R9·C16·X4 With TEAC the release by 90 min jumped from 20 to 94%, with TEC only from 12 to 20%. R9·C16 At 10% w/w and 10% DL, TEAC, TBC and GTB released LED completely, but TEC and TBAC released less than 30%. R10 The LED LoC then rose from 5 to 15% DL with TEAC or TBC and to 20% DL with GTB (Figure 6). R11·C17 At these LoCs the normalized release rates of LED were about 4.3 (TEAC), 7.0 (TBC) and 5.3 mg/min·cm² (GTB), with PVPVA at about 4.6, 6.2 and 4.8, all above neat PVPVA (Figure 7). R12·C17 Note by the AI writer: Figure 7 carries a second, unexplained dashed line at about 2.1 mg/min·cm² in addition to the neat-PVPVA line at about 2.7; the caption describes only one. X10 Citric acid, the only plasticizer that released 10% DL LED completely at 5% w/w, was not carried forward to higher loadings, so its LoC with LED is unknown. X16

The ritonavir control

If plasticizers help by restoring mobility, a drug that is already mobile should gain little. RTV has a $T_g$ of 50 °C and a binary LoC of 25% DL, so its molecular mobility "is expected to be less of a limitation". C19·R15 With 5% w/w GTB or TBC, RTV dispersions released completely at 30% DL but not at 35% DL (Figure S4), which the authors describe as "confirming the hypothesis of limited increase in LoC for the low Tg compounds". R14·C20 Figure 9 puts the three drugs side by side: TBC raised the LoC fivefold for ATZ, threefold for LED and 1.2-fold for RTV; GTB fivefold, fourfold and 1.2-fold. R16·C21 Note by the AI writer: the RTV result rests on two loadings 5% DL apart, so its true LoC lies anywhere between 30 and 35%; ATZ and RTV were tested with 5% w/w plasticizer and LED with 10% w/w; and all three binary LoCs come from earlier studies. The ranking is clear, the fold values are approximate. X9·X7

$T_g$ depression

Every plasticizer lowered the $T_g$ of the drug in 8:1 glasses, which the authors read as "suggesting miscibility". C22 For ATZ ($T_g$ 105.5 °C) the reductions were 32.7 °C with glycerol, 43.3 with GTA, 34.0 with TEC, 30.4 with TEAC, 33.3 with TBC and 30.9 °C with TBAC (Table II). For LED ($T_g$ 162 °C) they were 53.1 °C with TEC, 47.6 with TEAC, 51.1 with TBC and 48.9 °C with TBAC (Table III). R17·R18 The citrates thus gave similar drops, about 30 °C for ATZ and 50 °C for LED. C22 Note by the AI writer: Table III prints the acetylated citrates as "ATEC" and "ATBC"; these are the plasticizers called TEAC and TBAC elsewhere. The neat-drug $T_g$ values also vary slightly across the paper (ATZ 104 or 105.5 °C; LED 160, 161 or 162 °C). X2·X3 The plasticized dispersions also had lower $T_g$ values than the binary dispersions (Table S2), and PVPVA with GTB or TBC about 30 °C lower than neat PVPVA. R23·C30

The polymer alone, and water uptake

Two control experiments probe where the plasticizer acts. First, compacts of PVPVA without drug but with 10% w/w GTB or TBC dissolved faster than neat PVPVA (Figure 8): about 4.8 and 6.1 mg/min·cm² against about 2.7. R13 The authors describe this as "as much as twice" the neat rate and say that Figure 8 "confirms" that plasticizers increase the polymer dissolution rate. C18 Second, water vapour sorption showed only modest changes (Table D).

Table D. Water content at 200 min and the peak rate of mass change (dm/dT, axis "%mass") for PVPVA and ATZ dispersions, read from Figure 10 (approximate). R19·R20·R21

Water uptake fell as more ATZ was added, from about 32% at 5% DL to about 23% at 30% DL against about 41% for PVPVA, and the peak sorption rate, reached at about 6–7 min, fell with it; the authors attribute this to the lipophilicity of ATZ. R19·C23 With 5% w/w TBC, the 10, 20 and 30% DL dispersions held about 28, 27 and 27% water, which the authors call "similar" to the binary dispersions; the ternary values vary little with DL, whereas the binary ones fall, so at 30% DL the TBC dispersion holds about 4 points more water. R20·C24·X13 At 20% DL, TEC changed uptake hardly at all, TBC and TEAC increased it slightly and GTB more clearly, both in amount and rate. R21·C25 In drug-free PVPVA, 10% w/w TBC or GTB lowered uptake at 300 min from about 41.8 to about 34.4 and 32.0% (Figure 11), more than replacing a tenth of the polymer would explain; the authors expect this from the plasticizers' lipophilicity (Log P around 2.5). R22·C26 Overall the authors see "relatively minor changes" in the rate and extent of water uptake, and so look beyond faster water ingress for the mechanism. C34

Discussion: the proposed mechanism

The authors' explanation starts from the fact that, below the LoC, drug release is coupled to polymer dissolution, and that for PVPVA the slow step is "most likely" the disengagement of chains from the gel. C33·C41 At a "simplistic level", they write, a plasticizer "may increase the mobility of the ASD components", which promotes polymer dissolution, and "may also affect" the drug-rich barrier that earlier work postulates to form above the LoC. C31 Since water uptake changed only modestly, they set aside faster water ingress: the plasticizing gain in sorption rate "may be" offset by the plasticizers' unfavourable interaction with water. C34·C35

To make the link to chain mobility concrete, they turn to the polymer-dissolution work of Devotta and co-workers, who proposed that the chain disengagement rate $k_d$ is directly proportional to the time-dependent chain mobility $m_p$, and that the maximum mobility a chain can reach at a given concentration depends on the fractional free volume of the gel, $f_g$ (unnumbered equation, p. 788):

Here $A_d$ is a pre-exponential constant and $B_d$ a parameter related to the size of the mobile species. In plain terms: chain mobility rises steeply, exponentially, as the gel gains free volume. The free volume of the gel is taken as the sum of each component's free volume weighted by its volume fraction, so adding a low-$T_g$ plasticizer raises $f_g$, and the polymer "is predicted to dissolve faster"; where drug release is coupled to polymer dissolution, below the LoC, this "will translate" into faster drug release. E2·E3·E4·C40·C41 The same picture explains why release falls as drug loading rises: the drug is much less hydrophilic than PVPVA, so the gel holds less water, and a high-$T_g$ drug contributes less free volume than a low-$T_g$ component; both lower $f_g$. C42 The paper does not fit or evaluate the equation; it is used as a qualitative argument. E2

The evidence for this chain is indirect. The direct support is Figure 8: plasticizer alone makes drug-free PVPVA dissolve faster, by as much as twice, with a $T_g$ about 30 °C lower. R13·C18·R23 The authors find it "reasonable to suggest" that faster disentanglement is "one mechanism"; the conclusion section calls it "the suggested mechanism". C39 They also say that the free volume explanation is "consistent with" the faster polymer release in some plasticized systems but that "there are likely to be additional factors". C43 Two such factors are named: the lipophilic plasticizers "appear more effective" at higher drug loading, which "suggests that their interaction with and impact on the drug may also be important factors, particularly in the presence of water"; and the link between plasticizer and the drug-rich barrier needs "a more in-depth evaluation". C44·C32

What the paper suggests for formulation. The authors present plasticizers as "a potential new strategy to increase drug loading in ASDs for high Tg compounds with a low tendency to crystallize", one that "may help" with the high excipient burden of ASDs. C48 They see "tremendous potential" for plasticizers as an alternative to surfactants, and expect them to "serve dual functions" in PVPVA formulations made by hot-melt extrusion, where high-$T_g$ drugs otherwise need high process temperatures. C28·C29 They add that the release gain from a lower ASD $T_g$ needs "to be balanced with the increased risk for drug crystallization". C46

Note by the AI writer (not discussed in the paper): read as a formulation guide, the data say three things. A small amount of a lipophilic ester plasticizer can move a high-$T_g$ drug from 5% to 20–25% DL in PVPVA, a large practical gain. R1·R11 The amount needed grows with the drug's $T_g$ (5% w/w for ATZ, 10% w/w for LED), and neither Log P nor the $T_g$ drop of the drug:plasticizer glass picked the best plasticizer for every drug: TBAC was among the best for ATZ but released less than 30% of LED at 10% DL. R9·R10·X14·X15 And at the LoC itself several dispersions released after a lag and no faster than neat polymer, so the LoC is an edge rather than an operating point. R4·R7

Limitations

Limitations stated by the authors. The authors acknowledge that the water sorption experiment measured transport from the vapour phase and that penetration of bulk liquid water should be studied. C36 They state that "outstanding questions remain" about the lack of a precise correlation between the extent of plasticization and polymer release rates, and about the origin of the release decline above a certain drug loading. C45 They call the free volume explanation incomplete ("there are likely to be additional factors") and the connection to drug-rich barrier formation in need of further work. C43·C32 They also point out that lowering the ASD $T_g$ may raise the crystallization risk. C46

The AI writer's own judgement (1): what the LoC values can and cannot carry. The LoC gains are large and consistent, but three features of the design limit how precisely they can be read. The binary LoCs (ATZ 5%, LED 5%, RTV 25%) come from three earlier studies and were not re-measured with the same materials and protocol, so each fold increase divides a measured value by a cited one. X7 Whether loadings above 25% (ATZ) or 20% (LED) were tested is not stated, and the release profiles are in a supplement that was not available; the top LoCs may be lower bounds. X6 The RTV control rests on two loadings, at a lower plasticizer level than LED. X9 The ranking of drugs and plasticizers is solid; the exact fold values are not.

The AI writer's own judgement (2): the mechanism is plausible but untested. No quantity in the mechanistic chain was measured in the hydrated gel: not free volume, not chain mobility, not disengagement rate. C39·C41 The $T_g$ values come from 8:1 drug:plasticizer glasses without polymer, at about twice the plasticizer content of the dispersions, and they do not rank the LoC gains (GTA: largest drop, LoC 10%; TEAC: smallest drop, LoC 25%). X8·X14 The drug-free control shows that plasticizers speed PVPVA dissolution, but at the LoC several drug-containing dispersions released no faster than neat polymer, so faster polymer dissolution alone does not explain where each LoC falls. R13·R7·X5 The "lipophilic is better" reading has exceptions in both directions: TEAC (Log P 0.35) matched the most lipophilic esters with ATZ, TBAC (Log P 3.06) failed with LED, and only hydrophilic citric acid released 10% DL LED completely at 5% w/w. X15·X16

The AI writer's own judgement (3): smaller points in the source. The text rounds the LED $T_g$ drops to 10 and 14 °C where Table I gives 10.7 and 13.3 °C. X4 Figure 1 prints a molecular weight of 129 g/mol for citric acid, whose drawn structure corresponds to 192.12 g/mol, and its caption calls TBAC "tributyl citrate". X1·X2 The discussion cites Figure 8 for drug-containing dispersions where Figure 7 is meant, the Figure 10 caption omits TBAC, and Figure 7 has an unexplained second dashed line. X11·X12·X10 No statistics are reported, and the physical stability of the plasticized dispersions was not examined, although the strategy is limited to drugs "with a low tendency to crystallize". X20·X21 None of these changes the main finding.

Conclusion

The paper shows, for two high-$T_g$ drugs in copovidone, that a few percent of a non-surfactant plasticizer can raise the limit of congruency three- to fivefold: from 5% to 25% DL for atazanavir with 5% w/w TEAC, TBC, TBAC or GTB, and from 5% to 15–20% DL for ledipasvir with 10% w/w TEAC, TBC or GTB. R1·R11·R16 The hydrophilic glycerol, GDA and citric acid helped little, and the lower-$T_g$ ritonavir gained only from 25% to 30% DL. R1·R14 The plasticizers lowered the drug $T_g$ by 30–53 °C, made drug-free PVPVA dissolve faster and changed water vapour uptake only slightly. R17·R18·R13·C34 The authors propose, with the hedge "it is reasonable to suggest", that the plasticizer speeds up the disentanglement of polymer chains in the gel; this mechanism was not measured, and the size of the $T_g$ drop did not predict the size of the gain. C39·C45·X14 For formulators, the result opens a way to lower the excipient burden of high-$T_g$ drug ASDs, provided that each drug is screened for its own best plasticizer and that physical stability is checked. C48·C46