How drug loading switches ritonavir–copovidone dispersions from polymer-controlled to drug-controlled release

Background and motivation

Many drug candidates dissolve poorly in water, and amorphous solid dispersions (ASDs), in which the amorphous drug is mixed molecularly with a polymer, are a common way to deliver them. The amorphous form has a higher thermodynamic activity than the crystal, so an ASD can produce a supersaturated solution; the higher activity raises the rate of transport across a membrane, which "can potentially" improve absorption. Micellar solubilization and complexation also raise the dissolved concentration, but by raising the solubility of the crystal, so they do not raise the activity (Hancock and Parks 2000; Raina et al. 2014, 2015). C11

A second feature of some ASDs is liquid–liquid phase separation (LLPS) during dissolution. When the drug concentration passes the amorphous solubility, the excess drug separates into initially nano-sized amorphous drug-rich droplets. Because the amorphous solubility is the highest solute activity the drug can reach, the driving force for membrane transport is at its maximum from that concentration upward, and the nanodroplets can resupply absorbed drug as a reservoir, an effect earlier shown in a flow-through diffusion cell (Ilevbare and Taylor 2013; Indulkar et al. 2016). LLPS has mostly been seen with ASDs of low drug loading (DL) (Alonzo et al. 2011; Purohit and Taylor 2017; Jackson et al. 2016). C12

The polymer has two jobs. It can inhibit crystallization in the solid and in solution, and it can release the drug faster than the neat amorphous drug would dissolve, particularly at low DL, which helps the drug pass its amorphous solubility. At high DL impaired release, without droplets, has been seen for some compounds. C13 Water can undo the polymer's help: an otherwise miscible ASD can split into drug-rich domains inside a polymer-rich phase, a process called water-induced amorphous–amorphous phase separation (AAPS), whose extent depends on polymer, DL and water content; the polymer-lean domains are prone to crystallize, and a loss of release enhancement "would also be anticipated" (Rumondor et al. 2009, 2011; Purohit and Taylor 2015; Marsac et al. 2010). C14 For powdered ritonavir ASDs made with different polymers, Purohit and Taylor (2017) had shown that ASDs resistant to AAPS dissolved faster and gave LLPS, ASDs that underwent AAPS dissolved poorly and formed no nanodroplets, and the highest DL that still gave LLPS depended on the polymer. R23

Classic dissolution theory did not settle the matter. The Higuchi model for two-component solids predicts that drug and polymer leave together only at one "critical mixture ratio", and that otherwise the surface enriches in one component that then controls dissolution; for a drug and polymer of very different solubility, polymer control should be limited to very low drug fractions. Corrigan's review noted that these models "appear to" describe high-DL dispersions, where a drug-rich surface controls release, but not low-DL ones, where simultaneous release of drug and polymer had been reported over a range of polymer-rich compositions (Simonelli et al. 1969, 1976; Corrigan 1985). C23·C24

The authors therefore set out to probe how the mechanism of drug release from an ASD changes with drug loading: what role the polymer plays, how drug and polymer release rates are linked, and whether LLPS occurs. They used ritonavir (RTV) and polyvinylpyrrolidone/vinyl acetate (PVPVA, copovidone) and measured release from compacts in Wood's intrinsic dissolution apparatus. Holding the dissolving area constant removes the changing surface area of powders and disintegrating tablets, which the authors consider closer to tablet release in vitro and in vivo, and it leaves a partially dissolved surface that can be examined. C15

Background concepts

Congruent release and the limit of congruency. Release is congruent when drug and polymer leave the dosage form simultaneously, that is, at the same fractional rate, so that the percent-release curves of the two components overlap; incongruent release means they leave at different rates. The paper names the highest DL at which congruent release is observed the limit of congruency (LoC). The definition is qualitative: the paper gives no numerical tolerance for how far apart the two curves may be. C2·X7

Amorphous solubility and LLPS. The amorphous solubility is the highest concentration the amorphous drug can sustain in solution; above it the drug separates as a second, drug-rich liquid phase (LLPS), seen as nanodroplets. For ritonavir the authors use a value of about 30 µg/mL, cited from earlier work; they also cite a glass transition temperature of the water-saturated drug ("wet Tg") of about 0 °C, which makes amorphous ritonavir a supercooled liquid at 37 °C. M1

AAPS. Water-induced amorphous–amorphous phase separation is the splitting of a mixed drug–polymer glass into a drug-rich and a polymer-rich amorphous phase when water enters. The authors explain it in terms of interactions: the ASD stays mixed while drug–polymer attraction, helped by the entropy of mixing, outweighs drug–drug and polymer–polymer attraction; water interacts with the hydrophilic polymer and can disrupt drug–polymer contacts, so the hydrophobic drug self-associates into nano- to micron-sized domains. Water also lowers the glass transition temperature (Tg) and so raises molecular mobility. C27

Dissolution equations. The paper starts from the Noyes–Whitney equation (eq 1):

Here $C$ is the concentration in solution at time $t$, $C_s$ the solubility of the dissolving solid and $k$ a proportionality constant. In plain terms: a solid dissolves faster the further the solution is from saturation. Combined with diffusion-layer theory, $k$ becomes explicit in the Nernst–Brunner equation (eq 2): E1·C22

where $D$ is the diffusion coefficient of the solute, $A$ the surface area of the solid, $V$ the volume of the medium and $h$ the thickness of the diffusion layer. In plain terms: for a fixed area and stirring, the rate is set by the gap between the solubility of whatever solid form sits at the surface and the bulk concentration. The authors stress that $C_s$ must be the solubility of the surface form, the amorphous solubility for an amorphous surface, and that the rate falls to zero as $C$ approaches $C_s$. This is the basis of their test in drug-saturated medium. E2·C22

For the regime in which the polymer controls dissolution, the paper writes the drug's dissolution rate in terms of the polymer's (an unnumbered equation in the Discussion):

where $m_{\mathrm{d}\mathrm{r}\mathrm{u}\mathrm{g}}$ and $m_{\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{y}\mathrm{m}\mathrm{e}\mathrm{r}}$ are the amounts of drug and polymer present. In plain terms: when there is too little drug to form a layer of its own at the surface, the dissolving polymer carries the drug out in proportion to their amounts, so the drug can leave far faster than neat amorphous drug would dissolve. The authors write that this change of mechanism "is thought to occur because there is insufficient drug to form an intact layer at the surface". E3·C25·C26

Note by the AI writer (not discussed in the paper). The relation can be checked against the paper's own data in integrated form: for the 10% DL tablet, about 67 mg of PVPVA had left by 20 min, and 67 × 10/90 gives about 7.4 mg of ritonavir, close to the about 7.6 mg read from Figure 2A. E3·R1.a·R2.a

Methods in detail

The study has one drug, one polymer and seven drug loadings, and it combines three kinds of measurement: release of drug and polymer from constant-area tablets, release into a medium already saturated with drug, and characterization of the tablet surface and of thin films. A1·A2·A3·A4

Contribution 1: Release of drug and polymer from a fixed dissolving surface

Problem. Release from powders or disintegrating tablets mixes two effects, the mechanism and the changing surface area. Measuring the polymer as well as the drug is needed to say whether the two leave together. C15

What they did. RTV and PVPVA were dissolved together in methanol at 20% w/v solids, about 10 g per batch, and the solvent was removed in a rotary evaporator with a 50 °C bath, followed by 30 min more under vacuum and overnight drying in a vacuum oven; the powder was milled and sieved to 100–150 µm. ASDs were made at 10, 20, 25, 30, 35, 40 and 50 wt% drug. P1·A1 For each test 100 mg of powder was compressed at 1600 lbs in the 8 mm die of Wood's intrinsic dissolution apparatus, which also served as the holder, so that only one tablet face was exposed. The die was rotated at 100 rpm in 100 mL of 50 mM phosphate buffer, pH 6.8, at 37 °C. The compacts eroded from that face without disintegrating, so the 1 mL samples, replaced with fresh medium, were assayed without filtration. Three tablets were run per loading, and neat amorphous RTV and neat PVPVA tablets served as references. P2 Ritonavir was measured by HPLC with UV detection at 210 nm (C18 column, 0.1% trifluoroacetic acid in water and acetonitrile 55:45, linear from 0.1 to 10 µg/mL) and PVPVA by size exclusion chromatography with UV detection at 205 nm (A2500 column, phosphate-buffered saline pH 7.4 and methanol 70:30, linear from 5 to 200 µg/mL). P3·P4 Sampling ran to 120 min for most loadings, to 180 min for 40 and 50% DL and to 300 min for 35% DL; the paper does not state how percent release was normalized. P2·X12

Why it works. With a constant area, a difference in release rate between loadings reflects a difference in mechanism, and two separate assays show directly whether the drug follows the polymer. The partly dissolved face can be ejected and analysed. C15

Contribution 2: A drug-saturated medium as a test of what controls release

Problem. Release at the polymer's rate suggests polymer control, but a direct test is needed to tell it apart from drug control. C16

What they did. The 10, 30 and 50% DL tablets were run as above in a medium presaturated with ritonavir at its amorphous solubility, about 30 µg/mL. P5 The paper does not say how the medium was presaturated or how the dissolved drug already present was subtracted. X8 Separately, the medium of a 10% DL run was sampled at 30 min, when dissolution was nearly complete, diluted two-fold with buffer and sized by dynamic light scattering (backscatter at 173°, Z-average). P6

Why it works. By eq 2, a drug-rich amorphous surface has no driving force when the medium is already at the amorphous solubility, so drug-controlled release should stop; release governed by the dissolving polymer should be unaffected. E2·C42·C43

Contribution 3: The state of the tablet surface and of wet films

Problem. The release data alone cannot show whether the surface changes. C6

What they did. Tablets were ejected from the die at chosen times, blotted, and examined by attenuated total reflectance infrared spectroscopy (ATR-IR, 64 scans, 4 cm^-1 resolution) against amorphous RTV made by melt quenching and against PVPVA; the 20% DL tablet was examined at 0 and 5 min and the 50% DL tablet at 0, 2, 5 and 15 min. For scanning electron microscopy (SEM) the tablets were vacuum-dried overnight, coated with platinum and imaged at 5 kV (20% DL unexposed and after 5 min; 50% DL unexposed and after 15 min). P7·P8 The methods name 10, 30 and 50% DL for IR and 10 and 50% DL for SEM, while the results and Figures 5 and 6 show 20 and 50% DL. X1 To look for phase separation, RTV–PVPVA films with 0.1% w/w Nile red were spin-coated on quartz from methanol, dried, imaged by fluorescence microscopy, dipped in water for a few seconds, blotted and dried with compressed air, and imaged again; Nile red fluoresces weakly in hydrophilic surroundings and strongly in hydrophobic ones. Films were prepared at 10–50% DL; images of 20, 30 and 40% DL are shown. Dipped films were also checked for birefringence under polarized light. P9·P10·C20

How it differs from earlier work. The earlier ritonavir study cited by the authors used powdered ASDs; here the same constant-area tablets give drug release, polymer release and surface composition. R23·C15

Experimental analysis

The references: a fast polymer and an almost insoluble amorphous drug

A neat PVPVA tablet released about 46 mg in 20 min and about 81 mg in 40 min, its last plotted point, whereas a neat amorphous ritonavir tablet released essentially nothing in 120 min (Figure 2). R2.h·R1.h The text justifies reading Figure 8 at 60 min by saying that PVPVA alone is completely released in this time, but Figure 2B does not plot the neat polymer beyond 40 min. R2·X4

10–25% DL: drug and polymer leave together

At 10, 20 and 25% DL the drug and polymer curves in Figure 3A–C overlap, and both components were released completely. The 10% DL tablet was essentially done by 30 min (RTV about 99.5%, PVPVA about 94%), the 20% DL tablet by 40 min (about 105% and 94%) and the 25% DL tablet by 60–75 min (RTV about 97% at 60 min and 104% at 75 min, PVPVA about 100% and 105.5%). R3.a–f The polymer left these ASDs at about the dissolution rate of neat PVPVA: by 20 min the 10% DL tablet had released about 67 mg of PVPVA against about 46 mg for the neat polymer tablet (Figure 2B). R4·R2.a·R2.h The drug, which barely dissolved on its own, was carried out with the polymer: the 25% DL tablet had released about 24 mg of ritonavir by 60 min. R1.c·R1.h Within this range release slowed as loading rose; at 20 min about 77%, 78% and 46% of the drug had been released at 10, 20 and 25% DL. R3.a·R3.c·R3.e

Note by the AI writer (not discussed in the paper). At 25% DL the drug trails the polymer in mid-release (about 60% against 73% at 30 min, 85% against 99% at 50 min), and some values exceed 100%; with no numerical criterion for congruency, "congruent" here means overlapping curves within the method's spread. X7·X12

The cliff above 25% DL

Figure 8 compares the percentage released at 60 min (Table A). Release of both components is near complete up to 25% DL, drops at 30% DL, and nearly vanishes from 35% DL. The authors describe this as a dramatic drop, a "falling off a cliff" effect, above 25% DL, and take 25% DL as the LoC, the highest loading with congruent release. R11·C2·C31

Table A. Release at 60 min and the authors' three categories (read from Figure 8, approximate; categories from the Discussion). R11·C32

The authors stress that the range of loadings over which the mechanism changes "is very narrow" and that the change alters drug release by orders of magnitude; at 60 min, for example, about 97% of the drug had left the 25% DL tablet and about 0.8% the 40% DL tablet. C52·R11.c·R11.f Because 25% is the last congruent loading tested and 30% the first incongruent one, the threshold itself lies somewhere between the two. X7

30% and 35% DL: the polymer outruns the drug

The 30% DL tablet (Figure 3D) released slowly for the first 15 min, with a lag, then faster; drug and polymer stayed together to about 40 min (RTV about 14.5%, PVPVA about 20%). After that PVPVA pulled ahead (about 62% against 29% at 60 min) and was nearly complete at 120 min (about 96%), while the drug levelled at about 48.5%. The authors describe PVPVA leaving a drug-rich phase behind in the tablet after about 90 min. R5·R3.g–h The 35% DL tablet (Figure 3E) released both components slowly and incongruently over 300 min, reaching about 49% PVPVA and 11% RTV; its drug left faster than neat amorphous RTV but never reached the amorphous solubility. R6·R3.i–j

Two statements in the text do not match the figures. The Results say that less than 10% of each component was released at 40 and 50% DL, but the 40% DL polymer reached about 11% at 150 min and 13% at 180 min (Figure 3F). The Discussion says that less than 15% of each component was released by 180 min for 35–50% DL, but the 35% DL polymer had reached about 33.5% at 180 min (Figure 3E). R7·R8·X2 The 35% DL tablet is also placed differently in different parts of the paper: intermediate with 30% DL in the Results, in category 3 with "mixed behavior" in the Discussion, and left out of the high-loading group in the abstract. X13

40% and 50% DL: drug-controlled release

At 40 and 50% DL very little dissolved (Figure 3F, G): in 180 min the 40% DL tablet released about 3% of its drug and 13% of its polymer, and the 50% DL tablet about 0.65% and 6%. R3.k–n In mass terms both released at most about 0.5 mg of ritonavir in 120 min, close to the neat amorphous drug, which the authors describe as drug release similar to that of neat amorphous RTV. R1.f–h·R7·C4 The amorphous solubility was never reached and no nanodroplets formed. R7

Which loadings produced nanodroplets

Figure 2A plots the amorphous solubility as a dashed line at about 3.3 mg of ritonavir in the 100 mL medium. The 10, 20 and 25% DL tablets crossed it within about 5–10 min (for example about 3.5 mg at 10 min for 10% DL), the 30% DL tablet between 30 and 40 min, while the 35% DL tablet reached only about 1.5 mg by 120 min and the 40–50% DL tablets stayed near zero. R1·R20 The media of the 10–25% DL runs were visibly turbid, and DLS of the 10% DL medium at 30 min found nanodroplets of 218 ± 15 nm (Z-average). R9 No crystallization was seen in solution or on the compacts in any release experiment. R10 The authors conclude that LLPS was observed at 10–30% DL, where release was at least partly controlled by the polymer and congruent or close to congruent. C49·R20

Note by the AI writer (not discussed in the paper). Only the 10% DL sample was sized; nanodroplets at 20–30% DL are inferred from turbidity and from concentrations above the amorphous-solubility line. X5

The drug-saturated medium

In buffer presaturated with ritonavir (Figure 4, Table B) the 10% DL tablet released drug as fast and as completely as in plain buffer, and the 30% DL tablet kept releasing. The 50% DL tablet released "absolutely no" drug into the saturated medium, against a small release into plain buffer shown in the inset. R12·R13

Table B. Ritonavir released into plain and drug-saturated buffer (read from Figure 4 and its inset, approximate; the inset axis is unlabelled). R12

The authors read this as a clean split. At low loading, the drug concentration in solution does not affect release, "confirming" that the polymer controls it; at high loading, a ritonavir-rich surface has no driving force in a saturated medium, "in agreement with" eq 2. C43·C42

Note by the AI writer (not discussed in the paper). The text calls the 10 and 30% DL profiles in the two media similar, but at 30% DL the saturated medium gave more release (about 40.5% against 29.5% at 120 min). The plain-buffer 30% DL series in Figure 4 also released less than the 30% DL series of Figure 3D and Figure 8 (about 29.5% against 48.5% at 120 min), so the extent of release at this loading varied about 1.6-fold between runs. The qualitative point, that release continues in saturated medium, is unaffected. X3

The tablet surface and the wet films

The reference spectra separate the components: amorphous ritonavir has a peak at about 1625 cm^-1, which the authors assign to the amide linkage, and a band between 1570 and 1470 cm^-1; PVPVA has a peak at 1660 cm^-1 (the vinylpyrrolidone carbonyl), a shoulder at 1287 cm^-1 and further differences between 1400 and 1475 cm^-1. R14 The surface results are summarized in Table C. R15·R16·R17

Table C. Tablet-surface observations (from Figures 5 and 6 and the text; image features read by the AI). R15·R16·R17

The authors conclude that the 50% DL surface becomes enriched in drug and depleted in polymer within 5 min, and that the drug-rich surface layer "is confirmed" by SEM and IR, the porous surface being consistent with loss of polymer; for 20% DL they see no change in surface composition or morphology. C18·C19·C41·C17 In the films (Figure 7), all three loadings were uniform before water. After dipping, the 20% DL film stayed uniform, while the 30% and 40% DL films showed discrete domains of intense Nile red emission, sparse at 30% and dense at 40%; the authors read these as water-induced AAPS forming hydrophobic, drug-rich domains ("indicates"). No birefringence was seen, so no crystals had formed. R18·R19·C20·C21

Note by the AI writer (not discussed in the paper). The SEM tablets were vacuum-dried overnight, so the flaky texture may include drying effects; the IR spectra, taken on blotted wet tablets, independently show the ritonavir-rich surface. The low-loading surface was examined at 5 min only, while its release continued to about 40 min. X11·X9·X10

Discussion

The authors' mechanism: a race between phase separation and dissolution

The authors argue that AAPS and dissolution run in parallel at the hydrating surface and that their relative rates decide the outcome: AAPS slower than dissolution leaves release unaffected, AAPS faster reduces it. Higher-DL ASDs "can be expected" to separate more readily and faster because they are closer to the drug's miscibility limit in the polymer; low-DL ASDs may also separate with a very hydrophilic polymer, but more slowly, depending on drug–polymer interactions, and they may hold too little drug to form a barrier. C30·C10 They list the possible outcomes: a drug-enriched surface that releases at the rate of the neat amorphous drug (or of the crystal, should the drug-rich phase crystallize); a drug-rich phase too small to cover the surface, giving incongruent release; a fully miscible surface releasing congruently at the polymer's rate; or a partially miscible surface releasing at an average weighted by surface fraction ("AAPS is anticipated to impact dissolution"). C29 Figure 9 summarizes these scenarios. C53 Applied to ritonavir–PVPVA, the chain reads as follows.

For the low loadings, the authors write that congruent release "suggests the presence of interaction between the hydrophobic drug and the hydrophilic polymer" that is maintained during hydration, at least over the dissolution timescale, and hinders AAPS; release at the polymer's rate means dissolution is "largely controlled" by the polymer, with an interface composition that stays constant. The unchanged IR and SEM of the 20% DL surface and the uniform 20% DL film after water "support" this. C36·C37·C38

For the high loadings, release close to that of neat ritonavir "suggests that the initially miscible ASD undergoes AAPS during dissolution rather quickly", enriching the surface with drug at very early stages. The authors add that the drug-rich nanodomains formed first "can undergo coalescence to form large contiguous regions" covering the surface, ritonavir being a supercooled liquid at 37 °C with a wet Tg of about 0 °C. C39·C40·M1 At the solid–solution interface low-molecular-weight drug molecules "can move laterally" and enrich the dissolving surface, creating a barrier if there is enough drug to form a contiguous layer. C28

For 30% DL, the initial slow but congruent phase "suggests that some time is required for wetting", and this slow phase "may provide sufficient time for AAPS to commence"; drug release faster than neat ritonavir but slower than the polymer "suggests" partial AAPS with a drug-enriching surface, and the separated drug "may not be sufficient" to form a complete drug-rich interface. At 35% DL, the authors state, AAPS is faster than dissolution, so release stops at about 50% PVPVA and 10% RTV. C44·C45·C46·C47

What sets the switch point, in their reading, is drug–polymer interaction: "It is likely that key interactions between RTV and PVPVA in presence of water are saturated around 25-30% DL", leaving the further drug free to separate from PVPVA, and the cliff "can be attributed to relative faster rate of AAPS compared to dissolution at high DL". C34·C33 In the Conclusions the switch to drug-controlled dissolution "appears to correlate" with an increased tendency to undergo AAPS in the matrix. C56

LLPS needs polymer-controlled release

The authors draw a general rule for LLPS: it "can only occur during ASD dissolution if the dissolution rate of the ASD is controlled by components that dissolve faster than the drug alone". Drug-controlled release slows toward zero as the dissolved drug approaches its amorphous solubility (eq 2), so it can never overshoot into LLPS, as the 50% DL tablet in saturated medium shows. Polymer-controlled release keeps going past the amorphous solubility, after which LLPS follows, as the group had shown earlier for nifedipine ASDs. C48·C49·R24 They also note that this behaviour, in which the drug's dissolution rate is not affected by the drug concentration in solution, runs against "most commonly applied dissolution models". C55

What it suggests for formulation

Citing emerging evidence that LLPS nanodroplets matter for absorption in vivo, the authors recommend that ASDs release by polymer control: "pharmaceutical scientists should select drug-polymer combinations and loadings such that the release of the drug is mainly governed by the polymer dissolution" throughout, using soluble polymers that interact with the drug and resist AAPS, and that also inhibit crystallization. Hydrophilic polymers that enhance dissolution and resist AAPS "can ensure rapid and complete release". C51·C50 Given the large differences in rate, extent and speciation between the regimes and the sharp drop over a small loading increment, these observations "should help inform formulation strategies". C57

Note by the AI writer (not discussed in the paper). Two practical points follow from the data. First, following the polymer as well as the drug shows the loss of congruency at 30% DL even while some drug is still released, so it marks the edge before drug release collapses. Second, one release test in drug-saturated medium distinguishes polymer-controlled from drug-controlled release. For ritonavir–PVPVA the usable margin is narrow: 25% DL behaved well and 30% DL already failed. R3.g–h·R12·R11

Limitations

Limits the authors state or imply. The authors note that RTV–PVPVA ASDs are known to undergo AAPS, but that "the kinetics of this process in bulk water have not been reported". The mechanism rests on exactly that kinetics. R22 Ritonavir "did not crystallize during the release experiments", so the scenario in which a drug-rich surface crystallizes is discussed but not tested. C54·R10·C29 Several key steps are worded as hypotheses: interactions "likely" saturated near 25–30% DL, nanodomains that "can" coalesce, a switch that "appears to correlate" with AAPS. C34·C40·C56

The AI writer's own judgement.

• The release data are the strongest part: two components assayed separately from constant-area tablets in triplicate, with neat references, a large drop between 25 and 30% DL, and a decisive drug-saturated control. P2·R11·R12
• AAPS was imaged only in thin films dipped in water for a few seconds and imaged once, at 20, 30 and 40% DL; it was not observed in the dissolving tablets, and neither its rate nor the rate of dissolution was measured. The race between them is an inference. X6
• Drug–polymer interactions in the wet state, central to the proposed switch point, were not measured. C34
• Nanodroplets were sized at 10% DL only; LLPS at 20–30% DL is inferred from turbidity and concentration. X5
• Congruency has no numerical criterion, and the LoC lies somewhere between the tested 25 and 30% DL. X7
• Some text statements disagree with the figures: the release bounds for 35–40% DL, the loadings named for IR and SEM in the methods, and the "similar" 30% DL profiles in the two media; the 30% DL release extent also differed about 1.6-fold between two series. X2·X1·X3
• The surface evidence at low loading covers the first 5 min only, the SEM samples were dried, and the presaturated medium is not described. X9·X11·X8
• The study covers one drug, one polymer, one medium and compressed tablets made from rotary-evaporated powder.

Conclusion

This paper defines the limit of congruency and measures it through a simple but informative design: one drug, one polymer, seven loadings, tablets of fixed surface area, and separate assays for drug and polymer. Up to 25% DL ritonavir left with PVPVA at the polymer's rate, unaffected by drug already in the medium, and passed its amorphous solubility to form nanodroplets; from 30% DL the polymer left first, and at 40–50% DL a drug-rich surface formed within minutes and release nearly stopped, falling to zero in drug-saturated medium. C1·C2·R11·R12·R16·R20 The authors explain the switch as water-induced AAPS outpacing dissolution once ritonavir–PVPVA interactions are "likely" saturated near 25–30% DL; that part rests on film images and reasoning rather than on measured rates. C33·C34·C56·X6 The practical message they draw is to choose polymer and loading so that release stays polymer-controlled throughout. C51