When two amorphous drugs dissolve together: how a second drug lowers each drug's maximum concentration

Background and motivation

Amorphous solid dispersions (ASDs) are used to deliver poorly water-soluble drugs because the amorphous form dissolves to a higher, temporary concentration than the crystal, which can raise bioavailability (Newman et al., 2015; Shah et al., 2013). Their main weakness is crystallization during storage, which polymers with a high glass transition temperature (Tg) help to prevent through drug–polymer interactions and lower molecular mobility (Prasad et al., 2014; Trasi and Taylor, 2012). The polymer also shapes how the ASD dissolves: some polymers dissolve quickly and release the drug quickly, others release it slowly (Konno et al., 2008; Six et al., 2004). C1·C2·C3

Under non-sink conditions, where more drug is present than the medium can hold at the crystalline solubility, an ASD usually produces a supersaturated solution. If it releases enough drug to exceed the amorphous solubility, the solution separates into a water-rich phase and a disordered, drug-rich colloidal phase, a process called liquid–liquid phase separation (LLPS) (Ilevbare and Taylor, 2013). The amorphous solubility is therefore the ceiling on the free (molecularly dissolved) drug concentration for any formulation, and the free concentration, not the total, sets both the supersaturation and the rate of passive diffusion through a membrane (Raina et al., 2014, 2015). C4

Many therapies combine drugs. HIV, hepatitis, cancer and tuberculosis treatments often need more than one drug, and fixed dose combination (FDC) products bring two or more drugs into one dosage form to improve compliance; the FDA approved 131 FDC products between 1990 and 2013 (Kararli et al., 2014). One of them, Kaletra, contains amorphous ritonavir (RTV) and lopinavir (LPV) in one formulation. Whether the drugs share one solid, sit in separate particles of one capsule or bilayer tablet, or come from two products taken at the same time, they can dissolve in the gastrointestinal tract together. The authors note that physicochemical drug–drug interactions of this kind "could potentially" change dissolution rate and solubility, yet have been little studied. C5·R19·C6

The starting point is the group's earlier antisolvent study (Trasi and Taylor, 2015). By adding drug solutions in a water-miscible solvent to buffer and finding where LLPS began, it showed that the amorphous solubility of one drug fell when a second drug was present, and vice versa. That study also supplies the reference values used throughout this paper: amorphous solubilities of 25 µg/mL for RTV and 17 µg/mL for LPV, and crystalline solubilities of 1.7 and 2.7 µg/mL. C7·M1·M2

This paper asks whether the same reduction appears when amorphous formulations dissolve, which is the practically relevant case. The authors' hypothesis is that co-formulation or co-dissolution lowers the maximum achievable concentration of each drug under non-sink conditions because each drug lowers the other's amorphous solubility. They test it with drug-only amorphous films, with 20% (w/w) dispersions in two polymers that release at different speeds, and with two single-drug dispersions added to the same vessel. C8·A1·A2·A3·A4

Background concepts

Amorphous solubility and the drug-rich phase. A supersaturated solution of a poorly soluble drug can only become more concentrated up to a point. Beyond the amorphous solubility, the excess drug separates as a second liquid-like, drug-rich phase, often as nanometre-sized droplets ("nanodroplets"). The solution in contact with this phase stays at the amorphous solubility, so the drug-rich phase acts as a buffer for the free concentration. The authors use the terms LLPS concentration and amorphous solubility for the same quantity. C4·M1·M2

Why a second drug lowers the ceiling. When two drugs are miscible in the drug-rich phase, each is diluted by the other there. The authors explain the effect through chemical potential: mixing lowers the chemical potential of each drug in the drug-rich phase, so less of each drug needs to be in water to be in equilibrium with it. This holds, in the authors' words, "as long as the two components are miscible in the drug-rich phase". C36

The ideal-mixing ceiling (Eq. 1). If the two amorphous drugs form an ideal mixture, the maximum bulk concentration of each drug after the mixture dissolves follows a relation the authors take from Lindfors et al. (2006):

Here $C_{max}$ is the maximum drug concentration in the bulk aqueous solution, $S_a$ is the amorphous solubility of that drug on its own, and $x_d$ is the mole fraction of that drug in the amorphous mixture. In plain terms: each drug's ceiling is its own amorphous solubility scaled down by its share of the drug mixture, counted in molecules. A drug that makes up half of the molecules in the mixed phase can reach only half of its own amorphous solubility. E1·C38

Weight ratio and mole fraction. The films and dispersions are described by weight ratios of RTV to LPV (3:7, 5:5 and 7:3). Eq. 1 needs mole fractions, which depend on the two molecular weights. The paper uses mole fractions in Table 3 and Figure 8 but does not print the molecular weights; in Figure 8 the 5:5 dispersion is plotted at an RTV mole fraction of about 0.45, not 0.5. R5·R17·E1

Filtration versus ultracentrifugation. To measure the free drug concentration, any drug-rich droplets and undissolved particles must be removed before analysis. A 0.45 µm filter removes large particles but not nanodroplets; ultracentrifugation (here 40,000 rpm for 20 min) pellets them, so the supernatant contains only dissolved drug. This distinction matters in the paper, because the HPMCAS dispersions produced colloids that passed the filters. P5·P9·R6

Pyrene as a crystallization reporter. Pyrene is a fluorescent probe whose emission spectrum depends on the polarity of its surroundings. The authors use the ratio of the third to the first emission peak, I3/I1 (383 nm over 373 nm). When an amorphous drug-rich phase is present, pyrene partitions into it and I3/I1 is about 0.65; when that phase crystallizes, the probe is expelled into water and the ratio falls (Goddard et al., 1985; Jackson et al., 2014). A falling I3/I1 therefore marks the time at which the drug-rich phase crystallizes. P11·C32

Glass transition and miscibility. A homogeneous amorphous mixture shows one glass transition, at a temperature between those of its components; two separate amorphous phases show two. One Tg is therefore read as evidence that two components are mixed. The authors report Tg as the onset of the DSC event. P4·C13

Methods in detail

The study has four parts: a polymer-free reference (drug-only films), two kinds of real formulation (PVP and HPMCAS dispersions), a co-administration test (two single-drug dispersions in one vessel), and a short study of how stable the mixed drug-rich phase is against crystallization. Throughout, the medium is 100 mM phosphate buffer at pH 6.8. A1·A2·A3·A4·A5·M6

Contribution 1: drug-only amorphous films as a clean reference

Problem. In a dispersion, the polymer changes how fast the drug is released and whether nanodroplets form, which makes it hard to see the thermodynamic ceiling directly. The authors wanted reference values for the maximum concentration of each drug alone and with the other drug, with the composition of the amorphous phase fixed exactly. C15

What they did. Ritonavir and lopinavir (Attix Pharmaceuticals, crystalline as received) were dissolved in methanol, 5 mg of a single drug or 5 mg in total at RTV:LPV weight ratios of 7:3, 5:5 and 3:7, and deposited as a thin amorphous film on the inside of a scintillation vial by rotary evaporation. The film was then covered with 10 mL of buffer containing 10 µg/mL HPMCAS as a crystallization inhibitor and stirred at 300 rpm. Samples of 500 µL were filtered through a 0.45 µm glass filter at 10, 20, 30, 45 and 60 min and assayed by HPLC, and the volume was topped up with fresh medium. With 5 mg of drug in 10 mL, the conditions were non-sink for both drugs (Figure 3 caption). The temperature of the film experiments is not stated. M1·M2·P2·P6·P10·X5

Why it works. A film has a large surface area and contains the drugs in exactly the chosen ratio, so it should dissolve quickly up to the ceiling set by its composition. The single-drug films check how close the method gets to the known amorphous solubilities. C15·C17

Contribution 2: dispersions in a fast and a slow polymer, and two ways to measure dissolved drug

Problem. Real products are dispersions, and two common polymers behave differently: HPMCAS-MF (ShinEtsu) and PVP K-29/32 (BASF, Mw about 30,000 g/mol). The question is whether the ceiling seen in films also appears when the drugs are released from a polymer matrix. M3·M4·A2

What they did. Dispersions containing 20% (w/w) total drug, with RTV only, LPV only, or RTV:LPV 7:3, 5:5 and 3:7, were made by rotary evaporation from methanol and cryomilled. XRPD (5–40° 2θ) checked that they were amorphous, and DSC gave their onset Tg. For dissolution, 20 mg of dispersion was added to 40 mL of buffer at 25 °C and stirred at 300 rpm. Samples were taken over 60 min, filtered through 0.45 µm syringe filters and assayed by HPLC (Figure 5). This worked for PVP but not for HPMCAS, whose dispersions released amorphous particles so small that they passed the filters. For a fair comparison of both polymers, the suspensions were instead ultracentrifuged at 40,000 rpm for 20 min, and the supernatant was assayed at 20 min and 2 h (Figure 6). The sediment collected after 2 h was dissolved in methanol and assayed, which gave a mass balance and the mole fractions of RTV and LPV in the amorphous phase. Filtrates after 1 h of dissolution were examined by nanoparticle tracking analysis (NanoSight LM10, 532 nm laser), with camera settings chosen so that blank buffer showed no scattering. P1·P3·P4·P5·P8·P9·R6·R18

Note by the AI writer (not discussed in the paper). The method section names only the 3:7 dispersion and sampling "every 10 min", whereas Figure 5 shows all five compositions sampled at 5, 10, 20, 30, 45 and 60 min; the dose and set-up of the ultracentrifuged runs are not stated separately. The 20 mg in 40 mL condition is the most likely one for all runs, but this is an inference. X5

Contribution 3: two single-drug dispersions in the same vessel

Problem. Even if two drugs are formulated separately, a patient may take them together. Does the ceiling drop only when the drugs share one solid, or also when they meet in solution? C6·A4

What they did. An equal amount of RTV-PVP and LPV-PVP dispersion (equal weights of the two drugs) was added to the medium at the same time, and filtered samples were analysed over 60 min (Figure 5f). The same test was run with the two HPMCAS dispersions, giving a 1:1 weight ratio of RTV and LPV, with a single ultracentrifuged sample at 60 min. The amounts added are not printed. P7·R10·R11·X5

Contribution 4: crystallization of the drug-rich phase

Problem. A drug-rich phase may raise the risk of crystallization, because supersaturation is at its highest and the droplet surfaces "may act" as sites for heterogeneous nucleation. Does a mixed drug-rich phase behave differently from a single-drug one? C31

What they did. Drug solutions in methanol (10 mg/mL) were added to 10 mL of buffer containing 1 µM pyrene to create suspensions above the LLPS concentration: RTV 50 µg/mL, LPV 35 µg/mL, RTV 25 + LPV 17 µg/mL and RTV 50 + LPV 35 µg/mL. The mixtures correspond to about once and twice the single-drug LLPS concentrations. The suspensions were stirred at 300 rpm, and the pyrene emission (excitation 332 nm) was recorded for 5 h (Figure 7). Free drug in the supernatant was measured at 0 h and after 5 h (Table 2). P11·P12·R15·R16

Experimental analysis

Solid-state starting point. Both drugs were crystalline as received, and all ten dispersions, single-drug and combined, in both polymers, showed only a broad XRPD halo (Figure 2; the caption swaps the descriptions of panels a and b, which the legends identify as HPMCAS and PVP, respectively). RTV has a Tg of 46 °C and LPV of 77 °C; the co-melted drug mixtures gave single Tgs of 66, 60.5 and 54 °C at 3:7, 5:5 and 7:3, between the two drugs (Table 1). The PVP dispersions had Tgs of 125.5–140 °C and the HPMCAS dispersions 73–91 °C, each with a single event, which the authors read as miscibility of the drugs with each other and with the polymer. They attribute the higher Tg of the PVP dispersions "most likely" to the higher Tg of PVP itself, which is not printed. R1·R2·C11·C13·C14·X1

Main result 1: in films, each drug's plateau falls with its share. All films dissolved to supersaturated solutions, well above the crystalline solubilities (1.7 and 2.7 µg/mL). The pure RTV and LPV films reached 22.9 and 15.6 µg/mL after 60 min, close to the amorphous solubilities of 25 and 17 µg/mL, which the authors read as film dissolution having "nearly achieved" the amorphous solubility. In mixed films each drug levelled off lower, and the more of the other drug the film contained, the lower the plateau (Figure 3). Table A sets the measured values against Eq. 1. R3·R4·R5·C16·C17·C18·M1·M2

Table A. Maximum concentrations (µg/mL) from drug-only amorphous films and the ideal-mixing prediction of Eq. 1, from Table 3 of the paper (printed values). R5·E1

Five of the six mixed-film values lie within 0.5 µg/mL of the prediction; the largest gap is RTV from the 7:3 film, 15.2 measured against 16.7 predicted. The authors describe the reduction as "clearly demonstrated" and the agreement as "good", "indicating that RTV and LPV are miscible, form a close to ideal mixture" well described by Eq. 1. Figure 3a plots the pure RTV film at about 22.4 µg/mL at 60 min, slightly below the 22.9 µg/mL given in the text and Table 3 but within its error bar. R5·R3·C37·C39·X4

Note by the AI writer (not discussed in the paper). The paper does not print the molecular weights needed to turn weight ratios into mole fractions. With literature molecular weights (not taken from the paper), the AI writer reproduced all six calculated values of Table 3 within 0.1 µg/mL; for 5:5 the RTV mole fraction is about 0.466. The predictions use the cited amorphous solubilities (25 and 17 µg/mL); the pure films in this paper reached about 8% less. E1·X14

Main result 2: dispersions show the same drop, in both polymers. The filtered PVP data (Figure 5) already show it at 60 min: RTV reached about 20.4 µg/mL from the RTV-only dispersion but about 12.2, 10.8 and 6.0 µg/mL from the 7:3, 5:5 and 3:7 dispersions, and LPV about 18.0 µg/mL from the LPV-only dispersion against about 12.0, 9.1 and 5.9 µg/mL from 3:7, 5:5 and 7:3. All solutions stayed supersaturated with respect to the crystalline solubilities, and the RTV-PVP dispersion "did not appear to reach" the amorphous solubility within 1 h. The ultracentrifuged values at 2 h (Figure 6, Table B) show the same pattern for both polymers, the reduction depending on the relative amount of the second drug. The authors call the concentrations in combination "dramatically lower" than those from the single-drug dispersions. R9·R13·C9·C23·C24·C30·M1·M2

Table B. Free drug (µg/mL) in the ultracentrifuged supernatant 2 h after adding 20% (w/w) dispersions to pH 6.8 buffer, read from Figure 6c and 6d (approximate); error bars and n are not defined in the paper. R13·X5

Main result 3: the polymer changes the route, not the end point. HPMCAS dispersions turned the medium uniformly turbid within a few minutes. The authors read this as rapid dissolution "most likely" to above the amorphous solubility, forming a new scattering phase of drug-rich droplets. These colloids passed 0.45 µm filters: nanoparticle tracking found species of mean size approximately 500 nm in the RTV-HPMCAS filtrate and around 160 nm in the filtrate of the HPMCAS dispersion with both drugs (Figure 4 shows the 3:7 sample). PVP dispersions dissolved more slowly, gave less turbid suspensions and clear filtrates almost free of scattering species. In the ultracentrifuged data, HPMCAS dispersions were at their maximum within 20 min, while PVP dispersions were still well below it then (for example RTV about 16.5 against about 26.5 µg/mL from the single-drug dispersions; LPV from 3:7 about 7.4 against about 12.5 µg/mL) and needed about 2 h. By 2 h the values were similar, "indicating that the maximum concentration of drug attained is not dependent on the polymer type"; the exception the authors note is RTV-PVP, which stayed near 22 µg/mL. R6·R7·R8·R12·R13·R14·C19·C21·C28·C29·C45

Note by the AI writer (not discussed in the paper). Reading Figure 6c, most HPMCAS and PVP values at 2 h are within 1 µg/mL, but RTV from the 7:3 dispersion also differs (about 17.0 against about 14.9 µg/mL), in addition to the RTV-only pair the authors mention. X12

The authors add a methodological warning: because 0.45 µm filtration did not remove the colloids produced by the HPMCAS dispersions, filtered samples would count nanodroplets as dissolved drug, which "has important analytical implications" for anyone following supersaturation during non-sink dissolution of high-energy formulations. C22·C27

Main result 4: the dispersion data fall on the Eq. 1 lines. Figure 8 draws Eq. 1 as two straight lines against the RTV mole fraction in the drug-rich phase: RTV rising from zero at pure LPV to its amorphous solubility at pure RTV, and LPV falling from its amorphous solubility to zero. The lines cross near an RTV mole fraction of 0.39 at about 10 µg/mL. Onto these lines the authors place the free concentrations from the three HPMCAS dispersions. The RTV points lie on the RTV line (about 7.0, 12.0 and 17.0 µg/mL at mole fractions of about 0.27, 0.45 and 0.65). The LPV points (about 13.0, 8.0 and 5.0 µg/mL) lie close to the LPV line at 3:7 and about 13–15% below it at 5:5 and 7:3 (line values about 9.2 and 5.9 µg/mL). The authors call this "excellent agreement" and conclude that "irrespective of method of supersaturation generation", antisolvent addition or dissolution of an ASD, the solution phase behaviour is identical. R17·C43·C44·X8

Note by the AI writer (not discussed in the paper). Figure 8 does not state which time point is plotted; the values match the 2 h ultracentrifuged data of Figure 6. It also does not state whether the plotted mole fractions are those measured in the sediment or the nominal ones. The RTV line in Figure 8 reaches about 26 µg/mL at pure RTV, whereas Eq. 1 with the 25 µg/mL used for Table 3 gives 25. X7·X3

Main result 5: two separate dispersions give the same drop. When equal weights of RTV and LPV were added as two PVP dispersions, RTV first dissolved quickly, reaching about 15.2 µg/mL at 20 min, and then fell to about 12.2 µg/mL at 60 min as LPV rose past about 10 µg/mL (Figure 5f, Table C). The authors note that the final values were similar to those of the 5:5 ternary dispersion. With two HPMCAS dispersions added separately, the ultracentrifuged supernatant at 60 min held 11.2 (±0.1) µg/mL RTV and 9.1 (±0.15) µg/mL LPV, "in very close agreement" with both PVP results. R10·R11·R9·C25·C26

Table C. Co-formulation against co-administration at equal drug weights. Filtered values read from Figure 5 (approximate); HPMCAS values as printed in the text. R9·R10·R11

Main result 6: a mixed drug-rich phase resists crystallization for 5 h. In the LPV-only suspension (35 µg/mL), I3/I1 stayed at about 0.645 and 0.640 at 0 and 1 h and then fell to about 0.565 at 2 h and about 0.510 at 5 h, which the authors read as crystallization between 1 and 2 h. After 5 h the free LPV concentration was 4.3 µg/mL, far below its amorphous solubility and above its crystalline solubility. The RTV-only suspension (50 µg/mL) showed no such fall over 5 h (I3/I1 about 0.62–0.63 from 2 to 5 h) and still held 25.4 µg/mL free RTV after 5 h. Both mixtures kept I3/I1 at about 0.63–0.66 for 5 h, and their free drug concentrations did not change (Table D). The authors conclude that RTV "stabilized" LPV against crystallization. R15·R16·C33·C34·C35·M1·M2

Table D. Free drug (µg/mL, standard deviation in parentheses) from Table 2 of the paper, and pyrene I3/I1 read from Figure 7 (approximate). RTV-alone markers at 0 and 1 h are hidden in the original figure. R16·R15

Table D carries a second message. Both mixtures, although one contains twice as much drug as the other, held RTV at 14.2–15.8 µg/mL and LPV at 7.0–7.5 µg/mL, well below each drug's own amorphous solubility. This is the same depression of the ceiling seen in the dissolution experiments, now from the antisolvent side. R16·M1·M2

Discussion

How the authors connect the results. The argument runs from miscibility to a lower ceiling, and then splits into two consequences, one unfavourable and one favourable for a formulation. C13·C36·C39·C49·C51·C52

The first three steps are the authors' thermodynamic reading. They state the cause directly: "the underlying origin of the reduced amorphous solubility is the mixing of the two drugs" in the drug-rich phase, which lowers each drug's chemical potential. The Eq. 1 fit to the films (Table A) and the single Tg of each mixture support the premise of ideal mixing; the chemical potential itself was not measured. C36·C39·C13

Co-formulation and co-administration. The separate-addition runs carry the argument beyond fixed dose combinations. The authors conclude that the reduction "can also potentially occur when two drug formulations are administered independently" at the same time, because the drugs can mix in solution during dissolution and separate into a shared drug-rich phase; intimate contact in the formulation is not needed. In the conclusions this becomes "can also occur". They ask for the effect to be tested in "more biologically relevant dissolution media". C47·C48

The role of the polymer. The polymer did not change where each drug ended up after 2 h, but it changed the route. HPMCAS dispersions dissolved quickly to above the amorphous solubility and produced amorphous nanodroplets; PVP dispersions dissolved slowly with clear solutions. The authors suggest that "it may be preferable to use a polymer that promotes LLPS", because the drug-rich phase keeps supersaturation at its maximum while each drug is absorbed, an argument taken from Raina et al. (2014). C45·C46·C28

Crystallization as the upside. The authors expected, from thermodynamic models, that a lower chemical potential in a mixed phase lowers the maximum supersaturation and so "would be expected to impact the crystallization tendency" of each drug. They cite co-amorphous solids that crystallize less than the pure drugs (Löbmann et al., 2012) and attribute the stability of the mixed drug-rich phases, "at least in part", to the lower supersaturation. In the conclusions they call the lower driving force for crystallization "one potential advantage" of amorphous combinations. C49·C50·C51

What it suggests for formulation. The authors' recommendation is that, for combination products, "the potential for decreased bioavailability due to lowered effective drug concentrations should be considered", alongside the benefits for compliance and stability. C52

Note by the AI writer (not discussed in the paper). For a pair of drugs that mix almost ideally, Eq. 1 gives a quick estimate before any dissolution test: each drug's ceiling is its amorphous solubility times its mole fraction. The minor drug by mole loses most; in the 3:7 film RTV reached about a quarter of its pure-film value (6.3 against 22.9 µg/mL). A faster polymer or a higher total drug load does not lift this ceiling; it only changes how quickly it is reached. A dose that delivers enough free drug from a single-drug ASD may therefore fall short in a combination product, or when another amorphous drug that mixes with it is taken at the same time. The gain in crystallization resistance and the loss in free concentration come from the same cause, so they arrive together. R5·E1·R13

Limitations

Limitations stated by the authors. The authors limit their claims in several ways. The reduction is said to require that the two drugs are "miscible in the drug-rich phase". The co-administration effect is stated as something that "can also potentially occur", with testing in "more biologically relevant dissolution media" left for future work. The attribution of the crystallization resistance to lower supersaturation is made "at least in part". They also note that the RTV-PVP dispersion did not reach the expected amorphous solubility even after 2 h, and that filtration could not be used for HPMCAS dispersions. C36·C47·C48·C51·C29·P5

The AI writer's own judgement (1): what is shown directly and what is inferred. The central result is well supported. The film values are printed (Table 3), they agree with Eq. 1 within 0.5 µg/mL for five of six mixed values, and both polymers show the same ratio-dependent drop after ultracentrifugation. Three parts of the argument rest on thinner evidence. First, the shared drug-rich phase after separate addition was not shown: the PVP run was only filtered, PVP dispersions gave few colloids in nanoparticle tracking, and the HPMCAS run has a single 60 min value. Second, the crystallization study lasted 5 h, and the mixtures' I3/I1 also drifted down slightly (from about 0.657 to 0.645 and from about 0.656 to 0.632); whether crystallization is delayed or prevented, and how much of the effect is due to lower supersaturation rather than other effects of RTV in the droplets, is not resolved. Third, the bioavailability consequence is inferred: neither membrane flux nor in vivo exposure was measured. R5·R13·X9·X11·C52

The AI writer's own judgement (2): data presentation and conditions. Most dispersion and kinetic values are available only as plots, and the error bars and numbers of replicates in Figures 3, 5, 6 and 7 are not defined; Table 1 does not say what its values in parentheses are. Several conditions are missing: the temperature of the film, ultracentrifugation and pyrene experiments (only powder dissolution is stated at 25 °C), and the amounts used in the separate-addition runs. Several captions do not match their figures: the Figure 2 caption swaps the two panels; the Figure 3 and Figure 6 captions mention a red crystalline-solubility line that is not drawn; the Figure 5 caption calls the red line in panel b the crystalline solubility although it lies at about 17.8 µg/mL, the amorphous solubility of LPV; and the Eq. 1 line for RTV in Figure 8 ends at about 26 rather than 25 µg/mL. In Figure 8 the LPV points sit up to about 15% below the line, so "excellent agreement" is fair for RTV and generous for LPV. Finally, the paper's own Tg-based expectation that amorphous LPV "should be more stable" than RTV is not tested in the solid state, and in solution the order was reversed: the LPV drug-rich phase crystallized first. One drug pair, one drug load and a simple buffer were studied. X5·X6·X1·X2·X3·X8·X10·C12

Conclusion

The paper makes three contributions. First, with drug-only films it shows that ritonavir and lopinavir, which are miscible in the amorphous state, each reach a maximum concentration close to the ideal-mixing prediction Cmax = Sa × xd (Table 3, Eq. 1). Second, it shows that the same ceiling appears when the drugs are released from PVP or HPMCAS dispersions and even when two single-drug dispersions dissolve in the same vessel; the polymer changes the speed and whether nanodroplets form, not the end value after 2 h. Third, it finds that a drug-rich phase containing both drugs did not crystallize within 5 h, whereas the lopinavir-only phase crystallized within 1–2 h. In the authors' words, administering two amorphous drugs together "can result in a lowering of amorphous solubility of each component, and hence a reduction in the maximum achievable supersaturation". R5·E1·R13·R10·R11·R15·C53

For formulation, the message is a trade-off within one mechanism. A combination of miscible amorphous drugs gives each drug a lower ceiling for free concentration, which the authors say should be considered for bioavailability, and a lower driving force for crystallization, which they call a potential advantage. The authors ask for the effect to be tested in more biologically relevant media. C52·C49·C48 In the AI writer's view, membrane transport and drug pairs that mix less ideally, neither studied here, are the other tests that would show how far the simple Sa × xd rule carries.