Telling a drug-rich liquid from crystals with a fluorescent probe: ritonavir and lopinavir dispersions in Kollidon VA64

Background and motivation

Most new drug candidates dissolve poorly in water; the paper opens with the estimate that about 75% do, which limits their bioavailability. Amorphous solid dispersions (ASDs), in which the drug is dispersed in a polymer in its amorphous form, are a common answer, because the amorphous form has a higher free energy than the crystal and so can produce supersaturated solutions when it dissolves. The same excess energy also drives crystallization, from the solid or from the solution, and a polymer is added to slow that down. C1·C2

What happens in a supersaturated solution is often reported simply as "precipitation". The authors stress that the word covers two very different events. When the drug concentration passes the miscibility limit of the drug with water, which equals the predicted amorphous solubility, the solution splits into a continuous water-rich phase and a dispersed, non-crystalline drug-rich phase: this is liquid-liquid phase separation (LLPS). Crystals, by contrast, are a solid precipitate. C3·C4 The two outcomes have opposite consequences. Crystals use up the supersaturation. The drug-rich phase can act as a reservoir that replenishes drug as it is absorbed across a membrane and keeps the solution highly supersaturated, and drug-rich species "may enhance" bioavailability (Indulkar et al. 2016; Raina et al. 2014; Stewart et al. 2017). C5·C7

Whether an ASD produces a drug-rich phase on dissolution depends on whether the dissolved drug concentration exceeds the amorphous solubility. Jackson et al. (2016) showed for danazol dispersions that this depends on the polymer and on the drug:polymer ratio, and that the duration of supersaturation depends on how well the polymer inhibits crystallization; HPMC-AS kept a 10% drug-loading danazol dispersion supersaturated longer than PVP. C6·R26 Few studies have followed LLPS during ASD dissolution, and the authors note that telling an amorphous precipitate from a crystalline one in a turbid dissolution medium is analytically difficult. C8

The proposed tool is the fluorescence lifetime: the average time a fluorescent molecule stays in its excited state. Unlike steady-state intensity, the lifetime is considered "largely independent" of the excitation wavelength, the exposure time and the probe concentration, but it is sensitive to temperature, viscosity, polarity and molecular interactions, which makes it "potentially useful" in turbid samples (Berezin and Achilefu 2010). C9 In an earlier paper the same group used the lifetime to characterise drug-rich phases formed by solvent addition for felodipine, clotrimazole and efavirenz; the lifetime of PRODAN rose significantly above LLPS of clotrimazole and efavirenz, and the lifetime of the autofluorescent felodipine rose during its crystallization (Tres et al. 2018, J. Pharm. Sci.). R23·R24

This paper extends the method to ASD dissolution. The model drugs are the protease inhibitors ritonavir and lopinavir (Figure 1), whose supersaturation behaviour was already known: earlier UV-extinction measurements placed their LLPS onsets above 39 µM and 28 µM (Trasi and Taylor 2015a). M1·M2 The polymer is Kollidon VA64, a vinylpyrrolidone-vinyl acetate copolymer used in the commercial ritonavir-lopinavir dispersion Kaletra, which forms one-phase amorphous mixtures with both drugs. M3·C10 The authors set three aims: (i) find the LLPS onset of solutions made by solvent addition from the lifetime, (ii) follow LLPS during nonsink dissolution of dispersions with 10% and 50% w/w drug loading, and (iii) follow the crystallization of supersaturated solutions. Steady-state fluorescence, UV extinction and concentration, ultracentrifugation and nanoparticle tracking serve as cross-checks. A1·A2·A5

Background concepts

Amorphous solubility and LLPS. A crystal sets the equilibrium solubility. An amorphous solid, being less stable, can push the dissolved concentration higher; the highest concentration of molecularly dissolved drug that can be reached before a second, drug-rich liquid phase appears is called the amorphous solubility. Above it the solution separates into a water-rich phase, holding drug at about the amorphous solubility, and small droplets or colloids of a drug-rich phase. C4 In this paper the terms "LLPS onset concentration" and "amorphous solubility" are used for the same threshold. Two experimental routes to it appear: the onset of light scattering when drug is added stepwise from a stock solution (the 39 and 28 µM values, cited), and the drug concentration left in the supernatant after the drug-rich phase is spun down in an ultracentrifuge (measured here). M1·M2·P9

Fluorescence lifetime and PRODAN. PRODAN (6-propionyl-2-dimethylaminonaphthalene) is a naphthalene probe of moderate lipophilicity whose emission depends strongly on the polarity of its surroundings (Parasassi and Krasnowska 1998; Raina et al. 2015). M5 In water its decay is bi-exponential, with literature lifetimes of 0.7 ns (60%) and 2.0 ns (40%); the shorter component has been "attributed" to hydrogen-bonded probe-water complexes (Balter et al. 1988). M5·C15 The authors fit every decay with a bi-exponential model, whose two components are taken to represent two populations of probe molecules, and report the amplitude-weighted average lifetime as a single number "for a ready interpretation" (Sillen and Engelborghs 1998). E1·E2·C12 The working idea is that if a drug-rich phase forms, the lipophilic probe moves into it, sees a less polar environment and lives longer; if that phase later turns into crystals, the probe is pushed back into water and its lifetime falls. C17·C43

The steady-state emission ratio. In water PRODAN emits around 520 nm. In a less polar environment a second band appears at shorter wavelength (445 or 450 nm here). The authors follow the ratio of the two band intensities, which also reduces the effect of slow drifts in absolute intensity: P7·E5

In plain terms: $I_1$ is the emission from probe molecules in the aggregates and $I_2$ the emission from free probe in solution, so the ratio rises when a drug-rich phase takes up the probe. C13 The relation is given in words in the paper (Section 2.4).

UV extinction and apparent concentration. With in situ UV dip probes the drug is followed at its absorbance maximum (240 nm for ritonavir, 260 nm for lopinavir), while the signal at 450 nm, where neither drug absorbs, measures light lost to scattering particles, called extinction. P8·C14 The drug signal is corrected by subtracting the 450 nm signal: E4

In plain terms: $A({\lambda }_{max})$ is the signal at the drug's absorbance maximum and $A(450\mathrm{n}\mathrm{m})$ the scattering background taken at a wavelength without absorption; the difference is converted to an "apparent" concentration with a calibration in acetonitrile:water. E4·P8 The authors warn that this apparent value is not quantitative once a drug-rich phase is present, because drug-rich nanospecies both absorb and scatter (Van Eerdenbrugh et al. 2011). C37

Polymer-controlled and drug-controlled dissolution. Following Craig (2002), a solid dispersion in a water-soluble polymer can dissolve at the rate set by the polymer, carrying the drug with it, or at the much slower rate set by the drug itself. Which regime applies depends on the components and on the drug:polymer ratio. The authors use this framework to interpret their 10% and 50% drug-loading results. C23·C24

Nanoparticle tracking analysis (NTA). NTA follows individual scattering particles under a microscope and converts their Brownian motion into a hydrodynamic diameter with the Stokes-Einstein relation, which the paper names but does not print. P10·E3

Methods in detail

The study has three kinds of test, all in 100 mM pH 7.4 sodium phosphate buffer at 24 ± 1 °C with constant stirring and 0.9 µM PRODAN dissolved beforehand, and two end-point checks. P3·P4·P5 The flow below summarises them.

Preparing the dispersions

Ritonavir and lopinavir (ChemShuttle) were each combined with Kollidon VA64 (BASF) at drug:polymer ratios of 10:90 and 50:50 w/w, dissolved in ethanol:dichloromethane 30:70 v/v, and rotary-evaporated at 40 °C. The solids were dried overnight in a vacuum oven, ground with a mortar and pestle and sieved; the size limit is printed as "• 250 µm", with the relation symbol not rendered in the manuscript. P1·M1·M2·M3·X7 X-ray powder diffraction showed no crystalline material, but the patterns (Figure S-1) are in the Supplementary Information, which was not available. R10·P2

Test 1: finding the LLPS onset by solvent addition

Aliquots of a concentrated methanol stock were added to 10 mL of buffer that already contained 0.9 µM PRODAN and 100 µg/mL HPMC-AS, the latter to prevent crystallization during the measurement. P3·C11 The nominal concentrations, given in the legends of Figure 3, were 0, 5, 10, 20, 40, 59 and 69 µM for ritonavir and 0, 5, 10, 15, 20, 41, 46, 51 and 56 µM for lopinavir. P3 At each step the lifetime decay was recorded with an EasyLife V system (pulsed 370 nm LED, emission filter cut-off about 400 nm, 500 channels between 60 and 90 ns delay, instrument response from a Ludox scattering solution), and an emission spectrum was taken on a Shimadzu RF-5301PC (excitation 370 nm, emission 370-600 nm). P6·P7 For the amorphous-solubility reference, drug was added in 15 mL of buffer to above the expected onset, with either HPMC-AS at 100 µg/mL or Kollidon VA64 at 800 µg/mL, the latter matching the theoretical maximum polymer concentration after dissolution of a dispersion; the drug-rich phase was then pelleted at 209,875 × g for 20 min at 25 °C and the supernatant assayed by HPLC (Table 1). P9

Test 2: dissolving the dispersions under nonsink conditions

Dispersion powder was added to 50 mL of buffer with PRODAN but without HPMC-AS. The amount was chosen so that the theoretical maximum drug concentration exceeded the expected LLPS onset; the mass added is not printed. P4 Over about 60 min the lifetime and the emission ratio were followed, and in situ UV probes (1 cm path for ritonavir, 2 cm for lopinavir because of its low molar absorption) recorded spectra every 60 s. P4·P6·P7·P8·M2 At the end, the medium was ultracentrifuged and the supernatant assayed by HPLC; the 10% drug-loading media were filtered through a 1 µm glass-fibre filter and sized by NTA (NanoSight LM10, 532 nm laser, 30 s per sample). P9·P10 For the 50% lopinavir dispersion, the filtered medium was measured again for its lifetime, as a control (Section 3.2). P11 UV and HPLC values are the mean of two measurements with the range as error bar; no replicate number is given for the lifetimes. P8·P9·X8

Test 3: letting the drug-rich phase crystallize

Supersaturated solutions of each drug were made in 10 mL of buffer with PRODAN but without any HPMC-AS, at a concentration given only as "above the amorphous solubility", and the lifetime (and, for ritonavir, the emission ratio) was followed for about 213 min. P5·X4 The logic is the premise stated in Section 3.4: if the drug crystallizes, the drug-rich phase is depleted, the probe is expelled into water, and the lifetime decreases. C43 The authors contrast this with their earlier felodipine work, where the drug's own fluorescence lifetime increased on crystallization because the crystal holds the molecule in a more rigid framework. C42·R24

The design therefore uses one probe in three settings that differ in what can happen: a drug-rich phase that is protected from crystallization (Test 1, with HPMC-AS), a drug-rich phase that may or may not form depending on the dispersion (Test 2), and a drug-rich phase left free to crystallize (Test 3). A3·A5 What distinguishes it from the earlier solvent-shift study is the dissolution test with two drug loadings and the deliberate crystallization test read through the probe rather than through the drug itself. A2·R23

Experimental analysis

The lifetime marks the LLPS onset (Figures 2 and 3)

In the solvent-shift tests the average PRODAN lifetime stayed flat at low drug concentration and rose once a threshold was passed: from about 1 ns, the value of a solution containing only the probe, to 1.34 ns with 69 µM ritonavir and to 1.22 ns with 51 µM lopinavir (Figure 2). R1·R2·C16 Read point by point, the ritonavir lifetime was about 1.00, 0.98, 0.99 and 1.03 ns at 0, 5, 10 and 20 µM, then about 1.13 ns at 40 µM and 1.29 ns at 59 µM; for lopinavir it was about 0.97-1.03 ns from 0 to 20 µM and about 1.14, 1.19, 1.22 and 1.22 ns at 41, 46, 51 and 56 µM. R1·R2 The increase came from a longer second component, τ2 (Tables S-1 and S-2, not available here). R3 The authors attribute it to the presence of the probe in the newly formed drug-rich colloidal phase: being lipophilic, PRODAN preferentially sits in the more hydrophobic, disordered drug-rich phase, where the lower polarity lengthens its lifetime ("These increases are attributed to a longer τ2"). C17

The emission spectra (Figure 3) told the same story. Up to 40 µM ritonavir and 20 µM lopinavir the 520 nm band stayed in place, losing a little intensity; above those concentrations a second band appeared at 450 nm (ritonavir) and 445 nm (lopinavir), reaching about 149 a.u. at 69 µM ritonavir and about 219 a.u. at 56 µM lopinavir. R4·R5 The ratio I₁/I₂ stayed about 0.03-0.06 below these concentrations and rose to about 0.64 (ritonavir, 69 µM) and about 1.06 (lopinavir, 56 µM). R1·R2 The authors read the new band as the probe partitioning into the drug-rich phase ("these data suggest") and the ratio as "pointing to" aggregate formation. C19·C20

Table A. Where the probe signals change in the solvent-shift tests with 100 µg/mL HPMC-AS (read from Figure 2, approximate; 1.34 and 1.22 ns from the text). R1·R2·M1·M2

The authors call the lifetime and ratio results "in a good agreement" with the earlier UV-extinction onsets of 39 and 28 µM. C18·C20 Note by the AI writer (not discussed in the paper): the concentration steps bracket the rise only between 20 and 40 µM for ritonavir and between 20 and 41 µM for lopinavir, so the data are consistent with the cited onsets but do not locate them more finely. X2

Ritonavir dispersions: only the 10% loading crosses the onset (Figure 4)

The two ritonavir dispersions behaved very differently. With 10% drug loading the lifetime rose rapidly from approximately 0.95 ns, the value for probe alone, to 1.54 ns at 15 min and stayed constant to 60 min (about 1.52-1.56 ns); the emission ratio rose with it, to about 1.39-1.47 from 5 min on. With 50% drug loading the lifetime did not change significantly (about 1.01-1.06 ns) and the ratio stayed near zero. R6·C22 The extinction at 450 nm for the 10% dispersion rose to about 0.51 within 3 min and held at about 0.55-0.56, while the 50% dispersion stayed at about 0.01-0.02, which the authors call "indicative of" a second scattering phase only for the 10% loading. R7·C26

The UV apparent concentration of the 10% dispersion peaked at 96 µM after 3 min, above the amorphous solubility of 39 µM, and then fell to 44 µM at 59 min; the 50% dispersion rose slowly to 21 µM after approximately 60 min. R8·C27 The authors consider the fall "likely to be an artifact": the aggregates coalesce and grow, which changes how they scatter and absorb (Ilevbare and Taylor 2013). Recrystallization they consider "unlikely", because the solution holds a lot of polymer (ca. 800 µg/mL), ritonavir has a low crystallization tendency, and the lifetime never decreased. C28·C29

The end-point concentrations settled the question. After ultracentrifugation the 10% medium held 49 µM ritonavir, comparable to the amorphous solubility measured by solvent addition with Kollidon VA64 (48 µM), "confirming that crystallization had not occurred"; the 50% medium held 19 µM, close to the UV value of 21 µM. With HPMC-AS instead of Kollidon VA64 the amorphous solubility read about 47 µM in Figure 4d. R9·C30 Together, the authors conclude, the four read-outs show that only the 10% dispersion reached a concentration high enough for LLPS and a disordered drug-rich phase. C31

Lopinavir dispersions: the same split, and a lifetime artefact (Figure 5)

With 10% drug loading the lopinavir lifetime rose from about 0.95 ns to 1.23 ns at 13 min and stayed about 1.19-1.23 ns to 58 min; the emission ratio climbed from about 0.58 at 5 min to about 1.22 at 58 min. The authors read this as the formation of a disordered drug-rich phase, "confirmed" by the ratio. R11·C32 The extinction rose steadily and did not level off: about 0.21 at 10 min, 0.45 at 30 min and 0.70 at 54 min. R15·C35 The apparent UV concentration passed the amorphous solubility of 28 µM within minutes and kept climbing to about 466 µM at 54 min. R16·C36 The authors point out that UV values are not quantitative when drug-rich nanospecies are present. C37 Note by the AI writer (not discussed in the paper): at 800 µg/mL polymer a 10% dispersion supplies about 89 µg/mL of drug, roughly 141 µM of lopinavir (using its molar mass, which the paper does not print), so the UV reading exceeds all the drug that could have dissolved; it is a scattering signal, not a concentration. X6 The supernatant after ultracentrifugation held 36 µM, "in a good agreement" with the amorphous solubility by solvent addition (38 µM with Kollidon VA64; about 37.5 µM with HPMC-AS in Figure 5d). R17·C38

The 50% lopinavir dispersion gave the one misleading signal of the study. Its lifetime also rose, from about 0.99 ns to about 1.23 ns by 23 min, although its extinction stayed about 0.01 and its UV concentration reached only about 15 µM after 54 min. R11·R15·R16 Several undissolved solid particles were visible in the medium. After they were filtered off at the end of the test (1 µm), the lifetime was 1.00 ns, against 0.99 ns for a solution of probe alone (Table S-6, quoted in the text). R12·R13 The authors therefore attribute the rise to the dye interacting with undissolved dispersion particles rather than to a new drug-rich phase ("can be attributed"; "pointing to the absence" of a drug-rich phase). C33 The steady-state spectra of this run were "extremely noisy", "likely due to" the undissolved particles, so no ratio was calculated. R14·C34 The supernatant of the 50% medium held 18 µM, consistent with the UV value. R17·C38

Table B. Four read-outs for the four dispersions at 24 °C in pH 7.4 buffer (values without "about" are stated in the text; the rest are read from Figures 4 and 5, approximate; UV and HPLC are means of two). R6·R7·R8·R9·R11·R13·R15·R16·R17

Across both drugs, the authors state that only the 10% dispersions dissolved above the amorphous solubility and formed a disordered drug-rich phase. C39 They interpret the 10% result as "clearly polymer-controlled" dissolution, driven by the high aqueous solubility and hydrophilicity of the polymer, and the 50% result as "pointing to" drug-controlled dissolution, dictated by the low aqueous solubility and high hydrophobicity of the drug. C23·C24 They add that the same split had been seen for felodipine and bicalutamide dispersions with Kollidon VA64 at 5% and 50% loading. C25·R25 Release rates of drug and polymer were not measured here, so the control mechanism is an interpretation of where the concentration ended up. X10

Size of the drug-rich phase (Figure 6 and Table 2)

NTA of the filtered 10% media showed discrete scattering particles (Figure 6), with mean sizes of 696.4 ± 213.6 nm for ritonavir and 294.3 ± 67.7 nm for lopinavir (Table 2). R18·R19·C40 The authors call these amorphous precipitates and note that their size "greatly varies" with the compound. C41 The type of spread (±) and the number of measurements are not stated, and the amorphous nature of the particles rests on the supernatant concentrations matching the amorphous solubility, not on a solid-state measurement. X8·X5

Crystallization read through the probe (Figure 7)

Without any polymer, the two drugs parted ways. In the ritonavir solution the lifetime rose from about 1.00 ns to about 1.28 ns within the first minutes and stayed at about 1.31-1.37 ns up to 212 min; the authors take the near-constant lifetime as showing that ritonavir did not crystallize. R20·C44 In the lopinavir solution the lifetime first rose from about 1.00 to 1.31 ns, as the drug-rich phase formed, and then, after an induction period of 94 min, fell to 1.06 ns. R21 The authors attribute the decrease to PRODAN being expelled from the hydrophobic drug-rich phase, as crystals form, into the more hydrophilic aqueous environment ("can be attributed"); the value at 213 min, 1.06 ns, is close to that of PRODAN alone (1.00 ns). C45 They summarise that lopinavir's drug-rich phase was depleted within approximately 200 min by crystallization, while ritonavir's was preserved over the experiment. C46

Table C. Lopinavir lifetime in a supersaturated solution without polymer (read from Figure 7b, approximate; 1.31 and 1.06 ns from the text). The ritonavir lifetime under the same conditions stayed about 1.31-1.37 ns from 14 to 212 min. R21·R20

For ritonavir the emission ratio was also recorded. The authors describe it as "approximately constant"; read from Figure 7a it rose to about 1.57-1.64 in the first two hours and then declined to about 1.47 at 131-160 min and about 1.37 at 198-213 min. R20·C44 For lopinavir the spectra were too noisy, which the Figure 7 caption puts down to crystalline particles interfering with the emission. R22

Discussion

The authors' account links the results in one chain. Above the amorphous solubility the drug separates into a disordered drug-rich phase; the lipophilic probe moves into it and lives longer, and its emission shifts to the blue. C4·C17·C19 A 10% dispersion dissolves with its polymer and carries the drug past the amorphous solubility, so the drug-rich phase forms; a 50% dispersion dissolves at the pace of the drug and never gets there. C23·C24 Once formed, a drug-rich phase either persists (ritonavir) or crystallizes, and crystallization expels the probe back into water (lopinavir, without polymer). C44·C45·C46

The strength of this chain differs along its length. The 10% versus 50% contrast is the best-supported result: four independent read-outs agree for both drugs, and the supernatant concentrations measured after ultracentrifugation put the 10% media at the amorphous solubility and the 50% media well below it. R6·R7·R9·R11·R15·R17 The step from probe lifetime to "the probe is inside the drug-rich phase" is the authors' interpretation; it is consistent with the blue-shifted emission and the longer τ2 component, but the location of the probe was not measured directly. C17·C19·R3 The polymer-versus-drug-control explanation is drawn from the end concentrations, not from measured release rates. C23·C24·X10 And the crystallization of lopinavir rests on the lifetime alone. C45·X5

For formulation work, the authors' own messages are that the extent of supersaturation depends strongly on drug loading, that it matters which formulations produce LLPS because drug-rich species "may enhance" bioavailability, and that the lifetime can show whether a precipitate is amorphous or crystalline ("This study shows the utility"). C48·C7·C50 Note by the AI writer (not discussed in the paper): three practical points follow from the data themselves. First, for these two drugs in Kollidon VA64, the drug loading that still produces a drug-rich phase lies somewhere between 10% and 50%, and this study does not narrow it down. X10 Second, in situ UV alone would have misread both drugs once a drug-rich phase formed: a falling ritonavir signal that the authors attribute to aggregate growth, and a lopinavir signal far above any possible concentration; the ultracentrifuged supernatant was the reliable number. C28·C37·X6 Third, undissolved solid can raise the probe lifetime on its own, so a filtration control like the one used for the 50% lopinavir run is needed before a lifetime rise is read as LLPS. R13·C33

Limitations

Limits the authors state. The authors note that the UV apparent concentration "is not quantitative" when a drug-rich phase is present, and that the steady-state spectra could not be used when undissolved or crystalline particles were present (the 50% lopinavir dissolution and the lopinavir crystallization test). C37·R14·R22 They hedge their mechanistic statements: the lifetime increases "are attributed" to τ2 of probe in the drug-rich phase, the spectral data "suggest" partitioning, the UV decline is "likely to be an artifact", and the lopinavir lifetime decrease "can be attributed" to the probe being expelled. C17·C19·C28·C45

The AI writer's own judgement.

• Only two drug loadings were tested, so the switch from polymer- to drug-controlled dissolution is not located, and the release mechanism is inferred from end concentrations rather than measured. X10
• No replicate number or error bar is given for any lifetime or emission ratio, and the bi-exponential components behind the average lifetimes (Tables S-1 to S-8) were not available; lifetime differences of a few hundredths of a nanosecond should be treated as within unknown noise. X8
• The LLPS onset is bracketed by widely spaced concentrations (20-40 µM for ritonavir, 20-41 µM for lopinavir), so the "good agreement" with 39 and 28 µM is agreement with a bracket. X2
• Two different values serve as "the amorphous solubility": the cited UV-extinction onsets (39 and 28 µM, used for the dashed lines and some comparisons) and this paper's ultracentrifuge values (48 and 38 µM with Kollidon VA64), which are 9-10 µM higher; the difference is not discussed. X1
• Crystallization of lopinavir is inferred from the lifetime only, without XRPD, microscopy or a concentration measurement; the test ran without polymer and at a drug concentration that is not printed. X5·X4
• The ritonavir emission ratio in the crystallization test declined by about 15% over 213 min although the text calls it approximately constant; the lifetime, which stayed high, carries the conclusion. X3
• The probe-alone baseline differs between test series (about 1 ns, approximately 0.95 ns, 0.99-1.00 ns), and the 50% ritonavir run started at about 1.03 ns; lifetimes should be compared within a series. X9
• Some method details are misprinted or not rendered in the manuscript (sieve cut, XRPD symbols, HPLC column length). X7
• All tests ran at 24 °C in pH 7.4 phosphate buffer with two drugs and one dispersion polymer.

Conclusion

The paper shows that a polarity-sensitive probe can follow the fate of a supersaturated drug solution through its fluorescence lifetime. The lifetime of PRODAN rose above the LLPS onset of ritonavir and lopinavir, rose during dissolution only of the 10% (not the 50%) Kollidon VA64 dispersions, in agreement with extinction, ultracentrifuged concentrations and particle tracking, and fell back towards the probe-alone value when lopinavir's drug-rich phase, left without polymer, was read as crystallizing. R1·R2·R6·R9·R11·R17·R18·R21 The authors conclude that the extent of supersaturation is "strongly dependent" on drug loading and that the method "shows the utility" of the lifetime for telling an amorphous from a crystalline precipitate; the interpretation of the lifetime as probe partitioning, the polymer-versus-drug control of dissolution and the crystallization of lopinavir remain inferences that the paper does not test directly. C48·C50·C17·C23·C24·X5