B. Supersaturation from ASD dissolution

How it was shown

Alonzo and co-workers (2010) asked which crystallization route limits the concentration reached from an amorphous drug alonzo2010 C5. They melt-quenched felodipine and indomethacin, cryomilled them and kept the 100–300 µm fraction alonzo2010 P1. Felodipine was dissolved at 100, 500 or 1,000 µg solid per mL in 20 mL of 50 mM pH 6.8 phosphate buffer at 37 °C, and at 500 µg/mL at 25 °C, with fibre-optic UV every minute at 500 rpm; indomethacin at 300 µg/mL in 50 mL of 50 mM pH 2 phosphate buffer at 25 and 37 °C alonzo2010 P2, P3. Slurries of about 500–750 mg solid in 2 mL (felodipine) or 5 mL (indomethacin) of the same media were followed by in situ Raman spectroscopy, with a PLS model for felodipine crystallinity and, for indomethacin, the α band at 1,648 cm−1 and an amorphous-only band at 753 cm−1; wetted indomethacin was watched under polarized light alonzo2010 P5, P6, P7, P9. Raina and co-workers (2014) slurried about 500 mg of melt-quenched nifedipine in 2 mL of 50 mM pH 6.8 phosphate buffer, with or without 1 mg/mL PVP, HPMC or HPMCAS dissolved in it, or 1 g of 50:50 dispersions in 4 mL, at 25 and 37 °C; Raman spectra were taken every minute for 15 min and every 5 min up to 3 h and split into glass, γ and α by multivariate curve resolution (MCR), and particles were watched under crossed polarizers raina2014b P6, P7, P10, P4. A dissolution test of the neat glass and of the HPMC dispersion at 5 mg/mL solids and 37 °C gave the solution concentration raina2014b P11. Xie and Taylor (2015) melt-quenched celecoxib, sieved it to 106–250 µm, wetted particles with 3–4 drops of pH 6.8, 100 mM phosphate buffer, with or without 5 µg/mL HPMC or HPMCAS, under polarized light, slurried it in the same buffer at 37 °C with in situ Raman, and dissolved 8.8 mg in 400 mL at 37 °C and 300 rpm with in situ UV for 16 h, in triplicate, against the same mass of crystal xie2015 P2, P7, P3, P5. Wegiel and co-workers (2015) slurried about 300–500 mg of a 25% resveratrol–HPMCAS dispersion in 5 mL of pH 6.8 phosphate buffer, where HPMCAS dissolves, and of pH 1.2 HCl solution, where it does not, at 37 °C, followed the 1630/1605 cm−1 Raman ratio (0.76 for the crystal, about 1.1 for the amorphous dispersions; n = 3) and imaged the particles under polarized light after 24 h wegiel2015 P7, A3, R8, P9.

What the data show

Felodipine at 37 °C. The slurry was ≈54, ≈83 and ≈96% crystalline at 1, 5 and 24 min alonzo2010 R2. 100 µg/mL of amorphous solid followed almost the curve of the same mass of crystal (≈0.85 against ≈1.0 µg/mL near 120 min), and 500 and 1,000 µg/mL dissolved faster but plateaued at only ≈1.1 and ≈0.95–1.0 µg/mL, from ≈30 and ≈10 min alonzo2010 R1. Felodipine at 25 °C. The slurry was ≈10% crystalline at 1 min, ≈46% at 8 min and ≈89% at 29 min, and the solution peaked at ≈2.45 µg/mL at ≈16 min before falling to ≈0.8 µg/mL at 90 min alonzo2010 R2, R1. Indomethacin at 25 °C. Spectra after 120–240 min kept the amorphous band; dendritic needles appeared in the liquid between the particles within 60 min while the particles themselves did not crystallize; the solution peaked at ≈14.9 µg/mL at ≈25 min and was ≈2.7 µg/mL at 130 min alonzo2010 R11, R19, R7. Indomethacin at 37 °C. The slurry was largely amorphous at 20 min, clearly α at 90 min and like pure α at 120 min; the solution peaked at ≈11.2 µg/mL at ≈30 min, near the onset of conversion, and the powder agglomerated into a few large particles alonzo2010 R12, R13, R7, R10. Nifedipine. Particles were covered with birefringent crystals 15 min after wetting at 25 °C, in buffer and with each dissolved polymer; by MCR the glass fraction was ≈0.95, ≈0.61 and ≈0.35 at 3, 4 and 5 min and zero from 6 min; the neat glass gave ≈6 µg/mL in the dissolution test, on the crystalline-solubility line raina2014b R2, R6, R14. In the 50:50 PVP dispersion the particles shrank without turning birefringent and small plate-shaped crystals appeared in the surrounding solution within minutes; the authors state the same for the HPMC and HPMCAS dispersions without images raina2014b R13, X18. The HPMC dispersion reached ≈100 µg/mL at ≈22 min and fell to ≈41 µg/mL by 150 min raina2014b R14. Celecoxib. Wetted particles showed birefringent spots after 2 min and were widely crystalline at 10 min, with needles also in the surrounding liquid; in the slurry crystallization was complete within an hour xie2015 R8, R4. The amorphous solid dissolved faster at first (≈0.49 against ≈0.13 µg/mL at 20 min) but then followed the crystal, ending at ≈1.70 against ≈1.85 µg/mL at 16 h xie2015 R7. Resveratrol. The Raman ratio of the 25% HPMCAS dispersion fell from ≈1.05 to ≈0.96 at pH 6.8 and from ≈1.05–1.08 to ≈0.93 at pH 1.2 over about 1440 min, the curves overlapping within their error bars; at pH 1.2 the particles (≈200–300 µm) stayed intact with birefringent regions inside, at pH 6.8 they dispersed and left small crystalline particles wegiel2015 R17, R18. Posaconazole and delamanid in acid. Posaconazole–HPMCAS dispersions crystallized within 30 min in FaSSGF although the medium stayed below the crystalline solubility, at about 68% of the undissolved drug at 50% drug, and delamanid dispersions crystallized at their surfaces at pH 3.0 and 5.0 while the drug concentration in the acid hardly fell (F24) elkhabaz2019a R16, R20, R1; nguyen2023c R19, R23.

Figure

At 37 °C (open circles) the slurry is already above 50% crystalline at the first reading; at 25 °C (triangles) 50% takes about 8 min, the window in which some supersaturation could form alonzo2010 R2.

The dispersion particles shrink without birefringence while fine crystals multiply in the liquid around them between 1 and 10 min raina2014b R13.

The two curves overlap within their error bars although HPMCAS dissolves only at pH 6.8: the crystals form in or at the matrix whether or not the polymer leaves it wegiel2015 R17, C15.

Why

For felodipine at 37 °C the authors state that the amorphous solid "crystallizes more rapidly than it can dissolve", so that the crystal is what dissolves and no amorphous advantage appears alonzo2010 C8, C33. They explain the difference between 25 and 37 °C by the glass transition: dry felodipine has a Tg of about 45 °C and just below 35 °C after exposure to 75% RH (cited), so the hydrated solid at 37 °C is above Tg and "presumably" crystallizes faster alonzo2010 C34, M1. Indomethacin crystallized much more slowly although its Tg also falls with water and its driving force is larger, which they take as "an indication" that a single parameter such as Tg "will not fully explain" the behaviour alonzo2010 C42. For the falling indomethacin concentration at 25 °C they point to the needles in the liquid: crystallization from solution is the limiting factor there ("These observations indicate") alonzo2010 C31, C40. At 37 °C they combine agglomeration (less surface), conversion of the solid to α and weaker solution inhibition ("Presumably") alonzo2010 C41, C21. Fig. 16 of the paper sets the competition out as a scheme in which dissolution scales with surface area and a concentration difference and nucleation and growth rise with supersaturation alonzo2010 C43, E3, E4, E5. For nifedipine, a relatively fast crystallizer, Raina and co-workers find the matrix route "primarily" for the neat glass and argue that its inherent faster crystallization "appears" to carry over from earlier solid-state work raina2014b C10, C28, R17; in the dispersions the polymer keeps the hydrated matrix amorphous, so the route "has changed from the matrix to the solution phase" ("strongly suggest") raina2014b C21, C24. An alternative reading for felodipine at 25 °C is that the fall after the peak comes partly from loss of the dissolving amorphous phase, since the slurry was ≈70% crystalline at 15 min; the routes were not separated there alonzo2010 X6. For celecoxib, Xie and Taylor conclude that the neat amorphous drug "appears to undergo rapid crystallization", "primarily via the matrix route", which they relate to water lowering the Tg (not measured); the low concentration "indicates either" rapid complete matrix crystallization or rapid growth of seed crystals from the residual amorphous material xie2015 C26, C1, C32, C34. For resveratrol, a rapid crystallizer in the cited classification, Wegiel and co-workers infer from the alike kinetics at pH 6.8 and 1.2 that the route "may also be analogous", the crystals at pH 6.8 "most likely" forming from the matrix rather than from a supersaturated solution wegiel2015 R22, C14, C15; crystallization "appeared to occur in or at the surface of the matrix" wegiel2015 C16. (synthesis: in all five drugs the outcome follows from one comparison, how fast the wetted solid crystallizes against how fast it dissolves; the temperature effect for felodipine and the polymer placement for nifedipine both act through that comparison, and which polymer forms the dispersion decides the solid side of it, see F16.) For posaconazole in gastric fluid the authors argue that "Solution-phase crystallization cannot occur" in a subsaturated medium, so the crystals formed in the matrix elkhabaz2019a C34, and for delamanid the solution route was ruled out by a constant concentration in the acid and an ineffective solution inhibitor nguyen2023c C37, C39.

Conditions and limits

Neat melt-quenched powders and 50:50 dispersions; felodipine and nifedipine in pH 6.8 buffer, indomethacin in pH 2 buffer; 25 and 37 °C; up to 4 h alonzo2010 A1, A2; raina2014b A2, A3. The Raman slurries held about 250–375 mg/mL felodipine against 500 µg/mL in dissolution, so the polymer-to-solid ratio differed about 500–750-fold, and the nifedipine slurries held 125–250 mg/mL drug; slurry timings therefore rank systems rather than give the time scale in a dissolution vessel alonzo2010 X5; raina2014b X9. Replicates are not stated for any slurry run alonzo2010 X8; raina2014b X14. The two drugs of Alonzo were tested in different media and set-ups, so their comparison is qualitative alonzo2010 X9. The celecoxib micrographs do not always show one field, and the amount of neat amorphous celecoxib slurried is not stated xie2015 X13, X11; the resveratrol Raman ratio is uncalibrated, and the pH comparison has no statistical test wegiel2015 X13, C14.

Counterexamples and conflicts

Not every neat amorphous drug loses its advantage on wetting: in the crystallization-from-solution chapter amorphous ritonavir dissolved to 19.8 of a calculated 20.6 µg/mL, while celecoxib reached only 2.5 of 22.6 µg/mL, and amorphous efavirenz dissolved slowly without birefringence for 3 h ilevbare2013c R2, R5, C23, and in the LLPS chapter amorphous films of three weak bases reached 69–110% of their LLPS concentration (Amorphous solubility and LLPS F13). In acidified suspensions of celecoxib–HPMCAS particles, where the polymer barely dissolves, the particles stayed crystal-free for 48 h without surfactant and crystallized on or in the particles only after about 7 days chen2015a R14, R20, C33. No paper in the library contradicts the two-route frame itself.

Transfer check

Wet the neat amorphous drug with the dissolution medium at 37 °C and follow the solid (Raman, PXRD or polarized light) during the first 15–30 min while the concentration is read in a non-sink test alonzo2010 P5, P8; raina2014b P4. Above about 50% crystallinity within minutes, as for felodipine at 37 °C and nifedipine, the neat glass dissolves like the crystal and the polymer has to be in the solid; if the solid stays amorphous while the concentration falls, as for indomethacin at 25 °C, crystallization from solution is the problem and a solution inhibitor is the lever alonzo2010 C33, C40. Test at 37 °C: felodipine supersaturated at 25 °C but not at 37 °C alonzo2010 C6. For a dispersion of an enteric polymer, a slurry at a pH where the polymer stays undissolved (pH 1.2 for HPMCAS) shows matrix crystallization without the solution route wegiel2015 A3, C16.

History

The two ways in which an amorphous form loses its advantage were proposed in cited work alonzo2010 C2. Alonzo and co-workers (2010) were the first in the library to separate them in situ, for felodipine and indomethacin alonzo2010 C47. Raina and co-workers (2014) added nifedipine, the switch from the matrix route of the neat glass to the solution route of the dispersion, and a metastable crystal form on both routes (F3) raina2014b C21, C29. Wegiel and co-workers (2015) added resveratrol and used an undissolving medium to show the matrix route in a dispersion wegiel2015 C16; Xie and Taylor (2015) added celecoxib, whose neat glass crystallized mainly on its particles xie2015 C1. The title was widened from "where the polymer sits" to "which polymer is present and where it sits" when the papers of 2015–2018 showed that the dispersion polymer itself decides whether the wetted matrix crystallizes (F16). Later papers of the library cite Alonzo 2010 for matrix against solution crystallization during dissolution, among them work on celecoxib dispersions, on ezetimibe and posaconazole dispersions and on residual crystallinity (builds_on, as listed in the evidence index of this subcategory). Elkhabaz and co-workers (2019) and Nguyen and co-workers (2023) found the matrix route in gastric media, where an enteric polymer stays undissolved and the solution cannot be the source (F24) elkhabaz2019a C23; nguyen2023c C43.

How it was shown

Alonzo and co-workers (2010) slurried amorphous felodipine at 37 °C in pH 6.8 buffer containing 500 µg/mL PVP K29/32, HPMC 606 or HPMC-AS MF and quantified crystallinity by Raman PLS; wetted powder with 1 mg/mL PVP or HPMC-AS was examined by PXRD after 15 min; dissolution of 500 µg/mL amorphous felodipine was run with the same polymers predissolved at 500 µg/mL alonzo2010 P5, P6, P8, P2. Indomethacin slurries at 37 °C with 250 µg/mL PVP or HPMC were followed by Raman (data not shown) alonzo2010 R16. Raina and co-workers (2014) watched neat nifedipine glass in buffer with 1 mg/mL dissolved PVP, HPMC or HPMCAS under crossed polarizers and in Raman slurries at 25 and 37 °C, compared it with 50:50 dispersions of the same polymers in plain buffer, and placed the HPMCAS dispersion in pH 2 buffer to hold its matrix undissolved raina2014b P4, P6, P7, P5. Xie and Taylor (2015) watched amorphous celecoxib particles wetted with buffer containing 5 µg/mL HPMC or HPMCAS under polarized light for 60 min, slurried amorphous celecoxib with 22 µg/mL of a polymer, and dissolved 8.8 mg of it in 400 mL of buffer holding 5 µg/mL HPMCAS or HPMC at 37 °C for 16 h xie2015 P7, P3, P5, A5.

What the data show

Felodipine. Every polymer turned the crystallization curve sigmoidal: at 10 min the solid was ≈8% crystalline with PVP, ≈20% with HPMC and ≈32% with HPMC-AS against ≈90% without polymer, and all were ≈92–95% crystalline by ≈22–25 min alonzo2010 R6. By PXRD the wetted powder without polymer showed sharp felodipine peaks, with PVP none, and with HPMC-AS a largely amorphous pattern with a small peak at 10° 2θ alonzo2010 R5. In dissolution, PVP gave a peak of ≈7.3 µg/mL at ≈15 min that fell to ≈2.1 µg/mL at 130 min, HPMC ≈8.9 µg/mL at 40–50 min and ≈7.1 at 240 min, and HPMC-AS ≈12.0 µg/mL at ≈100 min and ≈10.2 at 240 min, while crystalline felodipine reached ≈1.1 µg/mL alonzo2010 R4. Indomethacin at 37 °C. Neither polymer substantially delayed the onset of crystallization of the slurried solid alonzo2010 C26, R16, although both held solvent-shift solutions far longer than buffer (F4) alonzo2010 R17. Nifedipine. With dissolved polymer, γ was the majority form at the first measurement at both temperatures and the γ profiles reached their maximum within ≈2–11 min, as in buffer raina2014b R7, R9. In the dispersions the particles did not become birefringent, and γ was detected in the slurry at 1 min with every polymer, which the authors read as crystallization from the solution raina2014b R13, R11, C21. At pH 2 the HPMCAS dispersion did not dissolve quickly, stayed non-birefringent and amorphous by Raman for 3 h, and only a few small crystals that did not grow substantially appeared in the solution after about 1 min raina2014b R15. Celecoxib. With 5 µg/mL HPMC a few small birefringent spots appeared on the particles at 20 and 60 min and none in the liquid; with HPMCAS spots at 5 min and birefringent regions on several particles at 60 min, again none in the liquid xie2015 R9, R10. In dissolution the concentration rose to ≈3.05 (HPMCAS) and ≈3.50 µg/mL (HPMC) at 16 h, against ≈1.70 µg/mL without polymer xie2015 R7. With 22 µg/mL of a polymer in the slurry, crystallization was "not substantially retarded" according to the results and lasted longer according to the discussion; neither is shown xie2015 R4, X7.

Figure

Without polymer (filled circles) the solid is crystalline at once; with PVP, HPMC or HPMC-AS the curves are sigmoidal with a lag of roughly 10–20 min, the window in which a supersaturated solution forms alonzo2010 R6, C14.

All four media (B–E) give birefringent particles: polymer in the medium does not protect the nifedipine glass raina2014b R2.

Why

Alonzo and co-workers "speculate" that crystallization starts at the surface of the amorphous particle, is made worse by water plasticization, and that the small amount of dissolved polymer interacts with that surface and delays surface crystallization, as a thin polymer film is known to do (cited); adsorption on the amorphous surface was not measured alonzo2010 C36, R22, R23. The felodipine result is "consistent with" an earlier report of HPMC in the medium protecting an amorphous macrolide during dissolution alonzo2010 C35, R21. Raina and co-workers note that amorphous nifedipine and felodipine have similar Tg and moisture sorption (cited), so the difference that appears only with polymers present "appears" to come from the faster inherent crystallization of nifedipine raina2014b C27, C28, R17. Inside a dispersion the polymer prevents crystallization of the hydrated matrix, and at pH 2 the slow dissolution of HPMCAS means low supersaturation and slow nucleation and growth from solution raina2014b C24, C25, C37. (synthesis: a polymer in the medium can act only on a surface it reaches before nuclei form; the nifedipine glass was unchanged for only about 2 min before MCR showed γ raina2014b R6, whereas the felodipine solid with polymer stayed below about 30% crystalline for 10 min alonzo2010 R6. This reading uses the authors' surface hypothesis, which neither paper tested.) For celecoxib, Xie and Taylor conclude that polymers dissolved in solution are "clearly unable" to inhibit crystal formation in the amorphous matrix, which "confirm[s] the need to prevent matrix crystallization", and that the polymer must be present in the dispersion to act xie2015 C35, C36, C16. A later celecoxib paper of this chapter reached the same conclusion for PAA dispersions in a medium holding HPMCAS: the dissolved drug stayed supersaturated but the drug left in the particles crystallized ("It therefore appears") xie2016a C31 (F16).

Conditions and limits

Polymer in the medium at 250, 500 or 1,000 µg/mL; no dispersion of felodipine or indomethacin was dissolved in this paper, and nifedipine dispersions were 50:50 only alonzo2010 A4; raina2014b A3. The PXRD check is one time point at twice the polymer level of the slurries and dissolution tests alonzo2010 X13. Early crystallinity values with polymer (about 5–12%) lie within the Raman model error (RMSEE 7.21) alonzo2010 X8. For the HPMC and HPMCAS dispersions at pH 6.8 the amorphous matrix is stated, not shown raina2014b X18.

Counterexamples and conflicts

A polymer in the dispersion does not always keep the matrix amorphous: bicalutamide–PVPVA particles crystallized within the particles during dissolution, and a telaprevir dispersion with 50% HPMCAS crystallized in the matrix and never phase separated moseson2021 C27; mosqueragiraldo2018b C40. Among the ritonavir dispersions of this chapter, HPMC in the dispersion gave no crystallization in 2 h of dissolution (F12) ilevbare2013d C15. At 35% drug the cellulosic dispersions of resveratrol crystallized in the matrix within the first hour, and only 80–95% polymer prevented it (F16) wegiel2015 R9, R14. No paper of the library reports a polymer in the medium protecting wetted nifedipine or celecoxib; in an ASD process study, polymeric additives in the medium were "largely ineffective" against matrix crystallization of bicalutamide dispersions, at best slightly delaying nucleation moseson2021 C72.

Transfer check

Wet the neat amorphous drug with medium that holds the candidate polymer at the concentration the dispersion would release, and follow crystallinity for 15–30 min at 37 °C alonzo2010 P5, P8. A lag of several minutes, as for felodipine, means the polymer can act from solution; no lag, as for nifedipine, means it must be molecularly mixed with the drug raina2014b C11, C24. For the dispersion itself, check the wetted particles for birefringence at the intended drug loading, since high loadings can crystallize in the matrix moseson2021 C27.

History

HPMC in the medium protecting an amorphous solid during dissolution was reported in earlier cited work alonzo2010 R21. Alonzo and co-workers (2010) showed it in the library for felodipine with three polymers and two methods alonzo2010 C13, C14. Raina and co-workers (2014) found that it fails for nifedipine and that the polymer has to sit in the dispersion, and checked this with an HPMCAS dispersion at pH 2 raina2014b C11, C24. Xie and Taylor (2015) found the same for celecoxib and added that the dissolved polymer still stops crystallization in the liquid xie2015 C26, C35; the title gained celecoxib then. Papers of the ASD process and Bioperformance chapters later describe matrix crystallization of high-loading dispersions (bicalutamide, telaprevir) moseson2021 C27; mosqueragiraldo2018b C40, and F16 collects what the dispersion polymer itself does. Nguyen and co-workers (2023) found that 1 mg/mL PVPVA in the medium, a known solution inhibitor of delamanid, left release after a pH 3.0 stage unchanged where the dispersions crystallized at their surfaces (F24) nguyen2023c R22, P16.

How it was shown

Raina and co-workers (2014) followed neat nifedipine glass (about 500 mg in 2 mL of pH 6.8 phosphate buffer, with or without 1 mg/mL dissolved polymer) and 50:50 dispersions (1 g in 4 mL) by Raman spectroscopy every minute for 15 min and every 5 min up to 3 h, at 25 and 37 °C raina2014b P6, P7. Forms were identified by the C−C−O band (glass 1214, α 1224, γ 1205 cm−1) raina2014b P8, M1. PCA showed that three components explained close to 96% of the variance, and MCR with non-negativity and closure split each series into the spectra of glass, γ and α and their relative amounts over time, without calibration mixtures; the γ reference was a time point at which only γ was present raina2014b R3, P10, C6, C7, E1. The resolved spectra matched measured reference spectra of the pure forms raina2014b R4, C9.

What the data show

Neat glass in buffer. At 25 °C γ was 1.0 from 6 to 13 min, then ≈0.88, ≈0.69, ≈0.47, ≈0.30 and ≈0.16 at 14, 15, 20, 25 and 30 min, and 0 from 40 min, mirrored by α (≈0.12 at 14 min, ≈0.53 at 20 min, 1.0 from 40 min); the authors read a crystallization step "complete within about 10 min" followed by conversion, with a γ half-life of about 20 min raina2014b R6, R7, C13. At 37 °C the γ plateau ended at ≈9 min and α was complete by 30 min raina2014b R8. The β form and form IV were never seen raina2014b R16. With dissolved polymer. At 25 °C the last time γ was still 1.0 was ≈13 min in buffer, ≈20 min with PVP, ≈60 min with HPMC and ≈110 min with HPMCAS; α reached 1.0 at 40, 60, 105 and 145 min raina2014b R9, R10. At 37 °C the plateau ended earlier (≈9, ≈12, ≈35 and ≈60 min), but α was complete only at 30, 80, 135 and 135 min raina2014b R9, R10. The peak-ratio summary gives the same order raina2014b R7. Dispersions. By peak ratios γ was detected at 1 min in every dispersion and α at 3, 25 and 30 min (25 °C) or 4, 7 and 9 min (37 °C) for PVP, HPMC and HPMCAS raina2014b R11. By MCR at 25 °C, γ vanished at 4 min in the PVP dispersion with α ≈0.99 at 5 min, and in the HPMC and HPMCAS dispersions γ declined from ≈5–7 min to 0 at 35 min while α rose to 1.0 at 40 min raina2014b R12. Solution-grown crystals were below 2 µm, against 1−300 µm for the crystallized glass raina2014b R13, R2.

Figure

The flat γ plateau lengthens from buffer to PVP, HPMC and HPMCAS, while the decline after it is similarly steep at 25 °C (A); at 37 °C (B) the plateau is shorter but the cellulose curves decline more slowly raina2014b R9, X11.

Why

The authors list three ways in which a polymer could slow the conversion: adsorption on the parent γ crystals that reduces heterogeneous nucleation of α on them, inhibition of α nucleation from solution, and inhibition of α growth (cited) raina2014b C31. Because the lag before α appears lengthens while the slopes after onset are similar, "particularly at 25 °C", they judge it "most likely" that the polymers interfere with nucleation of α; slower growth would have flattened the curves, as polymers do for anhydrate-to-hydrate conversions (cited) raina2014b C32, R19. For the faster conversion of the dispersions they propose that the dispersion dissolves faster and "may" reach a higher supersaturation, which "would" give faster nucleation and many small crystals whose larger surface dissolves faster and offers more sites for heterogeneous nucleation, "which may explain" the faster conversion ("most likely explained by the faster dissolution rate") raina2014b C33, C36. When nucleation and growth are fast, as they "appear to be" for nifedipine, dissolution controls the solution-mediated transformation raina2014b C35, C34. Metastable forms are common in melt crystallization because conversion is slow in a viscous melt and hard to catch in solvents where it is fast; the one-minute Raman sampling caught γ raina2014b C30. Supersaturation in the slurries was not measured: the only dissolution test used the HPMC dispersion at a 50-fold lower solid loading raina2014b X10. (synthesis: the nucleation reading rests on curve shapes alone, and the repeated values in three 25 °C profiles call for the original data before slopes are compared raina2014b X11.)

Conditions and limits

One drug, 1 mg/mL dissolved polymer, 50:50 dispersions only, 25 and 37 °C, pH 6.8, concentrated slurries (125–250 mg/mL drug) raina2014b A1, A2, A3, X9. MCR gives relative amounts under a closure constraint, checked only against peak ratios; γ and α do not always add up to 1, and the two analyses date the same events up to ≈15 min apart raina2014b X15, X6, X5. No MCR profiles of the dispersions at 37 °C are shown raina2014b X3. "Faster at 37 °C" holds for the onset of conversion, not for its completion with HPMC raina2014b X7.

Counterexamples and conflicts

The authors write that the dispersions "stabilized the metastable polymorph" raina2014b C41; at 25 °C γ was gone by ≈35 min in the cellulose dispersions, no later than for the neat glass in plain buffer (≈40 min), so the stabilization holds only relative to the PVP dispersion raina2014b X8, R12, R6. Indomethacin in this chapter crystallized in the wetted solid at 37 °C as the α form, with no other form reported alonzo2010 R12; a later indomethacin paper of the chapter describes α as the slightly less stable monotrope of the stable γ form (cited) xie2017 M1, so indomethacin too crystallized first as a metastable form, and it was again α, of 5–10 µm, that precipitated from the highly supersaturated solution of an EPO dispersion at pH 2.2, identified by SHG, Raman and PXRD xie2017 R23, R24, R25, C46 (F21). None found in the library against the γ-first pattern for nifedipine.

Transfer check

For a drug with a known metastable form, follow the solid by in situ Raman during dissolution with and without the candidate polymer, and use MCR or another calibration-free resolution if the metastable form cannot be isolated raina2014b C5, C43. Compare the solubility ratio of the forms: nifedipine's estimated 2.4 makes the lifetime of γ matter, whereas a ratio of 1.2 (mefenamic acid, cited) gave about equal plasma levels raina2014b C39, C40, R20. A lag before the stable form appears that lengthens with polymer, at unchanged slope, points to nucleation inhibition of the stable form raina2014b C32.

History

The scheme of two routes, each through a metastable form I to a stable form II, was adapted by Raina and co-workers (2014) from Alonzo and co-workers (2010) raina2014b C4. Polymer delay of an anhydrate-to-hydrate conversion in slurries (cited) was the model for the slope argument raina2014b R19, C32. None of the other papers of this chapter revisits nifedipine polymorphs; the dihydropyridine series of the crystallization-from-solution chapter found nifedipine precipitating within 0.25 h from a solvent shift raina2015b R6. Xie and co-workers (2017) found the metastable α form of indomethacin precipitating from a dissolving EPO dispersion, "expected" from the high supersaturation (cited), and showed that these small crystals redissolved faster than sieved α crystals when the pH rose xie2017 C45, C46, C4. The earlier wording of this finding, that indomethacin formed no metastable form, overlooked that α is the metastable indomethacin form and was corrected then.

How it was shown

Wegiel and co-workers (2015) made resveratrol dispersions by rotary evaporation (53–250 µm) with PVP, E100, HPMC, HPMCAS or CMCAB at 35% drug, and with the three cellulosic polymers at 60–95% polymer, slurried about 300–500 mg in 5 mL of pH 6.8 phosphate buffer at 37 °C, followed the 1630/1605 cm−1 Raman ratio every 5 min in the first hour and then less often up to 24–48 h (n = 3), checked the dried residues by XRPD and measured dissolved drug and polymer after about 24 h wegiel2015 P5, P7, A1, A2, P8. Xie and Taylor (2015) slurried 200 mg of 50% drug celecoxib dispersions (106–250 µm) in 2 mL of pH 6.8 buffer at 37 °C and followed the amorphous Raman shoulder near 1598 cm−1 every 15 min for 7 h xie2015 P1, P3, R3. Xie and Taylor (2016) dissolved PAA dispersions at 50% and 10% drug (dose 22 µg/mL in 400 mL, 37 °C) in buffer holding 5 µg/mL predissolved HPMCAS, retrieved the undissolved solid soon after the peak for polarized light, and watched wetted particles under polarized light from 20 s to 30 min xie2016a P6, P8. Li and Taylor (2018) suspended 50% drug lopinavir films with CA, CAP, EC, EUD, HPMCAS, HPMCP or HPMC in pH 3 citrate buffer at 25 °C without a crystallization inhibitor, imaged the particles under polarized light at 8–48 h and assayed the solution by HPLC at 2–48 h li2018 P6, P7.

What the data show

Resveratrol. The PVP dispersion stayed at a Raman ratio of ≈1.12–1.13 and E100 at ≈1.13, while HPMC fell from ≈1.07 to ≈0.87, HPMCAS from ≈1.05 to ≈0.84 and CMCAB from ≈1.03 to ≈0.81, mostly within the first hour, never reaching the crystal's 0.76; XRPD of the residues agreed wegiel2015 R9, R10. More polymer meant less crystallization: 85% HPMCAS stayed at ≈1.10 while 75% and 65% fell to ≈0.96 and ≈0.84; 95% CMCAB stayed flat while 90%, 85% and 65% fell to ≈0.99, ≈0.86 and ≈0.80; 80% HPMC stayed flat while 70% and 60% fell to ≈0.92 and ≈0.83 wegiel2015 R13. The minimum levels that stopped crystallization were 95% (CMCAB), 85% (HPMCAS) and 80% (HPMC); E100 and PVP were tested only at 65%, so their entries are upper bounds wegiel2015 R14, X7. The more the dispersion crystallized, the less it raised the concentration: the HPMCAS dispersions at 85, 75 and 65% polymer gave ≈1250, ≈885 and ≈460 µg/mL at 24 h wegiel2015 R15, C12. Celecoxib. The PVP-K12 dispersion lost its amorphous band between 2.5 and 4 h and showed a sharp band near 1550 cm−1 like the crystal; PVP-K29/32, HPMCAS and both HPMCAS ternaries kept the amorphous pattern for 7 h; the HPMC-containing dispersions were not reported xie2015 R4, X11. Wetted PAA 5:5 particles dissolved from 20 s and showed needle crystals at 5 and 10 min, too fast to tell matrix from solution crystallization, whereas HPMCAS 3:7 particles dissolved slowly and stayed crystal-free for 30 min xie2016a R7, C28, R8. With HPMCAS in the medium the PAA 5:5 dispersion peaked at ≈9.7 µg/mL within ≈5 min and still held ≈8.8 µg/mL at 16 h, the 1:9 one ≈11.6 and ≈10.8 µg/mL, of a possible 22 µg/mL; the residue retrieved after the peak was birefringent needles xie2016a R9, R10. PAA dispersions with 10% HPMCAS at 40–50% drug levelled at ≈15.9–16.8 µg/mL, which the authors take as continuing matrix crystallization ("it is apparent") xie2016a R14, C41. Lopinavir. EUD and CAP particles had marked regions at 8 h and coloured crystals at 16 h, CA and EC marked regions at 16 h, HPMCAS faint birefringence at 16 h, and HPMCP particles none at 48 h; amorphous lopinavir was a fibrous crystalline mass at 8 h li2018 R7. In solution EUD and CAP lost their supersaturation within about 10–12 h, like amorphous lopinavir, HPMCAS, CA and EC after about 12 h, and HPMCP stayed at ≈2.5–3.0 µg/mL, at or below its crystalline solubility, for 24 h li2018 R6, C25, X6.

Figure

The highest level of each polymer stays flat; every lower level falls within hours: the polymer content, not only the polymer, decides whether the wetted matrix crystallizes wegiel2015 R13.

Release stops at about half the dose and then does not fall for 16 h: the dissolved drug is protected, the drug left in the particles is not xie2016a R9, C31.

Why

For resveratrol, a rapid crystallizer whose dispersions had already differed in storage (PVP with strong hydrogen bonds and E100 with an ionic interaction stable, the cellulosics not; cited), Wegiel and co-workers think crystallization of the cellulosic dispersions is due to "insufficiently strong" drug–polymer interactions, its dependence on polymer content "most probably" reflecting the limited number of polymer groups able to hydrogen-bond with the drug, and a finer dispersion at low drug loading "will also" hinder it; none of these was measured in the wet state wegiel2015 R21, R22, C36, C37, C38. For celecoxib, Xie and Taylor find the drug in a dispersion "considerably more stable" on contact with buffer than the neat glass, with a higher PVP grade more resistant, but note that Raman alone does not identify the route by which the PVP-K12 dispersion crystallized xie2015 C37, C15, C38. For the PAA dispersions they conclude that matrix crystallization, which HPMCAS in solution cannot influence, competes with release and that once the matrix has crystallized no further drug can dissolve; since 10% drug did not help, PAA is "a very poor inhibitor of matrix crystallization, irrespective of the drug loading" xie2016a C31, C33. The inhibitor therefore has to stay in close contact with the drug during hydration and be released with it; in PAA–HPMCAS ternaries the two polymers "presumably" separated, leaving part of the drug without it (F20) xie2016a C54, C62. The extractor notes that the needles could also have nucleated in the concentrated layer around the dissolving particles xie2016a X6. For lopinavir, matrix crystallization followed the solution trends, and the authors expect crystallization of suspended dispersions to depend on the supersaturation, the solubility of the polymer and its inhibition in solution and matrix li2018 C26, C60. (synthesis: in the three drugs the protection of the wetted matrix followed the polymer in the particle and its share, not the polymer in the medium (F2); once crystals form in the particle they cap what the particle can release (F1) and seed the solution, and the crystallization-from-solution chapter finds that dissolved polymers at µg/mL levels rarely stop growth on seeds (Crystallization from solution F20); in the PAA case HPMCAS in the medium did slow growth enough to hold the dissolved drug xie2016a C36.)

Conditions and limits

Resveratrol at pH 6.8 (and 1.2, F1), 37 °C; celecoxib at pH 6.8, 37 °C; lopinavir at pH 3, 25 °C and only at 50% drug wegiel2015 A1; xie2015 P3; li2018 P6. The resveratrol and celecoxib slurries held about 60–100 and 100 mg/mL solids, and the solids loading of the resveratrol comparison is unresolved wegiel2015 X6; xie2015 P3. The HPMC-containing celecoxib dispersions were not checked in the slurry xie2015 X11. The amount and location of the crystals in the PAA dispersions were not determined xie2016a X6.

Counterexamples and conflicts

Polymer in the medium did delay crystallization of wetted amorphous felodipine (F2) alonzo2010 R6. The PVP-K12 celecoxib dispersion that crystallized in the slurry within 4 h released completely in dissolution, desupersaturated only from solution, and was fully protected by 5 µg/mL HPMCAS or HPMC in the medium, so a slurry result need not carry over to a dilute dissolution test xie2015 R4, R6, C39. Stopping matrix crystallization did not by itself give supersaturation: PVP and E100 kept resveratrol amorphous but gave no more dissolved drug than the crystal (F17) wegiel2015 C9. More polymer is not always safer: an enzalutamide dispersion with 90% HPMCAS was read as crystallizing in the matrix where a 50:50 one did not, an interpretation the extractor flags because no wetted solid was analysed wilson2018 C56.

Transfer check

Slurry the dispersion at the intended drug loading and at one or two higher polymer contents in the medium at 37 °C, and follow crystallinity by Raman, XRPD or polarized light for several hours wegiel2015 P7, A2. Dissolve it with a solution inhibitor predissolved: if release stops well below the dose while the dissolved drug holds, the matrix has crystallized, and the residue should be birefringent xie2016a R9, R10. Treat a very hydrophilic, fast-dissolving carrier (PAA) as a poor matrix inhibitor until shown otherwise xie2016a C33; elkhabaz2017 C34.

History

Alonzo and co-workers (2010) separated the two routes and Raina and co-workers (2014) showed that a dispersion moves nifedipine from the matrix to the solution route (F1, F2) alonzo2010 C47; raina2014b C21. Wegiel and co-workers (2015) were the first in the library to measure the polymer content needed to stop matrix crystallization during dissolution wegiel2015 C11. Xie and Taylor (2015, 2016) ranked celecoxib dispersions in slurries and showed that an inhibitor in the medium cannot replace the dispersion polymer against matrix crystallization xie2015 C37; xie2016a C31. Li and Taylor (2018) added lopinavir with six insoluble polymers at 50% drug li2018 C26. Later papers of the Bioperformance and ASD process chapters made matrix crystallization a main risk of high-loading dispersions mosqueragiraldo2018b C40; moseson2021 C72; moseson2022b C67. Elkhabaz and co-workers (2019) measured the loading dependence of matrix crystallization in gastric fluid for posaconazole–HPMCAS, and Nguyen and co-workers (2023) its polymer dependence for delamanid (F24) elkhabaz2019a C36; nguyen2023c C4.

How it was shown

Konno and co-workers (2008) made felodipine dispersions with PVP K30, HPMC 606 and HPMCAS AS-MF at 10–75% w/w polymer by solvent evaporation, dissolved powder holding 30 mg felodipine in 600 mL of pH 6.8 test fluid at 37 °C and 50 rpm (50 µg/mL if all dissolved), read UV through a 70-µm filter for 6 h in triplicate, and measured crystalline solubility with 1 mg/mL polymer over 24 h konno2008a P1, P3, P4, X15. Alonzo and co-workers (2010) dissolved amorphous felodipine (500 µg/mL) with 500 µg/mL polymer in pH 6.8 buffer at 37 °C, and amorphous indomethacin (300 µg/mL) with 250 µg/mL PVP or HPMC in pH 2 buffer at 25 and 37 °C alonzo2010 P2, P3. Alonzo and co-workers (2011) followed felodipine particle growth by light scattering after a methanol shift without polymer and during dissolution of 90:10 PVP and HPMC dispersions at 25 °C alonzo2011 P5, R10. Jackson and co-workers (2015) spin-coated danazol films with PVP K29/32, HPMC 606 or HPMCAS MF at 90:10, 70:30 and 50:50 polymer:drug and dissolved them in 50 mL of 10 mM pH 6.8 phosphate buffer with pyrene at 25 °C and 300 rpm, 20–30 µg/mL if all dissolved; the induction time was read where the UV concentration fell and the pyrene ratio rose jackson2015 P1, P2, P3. Ilevbare and co-workers (2013) added a ritonavir stock to 100 mM pH 6.8 buffer at 37 °C to 20 µg/mL with 15 additives at 5 µg/mL (F12) ilevbare2013d P6. Xie and Taylor (2015, 2016) pipetted a celecoxib methanol stock into 50 mL of 100 mM pH 6.8 phosphate buffer at 37 °C and 300 rpm to 22 µg/mL, about the cited amorphous solubility, with 5 µg/mL of PVP-K12, PVP-K29/32, HPMCAS, HPMC, PAA or pairs, and read the onset as a rise of extinction at 350 nm with a fall of absorbance at 249 nm, in triplicate for 8–16 h xie2015 P4, C11; xie2016a P3. Xie and co-workers (2017) pumped an indomethacin ethanol stock into pH 2.2 McIlvaine buffer at 37 °C with 100 µg/mL HPMC or EPO and followed absorbance and extinction for 4 h xie2017 P4. Trasi and co-workers (2016) followed a 50 µg/mL tacrolimus solution in 5 mM pH 4.5 phosphate buffer at 37 °C without and with 50 µg/mL HPMC, in triplicate trasi2016 P10. Que and co-workers (2018) timed unseeded paclitaxel at S ≈88 with and without 20 µg/mL HPMCAS que2018 P5. Duong and co-workers (2022) timed delamanid at its amorphous solubility (0.76 µg/mL) in PBS pH 6.5 with 300 µg/mL HPMC, PVPVA, Eudragit L100, HPMCP or HPMCAS duong2022a P4, P7. Polyzois and co-workers (2024) timed racemic and R-praziquantel at their amorphous solubilities in PBS pH 6.5 with 300 µg/mL of seven polymers over an 83-min run polyzois2024 P3.

What the data show

Felodipine dispersions. Maxima rose with polymer content, steeply for HPMCAS (≈1.7, 2.3, 4.4, 6.8 and 14.1 µg/mL at 10, 20, 50, 60 and 75%), little for PVP (≈0.8 to 2.3) and in between for HPMC (≈1.7 to 5.0); 10–50% PVP did not exceed the crystalline solubility konno2008a R10, R8. The 75% PVP dispersion dissolved fastest at first (≈0.272 against ≈0.092–0.115 µg/mL·min for the cellulosic ones) but gave the lowest supersaturation konno2008a R11. Felodipine with polymer in the medium. PVP ≈7.3 µg/mL at ≈15 min, ≈2.7 at 60 and ≈2.1 at 130 min; HPMC ≈8.9 at 40–50 min and ≈7.1 at 240 min; HPMC-AS ≈12.0 at ≈100 min and ≈10.2 at 240 min alonzo2010 R4. Felodipine particles. Without polymer the particles measured 450 nm at once and 936 nm at 7 min; from a 90:10 PVP dispersion they grew from ≈218 to ≈459 nm between 10 and 90 min, from HPMC dispersions to ≈280–307 nm alonzo2011 R10, R8. The 90:10 PVP apparent concentration fell from ≈36.5 to ≈9.2 µg/mL within an hour at 25 °C, the 90:10 HPMC one from ≈26.8 to ≈20.7 µg/mL alonzo2011 R5. Danazol. Induction times were ≈70, ≈56 and ≈60 min for the 90:10, 70:30 and 50:50 PVP dispersions, ≈200 and ≈188 min for 90:10 and 70:30 HPMC and ≈284 and ≈319 min for 90:10 and 70:30 HPMCAS; the 50:50 HPMC and HPMCAS dispersions did not crystallize within 1440 min, but they also dissolved slowly (F10) jackson2015 R5, R4. Indomethacin. With PVP the concentration was ≈23.7 µg/mL at 30 min and ≈23.8 at 240 min at 25 °C, and at 37 °C reached the same ≈23.7 at ≈80 min before falling to ≈13.7 µg/mL; with HPMC it rose slowly to ≈24.3 µg/mL at 240 min at 25 °C and reached only ≈14.5 at 37 °C, against ≈14.9 and ≈11.2 µg/mL peaks without polymer alonzo2010 R14, R15, R7. The crystalline α solubility was not changed by either polymer alonzo2010 R9. Ritonavir. With 5 µg/mL PVP, PVPVA, HPMC, HPMCAS and CAAdB 0.81 the induction times were 120.0, 119.2, 134.0, 146.0 and 242.8 min, against 119.3 min without additive ilevbare2013d R12. Celecoxib. Without polymer the concentration peaked at ≈20.8 µg/mL and fell to ≈4.1 µg/mL within 1 h; with PVP-K12 it held ≈20.9 µg/mL to 60 min and fell to ≈4.4 µg/mL by 8 h, with PVP-K29/32 it began to fall at about 2 h and was ≈9.1 µg/mL at 8 h, and with HPMCAS or HPMC, alone or HPMCAS with either PVP, it was ≈20.2–21.2 µg/mL at 8 h xie2015 R2. PAA left the curve unchanged (≈3.6 µg/mL at 60 min), whereas HPMCAS and HPMC kept ≈19.0 and ≈20.0 µg/mL at 16 h, also with PAA present xie2016a R3. Indomethacin at pH 2.2. With 100 µg/mL HPMC the absorbance plateau held from ≈16 to ≈110 min and then fell; with 100 µg/mL EPO the absorbance bent down and many small crystals appeared during the addition xie2017 R5, R8. Tacrolimus. Without HPMC the 50 µg/mL solution held for about 150 min and fell to ≈8 µg/mL by ≈680 min; with HPMC it held to ≈400 min and was ≈40 µg/mL at 800 min trasi2016 R9. Paclitaxel. Without polymer the unseeded solution crystallized within 6 min, with HPMCAS not within 16 h (F22) que2018 R3. Delamanid. PVPVA, HPMCP and HPMCAS prevented crystallization for the 1000 min observed, HPMC delayed it to about 50 min and Eudragit L100 had no effect (10.2 ± 3.1 against 10.7 ± 1.2 min) duong2022a R4, R2. Praziquantel. Induction times were 5.5 (none), 10.9 (PVPVA), 14.3 (HPC), 15.7 (PVA), 24.5 (HPMC 603) and 50.5 min (HPMCAS-LF), and HPMCAS-MF and HPMC E5 LV reached the end of the 83-min run; with R-praziquantel the same two gave 37 and 15.5 min (F29) polyzois2024 R3, R4. Posaconazole. A 10% HPMCAS dispersion dissolved directly in FaSSIF held about 21 µg/mL, near its amorphous solubility, for 180 min elkhabaz2019a R6.

Figure

The HPMCAS curve climbs steeply to 14.1 µg/mL at 75% polymer while PVP stays near the crystalline solubility up to 50% polymer konno2008a R10. Values above about 9.5 µg/mL lie above the paper's own amorphous-solubility estimate and are total UV-detected drug (F8) konno2008a X5.

PVP bars are shortest and HPMCAS longest at every loading; the onset concentrations on the right lie near 8 µg/mL whatever the polymer jackson2015 R5.

The polymer-free solution crashes within minutes, PVP-K12 falls after about 1 h and PVP-K29/32 after about 2 h, while HPMCAS, HPMC and the PVP plus HPMCAS pairs stay near 21 µg/mL xie2015 R1, R2.

Why

Konno and co-workers found that none of the polymers changed the crystalline solubility at 1 mg/mL, so that solubilization by complexation "can be discounted" and "it can be inferred" that the dispersions generate supersaturated solutions that the polymers maintain konno2008a C17, C22; they attribute the maintenance to slower crystal growth, which "most likely contributes" (F5) konno2008a C37. Alonzo and co-workers (2010) conclude that PVP is a poor inhibitor of felodipine crystallization from solution while HPMC and HPMC-AS inhibit it, "in excellent agreement" with Konno and co-workers alonzo2010 C11, C38; for indomethacin PVP "appears" to be an effective inhibitor of solution crystallization "at this temperature and supersaturation", and HPMC holding the 37 °C level better than PVP "could" reflect a lower driving force at its lower supersaturation alonzo2010 C22, C24, C25. Alonzo and co-workers (2011) read the faster growth of particles from PVP dispersions as HPMC inhibiting felodipine growth better alonzo2011 C18, C51. Jackson and co-workers conclude that the duration of supersaturation depends on the polymer and on the drug loading jackson2015 C15. For ritonavir the more hydrophobic, negatively charged cellulose polymers were effective and the more hydrophilic uncharged ones ineffective or less effective ilevbare2013d C30. For celecoxib, Xie and Taylor argue from classical nucleation theory that the polymer differences come neither from supersaturation, which they take as similar in all polymer solutions, nor from interfacial tension, which adsorbed polymer would lower, and suggest that the cellulose derivatives "may" interact more with dense liquid clusters of celecoxib than hydrophilic PVP, which also interacts with the solvent xie2015 C43, C44, C46; the effect "must be due to subtle molecular level interactions", and PAA's failure is "perhaps not particularly surprising" because the very hydrophilic polymer "is expected" to interact more with water than with celecoxib xie2016a C17, C18. For indomethacin HPMC's general effectiveness is attributed to its amphiphilic structure (cited), while EPO did not inhibit crystallization although it solubilizes the drug (F17) xie2017 C11, C2. For tacrolimus the induction times "suggest" that HPMC holds the supersaturation for a biologically relevant time trasi2016 C17. (synthesis: the six drugs were never compared on one scale; the crystallization-from-solution chapter relates growth and nucleation inhibition to the hydrophobicity of the polymer relative to the drug (Crystallization from solution F8, Crystallization from solution F13), which fits the failure of the very hydrophilic PAA and may be why the felodipine ranking does not carry over to indomethacin or ritonavir.) Duong and co-workers call PVPVA, HPMCP and HPMCAS "highly effective" for delamanid but find that PVPVA "shows disadvantages" because its drug-rich droplets grow (F12) duong2022a C12, C52. Polyzois and co-workers attribute the weaker effect of HPMCAS-LF and HPMC 603 than of the MF and E5 LV grades "likely" to functional group content and amphiphilicity, not measured polyzois2024 C25. Elkhabaz and co-workers judge HPMCAS "potentially an effective solution-crystallization inhibitor" for posaconazole when no matrix crystals are present elkhabaz2019a C52.

Conditions and limits

Felodipine at pH 6.8 and 37 °C (one dose, 30 mg in 600 mL), indomethacin at pH 2, danazol at 25 °C, ritonavir at 20 µg/mL with 5 µg/mL additive konno2008a X15; alonzo2010 P3; jackson2015 P2; ilevbare2013d P6. Polymer grades: PVP K29/32 or K30, HPMC 606, HPMCAS MF; molecular weights and substitution are not printed konno2008a M2, M3, M4; alonzo2010 M3, M4, M5. The felodipine maxima above about 9.5 µg/mL and the danazol UV peaks above 8 µg/mL include drug in particles (F8) konno2008a X5; jackson2015 C12. Celecoxib was tested at one polymer level (5 µg/mL) and one starting supersaturation, indomethacin at pH 2.2 only, tacrolimus at pH 4.5 with HPMC only xie2015 X8; xie2017 P4; trasi2016 P10; the crystalline solubility of celecoxib was measured with 22 µg/mL polymer, not the 5 µg/mL of the induction tests xie2015 X9.

Counterexamples and conflicts

For danazol the polymer changed how long supersaturation lasted, not how high the free drug went: every dispersion that formed a drug-rich phase gave 7.1–9.1 µg/mL by flux whatever the polymer (F7) jackson2015 R9. PVP in the medium delayed crystallization of the wetted felodipine solid at least as well as the cellulosics (F2) alonzo2010 C38. Among the 75% felodipine dispersions, PVP dissolved fastest at first konno2008a R11. In seeded growth PVP and PVPVA slowed celecoxib and efavirenz but not ritonavir growth, which agrees with the ritonavir result here ilevbare2013b R9. PVP is not without effect on celecoxib: it delayed nucleation by 1–2 h, longer for the higher molecular weight grade, and slowed the fall after nucleation xie2015 R1, C12. The enteric methacrylate Eudragit L100 had no effect on delamanid, and HPMC, effective for several drugs here, delayed delamanid only to about 50 min duong2022a R4. Even HPMCAS, which stopped paclitaxel nucleation for 16 h, only slowed growth on seeds (F22) que2018 C30. A ranking made with racemic praziquantel did not hold for its R-enantiomer (F29) polyzois2024 R4.

Transfer check

Rank the polymers for the new drug itself in a non-sink test at 37 °C, at the polymer concentration that complete release would give, after checking that the polymer leaves the crystalline solubility unchanged konno2008a P4, C6. Read the duration of supersaturation against the estimated amorphous solubility and check concentrations above it by a free-drug method (F7, F8). The felodipine order HPMCAS > HPMC > PVP is a hypothesis for another drug, not a rule: indomethacin and ritonavir gave other orders alonzo2010 R14; ilevbare2013d R12. Very hydrophilic polymers chosen for fast release (PAA, EPO in acid) should be expected to give no solution inhibition until tested xie2016a C14; xie2017 C2.

History

Before 2008 the same three polymers had been ranked for felodipine only in the solid state: alike when dry, HPMCAS and HPMC ahead of PVP with moisture (cited) konno2008a R18. Konno and co-workers (2008) ranked them first in the library by dispersion dissolution konno2008a C35. Alonzo and co-workers (2010) confirmed the felodipine ranking with the polymer in the medium and found PVP effective for indomethacin alonzo2010 C11, C22; Alonzo and co-workers (2011) added faster particle growth with PVP alonzo2011 C18; Ilevbare and co-workers (2013) found PVP without effect on ritonavir ilevbare2013d R12; Jackson and co-workers (2015) found the same order of duration for danazol at three loadings jackson2015 C19. Xie and Taylor (2015, 2016) found the cellulosic polymers far ahead of PVP and PAA for celecoxib, Trasi and co-workers (2016) found HPMC prolonging tacrolimus supersaturation, and Xie and co-workers (2017) found HPMC effective and EPO ineffective for indomethacin at pH 2.2 xie2015 C3; xie2016a C14, C15; trasi2016 C17; xie2017 C1, C2; the title gained celecoxib, PAA and EPO then. Que and co-workers (2018) found HPMCAS a "highly effective" nucleation inhibitor for paclitaxel, Duong and co-workers (2022) found PVPVA as good as the enteric cellulose esters and HPMC and Eudragit L100 poor for delamanid, and Polyzois and co-workers (2024) ranked seven polymers for praziquantel que2018 C13; duong2022a C12; polyzois2024 C7; the title gained praziquantel, PVPVA, Eudragit L100 and delamanid then.

How it was shown

Konno and co-workers (2008) added a small volume of felodipine in ethanol or methanol to 600 mL of pH 6.8 test fluid containing 150 or 50 µg/mL polymer (the levels that complete dissolution of 75% and 50% polymer dispersions would give) or none, to nominal 50, 15 or 9.5 µg/mL, at 37 °C and 100 rpm, with UV through a 70-µm filter or with second-derivative correction for scattering, n = 3 konno2008a P5, C6. From the initial linear part of each 9.5 µg/mL curve they took dMc/dt and, by Eq. 5, ranked growth rates assuming similar crystal surface areas konno2008a E5, P6, C28. The amorphous solubility was estimated at about 9.5 µg/mL (about 10 times the crystal) from the Hoffman equation with literature data konno2008a R17, E1, E2. Xie and Taylor (2016) suspended 7 mg of celecoxib seed crystals in 50 mL of pH 6.8 buffer at 37 °C with 0, 5, 11 or 22 µg/mL HPMCAS, added a methanol stock to 11 or 22 µg/mL and followed the fall in concentration by UV; the growth ratio R_0/R_p came from the initial slope without polymer and the slope over 16 h with polymer, and the nucleation ratio from these and the unseeded induction times xie2016a P4, P11, P12, E5, E6.

What the data show

50 µg/mL with 150 µg/mL polymer. HPMCAS ≈6.8 → ≈6.35 and HPMC ≈6.8 → ≈6.35 µg/mL over 240 min; PVP ≈6.3 → ≈1.9; no polymer ≈2.7 at 5 min → ≈1.05 konno2008a R12. 50 µg/mL with 50 µg/mL polymer. HPMCAS ≈6.4 → 6.0, HPMC ≈6.2 → 5.5, PVP ≈5.3 → 1.35 µg/mL konno2008a R13. 9.5 µg/mL. HPMCAS ≈9.1 → ≈8.5 and HPMC ≈8.95 → ≈7.8 µg/mL at about 210 min; PVP ≈8.1 → ≈2.8; no polymer ≈7.8 at 5 min, ≈4.3 at 30 min and ≈1.95 at the end konno2008a R14. 15 µg/mL. HPMCAS ≈11.1 → ≈10.45, HPMC ≈11.1 → ≈10.85, PVP ≈9.6 → ≈3.5, no polymer ≈3.1 → ≈1.8 µg/mL over 240 min konno2008a R15. The higher the start, the larger the drop by 5 min: about 86% of the nominal 50 µg/mL, 26% of 15 µg/mL and 4% of 9.5 µg/mL with HPMCAS (synthesis: (50 − 6.8)/50, (15 − 11.1)/15 and (9.5 − 9.1)/9.5 from the 5-min readings) konno2008a R12, R15, R14. The plotted buffer slope (≈7.8 → ≈4.3 µg/mL between 5 and 30 min) reproduces the printed ~0.14 µg/mL-min konno2008a E5. In the 2011 felodipine work, particles in methanol-shift solutions with HPMC grew from ≈183–203 nm at 5 min to ≈280–307 nm at 90 min at 25 °C, and precipitation at a supersaturation ratio of about 60 occurred even with HPMC present, which the authors read as nucleation not completely suppressed alonzo2011 R8, C30. Celecoxib. Without polymer seeded solutions fell from ≈22.5 to ≈4.0 µg/mL in 30 min with no lag; with HPMCAS they fell from ≈12.4 to ≈10.9 µg/mL in 4 h (11/11 µg/mL drug/polymer), to ≈19.8 (22/22) and to ≈16.0 µg/mL (22/5) xie2016a R11. The growth ratios were approximately 130, 158 and 70, less inhibition at the higher supersaturation and lower polymer level, and with no induction time seen in 16 h (taken as 960 min) the nucleation ratio J_0/J_p came to ~4000 xie2016a R12, R13, C37, by

\frac{J_0}{J_p} = \left(\frac{R_p}{R_0}\right)^{3}\left(\frac{t_{u,p}}{t_{u,0}}\right)^{4} xie2016a E6

which reproduces with an unseeded induction time of 5 min without polymer: (1/70)^3 \times (960/5)^4 \approx 3962 xie2016a E6.

Figure

The cellulose polymers stay almost flat while PVP and buffer decline; the initial slopes of panel (a) are the growth estimates of R16, from curves that start at different 5-min levels konno2008a R14, R15, X7.

Why

The authors "rationalize" the immediate drop as rapid nucleation, faster and giving more particles at higher concentration, which the polymers "do not appear" to have prevented konno2008a C24, C25. After the drop the solution has, "it appears", moved from the labile into the metastable zone, where nucleation is no longer spontaneous and only crystal growth could lower the concentration further; that the concentration stays up "strongly suggests" that the cellulose polymers inhibit growth konno2008a C27, C26. They add that a high theoretical solubility ratio, by classical nucleation theory, means easy nucleation, so that amorphous felodipine dissolving alone "will be accompanied by rapid crystallization" ("These data suggest") konno2008a C20, C21. The mechanism rests on concentration curves alone: no precipitate was isolated or examined konno2008a X4. The crystallization-from-solution chapter separates the two steps for felodipine with HPMC: an unseeded solution at S 10 with 1 µg/mL HPMC barely changed over 200 min while a seeded one fell from about 5.1 to 2.2 µg/mL in 58 min, so HPMC acted mainly by delaying nucleation (Crystallization from solution F6, Crystallization from solution F20) alonzo2012 R18, R17. (synthesis: the nominal 50 µg/mL start lies about five times above the estimated amorphous solubility of about 9.5 µg/mL konno2008a R17, the range in which later papers find a non-crystalline drug-rich phase separating before crystals (Amorphous solubility and LLPS F1); the immediate drop therefore need not be crystal nucleation, and the long flat plateaus with HPMC and HPMCAS fit nucleation inhibition at least as well as growth inhibition.) For celecoxib, Xie and Taylor find HPMCAS "a less effective inhibitor of growth at a higher supersaturation", in agreement with the models cited, and use the slow seeded growth to explain why crystals formed in dissolving PAA particles did not empty a medium holding HPMCAS (F16) xie2016a C37, C36; because growth from seeds is harder to stop than nucleation, they call it "critical" to prevent crystals forming in the dissolving matrix xie2016a C70.

Conditions and limits

One drug, pH 6.8 test fluid, 37 °C, 100 rpm against 50 rpm in the dissolution tests, two UV set-ups without saying which run used which konno2008a X6. The slopes come from one panel without uncertainty, from curves that start at ≈9.0, ≈8.1 and ≈7.8 µg/mL konno2008a X7. PVP and buffer solutions declined toward but did not reach the crystalline solubility in 4 h, and the 9.5 µg/mL run ended at about 210 min konno2008a X10, X16. Statements that buffer solutions precipitated "within 5 min" hold for the 15 µg/mL run only konno2008a X11.

Counterexamples and conflicts

The growth reading is contradicted for HPMC by the seeded and unseeded felodipine tests of the crystallization-from-solution chapter (growth at 0.607 of the polymer-free rate, estimated nucleation-rate ratio 7.05 × 10⁻⁴) alonzo2012 R5, R17. For celecoxib the growth effect was not small: HPMCAS slowed seeded growth 70- to 158-fold, so here both steps were inhibited, nucleation more xie2016a R12, C40. That chapter also finds growth inhibition weakening as supersaturation rises (HPMC's Kubota–Mullin factor on felodipine falling from 0.7 at S 3 to 0.2 at S 12), in line with the larger losses here from the higher starts (Crystallization from solution F9) alonzo2012 R9. The abstract's lead of HPMCAS in the solution tests holds only at 50 µg/mL polymer and from 9.5 µg/mL; at 150 µg/mL HPMC matched HPMCAS konno2008a X3.

Transfer check

To learn whether a polymer acts on nucleation or on growth for a new drug, run a seeded desupersaturation test and an unseeded induction-time test at the same supersaturation, below the amorphous solubility so that no drug-rich phase forms, and identify the precipitate by microscopy or diffraction (Crystallization from solution F5) konno2008a X4; alonzo2012 R18. A flat plateau after an initial drop does not by itself locate the effect.

History

Similar solution profiles of nifedipine with HPMCAS and HPMC had been reported (cited) konno2008a R19. Konno and co-workers (2008) gave the library's first solvent-shift profiles of felodipine with the three polymers and the growth reading konno2008a C36. Alonzo and co-workers (2011) showed particles forming and growing in such solutions alonzo2011 R8. Alonzo and co-workers (2012), in the crystallization-from-solution chapter, revised the mechanism for HPMC by separating nucleation from growth alonzo2012 R17. Xie and Taylor (2016) applied the same seeded and unseeded comparison to celecoxib with HPMCAS within this chapter and found nucleation affected far more than growth xie2016a C40; the title gained celecoxib and HPMCAS then. Que and co-workers (2018) found the same split for paclitaxel with HPMCAS, a nucleation inhibitor that only slowed growth on seeds, and related the remaining growth to a critical two-dimensional nucleus smaller than the spacing of adsorbed polymer (F22) que2018 C36, C38; the title gained paclitaxel then.

How it was shown

Konno and co-workers (2008) compared solvent-shift solutions with 150 or 50 µg/mL predissolved polymer at 100 rpm against powder dissolution of the dispersions at 50 rpm, both at 37 °C in pH 6.8 test fluid konno2008a P5, P3, X6. Jackson and co-workers (2015) compared their film dissolution runs with antisolvent solutions of 15 µg/mL danazol with predissolved polymer taken from a thesis (100 µg/mL polymer in the tables, 10 µg/mL in the Fig. 4 legend) jackson2015 P7, X2, X16. Chen and co-workers (2015) measured celecoxib induction times after a methanol shift to S = 5 in 50 mL at 300 rpm and 25 °C in eight media, and compared them with 10 mg/mL suspensions of celecoxib–HPMCAS spray-dried particles in 10 mL at 150 rpm, then repeated the solution test in three media with 3 µg/mL HPMCAS, the lowest level expected to leach from the particles chen2015a P9, P10, A3, C26, X13. Alonzo and co-workers (2010) ran solvent-shift tests of indomethacin at 25 and 37 °C with PVP or HPMC alongside dissolution of the amorphous solid alonzo2010 P4. Xie and Taylor (2015, 2016) set induction-time tests of 22 µg/mL celecoxib with 5 µg/mL polymer beside non-sink dissolution of 50% (and 30%) drug dispersions dosed to 22 µg/mL in 400 mL, both in pH 6.8 buffer at 37 °C xie2015 P4, P5; xie2016a P3, P5. Xie and co-workers (2017) measured the amorphous solubility of indomethacin with HPMC or EPO by titration at pH 2.2 and dissolved the matching dispersions in the same buffer xie2017 P4, P7.

What the data show

Felodipine. At 150 µg/mL polymer HPMC and HPMCAS held the solution equally; at 50 µg/mL HPMC was slightly lower (≈5.5 against ≈6.0 µg/mL at 240 min) konno2008a R12, R13, C14. From the dispersions, 75% HPMC gave 4.9 µg/mL and 75% HPMCAS 14.1 µg/mL konno2008a R7. Danazol. ASD induction times (90:10, 70:30) were ≈70 and ≈56 min (PVP), ≈200 and ≈188 min (HPMC) and ≈284 and ≈319 min (HPMCAS), against ≈111, ≈173 and ≈365 min with predissolved polymer; the 50:50 HPMC and HPMCAS dispersions did not crystallize within 1440 min jackson2015 R5. Celecoxib. Solution induction times at S = 5 were 7 min (Polysorbate 80), 12 (Poloxamer 188), 41 (SDS 0.25%), 88 (no surfactant), 273 (Triton X100) and more than 960 min (STC 0.75%) chen2015a R12. In the suspensions crystals appeared at 0.5 h with SDS and 1 h with Polysorbate 80 and not within 48 h in the other four media, including Poloxamer 188 chen2015a R14. With 3 µg/mL HPMCAS the induction time went from ≈12 to ≈50 min with Poloxamer 188, from ≈41 to ≈26 min with SDS and from ≈7 to ≈5.3 min with Polysorbate 80 chen2015a R19. The surfactant-free suspension held S ≈4.3–6.7 for 48 h without crystals, although the solution test in the same medium crystallized after 88 min at S = 5 chen2015a R17, R12, X6. Indomethacin. The solvent-shift tests predicted that both polymers would largely inhibit solution crystallization at the levels reached by dissolving the amorphous solid; at 25 °C dissolution with PVP held ≈23.8 µg/mL for 4 h, but at 37 °C it fell to ≈13.7 µg/mL by 240 min alonzo2010 C29, R14. Celecoxib dispersions. The PVP-K12 dispersion peaked at ≈21.0 µg/mL near 6 h and fell to ≈6.5 µg/mL at 16 h, PVP-K29/32 held ≈19.7 µg/mL at 16 h, and the HPMCAS dispersion reached ≈21.1 µg/mL without falling, in the order of the induction times (≈1 h, ≈2 h, more than 8 h) xie2015 R6, R1. The PAA dispersions peaked at ≈9–11 µg/mL within 3 min and fell to ≈2.7 µg/mL, "in excellent agreement" with PAA's lack of effect in solution; with 5 µg/mL HPMCAS in the medium, which holds the solution for 16 h, release still stopped at ≈9.7–11.6 µg/mL of 22 xie2016a R4, C25, R9. Indomethacin at pH 2.2. EPO let the drug crystallize during the titration and the EPO dispersion crystallized within about 20 min (8.8 µg/mL in the filtered supernatant at 1 h); HPMC held the solution about 1.5 h and the HPMC 3:7 dispersion did not crystallize in the acid stage xie2017 R8, R11, R5, R13.

Figure

Only the Poloxamer 188 bars (blue without, red with polymer) separate clearly: a trace of the ASD polymer changes the ranking a surfactant screen gives chen2015a R19.

Why

Konno and co-workers suggest a solution test "could be used" to screen polymers for a drug, find "a reasonable correlation" for PVP and HPMCAS, call HPMC "less predictable" and note the same behaviour of HPMC for nifedipine in earlier work konno2008a C30, C31, R19. Jackson and co-workers conclude that screening predicts the relative stability to crystallization "quite well" jackson2015 C20. Chen and co-workers suspected that leached HPMCAS would change the surfactant effect and conclude that the longer induction time "most likely explains" the stable Poloxamer 188 suspension, the real polymer level being "likely to be higher" chen2015a C25, C28. Alonzo and co-workers (2010) add that a polymer acts on two routes, so a test of one route (the solution) can misrank it on the other (the solid) (F2) alonzo2010 C38, C45. (synthesis: a solution test isolates what dissolved polymer or surfactant does to nucleation and growth at a fixed supersaturation; it leaves out how fast the dispersion releases drug, the supersaturation actually reached, and what dissolves out of the particles, which are where the HPMC, PVP and Poloxamer 188 discrepancies arise.) Xie and Taylor infer that, since the PVP and HPMCAS dispersions released completely, matrix crystallization was suppressed and the dissolution differences reflect solution crystallization, which the induction time measures ("it is apparent"); the HPMC-containing dispersions neither released fully nor were checked in the slurry xie2015 C39, X3. (synthesis: the PAA dispersions show the other blind spot of the solution test, crystallization of the wetted matrix, which no solution test can rank; see F16.) Polyzois and co-workers could not rank their two best polymers against each other because both reached the end of the 83-min run polyzois2024 K1. For delamanid, droplet growth with PVPVA, which the induction time does not show, led the authors to prefer the enteric polymers duong2022a C52.

Conditions and limits

The tests differ in volume, stirring, polymer level and temperature from the dissolution runs they are compared with konno2008a X6; jackson2015 X2; chen2015a X13. Error-bar type and replicates of the danazol induction times are not stated jackson2015 X8. The celecoxib HPMCAS comparison is one test per medium without statistics, and the SDS whiskers differ between two figures of the same mean chen2015a X18.

Counterexamples and conflicts

The crystallization-from-solution chapter finds induction times of one drug near its amorphous solubility differing up to about sixfold between protocols of the same group, so that inhibitor effects transfer only as ratios to a control run in the same protocol (Crystallization from solution F35). None found in the library in which a solution screen reversed the order of two polymers that the dispersion test separated widely, apart from HPMC against HPMCAS for felodipine and nifedipine konno2008a R19. A solid-state screen misled in the other direction: for resveratrol, PVP and E100, which had kept the dispersions amorphous at high humidity, kept them amorphous in buffer but gave no more dissolved drug than the crystal (42 and 46 against 42 µg/mL) wegiel2015 R21, R11, C9 (F17).

Transfer check

Screen with the polymer at the concentration the dispersion would release and, when screening surfactants or other excipients, include the dispersion polymer at the level that leaches from the particles konno2008a C6; chen2015a C26. Use the screen to rank, then confirm the top two or three candidates by dissolving the dispersion under non-sink conditions; express induction times as ratios to a control in the same protocol jackson2015 X8.

History

Konno and co-workers (2008) proposed the solution screen and found it imperfect for HPMC konno2008a C30, C32. Alonzo and co-workers (2010) showed that it covers only the solution route alonzo2010 C45. Jackson and co-workers (2015) confirmed the rank order for danazol and three loadings jackson2015 C20. Chen and co-workers (2015) showed that a surfactant screen without the ASD polymer misranks Poloxamer 188 chen2015a C28. Xie and Taylor (2015) found the induction times "quite predictive" for 50% celecoxib dispersions, and in 2016 that the test cannot see crystallization inside PAA particles xie2015 C48; xie2016a C31; Wegiel and co-workers (2015) found that a storage-stability ranking did not predict the concentration reached wegiel2015 C9, C35. Duong and co-workers (2022) added droplet size to the induction-time screen for delamanid duong2022a C13, and Polyzois and co-workers (2024) chose the praziquantel carriers by induction time and then found them far apart in release at the higher dose polyzois2024 C7, C17.

How it was shown

Polyzois and co-workers (2024) measured the crystalline solubility of R- and racemic praziquantel by shake flask in PBS pH 6.5 at 37 °C and the amorphous solubility by pumping a DMSO stock into PBS pH 6.5 or HCl pH 1.6 until light scattering rose polyzois2024 P1, P2. For the induction time they added the drug to its own amorphous solubility in PBS holding 300 µg/mL of one of seven polymers, and timed the appearance of scattering from crystals over an 83-min run (n = 3 for the racemate, n = 2 for R-praziquantel) polyzois2024 P3, X14. Solvent-evaporated dispersions of racemic praziquantel (20–50%) and R-praziquantel (30–40%) with HPMCAS-MF or HPMC E5 LV were released in 50 mL of PBS pH 6.5 at 667 µg/mL or 1.5 mg/mL with in situ UV (n = 3) polyzois2024 P4, P8. Alvarenga and co-workers (2025) made 30% dispersions of each form with each polymer by solvent evaporation and with HPMCAS by hot-melt extrusion, released them in buffers, fasted and fed media at 0.667 and 1.5 mg/mL in single- and two-stage tests, and looked for crystals by PLM and PXRD alvarenga2025 P3, P4, P9, P10, P6, P5. The racemate had a purity of 95%+ and R-praziquantel 99.9% alvarenga2025 M2, M1.

What the data show

Solubilities. Crystalline solubility in PBS pH 6.5 was 282.6 ± 3.3 µg/mL for R-praziquantel and 196.4 ± 0.3 µg/mL for the racemate, and amorphous solubility 1.4 ± 0.1 and 1.1 ± 0.1 mg/mL; at pH 1.6 both were 1.4 mg/mL polyzois2024 R1, R2. In the nine media of 2025 the amorphous-to-crystalline ratio was 4.0–5.8 for R-praziquantel and 5.6–7.7 for the racemate, lower because crystalline R-praziquantel is more soluble, and the amorphous solubilities differed significantly only in PBS pH 6.5 (1.39 against 1.17 mg/mL) alvarenga2025 R1, R3. Induction times. Racemate: none 5.5, PVPVA 10.9, HPC 14.3, PVA 15.7, HPMC 603 24.5, HPMCAS-LF 50.5 min, and HPMCAS-MF and HPMC E5 LV at least 83 min; R-praziquantel: about 5 min without polymer, 15.5 min with HPMC E5 LV, 37 min with HPMCAS-MF polyzois2024 R3, R4. Release from HPMCAS-MF. At 1.5 mg/mL both forms released 94% at 30% drug; at 40% drug R-praziquantel reached about 92.5% and the racemate about 82.5%, with racemate error bars of about ±11 polyzois2024 R12, R9, R13, X13. In 2025, at 1.5 mg/mL in PBS pH 6.5, R-praziquantel reached about 88% (maximum ≈1.4 mg/mL) against 71% (≈1.1 mg/mL) for the racemate, and after the pH 1.6 shift 91% against 73%; at 0.667 mg/mL and in fasted media the two released alike alvarenga2025 R8, R9, C43. After the pH 3.0 shift at 1.5 mg/mL R-praziquantel also ended higher (about 58 against 46%) alvarenga2025 R11, X13. Release from HPMC E5 LV. In 2024, R-praziquantel at 30% drug peaked at 77% at about 25 min and fell to 62%, while the racemate stayed at about 74–77% polyzois2024 R12, R9. In 2025, R-praziquantel–HPMC fell from 65 to 60% in PBS pH 6.5, from maxima of about 68% and 72% to about 58% after pH 3.0 and 5.0 starts, by 30% after 75% in FaSSIF, and from about 47 to 34% in FeSSIF-V2, whereas the racemate stayed level or rose in aqueous media alvarenga2025 R8, R11, R14, R18, C36. PLM showed birefringent needles and aggregates in R-praziquantel–HPMC samples at 60 and 120 min in all four aqueous protocols, and PXRD of HPMC residues gave clear R-praziquantel peaks against weak racemate peaks alvarenga2025 R20, R25, R26.

Figure

In (a) HPMC E5 LV and HPMCAS-MF reach the 83-min end of the run with the racemate; in (b) the same two polymers give only 15.5 and 37 min with R-praziquantel polyzois2024 R3, R4.

Few particles at 5 min and birefringent needles and aggregates at 60 and 120 min in all four protocols alvarenga2025 R20.

Why

Polyzois and co-workers give hedged reasons for the weaker inhibition of R-praziquantel: the racemate crystallizes as an anhydrate with syn conformation and R-praziquantel as hydrates with anti conformation, which "may be attributed" to a different drug–polymer interaction; water bound to an R-praziquantel carbonyl "may" reduce interaction with the polymer; and it is "not unreasonable to speculate" that the chiral cellulose polymers bind S-praziquantel more polyzois2024 C26, C27, C28. The fall of R-praziquantel release from HPMC E5 LV is attributed "likely" to crystallization, consistent with its short induction time with that polymer; no crystals were looked for in 2024 polyzois2024 C20. Alvarenga and co-workers read the HPMC curves as indicating that R-praziquantel crystallizes more readily from highly supersaturated solution ("indicate"), propose that molecules of one handedness "can" pack more efficiently and nucleate faster while in the racemate the counter-enantiomer acts as an "impurity", and that the driving force "likely" differs between the solid forms that crystallize alvarenga2025 C25, C40, C39. Bile salts "may" also have sped R-praziquantel crystallization in FaSSIF alvarenga2025 C37. The better release of R-praziquantel from HPMCAS in PBS pH 6.5 "is likely attributable" to its higher amorphous solubility there, and the two forms behaved alike in fasted media because bile salts raise the solubility of both alvarenga2025 C42, C43. Nucleation and growth rates were not measured, the crystal forms in the residues were not identified, and the two drugs differ in purity alvarenga2025 X14, X17. (synthesis: R-praziquantel crystallized sooner although it was less supersaturated against its own crystal at the amorphous solubility, about 1400/282.6 ≈ 5.0 against 1100/196.4 ≈ 5.6 for the racemate polyzois2024 R1, R2, so the difference does not come from a larger driving force on the crystal side. A polymer that discriminates between enantiomers in crystal growth was already shown for HPMCAS in a melt sato2015 C11, and the ranking of polymers depends on the drug as it does across drugs in Crystallization from solution F13; an enantiomer is, for this purpose, another drug.)

Conditions and limits

One drug pair, two cellulose polymers for the dispersions, one 83-min induction run per system; the two forms were tested at their own amorphous solubilities, so at different concentrations polyzois2024 X14. R-praziquantel induction times rest on n = 2 and no statistical test was applied to release polyzois2024 X12; alvarenga2025 X11. The 1.5 mg/mL aqueous and fasted curves of 2025 are given only in the text alvarenga2025 X1. Long-term physical stability of the R-praziquantel dispersions was not determined, and the 20% HPMCAS-MF racemate dispersion showed small crystalline peaks after 21 weeks at 40 °C/75% RH polyzois2024 C41, R7. R-praziquantel was not extruded in 2024 and only in two conditions in 2025 polyzois2024 C32; alvarenga2025 A3.

Counterexamples and conflicts

The R-praziquantel advantage from HPMCAS-MF at 40% drug lies within the racemate error bars polyzois2024 X13. In FeSSIF-V2 the authors report no significant difference between the forms, without a test, beside an HPMC pair at about 34 against 48% (F28) alvarenga2025 C38, X11. In the melt study of the Physical stability chapter the racemate grew more slowly than either enantiomer and each polymer inhibited the racemate less than the enantiomer sato2015 C12, whereas in solution here the polymers protected the racemate far longer polyzois2024 R3, R4.

Transfer check

For an enantiopure product, screen crystallization inhibitors with the enantiomer itself at its own amorphous solubility, not with the racemate polyzois2024 R4, C8. Measure crystalline and amorphous solubility of both forms in the test media, since the enantiomer's crystal may be more soluble and its gain may appear only in simple buffer polyzois2024 R1; alvarenga2025 R3, C43. Identify the solid form that crystallizes during release (PXRD of residues, polarized light), since hydrate and anhydrate forms differ between enantiomer and racemate polyzois2024 C26; alvarenga2025 R25, R26. Match the purity of the two forms before attributing differences to chirality alvarenga2025 X17.

History

In the Physical stability chapter, Sato and co-workers (2015) found HPMCAS discriminating between the enantiomers of a chiral compound in melt growth while PVP and HPMCP did not sato2015 C10, C11. Polyzois and co-workers (2024) were the first in the library to compare polymer inhibition and dispersion release for an enantiomer and its racemate, and found both best polymers much weaker for the enantiomer polyzois2024 C8, C39. Alvarenga and co-workers (2025) confirmed the HPMC failure with R-praziquantel by PLM and PXRD in all aqueous protocols and found the HPMCAS advantage of R-praziquantel limited to simple buffer and the pH 1.6 and 3.0 shifts alvarenga2025 C25, C42, C43, R11.

How it was shown

Alonzo and co-workers (2011) dissolved 90:10 and 50:50 HPMC–felodipine dispersions, pure amorphous felodipine and crystalline felodipine in a 300 mL donor at 37 °C; after the donor UV plateau (about 40 min for amorphous solids) a dialysis bag with 10 mL buffer containing 500 µg/mL HPMC and a second UV probe was immersed, and the receptor slope over 10–40 min was divided by the crystalline slope; methanol-shift solutions at nominal 5 and 10 µg/mL checked that flux scales with concentration alonzo2011 P4, C9. Solution ¹H NMR of 50-min samples, which sees only mobile dissolved molecules, compared the aromatic felodipine signal of the two dispersions alonzo2011 P7, C27. Jackson and co-workers (2015) dissolved danazol dispersions in the 200 mL donor of a side-by-side diffusion cell with a regenerated cellulose membrane (6–8 kDa) at 25 °C, took the flux from the receiver slope after the donor peaked and before crystallization, and converted it to free drug with a standard curve of solutions of known supersaturation jackson2015 P4. Trasi and Taylor (2015) dissolved 20% ritonavir and lopinavir dispersions in PVP and HPMCAS in 100 mM pH 6.8 buffer and assayed the supernatant after ultracentrifugation at 40,000 rpm for 20 min, at 20 min and 2 h trasi2015b P9, P5. Ilevbare and co-workers (2013) measured the ritonavir LLPS concentration by turbidity titration with 180 µg/mL polymer, the level a dissolved 10:90 dispersion would give ilevbare2013d P2. Trasi and co-workers (2016) measured the amorphous solubility of tacrolimus in 5 mM pH 4.5 phosphate buffer at 37 °C by UV scattering, pyrene fluorescence, 0.22-µm filtration, ultracentrifugation and film dissolution, dissolved the whole contents of each of six 5-mg capsules in 40 mL of the same buffer (125 µg/mL if all dissolved) with HPLC of 0.45-µm filtrates, and measured the mass flow rate across a regenerated cellulose membrane (12–14 kDa) from pure-drug solutions of 20–100 µg/mL and from capsule contents dissolved for an hour in 30 mL trasi2016 P3–P8, P11, P9. Li and Taylor (2018) cast lopinavir–HPMC films at 15–85% drug in vials and stirred them in 10 mL of pH 3 citrate buffer holding 20 µg/mL HPMC at 25 °C for 48 h, ultracentrifuging the samples, against an amorphous solubility from an HPMC film and from the LLPS onset li2018 P4, P5, P2, P3.

What the data show

Felodipine. The controls behaved as the flux equation requires: 10 and 5 µg/mL gave ratios of 10.0 ± 3.4 and 4.8 ± 0.39, slopes in the ratio 2.07 alonzo2011 R14, E1. The slopes were 0.0858 ± 0.0094 (90:10), 0.0782 ± 0.015 (50:50) and 0.0698 ± 0.015 (pure amorphous) against 0.0082 for the crystal (n = 1) alonzo2011 R14. In UV the 90:10 donor read ≈27.0 µg/mL and the 50:50 donor ≈8.9 µg/mL at 60 min, and the pure amorphous donor ≈8.6 µg/mL alonzo2011 R12. The NMR integrals were 1370 and 1014, a difference of about 26%, not threefold alonzo2011 R15. Danazol. Free drug was 7.3 ± 1.0 and 7.1 ± 1.1 µg/mL for 90:10 and 50:50 PVP, 8.7 ± 1.2 and 9.1 ± 1.6 for 90:10 and 50:50 HPMC and 8.2 ± 1.2 for 90:10 HPMCAS; antisolvent solutions gave 8.1 ± 1.2 without and 7.2–8.2 µg/mL with polymer; crystalline danazol 1.0 ± 0.1 µg/mL jackson2015 R9. The 50:50 HPMCAS dispersion, which did not form a drug-rich phase, gave 5.4 ± 0.8 µg/mL (F10) jackson2015 R9. Ritonavir and lopinavir. At 2 h the single-drug dispersions gave ritonavir ≈26.9 (HPMCAS) and ≈22.0 µg/mL (PVP) and lopinavir ≈17.8 and ≈15.9 µg/mL; drug-only films reached 22.9 and 15.6 µg/mL after 60 min trasi2015b R13, R5. The RTV–PVP dispersion was the one that stayed below its amorphous solubility, at about 22 µg/mL even after 2 h trasi2015b C29. Polymers and the ritonavir limit. With 180 µg/mL PVP, PAA, HPMC or HPMCAS the LLPS concentration was 18.3, 18.3, 18.7 and 18.2 µg/mL against 18.2 ± 0.35 µg/mL without polymer; CAAdP 0.85 gave 21.3 ± 1.22 µg/mL ilevbare2013d R10. In the LLPS chapter, a 1:9 ritonavir–PVP dispersion added in portions stalled at 18.3 and 18.0 µg/mL with 24.0 µg/mL added ilevbare2013a R17. Tacrolimus. The crystalline solubility was 1.4 µg/mL and the amorphous solubility 46 (scattering), 40–50 (pyrene), 50 (filtration), 49 (ultracentrifugation) and 48 µg/mL (film), a ratio of about 35, which the authors place near the middle of the 5–65 range of other drugs (cited) trasi2016 R4, R16, R17; the LLPS chapter finds the ratio running from about 2 to about 600 across drugs (Amorphous solubility and LLPS F6). At 5 min the products ranged from ≈26 (Prograf) to ≈41 µg/mL (Accord); all reached their maximum within about 45 min, read ≈46–50 µg/mL from 45 to 120 min and did not fall trasi2016 R10, C18, C30. The product containing sodium lauryl sulfate reached the same plateau trasi2016 M10, R10. Pure-drug flux rose from ≈0.0027 µg/min at 20 µg/mL to ≈0.0047 at 50 µg/mL and stayed at ≈0.0050–0.0055 µg/min from 60 to 100 µg/mL trasi2016 R14. Lopinavir with a dissolving polymer. HPMC dispersions gave ≈18.1–19.0 µg/mL at 3 h and ≈17.5–18.9 µg/mL at 48 h at 15–85% drug, against 17.4 ± 0.3 (HPMC film) and 18 ± 2 µg/mL (LLPS onset); a physical mixture of amorphous lopinavir and HPMCAS also reached ≈17–18.5 µg/mL li2018 R2, R3, R1, R14. Later papers. A paclitaxel–HPMCAS dispersion reached ≈31–33 µg/mL with or without seeds against an amorphous solubility of 31.6 µg/mL que2018 R13, R1, posaconazole–HPMCAS dispersions ≈21–24 µg/mL against an LLPS onset of about 20 µg/mL in FaSSIF elkhabaz2019a R6, R7, R2, and R- and racemic praziquantel–HPMCAS dispersions maxima of about 1.4 and 1.1 mg/mL against 1.39 and 1.17 mg/mL in PBS pH 6.5 alvarenga2025 R8, R1; these are total concentrations by UV or after moderate centrifugation, not free drug (F8).

Figure

The two controls show that the cell separates 5 from 10 µg/mL; the 90:10, 50:50 and pure amorphous ratios sit together near the predicted factor of 9 alonzo2011 R14.

Every row with a drug-rich phase lands at 7.1–9.1 µg/mL; only 50:50 HPMCAS, which never reached the amorphous solubility, is lower jackson2015 R9.

Pure-drug flux rises up to 50 µg/mL and then levels off; the band of the dissolved products lies on that plateau trasi2016 R14, R15.

Why

Jackson and co-workers state that the dispersions dissolve to a concentration above the amorphous solubility, but only briefly: a drug-rich phase in metastable equilibrium with the water-rich phase forms rapidly and fixes the free drug at the amorphous solubility, so that all dispersions that undergo LLPS "should" give the same flux and the extra solubility a dispersion might offer "cannot be realized" jackson2015 C27, C31, C32. Ilevbare and co-workers conclude that "it appears" there is an upper concentration of free drug that ASD dissolution can reach, and that the LLPS concentration "represents the upper limit of supersaturation" ilevbare2013d C35; the unchanged LLPS concentration with four polymers "suggests" that these polymers are not substantially incorporated into the ritonavir-rich phase ilevbare2013d C22. Alonzo and co-workers (2011) find the three felodipine activities "remarkably close" to the estimate, but add that "there is no reason to expect" a dispersion to give the same solution activity as the pure amorphous drug and that the equality "may be purely coincidental" alonzo2011 C22, C45, C48. Trasi and Taylor conclude that the maximum after 2 h is "not dependent on the polymer type" trasi2015b C28. The LLPS chapter frames the same cap as the amorphous solubility, reached but not exceeded by any solid form whose crystallization is slow enough (Amorphous solubility and LLPS F13), with free drug, activity and flux levelling off at that concentration (Amorphous solubility and LLPS F8). For tacrolimus, Trasi and co-workers state that membrane transport into a sink follows the activity of the free drug, not the total concentration, so that above the amorphous solubility added drug only forms more drug-rich phase and the mass flow reaches a maximum, "as observed for other drugs" trasi2016 C36, C37; the products reaching that maximum means that "the amorphous solubility advantage is achieved" and that the formulation components neither solubilize the drug nor change its amorphous solubility, "at least not at the concentrations present" trasi2016 C38, C31. They recall that polymers have little effect on the amorphous solubility, that small molecules mixing into the drug-rich phase lower it and micellar surfactants raise it, and that lactose, the main excipient, is "not anticipated" to do either trasi2016 C32, C33. For lopinavir, Li and Taylor read the unchanged ceiling with HPMC as the polymer dissolving and "not substantially mixed with the drug droplets due to its high aqueous solubility" li2018 C48. (synthesis: here the cap holds for dispersions and products of five drugs, at 10–85% drug, with polymers that dissolve; it is lowered by something that stays mixed with the drug, a second drug (F11), a polymer that enters the drug-rich phase, as for ibuprofen in the LLPS chapter, or a polymer that does not dissolve (F18); the measured total can rise above it with a polymer that solubilizes the drug (F17); and it is not reached when release is too slow (F10).)

Conditions and limits

The felodipine particle-blind test was run for HPMC dispersions at 37 °C only, and the claims for PVP–felodipine and both indomethacin systems rest on UV and light scattering alonzo2011 X16. The danazol fluxes are relative: the absolute values do not reproduce from the example profile, the flux and concentration columns are not proportional, and the flux window per formulation is not stated jackson2015 X4, X5, X6. The ritonavir and lopinavir values are plot readings with undefined error bars, and the cited amorphous solubilities come from an earlier paper trasi2015b X5, X14. Temperatures: felodipine 37 °C, danazol and the RTV/LPV dispersions 25 °C alonzo2011 P4; jackson2015 P4; trasi2015b P5. The tacrolimus products were tested in one buffer (pH 4.5) and 40 mL only; their drug loadings were not known, and the plateau was read on 0.45-µm PTFE filtrates whose retention of droplets was not tested trasi2016 P11, X4, X10; the 100 mL control the authors call sink is nominally 50 µg/mL, at the amorphous solubility trasi2016 X5. The lopinavir HPMC result is at pH 3 and 25 °C, with HPMC also in the medium li2018 P5.

Counterexamples and conflicts

CAAdP 0.85 at 180 µg/mL raised the ritonavir LLPS concentration by about 17%, unlike the other four polymers ilevbare2013d X7. In the LLPS chapter, dissolved HPMC and PVP-VA lowered the amorphous solubility of ibuprofen by about half by entering its drug-rich phase (Amorphous solubility and LLPS F19) ueda2020b R1; the ritonavir titrations here found no such lowering at 180 µg/mL ilevbare2013d R10. Konno's 75% HPMCAS felodipine dispersion reached 14.1 µg/mL against an estimate of about 9.5 µg/mL, unexplained in that paper and measured by UV (F8) konno2008a X5. The 2018 lopinavir study found the cap lowered when the polymer does not dissolve and is mixed with the drug: HPMCAS, CAP, HPMCP and EC dispersions stayed below 17.4 µg/mL at pH 3, down to ≈0.3–0.6 µg/mL at 15% drug (F18) li2018 R3; the title was narrowed to the dissolving polymer then. With a polymer that solubilizes the drug, the measured concentration can exceed the cap while free drug does not: HPMCAS–resveratrol solutions held up to ≈1610 µg/mL with at most about twice the crystal's flux (F17) wegiel2015 R19, R20, C41.

Transfer check

Above the estimated amorphous solubility, measure free drug by flux across a membrane that excludes droplets, by solution NMR or by ultracentrifugation; expect a plateau at the amorphous solubility alonzo2011 P4; jackson2015 P4; trasi2015b P9. A plateau clearly below it means slow release (F10), a second drug (F11) or a polymer that mixes into the drug-rich phase (LLPS chapter); a plateau above it means the method is counting particles (F8) or that a polymer solubilizes the drug (F17). Do not design for free drug above the amorphous solubility. For a marketed or multi-excipient product, compare its plateau and its maximal flux with the amorphous solubility and the flux plateau of the pure drug in the same medium, as the authors recommend for baseline metrics trasi2016 C46.

History

The LLPS chapter traces the cap to earlier dispersion work and to the ritonavir–PVP dispersion of 2013 (Amorphous solubility and LLPS F13). Within this chapter, Alonzo and co-workers (2011) gave the first particle-blind test of dispersions, which Ilevbare and co-workers (2013) cite as diffusion from dissolved ASDs being capped at the amorphous drug level (builds_on) ilevbare2013d R19. Ilevbare and co-workers (2013) named the LLPS concentration as the upper limit ilevbare2013d C35; Jackson and co-workers (2015) confirmed it by flux for three polymers and two loadings ("is confirmed here") jackson2015 C30; Trasi and Taylor (2015) by ultracentrifugation for two drugs and two polymers trasi2015b C28. Trasi and co-workers (2016) carried it to marketed multicomponent products, six tacrolimus capsules, by plateau and flux trasi2016 C42, C43; later tacrolimus papers of the library take the ≈50 µg/mL amorphous solubility as the design value of their non-sink and absorptive tests purohit2018 R4; trasi2017 M1; hate2019 M1. Li and Taylor (2018) confirmed the cap for lopinavir with HPMC and showed that undissolved polymers lower it (F18) li2018 C18, C67. Que and co-workers (2018), Elkhabaz and co-workers (2019) and Alvarenga and co-workers (2025) found dispersions reaching, not exceeding, the amorphous solubility measured in the same medium, by methods that do not exclude particles que2018 C20; elkhabaz2019a C7; alvarenga2025 R8.

How it was shown

Alonzo and co-workers (2011) followed dissolution with fibre-optic UV probes (270–450 nm every minute, second-derivative spectra calibrated in methanol) in 18 mL of buffer at 300 rpm, and compared the 90:10 solution spectrum at 60 min with a 10 µg/mL methanol solution, a 25 µg/mL crystalline felodipine nanosuspension and their sum scaled by 0.9 alonzo2011 P2, P10. Jackson and co-workers (2015) read UV at 288 nm with 350 nm as a no-absorption reference and compared it with free drug from flux jackson2015 P2, P4. Trasi and Taylor (2015) sampled through 0.45 µm glass syringe filters, examined filtrates by nanoparticle tracking, and switched to ultracentrifugation for the HPMCAS dispersions trasi2015b P5, P8, P9. Konno and co-workers (2008) sampled through a 70-µm filter on the dissolution tester and, in solvent-shift runs, either the same filter or second-derivative correction konno2008a P3, P5. Ilevbare and co-workers (2013) used extinction at non-absorbing wavelengths (280–450 nm) as the signal of a second phase ilevbare2013d P3. Xie and co-workers (2017) read indomethacin as the 319 nm absorbance minus the 450 nm extinction with a dip probe and checked selected runs by HPLC of filtered samples at 1, 3 and 8 h xie2017 E4, P9. Trasi and co-workers (2016) filtered a 60 µg/mL tacrolimus solution through 0.22-µm and 0.7-µm glass filters, examined 0.7-µm filtrates by nanoparticle tracking, and rinsed the filter of a dissolved capsule with acetonitrile to recover the undissolved drug trasi2016 P6, P13, P14.

What the data show

Felodipine. With more 90:10 dispersion the apparent peaks rose with the dose (≈75, ≈53 and ≈28 µg/mL for 90, 60 and 30 µg/mL added, 80–100% of the drug added) alonzo2011 R2. The summed reference spectrum gave a valley at ≈316 against ≈313 nm and a peak at ≈370 against ≈367 nm, and the shift appeared within 5 min of the start, not at any time for the 50:50 dispersion, pure amorphous felodipine or the controls alonzo2011 R21, R20. Particles of 200–400 nm absorb like the solution, and the cited work showed that this absorbance is not removed by the second derivative and grows relative to scattering as particles get smaller alonzo2011 C27, R25. The methanol-shift controls read ≈4.3 and ≈9.0 µg/mL against nominal 5 and 10 µg/mL even without particles alonzo2011 R12, X18. Indomethacin. The 90:10 PVP dispersion read ≈55.5 µg/mL against ≈22 µg/mL for 50:50 and a predicted 21 µg/mL; 90:10 HPMC–indomethacin fluctuated between 80 and 120 µg/mL near the linear limit of the probe; neither was checked by a particle-blind method alonzo2011 R16, R18, X16. Danazol. UV peaks above 8 µg/mL (≈9.4–16.2 µg/mL) contrast with flux values of 7.1–9.1 µg/mL jackson2015 R2, R9. Ritonavir and lopinavir. HPMCAS dispersions gave uniformly turbid solutions within minutes and turbid filtrates with many scattering species, of mean size about 500 nm (RTV–HPMCAS) and about 160 nm (both drugs); PVP filtrates were clear trasi2015b R6, R7, R8. Felodipine maxima of 2008. 14.1 µg/mL from the 75% HPMCAS dispersion and about 11 µg/mL held from a 15 µg/mL solvent shift, against an estimate of about 9.5 µg/mL konno2008a R7, R15, R17. Indomethacin. The EPO dispersion read ≈29.5 µg/mL by UV at 1 h while the filtered supernatant held 8.8 ± 0.6 µg/mL; after the pH rise UV and HPLC agreed better (≈33 against 32.1 µg/mL) xie2017 R11, R19, R10. The EPO–HPMC dispersion read ≈93.5 µg/mL at pH 2.2 against 80 µg/mL nominal, which the authors attribute to a changed extinction coefficient with nanodroplets xie2017 R15, C36. Tacrolimus. The 0.22-µm glass filter retained the droplets of a 60 µg/mL solution; 0.7-µm filtrates held 220 nm droplets trasi2016 R8, R11. The filtrates of dissolved capsules held particles of 210 nm both in 40 mL and in 100 mL, which the authors attribute to croscarmellose passing the filter, so tracking could not decide whether drug particles formed; the 0.7-µm filtrates held 45–50 µg/mL and the filter rinse ≈70 against 75 µg/mL expected trasi2016 R12, C39, R13, E3. Celecoxib. After crystallization the UV readings settled at ≈2.6–2.9 µg/mL, above the 1.5 µg/mL measured by HPLC after ultracentrifugation, and the crystal itself read ≈1.85 µg/mL at 16 h xie2016a R4, R1, X4; xie2015 R7, X12.

Figure

The black 90:10 spectrum has its valley at 312 nm, between the red solution (294 nm) and the blue nanosuspension (325 nm): the probe sees dissolved drug plus particles alonzo2011 R19.

The black 90:10 curve near 27 µg/mL is not all dissolved drug; flux and NMR put both dispersions at about the same dissolved level alonzo2011 R1, R14.

Why

Alonzo and co-workers give "The most likely explanation" for the contradiction between UV and the particle-blind methods: the submicron particles absorb light like the solution and are not corrected by second-derivative processing, so UV counts drug inside particles, whereas diffusion and NMR detect only highly mobile species alonzo2011 C27. Because absorption is additive, a spectrum that sums solution and nanosuspension "if our speculation is correct" should match, and it does alonzo2011 C35, C34. Jackson and co-workers warn that above the amorphous solubility nanospecies both absorb and scatter, and that filtering "may not be particularly useful" because the drug-rich phase may pass the filter and filtration may induce crystallization (cited) jackson2015 C12. Trasi and Taylor note that the failure of 0.45 µm filtration has "important analytical implications" and call ultracentrifugation "the optimum way" to determine dissolved drug for turbid dispersions trasi2015b C22, C27. Xie and co-workers note that the turbid suspension of tiny crystals interfered with the in situ UV, and that the very small crystals absorb light, which kept the absorbance well above that of a solution at the crystalline solubility xie2017 R11, C26. Trasi and co-workers caution that the concentration measured after dissolving a multicomponent formulation can include drug held in micelles or in a drug-rich phase, and rely on flux for the free drug trasi2016 C35, C36. (synthesis: the six papers point to one rule; any reading above the amorphous solubility estimate, or a plateau above the crystalline solubility after crystallization, is suspect until a method that excludes particles confirms it, because the excess is where the colloid or the fine crystals sit; and the filter that separates droplets has to be checked for the drug, since 0.7 µm passed and 0.22 µm held the tacrolimus droplets.) (synthesis: four later papers report readings above the amorphous solubility, or above 100% of the dose, without a particle-blind check: the paclitaxel UV excess, the posaconazole supernatants, the praziquantel plateaus and the delamanid values far above 1 µg/mL que2018 X4; elkhabaz2019a X3; polyzois2024 X3; nguyen2023c X9; by the reasoning above they are total drug, part of it in droplets or particles.)

Conditions and limits

The particle absorbance at the probe was not quantified; the nanosuspension reference comes from a thesis alonzo2011 K3, X7. The NMR comparison is a single pair of integrals alonzo2011 X6. Tracking sizes are given without the type of mean or distribution trasi2015b X13. The danazol UV readings were not corrected for the 2 mL withdrawn and replaced at each fluorescence time point jackson2015 X14. Replicate numbers of the indomethacin dissolution runs and HPLC checks are not stated xie2017 X6; the product(s) and replicates behind the tacrolimus filtration and tracking tests are not stated, and no croscarmellose-only control was run trasi2016 X6, X5.

Counterexamples and conflicts

Where no particles formed, UV agreed with the expected level: the 50:50 felodipine dispersion and pure amorphous felodipine read ≈8.9 and ≈8.6 µg/mL, near the predicted 9 µg/mL alonzo2011 R12, M1. In acidified celecoxib–HPMCAS suspensions, where the polymer barely dissolves, drug in the liquid was assayed after a 0.45 µm nylon filter without a colloid check chen2015a P11. The LLPS chapter compares onset methods and finds that absorbance alone and scattering alone can mislead (Amorphous solubility and LLPS F4).

Transfer check

If the medium turns cloudy or bluish, or extinction rises at a wavelength where the drug does not absorb, treat UV values as total drug ilevbare2013d P3; trasi2015b R6. Confirm the dissolved level by ultracentrifugation (for example 40,000 rpm for 20 min), by flux across a membrane that excludes droplets, or by solution NMR, and check the UV spectrum for a shift of peak or valley trasi2015b P9; alonzo2011 P4, R20. Do not rely on 0.45 µm filtration to separate nanodroplets trasi2015b C22; if a filter is used, test which pore size retains the drug's droplets in the medium (0.22 µm for tacrolimus) and close the mass balance with a filter rinse trasi2016 R8, R13. After crystallization in a turbid suspension, take the dissolved concentration from a filtered or centrifuged sample xie2017 R11.

History

Particle absorption by felodipine nanosuspensions had been described in work the 2011 paper cites, among it a paper of the library's Solution methods chapter (builds_on, as listed in the evidence index) alonzo2011 R25. Alonzo and co-workers (2011) were the first in this chapter to show the artefact for dispersions and to resolve it by flux and NMR alonzo2011 C27. Jackson and co-workers (2015) repeated the warning and used flux jackson2015 C12; Trasi and Taylor (2015) showed that 0.45 µm filters pass the HPMCAS colloid trasi2015b C22. Trasi and co-workers (2016) found the pore size that separates tacrolimus droplets and met the limit of particle tracking in a product with an insoluble excipient trasi2016 R8, C39; Xie and co-workers (2017) showed the overreading for a suspension of small crystals and checked it by HPLC xie2017 R11. Later papers of the chapter used in situ UV or moderate centrifugation without a particle-blind check, and their extractors flag the excesses que2018 X4; elkhabaz2019a X3; polyzois2024 X4; nguyen2023c X9.

How it was shown

Ilevbare and co-workers (2013) added a 10:90 ritonavir–PVP dispersion in about 3 mg portions to 25 mL of 100 mM pH 6.8 phosphate buffer at 37 °C, up to about 20 µg/mL, and diluted it to 10 µg/mL after about 60 min, following count rate and size by dynamic light scattering in a flow-through cell and extinction at non-absorbing wavelengths ilevbare2013d P4, P3. The dispersed phase from a dispersion dissolved in water was ultracentrifuged (274356g, 15 min) and scanned by DSC, a pyrene probe was followed after dispersion dissolution, and dried samples were imaged by SEM over 480 min ilevbare2013d P9, P7, P8. Jackson and co-workers (2015) read the pyrene I1/I3 ratio on aliquots during film dissolution and sized the scattering phase by nanoparticle tracking at the concentration maximum of the 90:10 dispersions jackson2015 P2, P3, P5. Alonzo and co-workers (2011) watched dispersion particles under crossed polars and sized the colloid by light scattering in dissolving dispersions and in methanol-shift solutions at 25 and 37 °C alonzo2011 P6, P5. Trasi and Taylor (2015) noted turbidity and tracked particles in filtrates of 20% ritonavir and lopinavir dispersions trasi2015b R6, P8. Xie and co-workers (2017) followed extinction at 450 nm, where indomethacin does not absorb, the drug's own fluorescence near 378 nm and DLS during titration with 100 µg/mL HPMC at pH 2.2, and filtered (0.45 µm) samples of dissolving dispersions for DLS and fluorescence xie2017 P4, P5, P6, P10. Trasi and co-workers (2016) tracked particles in a 60 µg/mL tacrolimus solution and in filtrates of dissolved capsules trasi2016 P13.

What the data show

Ritonavir. Count rate ≈150–230 kcps at 0 and 10 µg/mL, ≈4700 kcps at the first 20 µg/mL point, falling to ≈3550 kcps over 20 min; after dilution ≈190–260 kcps ilevbare2013d R1. The data bracket the onset between 10 and 20 µg/mL; the 18.3 µg/mL quoted for it is the separately titrated LLPS concentration with PVP ilevbare2013d X8, R10. The isolated phase showed a glass transition with onset −2.69 °C, inflection 1.05 °C and end 3.64 °C, against 50 °C for amorphous ritonavir ilevbare2013d R2, M1. I1/I3 fell from ≈1.91 to ≈1.50 within ≈8 min, stayed at ≈1.43–1.46 to 105 min and rose to ≈1.77 at 120 min ilevbare2013d R3. SEM showed smooth spheres mostly below 1 µm at 30 min, merging droplets at 60 min and needle clusters at 120 and 480 min; droplets from a solvent shift with 180 µg/mL PVP looked the same ilevbare2013d R4, R11. Danazol. The 90:10 PVP dispersion passed 8 µg/mL after ≈27 min and peaked at ≈9.4 µg/mL at ≈60 min as I1/I3 fell from ≈1.95 to ≈1.56; HPMC peaked at ≈9.9 µg/mL with I1/I3 ≈1.44–1.46, HPMCAS at ≈16.2 µg/mL with ≈1.39–1.57 jackson2015 R2. Onsets read from the probe were ≈6.0–10.4 µg/mL for eight dispersions against ≈8.1–8.3 µg/mL for antisolvent solutions jackson2015 R5. Tracking gave 256 ± 75 (PVP), 284 ± 81 (HPMC) and 246 ± 54 nm (HPMCAS) at 4.1–4.5 × 10⁹ per mL, against 205–267 nm and 2.9–3.2 × 10⁹ per mL for antisolvent solutions jackson2015 R6. Felodipine. The 90:10 HPMC particle had dissipated within 5 min and the medium turned uniformly cloudy, whereas a 50:50 particle kept its shape for at least 30 min alonzo2011 R4, R3. Particles from the 90:10 dispersion and from methanol-shift solutions with HPMC started at ≈183–203 nm and reached ≈280–307 nm by 90 min at 25 °C, and grew to ≈720 nm near 10 h; at 37 °C the dispersion-born particles started larger (≈259–274 against ≈194–219 nm) alonzo2011 R8, R6, R11, X19. 90:10 PVP–indomethacin gave 210 nm at 30 min; 50:50 PVP–indomethacin no correlation function alonzo2011 R17. Ritonavir and lopinavir, HPMCAS. Uniform turbidity within minutes, colloids of about 500 and 160 nm passing 0.45 µm filters; PVP dispersions dissolved more slowly with clear filtrates trasi2015b R6, R7. Indomethacin. With HPMC the 319 nm absorbance lost linearity and the 450 nm extinction rose from 36 µg/mL, DLS found droplets of 70–100 nm (PDI below 0.1) at or above 40 µg/mL, and the fluorescence band near 378 nm, negligible below 40 µg/mL, grew from ≈88 at 40 to ≈445 at 80 µg/mL xie2017 R4, R6, R7. Among dispersions, only the 10% HPMC one held droplets (65–80 nm, PDI about 0.24) in its 60-min filtrate, and the EPO–HPMC dispersion gave ≈75 nm species (PDI about 0.1) at 30 and 60 min, which crystallized after 3 h if the pH stayed at 2.2 xie2017 R14, R15. Tacrolimus. A 60 µg/mL solution held a second phase of 220 nm mean size after 0.7-µm filtration; for the dissolved capsules no drug colloid could be shown (F8) trasi2016 R11, C39. Posaconazole. After transfer to FaSSIF, 10% and 25% HPMCAS dispersions gave nanodroplets of about 304–339 and 328 nm and the tablet powder about 218 nm; the 10% droplets grew while their number fell (F26) elkhabaz2019a R13, R14. Delamanid. Drug-rich droplets formed at 50 µg/mL crystallized at once without polymer but stayed at about 440 (HPMCP) and 600 nm (HPMCAS) for 3 h, and salt dispersions produced stable nanodroplets during release duong2022a R7, R5, R25.

Figure

Both signals switch on between the 10 and 20 µg/mL plateaus and off again after dilution to 10 µg/mL, which crystals at ten times their solubility could not do ilevbare2013d R1, C6.

The pyrene ratio (blue) falls as the UV curve crosses the 8 µg/mL line and rises again when crystallization depletes the drug-rich phase jackson2015 R12.

Why

For ritonavir the authors argue that crystals would not dissolve at 10 µg/mL, far above the 1.3 µg/mL crystalline solubility, so the reversible scattering "strongly support[s]" liquid droplets; the low Tg is "consistent with" a ritonavir-rich phase holding a few percent of water, a supercooled liquid at 37 °C ("indicating"); and an LLPS origin of the colloids of rapidly dissolving dispersions is "highly likely" ilevbare2013d C6, C7, C8, C33. For danazol the pyrene drop is "consistent with" the probe entering a newly formed disordered drug-rich phase, and the scattering phase gives "additional support" for amorphous precipitation without loss of the amorphous advantage jackson2015 C13, C25. For felodipine the authors "speculate" that a 90:10 dispersion puts almost all its drug into a highly supersaturated state that immediately crystallizes; continuous growth, birefringence at about 1 µm, the spectra and the fast crystallization of amorphous felodipine in water lead them to conclude ("it would be reasonable to conclude") that the particles were crystalline, an origin in amorphous nanodomains "seems unlikely" alonzo2011 C29, C33, C36. No diffraction of the 200–300 nm particles was made alonzo2011 X7. Later papers of the library cite the 2011 work for nanodroplets formed by fast ASD dissolution at low drug loading (builds_on, as listed in the evidence index of this subcategory). For indomethacin, Xie and co-workers read the extinction rise with absorbance staying high as a liquid droplet phase rather than crystals, which would drop the absorbance, and the drug's fluorescence as "further evidence", its quantum yield being "anticipated" to rise when molecules are close together; the onset is "nearly identical" to the calculated amorphous solubility at 37 °C (cited) xie2017 C16, C17, C25, C23. In the EPO–HPMC dispersion, fast release by the ionized EPO carried the drug past the amorphous solubility and HPMC kept the excess from crystallizing ("attributed to") xie2017 C34, C36. Trasi and co-workers link the absence of a demonstrable colloid from the tacrolimus capsules to a 50% drug loading reported for the innovator, which they "believe" makes dissolution drug controlled (F10) trasi2016 C41. (synthesis: felodipine was the neat amorphous drug that crystallized fastest in water in this chapter (F1), which fits a crystalline colloid, but the evidence is indirect, and for ritonavir, danazol and indomethacin the same kind of colloid was read as non-crystalline.)

Conditions and limits

Ritonavir at 37 °C in 100 mM buffer or water; danazol at 25 °C in 10 mM buffer; felodipine at 25 and 37 °C ilevbare2013d P4, P9; jackson2015 P2; alonzo2011 A4. Tracking cannot tell droplets from crystals and was run on 90:10 danazol dispersions only jackson2015 X13. The ritonavir probe run is described in two media ilevbare2013d X1. Droplet sizes differ by method (SEM of dried samples about 500 nm where DLS in water starts at ≈240 nm) ilevbare2013d X12. Indomethacin at pH 2.2 and 37 °C; the non-crystalline state of its droplets rests on optical methods, valid within the first hour xie2017 X18. Tacrolimus at pH 4.5 and 37 °C trasi2016 P13.

Counterexamples and conflicts

The 50:50 dispersions of felodipine and indomethacin formed no detectable colloid, and slower PVP dispersions of ritonavir and lopinavir gave clear filtrates alonzo2011 R7, R17; trasi2015b R6. The ritonavir phase of the LLPS chapter, made by a solvent shift, had a Tg of 0.4 °C, close to the 1.05 °C here (Amorphous solubility and LLPS F2) ilevbare2013a R5. None found in the library that shows the 2011 felodipine particles to be amorphous by a direct measurement. Slowly releasing indomethacin dispersions (30% and 50% HPMC) formed no droplets in acid, and the six tacrolimus products, at a high drug loading, gave no sign of drug beyond the amorphous solubility xie2017 R14; trasi2016 C40; the indomethacin droplets of the EPO–HPMC dispersion crystallized after 3 h at pH 2.2 xie2017 R15.

Transfer check

Add the dispersion stepwise, then dilute below the onset: a scattering signal that disappears means droplets, not crystals ilevbare2013d P4, C6. Isolate the dispersed phase by ultracentrifugation for DSC, follow a polarity probe such as pyrene, and compare the onset with the LLPS concentration from a solvent shift in the same medium (Amorphous solubility and LLPS F4) ilevbare2013d P9, P7. For a fast crystallizer, check the colloid by diffraction or polarized light within the first minutes alonzo2011 X7.

History

Colloids from a β-carotene–PVP dispersion were reported in 1965 (cited) ilevbare2013d R22. Alonzo and co-workers (2011) reported submicron particles from 90:10 dispersions alonzo2011 C13. Ilevbare and co-workers (2013) were the first in the library to show that a dispersion colloid is a drug-rich liquid formed at the LLPS concentration ilevbare2013d C59. Jackson and co-workers (2015) extended it to danazol with three polymers and three loadings, and Trasi and Taylor (2015) to HPMCAS dispersions of ritonavir and lopinavir jackson2015 C29; trasi2015b C45. Xie and co-workers (2017) added indomethacin, detected without a probe by the drug's own fluorescence, and the redissolution of the droplets when the pH rose (F21) xie2017 C17, C5; the title gained indomethacin then. Elkhabaz and co-workers (2019) found nanodroplets after a pH shift from posaconazole dispersions that already held crystals from the acid stage (F26), and Duong and co-workers (2022) stable droplets from delamanid salt dispersions elkhabaz2019a C46; duong2022a C39.

How it was shown

Wegiel and co-workers (2015) stirred 20 mg of crystalline resveratrol in 20 mL of pH 6.8 buffer holding 5, 10 or 20 mg/mL HPMC or HPMCAS (or saturated E100 or CMCAB) for 24 h at 37 °C and measured the supernatant by UV; they added rising amounts of crystal (658–11,037 µg/mL nominal) to 20 mg/mL PVP and examined the precipitate by PXRD and IR; and they stirred 6–35 mg/mL of a 15:85 resveratrol–HPMCAS dispersion for 24 h and used the solutions as donors in a side-by-side diffusion cell (6–8 kDa membrane), against a saturated crystal suspension wegiel2015 P2, P3, P10, A4. The apparent supersaturation was the excess over the crystal solubility at the dissolved-polymer level, and supersaturation was taken as proportional to flux wegiel2015 E2, E3. Xie and co-workers (2017) shook γ and α indomethacin for 48 h at 37 °C in pH 2.2 McIlvaine buffer with 50–1000 µg/mL HPMC, EPO or both, and α at pH 5 with 100 µg/mL, and located the amorphous solubility by titration with extinction at 450 nm xie2017 P2, P4. The comparators measured crystalline solubility with 1 mg/mL polymer (felodipine), 22 µg/mL polymer (celecoxib, 48 h, ultracentrifuged) or an excess of each polymer (lopinavir) konno2008a P4; xie2015 P6; xie2016a P7; li2018 P1.

What the data show

Resveratrol with HPMCAS. Crystal solubility was 42 ± 6 µg/mL in buffer and ≈141, 414 and ≈1230 µg/mL with 5, 10 and 20 mg/mL HPMCAS; with HPMC ≈38, ≈51 and ≈85 µg/mL wegiel2015 R2, R3. The 65% HPMCAS dispersion gave 458 µg/mL with 4.7 mg/mL dissolved polymer, the 85% one ≈1250 µg/mL against ≈82 µg/mL crystal solubility at its polymer level; the polymer-corrected excess values (155 and about 1166 µg/mL) conflict between table, figure and text wegiel2015 R11, R15, X2, X3. In the loading series dissolved resveratrol rose from ≈530 to ≈1610 µg/mL (6.4 to 18.8 mg/mL loading) and then fell with a gelatinous precipitate, while flux fell from ≈4.4 to ≈2.6 µg/min and stayed at ≈2.8–3.1 µg/min beyond; the crystal gave 2.0 µg/min wegiel2015 R19, R20, X9. Resveratrol with PVP. Added at 658, 1795, 4231, 6744 and 11,037 µg/mL, it gave 709, 1610, 1193, 706 and 99 µg/mL in solution, with a growing non-crystalline precipitate containing resveratrol and PVP wegiel2015 R6, R7. The PVP dispersion "did not readily dissolve" and gave 42 µg/mL; E100, which does not dissolve at pH 6.8, gave 46 µg/mL wegiel2015 R12, R11, C28. Indomethacin with EPO. Crystalline solubility rose from 1.61 to 1.9, 2.3 and 6.2 µg/mL (γ) and from 2.8 to 2.8, 3.4 and 8.9 µg/mL (α) at 50, 150 and 1000 µg/mL EPO; HPMC left it at 1.35–1.5 (γ) and 2.47–2.9 µg/mL (α); at pH 5 the α solubility was 15 alone, 17 with HPMC and 27 µg/mL with EPO xie2017 R1, R2. The amorphous solubility was 36 µg/mL with 100 µg/mL HPMC, 41, 50 and 56 µg/mL at HPMC/EPO 150/50, 100/100 and 50/150, and above 80 µg/mL with 1000 µg/mL EPO; with EPO alone the drug crystallized during the titration xie2017 R3, R8. Predissolved EPO lifted the plateau of a 10% HPMC dispersion from ≈41 to ≈53, ≈69 and ≈77.5 µg/mL at 150, 500 and 1000 µg/mL, but only the highest level held to 300 min xie2017 R18, X19. Other pairs. Felodipine 0.98–1.01 against 0.94 µg/mL with 1 mg/mL polymer; celecoxib 1.72–2.14 against 1.5 µg/mL with 22 µg/mL polymer; lopinavir 2.6–3.6 against 3.1 µg/mL with excess polymer konno2008a R1; xie2015 R11; xie2016a R1; li2018 R1.

Figure

Concentration peaks near 18.8 mg/mL loading where flux is lowest; the dashed crystal flux of 2.0 µg/min lies not far below: most of the dissolved resveratrol is not free wegiel2015 R19, R20.

HPMCAS raises the solubility steeply above a few mg/mL, HPMC only modestly; the 20 mg/mL HPMCAS point exceeds the drug stated as added wegiel2015 R3, X5.

Why

Wegiel and co-workers equate solubilization by polymers with drug–polymer complex formation, and recall that complexed drug raises the total concentration but not the free drug, so that membrane transport is not faster and complexed solutions are thermodynamically "quite different" from supersaturated ones wegiel2015 C25, C23, C44. On the rising branch of the loading series more dispersion gave more total but less free resveratrol ("indicating"), and the gain from HPMCAS is "mainly due to" a soluble complex wegiel2015 C19, C41. PVP is known to form insoluble complexes with phenols above a critical ratio (cited), so the complexation with three-phenol resveratrol "can be readily rationalized"; the PVP dispersion's insolubility "appears" to arise from it, and such a complex "would be expected" to hinder crystallization wegiel2015 C31, C32, C29, C34. The authors suggest that soluble complexes "might explain" very high concentrations reported for other polyphenol dispersions wegiel2015 C43. For indomethacin, Xie and co-workers attribute the higher solubilities "presumably" to drug–polymer interactions in solution that lower the thermodynamic activity of the drug, so that LLPS begins at a higher concentration; where the crystalline solubility rises, the amorphous solubility "typically" rises too (cited) xie2017 C55, C56, C27; in the solid, EPO is expected to form a salt with the drug's carboxylic acid (cited) xie2017 C49. (synthesis: in both cases the solubilization grew with the polymer level, resveratrol from ≈141 to ≈1230 µg/mL between 5 and 20 mg/mL HPMCAS and indomethacin from 1.9 to 6.2 µg/mL between 50 and 1000 µg/mL EPO, while the near-unchanged values of the other drugs were measured at 22 µg/mL to 1 mg/mL; a polymer that does not solubilize at a screening level may do so at the mg/mL a high-polymer dispersion releases, 3.7–4.7 mg/mL in the resveratrol slurries wegiel2015 R11. The LLPS chapter shows the same pattern for a micellar solubilizer, which raises both solubilities together (Amorphous solubility and LLPS F12), and for the cationic EUD-E with ibuprofen (Amorphous solubility and LLPS F19).)

Conditions and limits

Resveratrol at pH 6.8 and 37 °C with mg/mL levels of dissolved polymer; flux was measured only for the 15:85 HPMCAS series, not for the 65% slurry, HPMC or PVP wegiel2015 C10. Indomethacin at pH 2.2 and 5 xie2017 P2. UV values in PVP solutions carry at least about 8% error wegiel2015 X12.

Counterexamples and conflicts

E100 gave no gain for resveratrol because it does not dissolve at pH 6.8, although the abstract names complexation wegiel2015 C28, X8. HPMC at 10 µg/mL raised the danazol crystalline solubility 2.4-fold (synthesis: 2.2/0.9) at a low level jackson2014 R1 (Crystallization from solution F19). For the six tacrolimus products the maximal flux showed no solubilization by the excipients, including sodium lauryl sulfate trasi2016 C38. A solubilizing excipient added to a supersaturated solution consumes supersaturation by the factor of its solubilizing effect, a Bioperformance paper finds raina2015c C33.

Transfer check

Measure the crystalline solubility at the dissolved-polymer level the dispersion actually releases (measure that level, as for resveratrol by HPLC with an aerosol detector), refer the supersaturation to it, and measure free drug by flux wegiel2015 P8, E2, P10. A flux that falls while the concentration rises means drug bound in solution wegiel2015 C19. For phenolic drugs with PVP watch for a precipitate as the drug-to-polymer ratio rises; for acidic drugs with a cationic polymer expect solubilization wherever the polymer dissolves wegiel2015 C31; xie2017 C55.

History

Konno and co-workers (2008) ruled out complexation of felodipine at 1 mg/mL polymer, and Alonzo and co-workers (2011) found no felodipine–HPMC complex by flux (F4, F7) konno2008a C17; alonzo2011 C23. Wegiel and co-workers (2015) were the first in the library to show soluble and insoluble drug–polymer complexes during dispersion dissolution wegiel2015 C49; Xie and co-workers (2017) found EPO raising both solubilities of indomethacin xie2017 C2. Later papers cite the resveratrol work for concentration not being free drug mosqueragiraldo2018b C5; jackson2016 C43, and the LLPS chapter measured how a cationic Eudragit raises ibuprofen solubility by complexes in the bulk ueda2020b C34.

How it was shown

Alonzo and co-workers (2011) made 90:10 and 50:50 polymer:drug dispersions of felodipine and indomethacin with HPMC or PVP by rotary evaporation, watched particles disintegrate under crossed polars, dissolved them under non-sink conditions with in situ UV, and dissolved three amounts of the 90:10 HPMC–felodipine dispersion alonzo2011 P1, P6, P2, A5. Jackson and co-workers (2015) dissolved spin-coated danazol films at 90:10, 70:30 and 50:50 with PVP, HPMC and HPMCAS at 25 °C, set to give 20–30 µg/mL if all dissolved, and measured free drug by flux for 90:10 and 50:50 jackson2015 P1, P2, P4. Trasi and Taylor (2015) followed 20% total-drug dispersions in PVP and HPMCAS by ultracentrifugation at 20 min and 2 h trasi2015b P9. Raina and co-workers (2014) dissolved neat amorphous nifedipine and a 50:50 HPMC dispersion at 5 mg/mL solids, 37 °C raina2014b P11. Chen and co-workers (2015) suspended 33.3% drug celecoxib–HPMCAS spray-dried particles at pH 3, where the polymer barely dissolves chen2015a P2, P10. Xie and Taylor (2015, 2016) dissolved celecoxib dispersions with PVP-K12, PVP-K29/32, HPMCAS, HPMC or PAA at 10–60% drug (106–250 µm) at a dose giving 22 µg/mL, about the amorphous solubility, in 400 mL of pH 6.8 buffer at 37 °C for 16 h xie2015 P5; xie2016a P5. Xie and co-workers (2017) dissolved indomethacin–HPMC dispersions at 10, 30 and 50% drug and an EPO dispersion, each holding 4 mg drug, in 50 mL of pH 2.2 buffer (80 µg/mL nominal) xie2017 P7. Trasi and co-workers (2016) compared the early dissolution of six tacrolimus products of unstated loading trasi2016 P11, X4; Li and Taylor (2018) sampled lopinavir films at 3 and 48 h li2018 P5.

What the data show

Felodipine. The 90:10 dispersion read ≈18.6 µg/mL at 5 min and ≈23.4 µg/mL at 10 min, the 50:50 dispersion ≈3.3 and ≈5.5 µg/mL alonzo2011 R1. Three amounts of 90:10 dispersion (90, 60 and 30 µg/mL of felodipine) peaked at ≈75, ≈53 and ≈28 µg/mL within about 10 min; at 90 µg/mL the reading then fell to ≈22.6 µg/mL by 50 min alonzo2011 R2. At 25 °C 90% PVP particles disappeared almost instantly, and 50% PVP particles stayed intact alonzo2011 R4. Only the 90% polymer media turned uniformly cloudy, with felodipine or indomethacin alonzo2011 R3. Danazol. Read from the 50:50 profiles, PVP gave ≈9.3 µg/mL at 60 min, HPMC ≈4.6 µg/mL at 60 min, ≈8.3 at 300 min and ≈9.85 at 1200 min, and HPMCAS ≈2.2, ≈5.3 and ≈6.1 µg/mL at 60, 300 and 990 min jackson2015 R13. The pyrene ratio of 50:50 HPMCAS drifted only from ≈1.93 to ≈1.82–1.89 and ≈1.745 at ≈890 min jackson2015 R2. By flux, 50:50 HPMC gave the highest value measured (9.1 ± 1.6 µg/mL) and 50:50 HPMCAS the lowest (5.4 ± 0.8 µg/mL) jackson2015 R9. Ritonavir and lopinavir. At 20 min the single-drug dispersions gave ritonavir ≈26.5 µg/mL from HPMCAS and ≈16.5 µg/mL from PVP; by 2 h ≈26.9 and ≈22.0 µg/mL trasi2015b R12, R13. Nifedipine. The 50:50 HPMC dispersion reached ≈100 µg/mL at ≈22 min against ≈6 µg/mL for the neat glass raina2014b R14. Celecoxib at pH 3. Without surfactant the liquid held only about 1.5–3.5 µg/mL HPMCAS over 24 h, and drug in the liquid rose slowly from S ≈4.3 to ≈6.7 over 48 h chen2015a R18, R17. Celecoxib at pH 6.8. HPMCAS dispersions reached ≈18.3, ≈10.5, ≈5.0 and ≈2.0 µg/mL at 1 h and ≈21.8, ≈20.4, ≈21.1 and ≈15.8 µg/mL at 16 h at 10, 30, 50 and 60% drug, without any fall; HPMC dispersions reached only ≈16.5 (50%) and ≈18.2 µg/mL (30%) at 16 h xie2016a R4, R17. At 50% drug the PVP dispersions gave ≈16–17 µg/mL at 2 h and ≈20 µg/mL by 4 h, the HPMCAS one ≈9.0 at 2 h and the HPMC one ≈5.8 µg/mL at 2 h xie2015 R6; the PAA dispersions peaked at ≈9–11 µg/mL within 3 min xie2016a R4. Indomethacin at pH 2.2. The HPMC dispersions read ≈5.5, ≈9 and ≈30.5 µg/mL at 10 min for 50, 30 and 10% drug, and ≈18.5, ≈24.5 and ≈39.5 µg/mL at 60 min; the 10% one was biphasic, fast to ≈30.5 µg/mL in 10 min and then slow, and "appeared to just reach" the amorphous solubility; the EPO dispersion peaked near ≈77 µg/mL at about 5 min xie2017 R10, R13. Tacrolimus. At 5 min the products ranged from ≈26 (Prograf) to ≈41 µg/mL (Accord) and by 20 min from ≈41 to ≈49 µg/mL, all at the same plateau later (F7) trasi2016 R10. Lopinavir at pH 3. From 3 to 48 h the HPMCAS, CAP and HPMCP films kept rising (for example HPMCAS at 33% drug ≈5.0 → ≈9.3 µg/mL), while HPMC, EC and CA films released fast li2018 R2, R3, C22.

Figure

The 50:50 particle (a, b) keeps its shape for 30 min; the 90:10 particle (c, d) has dissipated within 5 min alonzo2011 R4.

The UV curve levels near 6 µg/mL, below the red amorphous-solubility line, and the pyrene ratio barely moves: no drug-rich phase forms jackson2015 R2, R13.

The PAA dispersions spike and crash within the first hour; the HPMCAS dispersions rise more slowly the higher the drug loading, and the 60% one stays below the dose at 16 h xie2016a R4.

Why

Following cited work, Alonzo and co-workers distinguish two regimes: at high polymer loading there is too little drug to form a layer that limits dissolution, and the drug dissolves at a fraction of the polymer rate (Eq. 4); at moderate to high drug loading the polymer leaves the surface faster and a drug-rich interface, "possibly in the amorphous state", regulates dissolution; their data "appear to agree" alonzo2011 C37, C38, E4. At 10% drug this gives "very high local supersaturations", which they hold to be "most likely local" and depleted by rapid precipitation, while at 50% drug the drug-rich layer, if it stays amorphous, "would presumably" cap the concentration at the pure amorphous level alonzo2011 C39, C47, C46. Jackson and co-workers cannot tell whether the missing drug-rich phase of the 50:50 HPMCAS dispersion comes from a lower drug solubility in this formulation, a much slower release, or crystallization during dissolution ("it is not possible to determine"), and judge that 50% loadings "might not be expected" to perform comparably in membrane transport in vivo jackson2015 C34, C37. Trasi and Taylor conclude that the polymer influenced the tendency to undergo LLPS on dissolution trasi2015b C45. Chen and co-workers state that at this "relatively low" drug loading drug release "appears to be" controlled by the polymer chen2015a C31. Xie and Taylor recall the cited analyses in which drug and polymer dissolve together at low loading while at high loading the polymer may dissolve faster and leave a drug-rich layer xie2016a C4, E1; they attribute the fast start of PAA dispersions "presumably" to PAA being fully ionized (pKa 4.28) and very hydrophilic at pH 6.8 xie2016a C24. For indomethacin the faster release at lower loading is "indicative of" polymer-controlled dissolution, HPMC gelling and releasing "probably through a matrix-regulated diffusion mechanism", while EPO, ionized at pH 2.2, dissolves fast xie2017 C31, C64, C29. Trasi and co-workers "believe" that the high (50%) drug loading reported for the tacrolimus innovator makes its dissolution drug controlled, which would explain why no drug colloid formed; the loading is unreferenced and unknown for the generics trasi2016 C41, X4. (synthesis: the polymer that dissolves fastest in the medium carries the drug fastest, and its speed depends on the pH: PAA at pH 6.8 and EPO at pH 2.2 are ionized; HPMCAS at pH 6.8 dissolved fastest in the ritonavir and lopinavir work but slowest of the three in the 50:50 danazol films, so the polymer ranking for release speed is not fixed either and depends on drug loading. A fast polymer that does not inhibit crystallization loses the gain within minutes, see F16 and F20.) For praziquantel the gradual decline with loading "suggests that the release is polymer controlled", and no "falling-of-the-cliff" effect appeared; polymer release was not measured polyzois2024 C17, C16. For delamanid, HPMCAS-HF released almost no drug at pH 6.5 while about 30% of the polymer dissolved, so dissolved polymer did not guarantee released drug nguyen2023c R11, R12. Elkhabaz and co-workers read the poor 50% posaconazole dispersion as competitive crystallization during release, "highlighting" its susceptibility elkhabaz2019a C44.

Conditions and limits

Only 10, 30 and 50% drug were tested (no 30% felodipine), the release regimes were not checked by measuring polymer release (except HPMCAS leaching in acid), and the 70:30 danazol profiles were not shown alonzo2011 A1; jackson2015 R4; chen2015a P12. The 50:50 danazol films were partly crystalline before dissolution, so the high-loading results mix a loading effect with residual crystals jackson2015 X10. The felodipine runs at three amounts were single runs and rose faster than the triplicate run at the same dose alonzo2011 X15. In the celecoxib series the dose equalled the amorphous solubility, so "complete" means reaching about 22 µg/mL, and the 10% HPMCAS dispersion was cryo-milled, not sieved xie2016a X11, X10; the indomethacin series ran at pH 2.2 only xie2017 P7.

Counterexamples and conflicts

The 50:50 PVP danazol dispersion released like the 90:10 one and gave the same free drug jackson2015 X9, R9. The release-mechanism chapter of the library (written later) tests the polymer-controlled regime with normalized drug and polymer release rates saboo2020a C16; indulkar2019a C43. The Bioperformance chapter finds that a lower loading is not always better for reaching the amorphous solubility (Bioperformance F8).

Transfer check

Watch dispersion particles in the medium for the first 5–30 min and follow the drug under non-sink conditions at two or three loadings alonzo2011 P6; jackson2015 A2. A dispersion that disintegrates within minutes and overshoots the amorphous solubility will form a colloid (F9); one that dissolves slowly and stays below it, as 50:50 HPMCAS danazol did, gives less free drug (F7). Check the starting dispersion for crystals before ranking loadings jackson2015 R1. When comparing products of one drug (for example a brand and generics), read the early part of a non-sink test as well as the plateau, since the products may differ only in rate trasi2016 R10.

History

Polymer-controlled and drug-controlled dissolution were described in cited work on PVP coprecipitates alonzo2011 R26. Alonzo and co-workers (2011) linked the two regimes to colloid formation in ASD dissolution alonzo2011 C39. Jackson and co-workers (2015) found LLPS up to 30% danazol with three polymers and slow release at 50% jackson2015 C36, C37; Trasi and Taylor (2015) found HPMCAS faster than PVP for ritonavir and lopinavir trasi2015b R14. Xie and Taylor (2015, 2016) and Xie and co-workers (2017) found the same loading dependence for celecoxib and indomethacin with HPMCAS and HPMC and contrasted it with fast-dissolving PVP, PAA and EPO xie2016a C20; xie2015 C19; xie2017 C31; the title gained celecoxib and indomethacin then. Later papers of the library cite the 2011 and 2015 work for the dependence of LLPS on drug loading and polymer (builds_on, as listed in the evidence index of this subcategory), and an enzalutamide study cites the 2016 and 2018 work for soluble polymers releasing fast but letting the drug crystallize and poorly soluble polymers limiting release wilson2020 C6. Polyzois and co-workers (2024) found no release cliff for praziquantel up to 40–50% drug and hot-melt extruded dispersions releasing at least as well as solvent-evaporated ones, and Alvarenga and co-workers (2025) found release "minimally affected" by the process polyzois2024 C16, C31; alvarenga2025 C48.

How it was shown

Trasi and Taylor (2015) coated 5 mg of drug (RTV only, LPV only, or RTV:LPV 7:3, 5:5, 3:7 by weight) inside vials by rotary evaporation, dissolved the films in 10 mL of 100 mM pH 6.8 buffer with 10 µg/mL HPMCAS as crystallization inhibitor at 300 rpm, filtered samples through 0.45 µm glass filters and assayed them by HPLC up to 60 min trasi2015b P2, P6, P10. They made 20% total-drug dispersions in PVP K-29/32 and HPMCAS-MF by rotary evaporation, dissolved 20 mg in 40 mL at 25 °C, and assayed filtered samples (PVP) or ultracentrifuged supernatants (both polymers) at 20 min and 2 h trasi2015b P1, P5, P9. For separate addition, equal weights of RTV and LPV were added as two single-drug PVP dispersions (filtered sampling) or two HPMCAS dispersions (ultracentrifuged at 60 min) trasi2015b P7. DSC gave one Tg for every drug mixture and dispersion trasi2015b R2. Drug-rich phase suspensions were made by adding methanol stocks above the LLPS concentration (RTV 50, LPV 35, RTV 25 + LPV 17, RTV 50 + LPV 35 µg/mL) with pyrene, followed by the I3/I1 ratio for 5 h, and the supernatant was assayed at 0 and 5 h trasi2015b P11, P12. The ceiling was computed from Eq. 1 with amorphous solubilities of 25 (RTV) and 17 µg/mL (LPV) cited from the group's antisolvent study trasi2015b E1, M1, M2.

What the data show

Films. Each film released drug quickly over the first 20–30 min and then rose slowly; the predicted ceilings were 16.7, 11.6 and 6.8 µg/mL for RTV and 5.6, 9.0 and 12.3 µg/mL for LPV at 7:3, 5:5 and 3:7 trasi2015b R3, R4, R5. PVP dispersions, filtered, 60 min. RTV ≈20.4 µg/mL from the RTV dispersion against ≈12.2, ≈10.8 and ≈6.0 from 7:3, 5:5 and 3:7; LPV ≈18.0 against ≈12.0, ≈9.1 and ≈5.9 from 3:7, 5:5 and 7:3 trasi2015b R9. Ultracentrifuged, 2 h, HPMCAS / PVP. RTV ≈17.0 / ≈14.9 (7:3), ≈12.0 / ≈11.4 (5:5), ≈7.0 / ≈6.4 (3:7); LPV ≈13.0 / ≈12.0 (3:7), ≈8.0 / ≈8.4 (5:5), ≈5.2 / ≈5.6 (7:3) trasi2015b R13. Plotted against RTV mole fraction in the drug-rich phase, the HPMCAS RTV points (≈7.0, ≈12.0, ≈17.0 µg/mL) lie on the Eq. 1 line (≈7.0, ≈11.7, ≈16.9), and the LPV points (≈13.0, ≈8.0, ≈5.0) on or up to about 15% below it (≈12.2, ≈9.2, ≈5.9) trasi2015b R17, X8. Separate addition. With two PVP dispersions RTV rose to ≈15.2 µg/mL at 20 min and fell to ≈12.2 µg/mL at 60 min as LPV climbed to ≈9.7 µg/mL trasi2015b R10. Miscibility. Tg 46 °C (RTV), 77 °C (LPV), 66, 60.5 and 54 °C for 3:7, 5:5 and 7:3 trasi2015b R2. Crystallization of drug-rich phases. LPV alone: I3/I1 ≈0.645 at 0 h, ≈0.640 at 1 h, ≈0.565 at 2 h and ≈0.510 at 5 h, with 4.3 µg/mL free LPV at 5 h; RTV alone ≈0.62–0.63 with 25.4 µg/mL at 5 h; both mixtures ≈0.63–0.66, with free RTV 14.2–15.8 and LPV 7.0–7.5 µg/mL unchanged from 0 to 5 h, at nearly the same values for twice the total drug trasi2015b R15, R16.

Figure

The pure film (diamonds) reaches close to the blue amorphous-solubility line; each step toward a larger share of the other drug lowers the plateau trasi2015b R3, R4.

The RTV diamonds sit on the blue line; the LPV circles sit on or slightly below the orange line trasi2015b R17, X8.

Why

The authors trace the drop to mixing of the two drugs in the disordered drug-rich phase, which lowers the chemical potential of each and with it the concentration of each in the water, "as long as the two components are miscible in the drug-rich phase" trasi2015b C36. Assuming ideal mixing (cited), the ceiling is C_{max} = S_a x_d; the good agreement "indicat[es]" a close to ideal mixture, and the single Tg of each blend is read as miscibility trasi2015b C38, C39, C13. In the LLPS chapter the same relation is written from the ideal-mixing chemical potentials \mu_1 = \mu_1^{\circ} + RT\ln x_1 trasi2015a E2, E4. The solution phase behaviour is the same whether supersaturation comes from antisolvent addition or from dissolving a dispersion trasi2015b C44. After separate addition, the authors read the drop as mixing in solution followed by LLPS, so that intimate contact in the formulation is not needed ("can also potentially occur"); no colloid or drug-rich phase was analysed in those runs trasi2015b C47, X9. The later crystallization of the mixed phases is attributed "at least in part" to the lower supersaturation that follows from the lower amorphous solubility, presented as "one potential advantage" trasi2015b C51, C49. (synthesis: the polymer sets only how fast the lowered ceiling is reached (HPMCAS within 20 min, PVP in about 2 h); the composition of the drug-rich phase sets its height.) Li and Taylor (2018) apply the same reasoning to a polymer that does not dissolve: mixing with it lowers the chemical potential of the drug, and the lopinavir ceiling fell below the amorphous solubility, near or below the line of the amorphous solubility scaled by drug weight fraction for several polymers (F18) li2018 C9, C43, E7.

Conditions and limits

One drug pair that mixes almost ideally, 20% total drug, a simple buffer, 25 °C for powder dissolution and unstated temperatures for films, ultracentrifugation and pyrene runs trasi2015b X5. The Eq. 1 check uses cited amorphous solubilities; the pure films reached about 8% less trasi2015b X14. The mole fractions need molecular weights that the paper does not print trasi2015b E1. Crystallization was followed for 5 h only, and the attribution to lower supersaturation is not separated from other effects of ritonavir trasi2015b X11.

Counterexamples and conflicts

The LPV points of the dispersions lie up to about 15% below the ideal-mixing line trasi2015b X8. In the LLPS chapter an ionized partner (diclofenac) did not lower the ritonavir ceiling, because it stayed in the water (Amorphous solubility and LLPS F11) trasi2015a R11. None found in the library in which a miscible un-ionized second drug failed to lower the ceiling.

Transfer check

For two amorphous drugs in one product, or two amorphous products taken together, check that the pair is miscible (one Tg of the blend) and both un-ionized at the intestinal pH, compute each ceiling as amorphous solubility times mole fraction, and confirm it by dissolving films or the dispersion with ultracentrifugation trasi2015b R2, E1, P9. The minor drug loses most: in the 3:7 film ritonavir reached about a quarter of its pure-film value (synthesis: 6.3/22.9 = 0.28) trasi2015b R5. Weigh the lower ceiling against the slower crystallization of the mixed phase trasi2015b C49.

History

The group's antisolvent study, in the LLPS chapter, first showed that a second drug lowers the amorphous solubility of each trasi2015b C7; trasi2015a R7. Trasi and Taylor (2015) carried it to dissolving films and dispersions and to separate products taken together trasi2015b C9, C47. Later papers of the library cite it for the lower maximum supersaturation of binary amorphous combinations, for co-granules of ritonavir and lopinavir and for atazanavir with darunavir (builds_on, as listed in the evidence index of this subcategory) trasi2019 C38. Within this chapter, a lopinavir leaching study took the mutual lowering of amorphous solubility as the precedent for drug and polymer holding each other back in a dispersion li2016b C33, C34, and Li and Taylor (2018) cite it as the basis for an undissolved polymer lowering the ceiling (F18) li2018 C9.