Two ways an amorphous drug loses its advantage during dissolution, and how dissolved polymers slow each

Background and motivation

Many drug candidates are too poorly soluble in water to be absorbed well. Making the drug amorphous is one remedy: an amorphous solid has a higher free energy than its crystal and can produce solution concentrations many times above the crystalline solubility, and several in vivo studies had shown that this raises bioavailability (refs 3–9) C1. The same higher free energy makes the amorphous form metastable. A great deal of work had gone into keeping amorphous drugs from crystallizing during storage (refs 10–14), but much less into what happens while they dissolve, so the shape of a dissolution profile from an amorphous drug was poorly understood when this paper was written C2.

The introduction names two ways in which the advantage can be lost. First, an amorphous API in contact with water tends to crystallize by a solid-to-solid transition. If that happens quickly, the measured supersaturation is much lower than theory predicts, and if it is extremely rapid, "it is possible" that no supersaturation is seen at all (Hancock and Parks 2000, ref 15) C2. Second, once a supersaturated solution exists it has a thermodynamic driving force to crystallize; if the high concentration does not last long enough, the benefit of the amorphous form is "marginal at best" C3. Polymers are normally built into amorphous formulations to stop crystallization in storage (refs 20–23). The authors argue that keeping the solid stable during dissolution is "equally as important" and "should be as much of a consideration for formulators" C4.

The group had already shown, in a study of felodipine solid dispersions (Konno et al. 2008, ref 25), that felodipine crystallizes rapidly from aqueous solution at about 9 µg/mL without polymer, that HPMC and HPMC-AS substantially inhibit this, and that PVP is much less effective R20. That study dealt with the solution route. The present paper asks the broader question for neat amorphous drug: which of the two routes limits the concentration, and can dissolved polymer slow each one?

The working hypothesis is stated explicitly ("It is hypothesized"): the supersaturation reached when an amorphous solid dissolves depends on how fast the solid crystallizes and on how readily the supersaturated solution crystallizes, and the concentration–time profile is set by the rates of these transformations relative to the rate of dissolution C5. Two poorly soluble model drugs, felodipine and indomethacin, were dissolved as neat amorphous powders at 25 and 37 °C, with and without PVP, HPMC or HPMC-AS dissolved in the medium beforehand A1·A2·A4. Crucially, the solid was followed in parallel with the solution, so each loss of concentration can be traced to one route or the other P5·P8·P9.

Background concepts

Supersaturation and the amorphous solubility. A solution is supersaturated when its concentration exceeds the equilibrium solubility of the stable crystal. An amorphous solid can drive the concentration up to its own, higher solubility, the "amorphous solubility", but no further. The paper estimates that ceiling from thermal data. Equation 1 relates the solubility ratio to the free-energy difference between the two forms (taken from Hancock and Parks, ref 15) E1:

In plain terms: the amorphous form is more soluble than the crystal by a factor that grows exponentially with how much more free energy it has. Here ${\sigma }^{\mathrm{a}\mathrm{m}\mathrm{o}\mathrm{r}\mathrm{p}\mathrm{h}}$ and ${\sigma }^{\mathrm{c}\mathrm{r}\mathrm{y}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{l}}$ are the solubilities of the amorphous form and of the stable crystal, $\Delta G$ is the free-energy difference between them, $R$ is the gas constant and $T$ the absolute temperature E1. The free-energy difference is estimated with the Hoffman equation (Equation 2, ref 24) E2:

In plain terms: the further the working temperature lies below the melting point, and the larger the heat of melting, the larger the free-energy excess of the amorphous form. $\Delta H_f$ is the enthalpy of fusion, $T_m$ the melting temperature and $\Delta T=T_m-T$ E2. With the measured values for felodipine (30.12 kJ/mol, 143.41 °C) and a crystalline solubility of 1 µg/mL, the estimate at 37 °C is about 9 µg/mL M1·R3. For indomethacin (32.48 kJ/mol, 153.4 °C) the authors give 17–25 µg/mL at 25 °C, using a measured α-form solubility of 1–1.5 µg/mL, and about 37 µg/mL at 37 °C, using 3 µg/mL M2·R8. These estimates are the yardsticks against which every measured peak in the paper is judged.

Two crystallization routes. The paper's central picture (Figure 16) has three arrows leaving the dissolving amorphous particles C43. Dissolution feeds a supersaturated solution; from that solution, crystals can nucleate and grow, which pulls the concentration back down. The third arrow, "crystallization from matrix", is the amorphous particle itself turning into crystal on contact with the medium. The first two together are a solution-mediated transformation: the crystals form in the liquid. The third is a solid-state conversion. Figure 16 prints a proportionality for each step E3·E4·E5:

In plain terms: the dissolution rate scales with the exposed surface area $A$ and with the gap between the solution concentration $C$ and the equilibrium concentration $C_{eq}$ (a modified Noyes–Whitney equation, ref 35) E3.

In plain terms: the nucleation rate $J$ climbs steeply as the supersaturation $S$ rises (ref 36). The figure prints this term with square brackets and without the square on $\ln S$ that the classical expression carries, and the text describes $J$ simply as "proportional to the degree of supersaturation" E4·X3.

In plain terms: crystals grow faster the further the concentration lies above the equilibrium concentration (ref 37) E5. The authors use the three relations only qualitatively. Their message is that the higher the supersaturation, the higher the nucleation and growth rates, so the more an amorphous drug supersaturates, the more likely the solution is to crystallize C43.

Non-sink conditions. All dissolution tests added far more amorphous solid than the medium could dissolve at the crystalline solubility. Under these conditions the profile shows the highest concentration the solid can actually generate, rather than how fast a small dose dissolves P2·P3·R1.

Solvent shift. To study the solution route on its own, the authors made indomethacin solutions supersaturated by adding a small volume of a methanol stock to buffer, with no solid present. Any loss of concentration then comes from crystallization in the solution P4.

Following the solid in situ. Raman spectroscopy with an immersion probe records spectra of a slurry every minute. For felodipine the spectra were converted to % crystalline by a partial-least-squares (PLS) model calibrated on amorphous/crystalline mixtures; for indomethacin, whose spectra had fluorescent backgrounds, the authors tracked one band of the α crystal form (1,648 cm⁻¹) and one band found only in the amorphous form (753 cm⁻¹) P6·P7. Powder X-ray diffraction (PXRD) of wetted powder and polarized-light microscopy, which shows crystals as bright, birefringent objects, gave independent checks P8·P9.

Methods in detail

The study has no new model; its contribution is experimental design. Each drug was examined along two tracks at once, the concentration in solution and the state of the solid, under matched media and temperatures, and the solution route was then isolated with solvent-shift tests P2·P3·P4·P5. The flow below summarizes the work.

Materials and the amorphous solid

Felodipine came from AstraZeneca and indomethacin from Hawkins Pharmaceutical; PVP K29/32 from Sigma Aldrich; HPMC (Pharmacoat 606) and HPMC-AS (MF grade) from Shin-Etsu M1·M2·M3·M4·M5. Both drugs were made amorphous by heating the crystal to 10 °C above its melting point and quench-cooling, then cryo-milling (30 s at 3 impacts per second, 3 cycles) and sieving, keeping the 100–300 µm fraction. The powder was stored over desiccant in a refrigerator or freezer and checked to be amorphous by PXRD or cross-polarized microscopy before use P1. No solid dispersion was made: in every experiment the polymer was dissolved in the medium before the drug was added A4.

Dissolution with in situ UV

Felodipine was dissolved in 20 mL of pH 6.8, 50 mM phosphate buffer in a pION µDiss Profiler, which holds six fiber-optic UV probes and stirs magnetically at 500 rpm. Amorphous solid was added at 100, 500 or 1,000 µg per mL at 37 °C and at 500 µg/mL at 25 °C; crystalline felodipine at 37 °C served as reference. When polymer was used it was pre-dissolved at 500 µg/mL (0.05% w/v) P2. Indomethacin was dissolved in 50 mL of pH 2, 50 mM phosphate buffer in a jacketed flask with a single fiber-optic probe and a stir plate whose speed is not given, always at 300 µg solid per mL, with PVP or HPMC at 250 µg/mL (0.025% w/v) P3. Spectra were taken every minute, and second-derivative spectra were used for calibration and samples to suppress scattering from particles P2·P3. The paper does not state the number of replicates or what its error bars represent X8.

Following the solid: Raman, PXRD and microscopy

For Raman, an immersion probe sat in a slurry of about 500–750 mg solid in 2 mL (felodipine) or 5 mL (indomethacin) of the same medium, at 25 or 37 °C, with spectra every minute until the transformation finished or 4 h had passed P5. The felodipine PLS model used calibration mixtures of 0, 10, 25, 50, 75, 90 and 100% crystalline drug, three independent sets made by cryo-milling; one principal component described 96.4% of the variation, with Q2(cum) 0.959 and a root mean square error of estimation of 7.21 P6. For indomethacin the intensity at 1,730 cm⁻¹ was subtracted from the α band at 1,648 cm⁻¹ after unit normalization, and the corrected α intensity was plotted against time as a qualitative profile P7. PXRD was recorded on amorphous felodipine wetted with 3–4 drops of medium, without polymer or with 1 mg/mL PVP or HPMC-AS, starting 15 min after wetting P8. Polarized-light microscopy (10×, crossed polarizers with a 530 nm tint plate, room temperature) followed amorphous indomethacin on a slide exposed to pH 2 buffer, with and without polymer P9.

Note by the AI writer (not discussed in the paper): the slurries are very different from the dissolution vessels. A felodipine slurry held about 250–375 mg/mL of solid against 500 µg/mL in the dissolution test, with the same 500 µg/mL of polymer in the liquid, so there was roughly 500–750 times less polymer per unit of solid surface in the slurry X5. The slurry tells us how the solid behaves in that medium; it does not give the time scale inside the dissolution vessel.

Isolating the solution route: solvent shift

Indomethacin dissolved in less than 1 mL of methanol was added to pH 2 buffer to give about 40 µg/mL at 37 °C and about 25 µg/mL at 25 °C, levels chosen from the estimated amorphous solubility, and the concentration was followed for up to 4 h without polymer or with PVP or HPMC P4. The methods and results text give the polymer level as 250 µg/mL, but the captions of Figures 13 and 14 say 500 µg/mL; the paper does not resolve this X1. No solvent-shift test was run for felodipine; the paper relies on the earlier felodipine data of ref 25 instead R20.

The yardstick: estimated amorphous solubility

Enthalpies of fusion were measured by DSC (TA Q2000), and the solubility of crystalline α-indomethacin was measured in the pH 2 media with and without polymer; the polymers did not change it at either temperature P10·P11·R9. The felodipine crystalline solubility (1 µg/mL at 37 °C) was taken from ref 25 M1. Equations 1 and 2 then give the estimates quoted above R3·R8. Note by the AI writer (not discussed in the paper): recomputing with the printed inputs reproduces the felodipine value (9.2 µg/mL against "around 9") and the 37 °C indomethacin value (36.5 against "around 37"), but for indomethacin at 25 °C the printed inputs give 15.8–23.6 µg/mL rather than the printed 17–25 µg/mL E1·E2·X2. The paper also does not say which crystal form of indomethacin was melted for the DSC data, and it gives no felodipine solubility at 25 °C, so the 25 °C felodipine supersaturation cannot be expressed as a ratio X12.

Experimental analysis

Felodipine without polymer: the solid wins the race

At 37 °C amorphous felodipine showed no advantage over the crystal (Figure 2). With 100 µg/mL of amorphous solid, the concentration crept up along almost the same curve as the same mass of crystal and levelled near 1 µg/mL, the crystalline solubility; five or ten times more amorphous solid dissolved faster but still levelled at about 1 µg/mL R1·C6. At 25 °C the picture changed: 500 µg/mL of amorphous solid produced a peak of about 2.5 µg/mL at about 16 min, followed by a fall to below 1 µg/mL R1.

Table A. Felodipine dissolution without polymer, pH 6.8 buffer (read from Figure 2, approximate) R1.

The Raman data explain the difference (Figure 3). At 37 °C the slurried solid was already about 54% crystalline at the first reading, 1 min after contact, about 83% at 5 min and about 96% at 24 min. At 25 °C it was about 10% crystalline at 1 min and needed about 8 min to reach 50%, reaching about 89% at 29 min R2. The authors' reading is direct: at 37 °C "the amorphous solid crystallizes more rapidly than it can dissolve on contact with the dissolution medium", so what dissolves is effectively crystal, while at 25 °C "this delay in crystallization at 25°C enables some of the amorphous material to dissolve" before it converts C8·C9. They attribute the fall after the 25 °C peak to crystallization of felodipine from the solution phase C7.

Note by the AI writer (not discussed in the paper): by the time of the 25 °C peak the slurried solid was about 70% crystalline, and it kept converting while the concentration fell, so loss of the dissolving amorphous phase may also contribute to the fall. The paper did not separate the two routes for this case X6.

Felodipine with polymer: delayed solid crystallization, then a test of the solution

With 500 µg/mL of PVP, HPMC or HPMC-AS in the medium, amorphous felodipine at 37 °C did supersaturate (Figure 4) R4·C10. Each polymer changed the crystallization of the slurried solid from immediate to sigmoidal, with a lag, a transformation period and a plateau (Figure 6) R6. PXRD 15 min after wetting showed sharp felodipine peaks without polymer, no crystalline peaks with 1 mg/mL PVP, and a largely amorphous pattern with one small peak at 10° 2θ with 1 mg/mL HPMC-AS, which the authors read as crystallization "commencing" (Figure 5) R5·C12.

Table B. Felodipine with polymer pre-dissolved at 500 µg/mL (PXRD at 1 mg/mL). Slurry times read from Figure 6, concentrations read from Figure 4 (approximate) R6·R5·R4·R1.

Two points stand out in Table B. First, every polymer delayed crystallization of the solid R6. Note by the AI writer: the ranking of the delay in the slurry (PVP longest, HPMC-AS shortest) is the reverse of the ranking of the peak concentrations, so the slurry delay alone does not predict how high the solution goes R6·R4. The authors conclude that "it appears that the polymers delay crystallization of the amorphous solid for a sufficient length of time" for a supersaturated solution to form, and that PVP is "at least as effective as the other polymers at delaying crystallization of the solid material" C14·C38. Second, what happens after the peak depends on the polymer's power to stop crystallization from solution. PVP lost most of its supersaturation within an hour, whereas HPMC and HPMC-AS kept the concentration above about 7 and 10 µg/mL for 4 h R4. The authors read this as PVP being "a poor inhibitor of crystallization from a supersaturated solution" for felodipine, in "excellent agreement" with the earlier solvent-shift data of ref 25 C11·R20.

The peaks are compared with the estimated amorphous solubility of about 9 µg/mL. The authors describe the maximum with each polymer as "close to" the estimate and, in the discussion, the peaks as falling in "the range 8–12 µg/mL", in "reasonable agreement" with it C10·C37. Note by the AI writer: the PVP peak (about 7.3 µg/mL by reading, about 7.5 by the text) lies below that range, and HPMC-AS rose above the estimate, to about 12 µg/mL X4. The authors' summary for felodipine is that crystallization of the solid and crystallization from solution both limit the concentration, and that when both are substantially delayed the concentration comes close to the theoretical value C39.

Note by the AI writer (not discussed in the paper): in the slurries every polymer let the solid reach about 92–95% crystallinity within about 22–25 min, yet with HPMC-AS the solution concentration kept rising until about 100 min R6·R4. This is not a contradiction once the solid loading is considered; the dissolution test had far more polymer per unit of solid than the slurry, so the solid there may have stayed amorphous longer than the slurry suggests X5. It does mean that the slurry lag times rank the polymers but do not tell how long the solid survived in the vessel.

Indomethacin without polymer: the solution loses

Amorphous indomethacin in pH 2 buffer behaved differently (Figure 7). It supersaturated at both temperatures: about 15 µg/mL at 25 °C (peak at about 25 min) and about 11.5 µg/mL at 37 °C (peak at about 30 min), after which the concentration fell R7. The estimated amorphous solubility at 25 °C, 17–25 µg/mL, is "just above" this peak, while at 37 °C the estimate of about 37 µg/mL is "substantially higher" than anything observed C15·R8. The fall after the peak suggests, in the authors' words, "that indomethacin is crystallizing from the supersaturated solution" C16.

Table C. Amorphous indomethacin, 300 µg/mL in pH 2 buffer, no polymer. Concentrations read from Figure 7 (approximate); solid state from Figures 8, 9 and 15 R7·R11·R12·R19·R10.

At 25 °C the solid did not crystallize. Raman spectra after 120, 180 and 240 min kept the band at 753 cm⁻¹, present only in the amorphous form, and the amorphous carbonyl pattern between 1,600 and 1,800 cm⁻¹, and cross-polarized microscopy agreed (Figure 8) R11·C19. Under the polarized-light microscope, amorphous indomethacin in pH 2 buffer showed no crystals at first; after 60 min dendritic, needle-like crystals had appeared in the liquid between the particles, while the particles themselves showed no crystallization (Figure 15) R19. The authors take this as crystallization "from the solution phase" and conclude that "the limiting factor in the dissolution of amorphous indomethacin at 25°C is crystallization from solution" ("These observations indicate") C31·C40.

At 37 °C the solid did crystallize. The slurry was still largely amorphous at 20 min, the α polymorph was clearly present at 90 min, and the spectrum resembled pure α at 120 min (Figure 9) R12·C20. Tracking the α band gives a lag of about 30 min and a plateau at about 120 min, and the concentration maximum of the dissolution test falls near the onset of this conversion (Figure 10) R13. The authors propose that crystallization of the solid "would result in a reduction in the dissolution rate", which, combined with crystallization from solution, lowers the concentration ("potential correlations") C21. At 37 °C the amorphous powder also agglomerated quickly into a few large particles, which the authors note gives "an uncontrolled surface area with an unknown impact on the dissolution profile" R10·C18.

Indomethacin with polymer: dissolution and solvent shift

With 250 µg/mL PVP in the medium (Figure 11), dissolving amorphous indomethacin reached about 23.5 µg/mL at 25 °C, inside the estimated amorphous solubility range, and held it for the full 4 h; the authors conclude that "it appears that PVP is an effective inhibitor of solution crystallization at this temperature and supersaturation" R14·C22. At 37 °C the same maximum was reached, more slowly (about 80 min), but it was not maintained and fell to about 14 µg/mL by 240 min, still far below the 37 °C estimate R14·C23. With 250 µg/mL HPMC (Figure 12) the 25 °C concentration rose slowly to about 24 µg/mL at 240 min, with wide error bars; at 37 °C it reached only about 14.5 µg/mL, but HPMC "appeared to maintain the solution concentration more effectively at 37°C than PVP, albeit at a lower level of supersaturation" R15·C24. Several factors "could" explain this, among them a difference in the driving force for crystallization, which rises with supersaturation C25. Raman of 37 °C slurries with either polymer (data not shown) found no substantial delay of crystallization of the solid, with the caveat that "a more rigorous study is needed" R16·C26·C27.

Table D. Solvent-shift indomethacin in pH 2 buffer, read from Figures 13 and 14 (approximate). Polymer level 250 µg/mL in the text, 500 µg/mL in the figure captions R17·R18·X1.

The solvent-shift tests (Table D) isolate the solution route. Without polymer, a solution of about 42 µg/mL at 37 °C fell to about 7 µg/mL within 30 min, and a solution of about 27 µg/mL at 25 °C held for about 25 min and then crashed to about 5 µg/mL by 90 min R17·R18·C30. With HPMC the 37 °C solution kept its initial concentration for about 2.5 h; with PVP it dropped at once to about 33 µg/mL, stayed fairly steady until 160 min and then declined R17·C28. At 25 °C both polymers allowed only a very slow decline, to about 20 µg/mL after 4 h R18·C30. Under the microscope, needle crystals did not form at all with either polymer, the solution-mediated transformation being "completely arrested" R19·C31. The authors conclude that "it is clear" both polymers inhibit crystallization of indomethacin from supersaturated solution at both temperatures and expect them to "largely inhibit" it at the supersaturations reached by dissolving the amorphous solid C28·C29.

Note by the AI writer: with PVP the 37 °C solvent-shift solution started lower (about 33–35 µg/mL) than with HPMC (about 39) or without polymer (about 42), so the PVP and HPMC curves at 37 °C did not start from the same supersaturation R17. The 25 °C HPMC dissolution curve is also called "similar" to the PVP one, but it rose much more slowly, reaching its final level only at 4 h X10.

Discussion

The two-route framework

The discussion opens with the general statement that both the dissolving amorphous solid and the supersaturated solution are "thermodynamically metastable or unstable", and that "the kinetics of these phase changes relative to kinetics of dissolution will in large part shape the observed concentration-time profile" C32. The data of the paper sort neatly into the cases this framework allows. The flow below follows the authors' account C33·C40·C43·C44·C39.

The authors spell out the fast-conversion case: if the amorphous material crystallizes fast enough on contact, "it will not be possible to generate a solution concentration higher than that produced by the crystalline material", and felodipine at 37 °C is the example C33. If the solid converts slowly or not at all, any shortfall comes from crystallization of the supersaturated solution; this was, in their account, "the dominant mechanism" for indomethacin at 25 °C and for felodipine with pre-dissolved polymer C43·C44. Because nucleation and growth speed up with supersaturation, the more an amorphous solid supersaturates the solution, the more likely the solution is to crystallize, so a high amorphous solubility is in part self-defeating unless something inhibits nucleation and growth C43.

Note by the AI writer (not discussed in the paper): for felodipine with polymer, the route of the later loss was not measured directly. The solid in the dissolution vessels was not monitored and no felodipine solvent-shift test was run in this paper, so the assignment to the solution route rests on the earlier felodipine study and on the PVP contrast X7.

Why felodipine crystallizes faster at 37 °C

The authors explain the temperature effect through the glass transition. Dry amorphous felodipine has a $T_g$ of about 45 °C and felodipine exposed to 75% relative humidity a $T_g$ just below 35 °C (ref 28), so "the hydrated amorphous material at 37°C will be above Tg and therefore presumably subject to faster crystallization kinetics than the sample at 25°C" M1·C34. The argument rests on cited $T_g$ values for humidified powder, not on a $T_g$ measured in contact with the medium C34.

The same reasoning fails between the two drugs, and the authors say so. The $T_g$ of indomethacin also falls rapidly with water (ref 33), and indomethacin has the larger driving force for crystallization, which "would suggest that it should at least crystallize near the same rate as felodipine". Yet the data "show dramatic differences" in crystallization rate during dissolution, which the authors take as "an indication of the complexity of the processes taking place in these systems and that a single parameter like the glass transition temperature will not fully explain the observed behavior" (ref 34) R24·C42. Note by the AI writer: the recomputed free-energy differences support the driving-force statement (about 6.4 kJ/mol for indomethacin against 5.7 kJ/mol for felodipine at 37 °C from the printed inputs), but the two drugs were run in different buffers (pH 2 and pH 6.8), volumes, stirring set-ups and polymer levels, so the cross-drug comparison is qualitative E1·E2·X9.

How might dissolved polymer protect the solid?

That pre-dissolved polymer delays crystallization of the amorphous solid had been reported once before, for HPMC and 9,3″-diacetylmidecamycin (Sato et al., ref 29), and the felodipine data agree R21·C35. For a mechanism the authors offer a hedged chain. Crystallization of amorphous solids is known to start at the surface for other compounds (refs 12, 30, 31), and here it "would be exacerbated through plasticization by water"; given the low polymer concentrations, "it would seem reasonable to speculate that the polymers act to inhibit crystallization of the solid material by interacting with the surface and delaying surface crystallization" R23·C36. Work by Yu and co-workers showing that a thin polymer coating inhibits surface crystallization of amorphous indomethacin is cited as support (ref 32) R22. Polymer adsorption on the particles was not measured, and neither was the site where crystallization begins C36.

For indomethacin at 37 °C the protective effect was absent: the polymers "do not appear to be very effective at delaying the solid phase crystallization" C41. The lower supersaturation at 37 °C is "presumably" explained by delayed dissolution through agglomeration, which reduces surface area, plus crystallization of the solid; the polymers "appeared to be less effective at inhibiting solution crystallization at 37°C", which "may also have an impact" C41.

What it suggests for formulation

The practical point the authors draw is that a polymer should be judged on each route separately. PVP was at least as good as the cellulosic polymers at delaying crystallization of the felodipine solid but much worse at keeping felodipine in solution C38. Since "it is likely that both routes of crystallization will be important for many compounds", knowing their relative contribution would explain observed profiles and let the stabilizing potential of formulation polymers be assessed on each route C45. From this they propose, explicitly as speculation ("it would be resonable to speculate"), that "certain amorphous APIs might be more effectively formulated using multiple polymers: one polymer optimal for stabilizing the amorphous solid with the other being the most effective solution crystallization inhibitor" C46. No two-polymer formulation was made in this paper C46.

Note by the AI writer (not discussed in the paper): the polymers here were dissolved in the medium before the drug arrived, at fixed concentrations, rather than released from a solid dispersion. In a dispersion the polymer concentration near the dissolving surface changes with time and drug loading, so the results show what each polymer can do once dissolved, not how much of it a formulation delivers to the right place.

Limitations

Limits stated by the authors. The 37 °C indomethacin data are "complicated by a rapid agglomeration of the amorphous material", giving an uncontrolled surface area with unknown effect on the profile C18. The onset of crystallization of indomethacin with polymer at 37 °C needs "a more rigorous study" C27. The mechanism of polymer protection of the solid is framed as speculation, and the dramatic difference between the two drugs is left as a sign of complexity rather than explained C36·C42.

The AI writer's own judgement (1): what is solid. The strength of the paper is the pairing of in situ solution and solid-state measurements. For felodipine at 37 °C (over half crystalline within a minute, no supersaturation at three solid amounts) and for indomethacin at 25 °C (solid amorphous for 4 h while needles grow in the liquid and the concentration falls) the route of loss is shown directly R1·R2·R11·R19·R7. The polymer delay of felodipine crystallization is seen by two methods, Raman and PXRD, and the felodipine solution results agree with an independent earlier study R6·R5·R20.

The AI writer's own judgement (2): what to discount. The Raman slurries held hundreds of times more solid per unit of polymer than the dissolution tests, so the lag times rank the polymers but may not hold in the vessel X5. The route of loss for felodipine at 25 °C and for felodipine with polymer is inferred rather than measured X6·X7. Replicate numbers and the meaning of error bars are not given, and the Raman model error (7.21) is as large as the early % crystalline values with polymer X8. The polymer level of the solvent-shift tests is printed as both 250 and 500 µg/mL X1. The abstract's statement that polymers retarded crystallization from supersaturated solutions "of both indomethacin and felodipine for up to 4 h" holds for HPMC and HPMC-AS with felodipine and for both polymers with indomethacin at 25 °C, but not for PVP with felodipine, which fell to about 2 µg/mL by 130 min X11·R4. The 25 °C indomethacin estimate does not reproduce exactly from its printed inputs, and the nucleation term in Figure 16 differs from its classical form X2·X3.

Conclusion

The paper makes one clear distinction: an amorphous drug can lose its solubility advantage during dissolution either because the solid crystallizes on contact with the medium or because the supersaturated solution crystallizes, and the two routes can be told apart by watching the solid and the solution at the same time C47·C32. Felodipine at 37 °C is the clean example of the first route and indomethacin at 25 °C of the second C8·C40. Polymer dissolved in the medium slowed both routes for felodipine, letting it approach its estimated amorphous solubility of about 9 µg/mL, but the polymers differed in which route they protected: PVP guarded the solid, HPMC and HPMC-AS also guarded the solution C14·C37·C38. The authors conclude that "polymeric additives can potentially retard both of these crystallization routes" and speculate that a formulation could use one polymer for each C50·C46.