How far a weakly basic drug supersaturates after it leaves the stomach: a pH-solubility phase diagram for pazopanib

Background and motivation

Many oral drugs are poorly soluble, and for compounds whose absorption is limited by solubility (class IIb of the developability classification system), formulators either raise the crystalline solubility, for example with surfactants or cyclodextrins, or generate a supersaturated solution with an amorphous solid dispersion, a prodrug, a cocrystal or a salt (refs 1–5). There is a ceiling to supersaturation. Once the dissolved concentration exceeds the amorphous solubility, the solution separates into a drug-rich and a drug-lean phase, which is called liquid–liquid phase separation (LLPS); when the drug-rich phase is a glass, that is, below the glass transition temperature (Tg) of the water-saturated drug-rich phase, the same process is called glass–liquid phase separation (GLPS) (Taylor and Zhang 2016, ref 6). C2

This ceiling matters for absorption. Earlier membrane-transport work from the same group showed that flux across an artificial membrane rises linearly with concentration up to the LLPS/GLPS concentration and does not rise significantly beyond it (refs 7–9), and supersaturated solutions "are thought to" raise the rate of passive absorption (refs 10, 11). Formulations that supersaturate quickly and then sit at the LLPS/GLPS concentration, with drug-rich colloids present and no crystallization, should therefore perform well; ref 6 calls this "spring and plateau" behaviour. C3

Weakly basic drugs reach supersaturation without any special formulation. With a pKa between 5 and 8 they are ionized and dissolve in the stomach, but in the higher pH of the small intestine they revert partly or fully to the less soluble un-ionized form. If the concentration then exceeds the amorphous solubility at that pH, LLPS or GLPS follows (Hsieh 2012, ref 12; Psachoulias 2011, ref 13; Indulkar 2015, ref 14). Small amorphous colloids can replenish drug as it is absorbed, whereas crystal nucleation and growth remove dissolved drug (ref 12). C4 A polymer that inhibits nucleation or crystal growth in solution is therefore wanted, and the hydrophobicity of the polymer relative to the drug "is considered" essential for growth inhibition (Ilevbare 2013, refs 17, 18). C5 The group had already shown that a pH-solubility diagram carrying both the crystalline and the amorphous boundary captures where a weak base is subsaturated or supersaturated (Indulkar 2015, ref 14). C6

Pazopanib is a tyrosine kinase inhibitor and a weak base whose strongest basic group has a pKa of 6.4. It is given at 800 mg once daily as the HCl salt (ref 19). M1 Like other kinase inhibitors, its intrinsic solubility is low and its bioavailability is low and variable: about 21% for pazopanib HCl (ref 30), with exposure raised up to 3.4-fold by food (ref 31) and by 46% when tablets are crushed (ref 19). R24·C7 The authors read these clinical effects as signs that food and the way the dose is presented change solubility or dissolution ("suggest"). C44 The study therefore asks how much supersaturation pazopanib HCl generates when it dissolves at gastric pH and meets intestinal pH, and whether a pH-solubility diagram with both boundaries explains what then forms. To do this the authors dissolved 1, 0.1 and 0.01 times the clinical dose in simulated gastric fluid, shifted the pH with or without the crystallization inhibitor hydroxypropylmethyl cellulose (HPMC), and followed the solution and the particles. P6

Background concepts

Crystalline and amorphous solubility. The crystalline solubility is the equilibrium concentration over the most stable solid; for a weak base it rises steeply as pH falls below the pKa. The amorphous solubility is the highest concentration of molecularly dissolved drug that can exist before an amorphous drug-rich phase separates; it is higher than the crystalline solubility because the amorphous solid has a higher free energy. Any concentration between the two is supersaturated with respect to the crystal but cannot separate into an amorphous phase; any concentration above the amorphous solubility can. C13

The Henderson–Hasselbalch relation for a weak base. For a base with one relevant ionizable group, the total solubility at a given pH is the solubility of the un-ionized form multiplied by a factor that accounts for the ionized form (Eq. 1):

Here $C_T$ is the predicted total solubility, $C_U$ the concentration of the un-ionized form, and $X$ the ratio set by the difference between pH and pKa. In plain terms: far above the pKa, $X$ is large and the total solubility equals the un-ionized solubility; each pH unit below the pKa raises it about tenfold. The authors used the solubility at pH 8.4, where pazopanib is "essentially completely unionized", as $C_U$, with a pKa of 6.4, to draw a predicted curve for both the crystalline and the amorphous form. E1·C9·P11

Supersaturation ratio. The paper expresses supersaturation as a concentration ratio (Eq. 2):

Here $C$ is the amorphous solubility and $C^{*}$ the equilibrium solubility of the crystalline free base at the same pH. In plain terms: $S_C$ is the largest supersaturation that a solution can reach before amorphous phase separation caps it, so it also measures how much an amorphous or supersaturating form could raise the driving force for passive absorption over the crystal. E2·C20

Measuring the amorphous solubility. The authors used a solvent-shift method: a concentrated DMSO solution of the drug is pumped slowly into stirred buffer, and light scattering (extinction at 400 nm, a wavelength where the drug does not absorb) is recorded. The concentration at which the extinction starts to rise marks the appearance of the drug-rich phase and is taken as the amorphous solubility, also called the GLPS concentration. P4·C8

The two-stage pH-shift test. The drug is first dissolved in simulated gastric fluid (SGF, pH 1.2) and then transferred into a more concentrated neutral buffer, so that the pH jumps as in gastric emptying. The concentration in the second ("intestinal") compartment is followed in situ with a fiber-optic UV probe. The figures label this reading "apparent concentration"; the medium can contain particles at the same time. P6

Supersaturation maintenance. After the pH shift a supersaturated solution stays near its starting concentration for some time before crystallization makes the concentration fall. The paper reports this time as "supersaturation maintenance" in Figure 8b and calls it the induction time in the text. R22

Methods in detail

The study has three parts: build the two solubility boundaries against pH, run a pH-shift test at three doses with and without HPMC, and identify what precipitates. The figure below shows how they fit together; the last step places each dose on the phase diagram (Figure 9).

Part 1: the crystalline floor and the amorphous ceiling

Problem. The extent of supersaturation possible on a pH shift depends on two curves that move with pH, and for pazopanib neither was known. C19

What they did. The crystalline free base was made by dissolving pazopanib HCl in water and precipitating the base with excess sodium hydroxide; it was crystalline by X-ray diffraction (data not shown) and was dried under vacuum for 12 h. P1 Excess free base was equilibrated at 37 °C for 72 h in buffers chosen for their buffering capacity: 0.1 M citrate from pH 5 to 6 and 0.1 M phosphate from pH 6.4 to 8.4 (Table 1). The solid was removed by ultracentrifugation at 35,000 rpm and the supernatant assayed by HPLC (C18 column, pH 2.5 phosphate buffer/acetonitrile 75/25, UV at 270 nm, detection limit about 10 ng/mL). Figure 1 also shows crystalline points near pH 1.2 and at pH 4, for which the medium is not stated. P2·P3·X10 For the amorphous solubility, a DMSO stock of the salt (20 mg/mL at pH 5, 10 mg/mL elsewhere) was pumped at 5–20 µL/min into 15 mL of degassed, stirred buffer at 37 °C containing 5 µg/mL HPMC; the final DMSO content was 1.5–7 µL/mL. At pH 8.4 the value was checked independently: stock was added to buffer containing 50 µg/mL HPMC, the precipitate was removed at once by ultracentrifugation and the supernatant assayed by HPLC. P4·P5

Why it works. The crystalline value is a true equilibrium, while the extinction onset catches the moment the solution can no longer hold more drug in one phase; together they bracket every possible dissolved state at each pH. The Henderson–Hasselbalch curves (Eq. 1) then test whether each data set behaves as expected for a base with a pKa of 6.4. E1·C10 A small amount of HPMC in the amorphous-solubility media slows crystallization so that the amorphous limit can be reached before crystals appear; the authors note that this becomes harder as pH falls. C16

Part 2: the pH-shift test at three doses

Problem. Whether a given dose ends up subsaturated, supersaturated or above the amorphous solubility after gastric emptying depends on how much dissolves in the stomach and on dilution in the intestine. The authors therefore scaled the clinical dose to the test volume and varied it tenfold twice. P6

What they did. The clinical dose (800 mg, assumed taken with 250 mL of water) scaled to 3.47 mg/mL of pazopanib HCl; the other doses were 0.35 and 0.035 mg/mL. The salt was added to SGF without pepsin containing 0, 15, 60 or 150 µg/mL HPMC, and the gastric compartment was sampled at 3, 5, 10, 15 and 20 min, filtered (0.22 µm) and assayed by HPLC. At 20 min, 7 mL was pipetted into 14 mL of concentrated pH 6.5 phosphate buffer (a three-fold dilution, so the intestinal HPMC levels were 0, 5, 20 or 50 µg/mL) and 12 M NaOH was added to bring the pH to 6.5; the final pH was 6.47–6.53. The UV probe then recorded the apparent concentration for 180 min, at 37 °C and 300 rpm. P6 In the paper's terms, complete dissolution and three-fold dilution would give 54, 5.4 and 0.5 times the amorphous solubility at pH 6.5 for the three doses. P6·R23

Card check: the paper does not show how the 54-fold value was obtained. With printed numbers only, 3.47 mg/mL of salt diluted three-fold is 1157 µg/mL, 58 times 19.8 µg/mL; the stated 54-fold follows if the salt mass is converted to free base, which needs molar masses the paper does not print (extractor note X11). X11

Why it works. The design spans all three regions of the phase diagram with one drug and one medium, and the paired runs with and without HPMC separate what thermodynamics allows from what crystallization kinetics then does. The intermediate dose was also run at 5 and 20 µg/mL HPMC to see how the duration scales with polymer concentration. A3·A4

Part 3: identifying the precipitate

What they did. Aliquots were placed on glass slides and viewed at 20× under crossed polarizers 2, 10 and 60 min after the pH shift, distinguishing birefringent crystals from non-birefringent (non-crystalline) material. P7 Immediately after the shift, the intermediate-dose solutions were sized by dynamic light scattering (backscatter at 173°, Z-average and zeta potential recorded) and by nanoparticle tracking analysis (NTA), which was run at concentrations below and above the GLPS concentration. P8·P9 The precipitate was also collected by filtration, air-dried and its glass transition measured by DSC; the instrument, heating rate and the dose or HPMC level of that sample are not stated. P10·X5

How it differs from earlier work. The earlier pH-dependent LLPS study (ref 14) built the two-boundary diagram; here the same diagram is used to predict and then test, dose by dose, what happens in a dynamic two-stage dissolution test that starts from the marketed salt form. C6·C45

Experimental analysis

The two boundaries and the 500–600-fold gap

Table A. Solubility of pazopanib at 37 °C. Amorphous values from Table 2 (UV extinction with 5 µg/mL HPMC; HPLC check with 50 µg/mL HPMC); crystalline values from the text at pH 1.2, 6.5 and 8.4 and otherwise read from Figure 1 (approximate); S_C read from Figure 2 (approximate). Replicate numbers are not stated. R1·R2·R4·X15

Both solubilities follow the ionization of a base. The crystalline free base dissolves to only 18 ng/mL at pH 8.4 and about 33 ng/mL at pH 6.5, but to 0.571 mg/mL in the gastric medium, a fall of more than four orders of magnitude between pH 1.2 and 6.5 that the authors see as a "huge potential" for supersaturation on transfer. R2·C19 The crystalline points follow the Henderson–Hasselbalch curve from pH 5 to 8.4; at pH 4 the measured value (about 7.1 µg/mL) lies a little above the curve (about 4.8 µg/mL), and the curve is not drawn below pH 4. R2.b·R3·C10 Recomputing Eq. 1 at pH 4 from the pH 8.4 value gives 4.5 µg/mL, the same as the drawn curve within plot reading. E1 The amorphous solubility rises from 9.2 ± 0.7 µg/mL at pH 8.4 to 60 ± 1 µg/mL at pH 5 and follows its own predicted curve from pH 6 upward, but at pH 5 and 5.5 it lies well below it (60 and 40 µg/mL against about 250 and 80 µg/mL on the curve). R1·R3 The authors attribute this to rapid crystallization during the measurement, before the amorphous limit is reached (results: "due to"; next paragraph: "likely"). C11 At pH 8.4 the HPLC check (11.4 ± 0.1 µg/mL with 50 µg/mL HPMC) and the extinction method (9.2 ± 0.7 µg/mL with 5 µg/mL HPMC) give similar values, which the authors call good agreement. R1.g·R1.h·C12

The gap. The ratio of the two solubilities is about 505, 603, 528 and 501 at pH 6.4, 6.5, 7.0 and 8.4, and falls to about 379, 304 and 103 at pH 6.0, 5.5 and 5.0. R4 Recomputing Eq. 2 at pH 6.5 gives 19.8/0.033 = 600, which matches the bar in Figure 2. E2 The abstract summarizes this as "approximately 600-fold" at pH 6.5, the results and discussion as "approximately 500-600" near neutral pH, and the conclusions as "approximately 500 times"; all three describe the same data at different pH values. C14·X1 It is the highest amorphous-to-crystalline ratio the authors' laboratory had measured; other compounds had given 4–80-fold. C15·R5 The authors note a high experimental uncertainty, because the crystalline solubility is so low and the amorphous solubility is hard to measure before crystallization, especially at lower pH. C16 They could not check the ratio theoretically: the usual estimate needs the melting enthalpy, the melting temperature and the water content of the amorphous drug at a water activity of 1, and pazopanib degrades as it melts. A melting point near 300 °C, with degradation, "lends support" to a high ratio, since the melting point is a key determinant of it. C17·C18·M1

What happens after the pH shift

Table B. Outcome of the pH-shift test by dose. Gastric values from the text and Figures 3, 5 and 7; intestinal values read from Figures 3, 5 and 7 (approximate); the pH shift is at 20 min of total test time; the amorphous solubility at pH 6.5 is 19.8 ± 0.5 µg/mL and the crystalline 0.033 ± 0.001 µg/mL. R6·R7·R8·R9·R11·R12·R13·R14·R15·R16·R18·R19·R20·R1.e·R2.g

The clinical dose. Pazopanib HCl dissolved rapidly but only in part in SGF, reaching a plateau of 2.1 mg/mL within the first minute, so a suspension of the salt was transferred. R6·C21 After the shift the concentration was at first similar to the amorphous solubility (19.8 ± 0.5 µg/mL), which the authors say "aligns well" with the expectation that the maximum free drug concentration equals the amorphous solubility. R8·C22 With HPMC the plateau of about 17 µg/mL lasted only a few minutes, followed by a gradual fall to about 0.2 µg/mL at the end, still well above the crystalline solubility. R8 Without HPMC the concentration fell quickly: the last reading, about 0.07 µg/mL, was taken about 13 min after the shift, and the text puts later values below the UV detection limit (about 0.05 µg/mL). R9·X4 A filtered sample assayed by HPLC gave 0.18 ± 0.03 µg/mL, similar to the probe; the paper does not say which run or time point it came from, but its value matches the end of the run with HPMC. R10·C23·X3

Under the polarized light microscope, 2 min after the shift both runs contained a few birefringent salt crystals and a large amount of non-birefringent material, which was far more abundant with HPMC (Figure 4). By 10 min the run without HPMC showed only new crystals, while the run with HPMC still held a large non-crystalline aggregate; by 60 min crystals dominated in both. R11 The authors attribute the early solids to undissolved salt and to amorphous free base formed by GLPS ("can be attributed"), and they assume that the new crystals are pazopanib free base. C24·C25 The abstract states that microscopy "confirmed" that HPMC delayed the free-base crystals. C26

Card check: no diffraction or other solid-state analysis of the precipitates is reported, so the crystal identities rest on appearance under polarized light (extractor note X6). X6

The intermediate dose. The salt dissolved completely in SGF, so all the drug was in solution at transfer. R7·C21 With HPMC the text describes a concentration "equivalent to the amorphous solubility" held for 40 min, then a gradual fall, ending at 0.34 ± 0.03 µg/mL against a crystalline solubility of 0.033 ± 0.001 µg/mL. R12 In Figure 5 the plateau sits at about 22–23 µg/mL and rises to about 30 µg/mL about 32 min after the shift before a sharp fall to about 4 µg/mL, and the plotted trace ends at about 0.24 µg/mL. R12

Card check: the plateau lies up to about 50% above the 19.8 µg/mL amorphous solubility, and the plotted end value (about 0.24 µg/mL) differs from the 0.34 ± 0.03 µg/mL in the text; the paper describes no scattering correction for the in situ UV values (extractor notes X7, X2). X7·X2

Without HPMC the concentration fell from about 20 to about 0.06 µg/mL within about 7 min of the shift. R13 Immediately after the shift the solution looked milky with a bluish tinge. R14 DLS found a single population of particles with a Z-average of 495.9 nm (polydispersity index 0.027, zeta potential +4.16 mV) without HPMC and 145.4 nm (0.140, −0.65 mV) with HPMC; NTA with HPMC found particles of about 50–650 nm, most of them 100–400 nm (Figure 6). R15·R16 The authors take these results, with the concentration profiles, to "indicate" that GLPS into a water-saturated colloidal amorphous drug-rich phase occurred immediately on the shift, as expected once the amorphous solubility is exceeded. C27 The filtered, air-dried precipitate had a Tg of about 83 °C, which they say "confirms" that the particles are glassy at 37 °C and that the process is GLPS rather than LLPS. R17·C28 The positive zeta potential is "not unexpected" for a base with pKa 6.4 in this medium, and its drop with HPMC is read as a "qualitative indicator" that HPMC adsorbs on the particle surface. C29·C30 Both sizes fall within the 100–500 nm usually reported for such drug-rich species, and the authors conclude that HPMC was "clearly instrumental" in producing smaller particles. R25·C38

Card check: the DLS size axes in Figure 6 are labelled "Size (r.nm)", so whether the reported sizes are radii or diameters is not stated (extractor note X14). X14

The low dose. The salt dissolved completely, and after the shift the intestinal concentration started at about 11 µg/mL, between the crystalline and the amorphous solubility. R7·R18 NTA found no scattering species immediately after the shift, "as expected" because GLPS cannot occur below the amorphous solubility. R20·C31 Without HPMC the concentration fell from about 13 to about 0.04 µg/mL within 10 min; with HPMC it stayed near 11 µg/mL for about 15 min and then declined slowly, to about 6.6 µg/mL 40 min after the shift and about 0.8 µg/mL after 180 min. R18·R19

How much HPMC is needed

Table C. Effect of HPMC at the intermediate dose (5.4×). Maintenance from the text (0 µg/mL) and read from Figure 8b (approximate; criterion, n and error-bar type not stated); trace values read from Figures 5 and 8a (approximate). R12·R13·R21·R22·X9

Without HPMC supersaturation lasted 5 ± 2 min; even 5 µg/mL HPMC produced a "notable" improvement, and the highest concentration, 50 µg/mL, prolonged it about 8-fold (Figure 8). R22·C32 All runs with HPMC ended between about 0.16 and 0.24 µg/mL, well above the crystalline solubility. R21·R12 The authors conclude that HPMC "clearly" extends supersaturation, that it is an effective crystallization inhibitor both with and without amorphous particles present (the intermediate and the low dose), and that it also slows desupersaturation. C39 They add that, with HPMC, the desupersaturation rate increased with dose. C40

Card check: no rates were calculated; the ranking holds for when supersaturation is lost (about 12 min after the shift at the high dose, about 35 min at the intermediate dose, a slow decline at the low dose), while the steepest single drop occurs at the intermediate dose (extractor note X8). X8

Discussion

Reading the outcomes from the phase diagram

The discussion turns the two curves into a map with three regions (Figure 9) and places each dose on it by its theoretical intestinal concentration at pH 6.5. R23·C45 The chain of reasoning is shown below; the split at the bottom is the role of the crystallization inhibitor.

Region 1. Below the crystalline solubility the solution is subsaturated, no crystallization can occur, and passive absorption "would be expected" to rise linearly with dose. Given the 800 mg dose and the low free-base solubility, the authors do "not expect" pazopanib to be in this regime in practice, even allowing for the extra solubilizing capacity of intestinal fluids. C33

Region 2. Between the curves a supersaturated solution forms and crystallization is the only possible phase change; high passive absorption is "anticipated" while supersaturation lasts. The low dose explores this region, and the authors write that "it is clear" its duration of supersaturation depends highly on whether an inhibitor is present. C34·C35·R18·R19

Region 3. Above the amorphous solubility, unless crystallization comes first, the drug separates into an amorphous phase and the solution concentration stays at the amorphous solubility until enough crystallization has depleted the supersaturation. The high and intermediate doses fall here, and the authors write that Figures 3 and 5 "clearly demonstrate" that, with an inhibitor present, the pH increase gives a concentration equal to the amorphous solubility, with polarized light, DLS and NTA showing amorphous precipitate. C36·R8·R12·R11·R15·R16 Maximum passive diffusion is "expected" while the concentration is held there, because amorphous particles are "expected" to replenish absorbed drug, whereas crystallization reduces the dissolved concentration and so absorption. C35·C43

From the test tube to the intestine

The authors call the in vitro test a closed, non-sink, low-volume compartment; in vivo, absorption removes dissolved drug and "potentially" reduces precipitation. C41 They argue, however, that once a dose exceeds the amorphous solubility after the pH change it phase-separates at once, and absorption "is unlikely" to be fast enough to prevent this in vivo; whether the boundary is crossed depends on the dose, the stomach pH and volume, the transfer rate and the intestinal pH, volume and solubilizing capacity. C42 Intestinal solubilizers would raise both boundaries, and slower transfer or larger dilution in vivo mean the in vitro test "most likely represents the worst case scenario"; another salt form may change the gastric solubility, the supersaturation generated and the precipitation. C37 For pazopanib, variation in phase behaviour driven by formulation and by physiology "could impact" the extent of absorption, which the authors connect with the clinical food and crushing effects. C43·C44

What it suggests for formulation

The authors' own conclusion is that such in vitro evaluations, read against a pH-solubility diagram, "are likely to be helpful" in optimizing formulations of poorly soluble weak bases and in understanding where absorption variability comes from. C46

Note by the AI writer (not discussed in the paper): three practical points follow from the data. First, for pazopanib the free concentration at pH 6.5 cannot rise above about 20 µg/mL however much salt dissolves in the stomach; a larger dose adds colloids, not free drug, so the formulation task is to hold that level rather than to raise gastric dissolution. Second, the polymer level matters at very low concentrations: 5 µg/mL HPMC already nearly doubled the maintenance time and 50 µg/mL raised it about 8-fold in this test. Third, a 500–600-fold window between amorphous and crystalline solubility is unusually wide, but without an inhibitor it closed within minutes, so crystallization inhibition rather than solubilization is the limiting step for this drug. R1.e·R22·R4

Limitations

Limitations stated by the authors. The amorphous-to-crystalline ratio carries a high experimental uncertainty, because the crystalline solubility is very low and the amorphous solubility is hard to measure before crystallization, especially as pH decreases; it could not be checked theoretically because pazopanib degrades on melting. C16·C17 Below pH 6 the measured amorphous solubility is lower than predicted, which the authors attribute to crystallization ("likely"). C11 The in vitro test is closed and non-sink, and both boundaries will shift upward in intestinal fluids; slower transfer and larger dilution in vivo mean the test "most likely" represents the worst case. C37·C41

The AI writer's own judgement (1): what the data show directly. The two solubility curves are measured directly, two methods agree on the amorphous solubility at pH 8.4, and the supersaturation ratios reproduce from the tabulated values. R1.g·R1.h·E2 The central dynamic result, that the two doses above the amorphous solubility level off near it with HPMC while the dose below it forms no colloids, also rests on direct measurements. R8·R12·R20 But each pH-shift condition is shown as one trace, and replicate numbers are not stated for the solubilities, the traces, DLS or NTA, so small differences between traces should not be over-read. X15 The in situ UV values are apparent concentrations recorded in media containing colloids; at the intermediate dose the plateau is up to about 50% above the amorphous solubility, and no scattering correction is described. X7 Some values are stated differently in different places: the ratio as 600 or 500, and the intermediate-dose end value as 0.34 ± 0.03 µg/mL in the text against about 0.24 µg/mL on the plot. X1·X2

The AI writer's own judgement (2): what is inferred. The identity of the crystals that appear after the shift is assumed (free base) and the salt crystals are identified by appearance; no diffraction of the precipitates is reported. C25·X6 The glassy state at 37 °C, which defines GLPS, is inferred from the Tg of filtered, air-dried precipitate; GLPS concerns the Tg of the water-saturated drug-rich phase, which water would lower, and that value was not measured. C28·X5 The supersaturation maintenance times have no stated end-point criterion. X9 Both boundaries were measured in simple buffers with 5 µg/mL HPMC, not in the pH-shift medium (a concentrated phosphate buffer mixed with SGF), and below pH 6 the amorphous boundary drawn in Figure 9 follows the model curve rather than the measured points. X10·X12 Finally, no flux or absorption was measured: the statements about absorption in each region, and about the reservoir role of the colloids, are expectations drawn from earlier work. C35·C42·C43

Conclusion

The paper makes three contributions. It measures the pH-dependent crystalline and amorphous solubility of pazopanib at 37 °C and finds that near neutral pH the amorphous form can reach about 500–600 times the crystalline solubility, a ratio the authors had not seen before. R1·R2·R4·C14·C15 It shows, in a two-stage pH-shift test starting from the marketed HCl salt, that doses whose intestinal concentration would exceed the amorphous solubility stop near it and form submicron glassy drug-rich colloids, while a dose below it forms no colloids and can only crystallize. R8·R12·R15·R17·R20·C27·C31 And it shows that HPMC is what keeps the supersaturation: without it every dose crystallized within minutes, and 5–50 µg/mL lengthened supersaturation up to about 8-fold and gave smaller colloids. R22·C32·C38·C39 In the authors' words, the phase behaviour after the pH shift "could be readily explained" by the pH-solubility phase diagram together with the kinetics of crystallization. C45

The approach is simple: two solubility curves and a dose placed on them predict which phase changes are possible, and a short pH-shift test with and without a polymer shows how long the useful state lasts. For pazopanib it was tested with one polymer, three doses and simple buffers, and the link to absorption remains to be measured.