Which polymer keeps dissolving felodipine supersaturated: PVP, HPMC and HPMCAS compared

Background and motivation

An amorphous drug dissolves to a higher concentration than its crystal, which is why amorphous solid dispersions (ASDs) are used for poorly water-soluble drugs (refs 1–5). The amorphous form is, however, unstable and may crystallize, both on storage and during dissolution, and the advantage is then lost C1. Mixing the drug with a carrier polymer at the molecular level often stabilises the solid; the stabilising effect has been attributed to antiplasticization, drug–polymer interactions, coupling of drug and polymer mobility and a higher activation energy for nucleation (refs 8–15) C2.

The authors argue that the solid state is only half the problem: once the dispersion dissolves, how long the supersaturation it creates survives is "also of critical importance" C3. Hancock and Parks (ref 3) had shown that the solubility predicted for amorphous drugs is often much higher than what is measured, a gap that "may be attributed, at least in part" to rapid crystallization of a less soluble form from the supersaturated solution C4. Dispersions often give concentrations far above crystalline solubility, sometimes higher than the neat amorphous drug gives (Gupta 2004, ref 16), and two explanations were in use: the polymer inhibits crystallization in solution (refs 16, 17), or it raises the equilibrium solubility of the drug by complexation (refs 18–20) C5.

Felodipine had been studied by the same group in the solid state. Without moisture, its nucleation rate in dispersions did not depend on whether the polymer was PVP, HPMC or HPMCAS (Konno 2006, ref 21); with atmospheric moisture the three differed, inhibition correlating with the amounts of polymer and of water, and HPMCAS and HPMC beat PVP at high relative humidity (Konno 2008, ref 22) R18. This paper asks the next question: how do the same three polymers, chosen because they are widely used in solid dispersions, change what happens when the dispersions dissolve C6? To answer it, the authors set the dissolution profiles of the dispersions against concentration–time profiles of supersaturated felodipine solutions made in media that already contain the same polymers, and against crystalline solubility in polymer solutions P3·P4·P5.

Background concepts

Supersaturation. A solution is supersaturated when it holds more drug than the equilibrium solubility of the stable crystal. The paper uses the supersaturation ratio (Eq. 4):

Here $c$ is the solution concentration and $c^{*}$ the equilibrium solubility E4. In plain terms: $S=1$ is a saturated solution; $S$ above 1 means there is a thermodynamic driving force for the excess drug to crystallize. For crystalline felodipine in the pH 6.8 test fluid at 37 °C, $c^{*}$ was 0.94 μg/mL R1.a.

How much an amorphous form can gain. The ratio between the solubilities of the amorphous and the crystalline form follows from the free energy difference between them (Eq. 1):

Here $\Delta G_c$ is the free energy difference between amorphous and crystalline solid, $R$ the gas constant, $T$ the temperature, $a$ the activity and $S$ the solubility (in this equation $S$ is a solubility, not the supersaturation of Eq. 4) E1. In plain terms: the more the amorphous form lies above the crystal in free energy, the higher its solubility relative to the crystal, and the ratio grows exponentially with $\Delta G_c$. The Hoffman equation estimates $\Delta G_c$ from melting data (Eq. 2):

Here $\Delta H_f$ is the heat of fusion and $T_m$ the melting temperature of the crystal E2. In plain terms: a compound that melts high and with a large heat of fusion has a large free energy gap at body temperature, and so a large expected amorphous advantage ("It is clear" from Eq. 2, in the authors' words) C19. The Hoffman equation had earlier been found to estimate the free energy difference of felodipine well (Marsac 2006, ref 15) R20. Using it with the thermodynamic data of ref 15, the authors expect amorphous felodipine to reach about 10 times the crystalline solubility at 37 °C, that is about 9.5 μg/mL R17. The melting data themselves are not printed in this paper; 10 × 0.94 μg/mL gives 9.4 μg/mL, printed as "about 9.5" E1·E2.

Nucleation. New crystals form by nucleation, whose rate in classical nucleation theory is (Eq. 3, as printed):

Here $J$ is the nucleation rate, $A$ a pre-exponential factor (not defined in the paper), $\gamma$ the interfacial tension between crystal and solution, $v$ the molecular volume, $k$ the Boltzmann constant, $T$ the temperature and $S$ the supersaturation of Eq. 4 E3. In plain terms: the nucleation rate rises extremely steeply with supersaturation, so a strongly supersaturated solution nucleates fast and a mildly supersaturated one hardly at all. The paper prints the denominator with $\ln S$; the usual classical form has $(\ln S{)}^2$ X8. The paper uses the equation only qualitatively, so no result depends on the exponent X8. From it the authors expect systems with a high amorphous-to-crystal solubility ratio to "have a greater tendency to nucleate" when the amorphous form dissolves C20.

Labile and metastable zones. Above a critical supersaturation, nuclei form spontaneously and quickly (the labile zone). Below it, in the metastable zone, existing crystals can still grow but new ones form only slowly. The authors use this picture to explain why their supersaturated solutions first drop and then level off C27.

Crystal growth. Once crystals exist, the drug leaves solution by growing onto them. The overall mass growth rate is (Eq. 5):

Here $R_G$ is the overall mass growth rate, $A_c$ the effective crystal surface area, $M_c$ the mass incorporated into the crystals and $t$ time E5. In plain terms: growth rate is the mass gained by the crystals per unit time and per unit surface. Because the drug that grows onto crystals is the drug that leaves solution, the slope of the falling concentration gives $d{M}_c/dt$; if all systems are assumed to have similar crystal surface areas, the slopes rank the growth rates C28.

Solubilisation by complexation. A dissolved polymer can bind drug molecules in solution and so raise the apparent equilibrium solubility. If that happened here, the high concentrations from the dispersions would not be supersaturation at all, and the driving force for crystallization would be smaller (ref 18) C5·C22. The paper tests this directly by measuring crystalline solubility in polymer solutions P4.

Methods in detail

The study has three parts that are designed to be read together: dissolution of the dispersions themselves, a solubility test that checks whether the polymers solubilise felodipine, and solvent-shift solutions that show how each polymer affects a supersaturated solution without any solid dispersion present. Two calculations, the amorphous solubility estimate and the growth-rate slopes, connect them.

Part 1: dissolution of the dispersions

Problem. To compare polymers, the dispersions must differ only in polymer type and amount. How. Felodipine was dissolved in ethanol, the polymer added, and the solvent removed under vacuum for 8 h; the solid was then held at 160 °C for a few seconds "to ensure" it was fully amorphous, milled and passed through a 30 mesh sieve P1. The polymers were PVP K30 (BASF), HPMC (Pharmacoat 606, Shin-Etsu) and HPMCAS (AQOAT AS-MF, Shin-Etsu), each at 10, 20, 50, 60 and 75% w/w of the dispersion M2·M3·M4·P1. All dispersions had a similar mean particle size, 150–200 μm (n = 50), and DSC at 10 °C/min showed no melting peak of felodipine R2·R3·P2.

Dissolution followed the paddle method of the Japanese Pharmacopoeia: powder containing 30 mg felodipine in 600 mL of a pH 6.8 test fluid (the USP 26 simulated intestinal fluid without pancreatin), 37 °C, 50 rpm, triplicate runs over 360 min, with UV readings through a 70-μm filter on the sampling line P3·M7. From each profile the authors took the highest concentration in the 6 h (Figure 3) and an "initial dissolution rate" from the concentration after 5 min (Figure 4) P3·E6. The medium is named in two ways: simulated intestinal fluid in the methods and "JP 2nd fluid" in the Table 1 caption; no recipe is printed X9.

Note by the AI writer (not discussed in the paper): 30 mg in 600 mL is 50 μg/mL if all drug dissolved, about 50 times the crystalline solubility, so these are strongly non-sink conditions X15. Even the best result, 14.1 μg/mL, corresponds to about 28% of the dose in solution X15.

Part 2: does the polymer raise crystalline solubility?

Problem. If a polymer complexed felodipine and raised its solubility, a dispersion with that polymer would show higher concentrations without any real supersaturation. How. 30 mg crystalline felodipine was stirred in 600 mL test fluid containing 600 mg polymer (1 mg/mL) or no polymer, at 37 °C and 50 rpm for 24 h, in triplicate P4. The authors note that 1 mg/mL is far more polymer than is ever present in the other experiments (50–150 μg/mL), so a null result here covers them C17.

Part 3: supersaturated solutions without a solid dispersion

Problem. Dissolution of a dispersion mixes two things: how fast drug and polymer leave the particle, and what the polymer then does in solution. To isolate the second, the authors made supersaturated solutions directly. How. Felodipine dissolved in a small volume of ethanol or methanol was added to 600 mL test fluid in which 90 or 30 mg polymer had been dissolved (150 or 50 μg/mL), or to test fluid without polymer P5. These polymer levels are what complete dissolution of a dispersion with 75% or 50% polymer would give C6. The nominal starting felodipine concentration was 50, 15 or 9.5 μg/mL; the solutions were held at 37 °C and stirred at 100 rpm, and concentration was followed by UV for up to 4 h in triplicate P5. Two UV set-ups were used: one with a 70-μm filter and one that removed scattering from crystallizing particles by second-derivative processing; the paper does not say which run used which P5·X6.

The growth-rate estimate. For the 9.5 μg/mL runs (Figure 6a), the authors took $d{M}_c/dt$ from the initial linear part of each concentration–time curve and, assuming similar effective crystal surface areas in all four systems, used Eq. 5 to rank the growth rates P6·C28. The printed buffer slope can be checked from the plot: about 7.8 μg/mL at 5 min and about 4.3 μg/mL at 30 min give about 0.14 μg/mL-min, consistent with the printed ~0.1399 E5·R14.d.

Experimental analysis

Dissolution: HPMCAS highest, PVP lowest, and the gap grows with polymer content

Most dispersions produced concentrations above the crystalline solubility; the exceptions were the dispersions with 10–50% PVP, which stayed at about crystalline solubility R8·C7. Figure 1 gives the full profiles for each polymer and Figure 2 replots them by polymer at 10, 50 and 75% polymer R4·R5·R6·R7. Table A collects the highest concentrations from Figure 3.

Table A. Highest felodipine concentration (μg/mL) during 0–360 min of dissolution in pH 6.8 test fluid at 37 °C; read from Figure 3 (approximate), except the 75% values, which are given in the text; Figure 3 reads about 5.0 for 75% HPMC. Crystalline solubility 0.94 μg/mL R10·R7·R1.a.

Three patterns stand out. First, at 50–75% polymer the order is always HPMCAS > HPMC > PVP, and the authors summarise that HPMCAS "resulted in the highest supersaturation" while PVP was "only slightly effective" C11·R10. Second, the maximum rises with polymer content for each polymer, steeply for HPMCAS and little for PVP C8·R10. The rise is not step by step: PVP barely changes from 10 to 20%, and HPMC not at all from 50 to 60% X14. Third, at 10% polymer the authors see "little difference" between polymers, with maxima close to crystalline solubility; the plots, however, show both cellulosic polymers at about 1.7 μg/mL against about 0.8 μg/mL for PVP, about twice as high C10·X1.

The 75% HPMCAS profile rises to about 13.4 μg/mL at 180 min and 14.1 μg/mL at 300 min, then falls to about 11.1 μg/mL at 360 min, with a wide spread between replicates at the last point R5.e. The 75% PVP dispersion, by contrast, is at about 2.06 μg/mL after 60 min and stays near 2.1–2.3 μg/mL R4.e.

How fast the dispersions dissolve. The initial dissolution rate, from the 5-min concentration, was about 0.272 μg/mL min for 75% PVP, about 0.115 for 75% HPMC and about 0.092 for 75% HPMCAS; every dispersion with 10–60% polymer was below about 0.05 μg/mL min (Figure 4) R11. The rate for 75% PVP matches its 5-min concentration of about 1.36 μg/mL divided by 5 min E6·R4. The authors note that the rate fell as polymer content fell C12. They also state that PVP dispersions reached their maximum within 60 min while cellulosic ones peaked at 180–240 min C9. In Figure 1 this holds for 75% PVP only; the other PVP dispersions peak at 180–240 min, and several cellulosic profiles, including 75% HPMCAS and all HPMC dispersions, peak at 300–360 min R9·X2. A fast start, then, did not mean a high level: the fastest-dissolving dispersion, 75% PVP, reached the lowest plateau of the three 75% dispersions R11·R7.

Solubility: the polymers do not solubilise felodipine

Table B. Crystalline felodipine solubility after 24 h at 37 °C in pH 6.8 test fluid, mean with standard deviation in parentheses, n = 3; from Table 1 R1.

The four values lie within one standard deviation of each other R1. The authors conclude that the observed concentrations "cannot be accounted for" by the polymers enhancing solubility, so the complexation explanation "can be discounted" C17. "Thus it can be inferred", they write, that supersaturated solutions are generated by the dispersions and maintained C22.

Supersaturated solutions: cellulosic polymers hold, PVP lets go

After adding felodipine to a nominal 50 μg/mL, the first reading, at 5 min, was about 6 μg/mL with all three polymers and about 2.7 μg/mL without polymer: most of the added drug had already left solution R12·R13. What happened next depended on the polymer (Figure 5, Table C).

Table C. Felodipine concentration (μg/mL) at 5 min → at the end of the run (240 min; about 210 min for the 9.5 μg/mL run), all values read from Figures 5a, 5b, 6a and 6b (approximate), n = 3. The no-polymer curves of Figures 5a and 5b are the same data R12·R13·R14·R15.

With HPMC or HPMCAS the concentration measured at 5 min "decreased only marginally" over 4 h C13. At the lower polymer level (50 μg/mL) HPMC "appeared to be slightly less effective" than HPMCAS, but at 150 μg/mL no difference was apparent C14. With PVP the concentration fell steadily throughout, and the authors describe it as "eventually reaching a value close to the equilibrium solubility" C15. At 240 min the PVP solutions still read about 1.9 and 1.35 μg/mL, against the crystalline 0.94, so "toward" fits the plots better than "reaching" X10. These results are, in the authors' words, "qualitatively similar" to earlier results for nifedipine (Tanno 2004, ref 17) C15·R19.

From lower starting levels (Figure 6, 150 μg/mL polymer) the cellulosic polymers kept far more of the added drug C16. From 15 μg/mL they held about 11 μg/mL for the whole 4 h R15. From 9.5 μg/mL the authors give average concentrations of 7.8–8.9 μg/mL with HPMC and 8.5–9.1 μg/mL with HPMCAS; this run ended at about 210 min, not 4 h R14·X16. PVP again declined, to about 3.5 and 2.8 μg/mL R15·R14. Without polymer, the text states that "significant precipitation had occurred within 5 min" C16. That holds from 15 μg/mL (about 3.1 μg/mL at 5 min), whereas from 9.5 μg/mL the reading at 5 min was still about 7.8 μg/mL and the fall took about an hour X11·R14.d·R15.d. The authors use this 9.5 μg/mL run without polymer, started at the estimated amorphous solubility, as evidence that crystallization "commences immediately" and that "the dissolution of amorphous felodipine will be accompanied by rapid crystallization" ("These data suggest") C21.

Note by the AI writer (not discussed in the paper): the no-polymer solutions level off at about 1.05–1.1 μg/mL (Figure 5), about 2.0–2.5 μg/mL (Figure 6a) and about 1.8 μg/mL (Figure 6b), all above the measured crystalline solubility of 0.94 μg/mL; even without polymer, none of the runs returned fully to equilibrium within the experiment X10.

Growth-rate estimates

Table D. Initial slopes printed in the text (p.498) for the curves of Figure 6a, with the 5-min readings of the same curves (read from Figure 6a, approximate) R16·R14.

Growth in buffer or with PVP is, by these slopes, far faster than with HPMC or HPMCAS R16. With the cellulose polymers the slope is "close to zero", which, "it seems", means that crystal growth is "greatly inhibited" C29. In the conclusions this becomes a statement that at equivalent supersaturations all three polymers reduced crystal growth rates relative to the drug alone, HPMCAS and HPMC most and PVP much less (abstract: "appeared to reduce") C36. The slopes are relative, rest on the equal-surface-area assumption and come from one panel; and, as Table D shows, the four curves do not start from the same concentration at 5 min, so "equivalent supersaturation" holds only approximately X7.

Comparing the solution test with dissolution

For PVP and HPMCAS the two tests agree: PVP is inferior in both, "a reasonable correlation" C31. HPMC breaks the pattern. In solution it held felodipine as well as HPMCAS (about 10.85 against 10.45 μg/mL after 4 h from 15 μg/mL), but from dispersions with 75% polymer it reached 4.9 against 14.1 μg/mL R15·R7. The authors call HPMC "less predictable", note that Tanno et al. saw the same for nifedipine (ref 17), and state that "at present the reasons behind these differences are not clear" C31·C32·R19.

The abstract says that HPMCAS maintained the highest supersaturation for the greatest length of time "for both the dissolution and solution crystallization experiments" C35. The solution data do not show an HPMCAS lead at 150 μg/mL polymer (Figures 5a and 6b), and the discussion itself calls HPMC comparable to HPMCAS there; HPMCAS is ahead only with 50 μg/mL polymer and from the 9.5 μg/mL start X3·C14·R14.

Finally, the authors compare with the solid state. Dry, the three polymers did not differ in stabilising felodipine dispersions (ref 21); with absorbed moisture, HPMCAS and HPMC were alike and better than PVP (ref 22); in dissolution, HPMCAS generated more supersaturation than HPMC R18·C33. They conclude that "different crystallization inhibition mechanisms are of importance" and "may vary with the amount of water" ("It thus appears") C34.

Discussion

The authors' chain of reasoning

The discussion (Section 5) builds one argument from the pieces above. Amorphous felodipine should dissolve to about 9.5 μg/mL, and because nucleation rises steeply with supersaturation, its dissolution alone would be expected to end in rapid crystallization R17·C20·C21. Yet dispersions, above all with HPMCAS, gave concentrations above crystalline solubility that lasted for hours C22. Since the polymers do not raise crystalline solubility, these must be supersaturated solutions that the polymers maintain C17·C22. The solvent-shift results then show how they are maintained:

Each step carries the authors' hedge. The immediate drop is "rationalized" as nucleation that is faster at higher concentration C24. The polymers "do not appear to have prevented" it C25. That concentrations then stay above equilibrium "strongly suggests" that the polymers inhibit crystal growth, as polymers had been seen to do before (refs 29–33) C26. After the drop, "it appears" the solution has moved from the labile into the metastable zone, so only growth can lower the concentration further, and if growth is inhibited the concentration "will presumably remain elevated" C27. The abstract closes the loop: all polymers "appeared to reduce" the growth rate at an equivalent supersaturation, and "this mechanism most likely contributes" to the high concentrations reached when the dispersions dissolve C36·C37.

What the argument explains and what it leaves open

The chain explains the solvent-shift data well: in every run the cellulosic polymers kept the level reached after the first drop, and PVP did not R12·R13·R14·R15. It also explains the ranking of PVP against HPMCAS in dissolution C31. It does not explain HPMC, which inhibits growth in solution about as well as HPMCAS yet gives far less supersaturation from dispersions (4.9 against 14.1 μg/mL at 75% polymer) R7·C31. The authors leave this open ("not clear") and point instead to the solid-state comparison, where HPMCAS and HPMC also behaved alike under moisture, concluding that different inhibition mechanisms matter in different settings and "may vary" with water C32·C33·C34.

Note by the AI writer (not discussed in the paper): the dissolution test and the solvent-shift test differ in more than the presence of a solid. In a dissolution vessel the polymer and drug are released together from each particle, so the local drug concentration and the local polymer concentration near the dissolving surface are very different from the uniform solution of the solvent-shift test; how fast each polymer dissolves and releases the drug is part of the result. The PVP dispersions, for example, released drug fastest in the first minutes but levelled off lowest R11·R7. A solution test that ranks crystal-growth inhibition therefore cannot capture the release side of the problem, which is one plausible reason for the HPMC mismatch; the paper does not test this.

Implications for formulation

The authors' recommendation is to select the polymer for a dispersion "after considering both" its solid-state stability and the stability of the supersaturated solution it creates on dissolution C38. They also suggest that solution tests "could be used to screen for the best polymers for a particular drug", with the HPMC result as a caution C30·C31.

Note by the AI writer (not discussed in the paper): for felodipine in this set, only HPMCAS at 60–75% polymer reached several times the crystalline solubility; at 10–20% polymer all dispersions stayed within about 2.5 times of it, so low polymer contents bought little in solution R10. The fastest-dissolving dispersion (75% PVP) gave the lowest plateau of the 75% dispersions, a reminder that dissolution rate and maintained supersaturation are separate properties R11·R7.

Limitations

Limits the authors state. The reasons why HPMC behaves differently in the two tests are "not clear" at present C32. The growth-rate comparison holds "if it is assumed" that the four systems have similar effective crystal surface areas C28. The mechanism claims carry the hedges quoted above ("do not appear", "strongly suggests", "it appears", "presumably", "most likely contributes") C25·C26·C27·C37.

The AI writer's own judgement (1): what precipitated was never examined. Every statement about nucleation and crystal growth rests on UV concentration–time curves; no precipitate was isolated, imaged or analysed by diffraction, and no particles were counted X4. The drop in concentration is consistent with crystallization, but the paper does not show which solid formed or how many particles there were, so "nucleation was not prevented, growth was" remains an interpretation of the curve shapes X4. The growth-rate ranking adds an assumption (equal crystal surface areas) and starts from different 5-min levels in a single experiment X7.

The AI writer's own judgement (2): concentrations above the amorphous estimate. The paper's own estimate of the amorphous solubility is about 9.5 μg/mL R17. The 75% HPMCAS dispersion reached 14.1 μg/mL, and the cellulosic polymers held about 11 μg/mL after the 15 μg/mL addition, both above that estimate; the paper does not discuss this R7·R15·X5. The UV readings were taken through a 70-μm filter or corrected by second-derivative processing, neither of which separates submicron particles from dissolved drug, so part of the measured drug may not have been molecularly dissolved; alternatively the estimate itself may be low X5. Either way, the values above about 9.5 μg/mL should be read as total UV-detected drug X5.

The AI writer's own judgement (3): text versus plots, and test conditions. Several statements in the text are stronger than the figures: the "little difference" between polymers at 10% polymer, PVP maxima "within 60 min", PVP solutions "reaching" crystalline solubility, buffer precipitation "within 5 min" in both Figure 6 runs, and the abstract's HPMCAS lead in the solution tests X1·X2·X10·X11·X3. None of these changes the main ranking. The dissolution test was stirred at 50 rpm and the solution tests at 100 rpm, and the medium is named in two ways X6·X9. There was no dissolution of neat amorphous felodipine, so the comparison between dispersions and the amorphous drug alone is inferred, not measured X12. Equation 3 is printed without the square on $\ln S$, which does not affect any result X8. One drug, one dose and one medium were studied.

Conclusion

For felodipine dispersions dissolving in pH 6.8 test fluid at 37 °C, HPMCAS produced the highest and most persistent supersaturation, HPMC an intermediate one and PVP the least, and the differences grew with polymer content C35·R10. The polymers did not change crystalline solubility, so the higher concentrations are supersaturation, not solubilisation C17·R1. In supersaturated solutions made without a dispersion, HPMC and HPMCAS held felodipine near its post-drop level for hours while PVP and buffer let it decline; the authors read this as inhibited crystal growth rather than inhibited nucleation, and propose that this "most likely contributes" to what the dispersions achieve C23·C36·C37. The solution test predicted the dissolution ranking for PVP and HPMCAS but not for HPMC, and the authors advise choosing a polymer on both solid-state and solution behaviour C31·C38.