How humidity and small amounts of PVP or HPMCAS change the growth of felodipine crystals from the amorphous drug

Background and motivation

An amorphous drug dissolves faster and reaches higher concentrations than its crystal, which is why amorphous delivery is a generally accepted way to improve the solubility, dissolution rate and bioavailability of poorly soluble drugs. The benefit lasts only as long as the drug does not crystallize, so a polymer is usually mixed in at the molecular level as a crystallization inhibitor, giving an amorphous solid dispersion (ASD) (Leuner and Dressman 2000; Serajuddin 1999). C1 How the polymer stabilizes the drug has been attributed to three effects: lower molecular mobility, a smaller thermodynamic driving force for crystallization, and disruption of the molecular recognition events that crystallization needs (Bhugra and Pikal 2008; Marsac 2007). C2

The weak point is water. Many of the polymers used in ASDs are hygroscopic, so a dispersion of a hydrophobic drug takes up more moisture than the amorphous drug alone (Crowley and Zografi 2002; Rumondor et al. 2009). C3·R17 Absorbed water in turn generally promotes crystallization: Andronis and co-workers showed this for amorphous indomethacin, and Shamblin and Zografi for sucrose–PVP and sucrose–PVP/VA dispersions. C4 A formulation scientist therefore faces a trade: the polymer that protects the drug may also bring in the water that undoes the protection.

Crystallization is generally considered to proceed in two steps, nucleation and growth, and a polymer could act on either. The authors argue that the effect of polymers on crystallization in the presence of water cannot be understood without studying the two steps separately. C5 For felodipine, the nucleation step had been studied: Konno and Taylor (2008) measured how polymers and sorbed water change the nucleation rate of felodipine from amorphous dispersions. Crystal growth rates had not been measured. C5 The same group had also measured overall (bulk) crystallization of felodipine dispersions at different humidities, at higher polymer contents than used here. R15

This paper fills the growth gap. It uses felodipine as the model drug and two polymers chosen because they differ in hygroscopicity and chemistry: poly(vinylpyrrolidone) (PVP) and hypromellose acetate succinate (HPMCAS). C6 Films contained 0, 2.5, 5 or 10% w/w polymer and were stored at relative humidities (RH) from 13 to 93%. P2·A2·A3 The question is simple: how fast do felodipine crystals grow at each humidity, and how does each polymer change that rate?

Background concepts

Crystal growth from a thin amorphous film. In a thin film of amorphous drug, a crystal that has nucleated grows outward, usually as a spherulite: a round aggregate of crystallites radiating from a centre. The speed at which its edge advances, the change in radius per unit time, is the linear growth rate, given here in mm/min. P5 The films in this study are uncovered, so growth happens at the free surface; earlier studies found crystal growth at a surface to be much faster than growth inside the bulk (Zhu et al. 2008; Wu and Yu 2006; Wu et al. 2007). R12

Glass transition and the plasticizing effect of water. Below its glass transition temperature ($T_g$) an amorphous solid is a glass and its molecules move slowly. Water absorbed into an amorphous solid increases molecular mobility, which shows up as a lower $T_g$; this plasticization has been reported for indomethacin and its sodium salt, for glucose, trehalose and dextran, and for felodipine. C23·R19 For amorphous felodipine the paper takes $T_g$ values of 317, 316, 314, 310, 308 and 306 K after storage at 0, 13, 33, 52, 64 and 75% RH from earlier work (Marsac et al. 2008; Mao et al. 2006). M1 Figure 8 reproduces related data from the group's earlier papers: amorphous felodipine takes up only about 0.10 g water per 100 g solid at 11% RH and about 1.48 g at 94% RH, and its $T_g$ falls from about 45.7 °C dry to about 29.6 °C at 94% RH, both along smooth curves. R9·R10 Even after equilibration at 93% RH, felodipine's $T_g$ (30 °C) stays above the 22–25 °C of the experiments, so amorphous felodipine alone is below its $T_g$ in every condition tested. R10·C31

Molecular mobility from the Adam–Gibbs–Vogel equation. To turn $T_g$ into a mobility, the authors use the Adam–Gibbs–Vogel (AGV) equation for the mean molecular relaxation time $\tau$ of a glass at temperature $T$ (Eq. 1):

Here ${\tau }_0$ is a pre-exponential factor, assumed to be of the order of the lifetime of atomic vibrations (${10}^{-14}$ s); $D$ and $T_0$ are material constants; and $T_f$ is the fictive temperature, the temperature at which the frozen-in structure of the glass would be at equilibrium. For felodipine the paper uses $D$ = 9.2, $T_0$ = 253.4 K and $T_f$ = 310 K (Mao et al. 2006). E1·M1 In plain terms: the relaxation time grows exponentially as the glass is taken further below the temperature region where it was formed, and the constants set how steeply. The authors compare $\tau$ at the storage temperature with ${\tau }_{T_g}$, the relaxation time at $T_g$, and assume that water changes only $T_g$ and none of the other parameters. C25·P8

Coupling between mobility and crystallization. If crystal growth followed molecular mobility one to one, the logarithm of the growth rate would change in step with the logarithm of the relaxation rate. In practice crystallization is often less sensitive than relaxation, and the relation is written with an exponent, the coupling constant $\xi$. Figure 7 plots, on one scale,

against the reduced temperature $T/T_g$, together with log(growth rate) + 1 for the drug alone. E2·R7 In plain terms: $\xi$ = 1 means growth speeds up exactly as fast as molecules do (the AGV line); $\xi$ below 1 means growth changes less steeply than mobility. Aso and co-workers needed $\xi$ = 0.75 to relate relaxation and crystallization of amorphous nifedipine, and Zhou and co-workers saw the same trend for griseofulvin. R11

Drug–polymer interactions and amorphous–amorphous phase separation. A polymer that forms specific interactions with the drug, such as hydrogen bonds, can slow crystallization by tying up the molecules that would otherwise join a crystal. Earlier work by the group found that PVP forms stronger specific interactions with felodipine than HPMCAS does, and that HPMCAS also interacts specifically with felodipine. R14 Water can compete for those interaction sites: PVP has a very strong affinity for water. R17 At high enough humidity, sorbed water can also make a drug–polymer dispersion split into two amorphous phases, one rich in drug and one rich in polymer; this amorphous–amorphous phase separation is known for felodipine–PVP dispersions at high RH. R18 The composition of each phase then no longer depends on the starting composition; only the amounts of the two phases do. C43

Methods in detail

The study has one measurement at its core, the growth rate of existing felodipine crystals in thin films held at fixed humidity, and two supporting analyses: micrographs and Raman spectra of the crystals, and a comparison of the drug-alone growth rates with molecular mobility estimated from cited $T_g$ data. The flow below shows how the parts fit together.

Films and the nucleation step

Felodipine, alone or with PVP K29-32 or HPMCAS AS-MF at 2.5, 5 or 10% of the dry weight, was dissolved in a 1:1 (by weight) mixture of ethanol and dichloromethane. One or two drops were placed on a microscope slide and spun on a two-stage spin coater, 500 rpm for 18 s and then 1000 rpm for 30 s. M1–M5·P1·P2 The slide was then put on a hot plate at 90 °C, which removes residual solvent and lets felodipine crystals nucleate and grow to a visible size; as soon as spherulites appeared the slide was taken off. P1·C7 This step matters for the design: growth is followed from crystals that already exist, so the measured rate is a growth rate and is not mixed with the time it takes a crystal to nucleate. The solution concentration and the film thickness are not given. P1

Two ways of holding the humidity

The drug-alone films and the polymer films were stored differently. Drug-alone films sat in a controlled-humidity microscope stage at 25 °C mounted on a polarizing microscope; the RH levels of this stage are not listed in the text, but Figure 1 plots points at about 13, 33, 52, 67, 75, 80, 84, 87, 90 and 93% RH. P3 The polymer films were placed in airtight holders over saturated salt solutions at room temperature: LiCl (13% RH), MgCl₂ (33%), Mg(NO₃)₂ (52%), CoCl₂ (64%), NaCl (75%), KCl (86%) and KNO₃ (93%). P4 The paper gives the temperature range of the experiments as 22–25 °C. R10 Note by the AI writer (not discussed in the paper): the two set-ups and their RH levels are not identical, and in Figures 3 and 5 the polymer points of the CoCl₂ condition are plotted at about 67% RH although the text calls it 64%; comparisons between drug alone and polymer films near a given RH are therefore approximate. X4·X5

Measuring the growth rate

Crystals were photographed periodically without removing the films from their holders. The diameter of each crystal was read either against microruler standards or from the area of the round spherulite in ImageJ, and the growth rate was the change in radius with time. P5 Each crystal was followed over at least four time points, and each reported rate is the average of three or more crystals from different samples; Figure 1 states n = 3. P5·R1 The type of error bar is not stated. X7

Crystal form and shape

Raman microscopy with a 783 nm laser (60 s exposure, three spectra averaged, 2000–750 cm⁻¹) compared crystals of the two shapes, and FT-Raman with a 1064 nm laser gave reference spectra of amorphous and crystalline felodipine powders. P6·P7 Photomicrographs (Figures 2, 4 and 6) record the shapes of crystals grown at low and high humidity with and without polymer. R2·R4·R6

Comparing growth with molecular mobility

For the drug alone the authors computed the AGV relaxation time (Eq. 1) at each storage condition, using the cited constants and the cited $T_g$ at each RH, and plotted log(${\tau }_{T_g}/\tau$) against $T/T_g$ with $\xi$ = 1 and $\xi$ = 0.5. On the same chart they plotted the growth rate as log(growth rate) + 1, shifted so that both fit one axis. P8·R7 Note by the AI writer: the printed constants reproduce the plotted $\xi$ = 1 line when $T_g$ = 317 K is used, giving about −1.73 at $T/T_g$ = 0.91 and −0.92 at 0.95, as read from Figure 7. E1

Experimental analysis

Drug alone: a gentle rise, then a jump

Figure 1 is the central result for the drug alone. Between 13 and 80% RH the growth rate rose from about 0.030 to about 0.066 mm/min, which the authors describe as an approximately 2-fold increase and an "approximately log−linear" relationship with RH. R1·C8 Most of that rise happened below about 67% RH; between about 67 and 80% RH the rate was nearly flat (about 0.061–0.066 mm/min). R1 The error bars are wide at low RH: at 13% RH the bar spans about 0.015–0.045 mm/min. R1

Table A. Growth rate of felodipine crystals from drug-alone films on the humidity stage at 25 °C, n = 3, read from Figure 1 (approximate). "Fast form" marks the points above 84% RH, where the text reports needle-like crystals. R1·R2

Above 80% RH the behaviour changed abruptly. At the next condition, seven crystals were followed: five grew as tight spherulites at about 0.048 mm/min and two as needles at about 1.3 mm/min. At about 87, 90 and 93% RH the rate was about 2.7, 2.8 and 2.3 mm/min. R1·R2 The results section calls this a 40-fold increase at 87% RH; the abstract and conclusions give "15−40-fold". C9 The authors note that, "To our knowledge", such an abrupt increase in growth rate with storage RH had not been reported before. C10 The jump came with a change of shape, from round spherulites (Figure 2a, 75% RH) to needle-like crystals (Figure 2b, 93% RH). R2 Note by the AI writer: relative to about 0.066 mm/min at 80% RH, the plotted means give about 20-fold for the needles at about 84% RH and about 35–42-fold at 87–93% RH; the basis of the lower bound "15" is not stated. The RH of the split condition is also given three ways: 86% in the Figure 1 caption, about 84% in the plot and 87% in the text. X2·X1 At that condition most crystals (five of seven) still grew slowly, so it is better read as a transition point with both forms. X13

Raman spectra of a spherulite and of a needle-like crystal grown at 93% RH matched those of crystalline felodipine in peak positions; the relative peak heights differed, which the authors attribute to preferred orientation, and they conclude that the same crystal phase grows at both humidities ("suggesting"). R8·C34 The text names crystals grown at 75% and 90% RH for this comparison, while the Figure 9 caption names a spherulite nucleated on a slide and a needle grown at 93% RH. X6

Growth compared with mobility

Figure 7 places the drug-alone growth rates on the reduced-temperature scale. The five points from 13 to 75% RH run from $T/T_g$ about 0.942 to about 0.974, and log(growth rate) + 1 rises over them from about −0.59 to about −0.23. R7 The AGV line ($\xi$ = 1) rises much more steeply, from about −1.73 at $T/T_g$ = 0.91 through about −0.92 at 0.95 to 0 at 1.00, so the crystals grew faster below $T_g$ than the AGV mobility predicts ("The results show"). R7·C26 With $\xi$ = 0.5 the line passes through the five points, and the authors report a "good correlation" with this coupling constant. R7·C27 The four high-RH points, at $T/T_g$ about 0.977–0.983, lie at about 1.10–1.44, far above both lines. R7 Note by the AI writer: no fit statistic is given, and the five lower points of Figure 7 sit about 0.05 log units below the logarithm of the Figure 1 means plus one, while the high-RH points agree; the difference is small and does not change the reading of the figure. X9·X10

HPMCAS films: slow growth, almost independent of humidity

Figure 3 shows growth from films with 2.5, 5 and 10% HPMCAS. With 2.5% HPMCAS the rates at 13, 33 and 52% RH were about 0.034, 0.055 and 0.025 mm/min, close to the drug alone; the authors report "no significant difference" up to 52% RH and lower rates above it (about 0.011–0.022 mm/min from 64 to 93% RH). R3·C15 More polymer gave slower growth over the whole RH range: 5% HPMCAS gave about 0.0022–0.0078 mm/min and 10% HPMCAS about 0.00037–0.0024 mm/min, at least an order of magnitude below the drug alone. R3·C16 For no polymer content was there a consistent trend with RH; the authors call the rates "remarkably insensitive" to storage RH, matching earlier findings for the nucleation rate of felodipine from HPMCAS films and for the bulk crystallization of felodipine–HPMCAS dispersions (the latter at higher polymer content). C13·C36·R13·R15 At 86% and 93% RH, where the drug alone grew at about 2.3–2.8 mm/min, every HPMCAS film stayed at or below about 0.022 mm/min: HPMCAS at any level removed the high-RH jump. R3·R1·C17

The crystals themselves looked different. Instead of smooth spherulites, the HPMCAS films gave disrupted spherulites with irregular boundaries, most clearly at 10% HPMCAS (Figure 4c, d). R4 The authors note that these irregular edges enlarged the error bars, which "may hide" subtle differences in growth rate. C39

PVP films: content matters when dry, humidity matters when wet

Figure 5 shows a different pattern for PVP. At 52% RH and below, the PVP content mattered greatly: at 13% RH the growth rate was about 0.0070, 0.00067 and 0.000058 mm/min for 2.5, 5 and 10% PVP, a range the authors give as "approximately 2 orders of magnitude" between 2.5 and 10% PVP. R5·C18 At every PVP content the growth rate increased with RH, and most steeply for 10% PVP, which went from about 0.000058 mm/min at 13% RH to about 0.014 at 75% and about 0.073 mm/min at 93% RH. R5·C19 From the 64% RH condition upward the three PVP levels drew together (about 0.015, 0.0040 and 0.0095 mm/min at 64% RH), and the 2.5% and 5% films became more sensitive to RH. R5·C21 PVP slowed growth relative to the drug alone at every RH except 86%, where the 2.5% and 10% PVP films (about 0.048 and 0.046 mm/min) grew as fast as the drug-alone spherulites. C20·R5·R1

At 64% RH and below, crystals in PVP films showed disrupted spherulitic morphology, with separate spherulitic segments, most clearly at the higher polymer level (Figure 6a, c); from 75% RH upward the overall circular shape of the spherulites was better kept (Figure 6d shows a large round spherulite in 10% PVP at 93% RH). R6

Table B. Growth rate of felodipine crystals (mm/min) with and without polymer, read from Figures 1, 3 and 5 (all values approximate). Drug-alone values come from the humidity stage at 25 °C and slightly different RH; at 86% RH the drug-alone column gives the spherulites / needles measured at about 84% RH. Four markers are partly hidden behind others in the original figures (5% HPMCAS at 86%, 5% PVP at 52%, 2.5% PVP at 86% and 93% RH). R1·R3·R5

Which polymer inhibits better?

The authors summarize the comparison as: below 52% RH PVP is the better crystal growth inhibitor, above 64% RH HPMCAS is (abstract), with the discussion wording that PVP "appears to be a better crystal growth inhibitor" under low RH. C14 The data at the two ends support this at every polymer level. At 13% RH even 2.5% PVP cut growth to about 0.0070 mm/min against about 0.030 for the drug alone, which 2.5% HPMCAS (about 0.034) did not do, and 5% and 10% PVP gave lower rates than the matching HPMCAS films. C22·R1·R3·R5 At 93% RH the order is reversed at every level: 10% HPMCAS gave about 0.0024 mm/min against about 0.073 for 10% PVP. R3·R5 In contrast, the nucleation rate of felodipine at low RH did not differ between PVP and HPMCAS at a given concentration in earlier work. C41·R14

Table C. Which polymer gave the slower growth at each RH and polymer content, compared by the AI writer from the readings in Table B (error bars not considered). R3·R5·X8

Note by the AI writer: Table C shows that the crossover is not sharp. At 52% and 64% RH the order depends on the polymer content, so the switch lies somewhere between 52 and 75% RH. X8 The abstract's statement that crystals from felodipine alone "had the fastest growth rate under all conditions" also does not hold point by point: 2.5% HPMCAS plots slightly above the drug alone at 13 and 33% RH (within overlapping error bars), and PVP films match the drug alone at 86% RH. C11·X3

Discussion

The drug alone: water, mobility and surface growth

The authors' starting point is that crystal growth depends on molecular mobility, and that water raises mobility by plasticizing the amorphous drug; growth "is expected" to speed up as RH rises, and this was "indeed observed" for the drug alone up to 80% RH. C24 Figure 7 shows, however, that growth rose less steeply than the AGV mobility: below $T_g$ the crystals grew faster than the relaxation time predicts, and a coupling constant of 0.5 described the 13–75% RH data. C26·C27 The authors offer two "possible explanations". One is a decoupling between drug diffusion and viscosity, reported for several compounds. The other is that the uncovered thin films report growth at the surface, which other studies found to be much faster than growth in the bulk; if so, it "may not be reasonable" to correlate bulk mobility estimated from $T_g$ with surface growth. C28·C29·R12 They add that crystallization "likely" starts at the surface anyway, so the thin-film format is relevant in practice. C30

The jump above 80% RH remains unexplained

Neither water uptake nor $T_g$ of amorphous felodipine shows a discontinuity between 80 and 93% RH (Figure 8), and $T_g$ after 93% RH (30 °C) is still above the experiment temperature. C31·R9·R10 The abrupt change in growth rate, together with the change in crystal shape, therefore "suggest a fundamental change in the mechanism of crystal growth at high RH". C32 Raman shows the same crystal form in both shapes. C34 The reason for the jump is stated as "currently unknown" in both the discussion and the conclusions. C33

How the authors explain the polymer effects

The HPMCAS result cannot come from a simple mobility (Tg) effect, the authors argue: earlier work showed that HPMCAS at low levels does not raise the calorimetric $T_g$ of felodipine, and dispersion films take up more water than the drug alone, so they should have no $T_g$ advantage. C35·R14·R17 A similar lack of correlation between nucleation rate and $T_g$ had been found for felodipine–HPMCAS films. R13 The authors then offer three possible reasons for the slow, RH-insensitive growth with HPMCAS. First, specific felodipine–HPMCAS interactions "may" couple the motions of the drug with those of the polymer side chains, and water may barely affect this because HPMCAS carries several hydrogen-bond donors and acceptors per repeat unit that can take up water without breaking the drug–polymer contacts. C37 Second, the polymer "may" interact with the crystal surface and block growth sites; HPMCAS is known to inhibit crystallization of felodipine from dilute supersaturated aqueous solution. C38·R16 Third, the irregular crystal edges in HPMCAS films widen the error bars, which "may hide" small differences. C39

For PVP the argument runs through interaction strength and water. At low RH, PVP's better inhibition "may arise because" PVP forms stronger specific interactions with felodipine than HPMCAS does. C40 PVP also has a very strong affinity for water; if the felodipine–PVP interactions are what inhibit growth, water absorbed at higher RH "may result in disruption" of them and weaken the inhibition. C42·R17 At high RH, sorbed water further induces amorphous–amorphous phase separation in felodipine–PVP dispersions (cited). The drug-rich domains that form have a composition independent of how much PVP was added, so growth from them "would be expected" to be independent of PVP content, as observed from 64% RH; the abstract states this cause with "probably" and the conclusions with "may result from". C43·R18 The conclusions draw these strands together under an explicit hedge: "It is speculated that the strength of the drug−polymer interactions may play a role" in how effective a polymer is as a growth inhibitor. C44 The chain below lays out this proposed mechanism; none of its links was measured in this paper. X12

What it suggests for formulation

The authors' own conclusion is general: polymeric additives inhibit the crystal growth of felodipine in the presence of moisture, "with the extent of the inhibition dependent on a number of different factors". C45 Note by the AI writer (not discussed in the paper): for a formulator, the most useful result is that the ranking of two inhibitors flipped between dry and humid storage. A polymer chosen because it slows crystallization best in dry screening (here PVP) may perform worst once the product sees high humidity, while HPMCAS, weaker in dry air, kept its effect at every RH. Screening growth inhibition at the intended storage RH would catch this. The study used only 2.5–10% polymer, far below typical ASD loadings, and surface growth on uncovered films; whether the same order holds at higher polymer contents, where water uptake and phase separation behave differently, is open.

Limitations

Limits the authors state. The cause of the abrupt growth increase above 80% RH is "currently unknown". C33 Bulk mobility estimated from $T_g$ "may not" be the right yardstick for growth at the film surface. C29 The irregular crystal boundaries in HPMCAS films gave larger error bars that "may hide subtle differences" in growth rate. C39 The link between interaction strength and inhibitor performance is offered as speculation ("It is speculated"). C44

The AI writer's own judgement (1): what was and was not measured. The growth rates themselves are direct measurements, and the largest effects (the jump of the drug alone above 80% RH, and the two-decade spread between PVP levels at 13% RH) are far larger than the error bars. R1·R5 Everything that would explain the polymer effects, however, comes from earlier work: the drug–polymer interactions, the water content and $T_g$ of the films, and the amorphous–amorphous phase separation of felodipine–PVP were not measured on these films. X12 The mechanisms in the Discussion are therefore hypotheses consistent with the data, not results of this paper.

The AI writer's own judgement (2): statistics and labels. The error-bar type is not stated, n is given only for the drug alone (n = 3), and no statistical test stands behind "no significant difference" or "significant reduction"; at 52% RH the error bars of 2.5% HPMCAS and of the drug alone overlap. X7 Several RH labels disagree: the split condition is 84, 86 or 87% RH depending on where one looks, the CoCl₂ condition is 64% in the text but plotted at about 67%, and the MgCl₂ condition is 33% or 32%. X1·X5 The drug-alone films and the polymer films were stored on different humidity set-ups. X4 The abstract's "15−40-fold" has an unexplained lower bound, and at the split condition five of seven crystals still grew slowly. X2·X13 The coupling constant of 0.5 has no fit statistic and excludes the high-RH points, and the $T_g$ values used for $T/T_g$ (317 K dry) differ by about 2 K from the cited data shown in Figure 8. X9·X11 None of these changes the main reading, but they limit how precisely the RH thresholds (80%, 52%, 64%) can be quoted.

The AI writer's own judgement (3): scope. One drug, two polymers at 2.5–10% w/w, near room temperature, growth at the surface of uncovered films. Real dispersions usually contain far more polymer, and growth inside a bulk solid or a compact may respond to water differently. P2·C30

Conclusion

The paper isolates the growth step of felodipine crystallization and shows how water and two common polymers act on it. For the amorphous drug alone, growth rises slowly with RH up to 80% RH and then jumps by more than an order of magnitude with a change from spherulites to needle-like crystals, for reasons the authors could not identify. C8·C9·C33 Below $T_g$ the growth is faster than the AGV mobility predicts, and a coupling constant of 0.5 describes the 13–75% RH data. C26·C27 Both polymers slow growth, more at higher content; PVP is the stronger inhibitor in dry air but loses its effect as RH rises, while HPMCAS keeps growth slow and almost independent of RH, so that below 52% RH PVP and above 64% RH HPMCAS is the better inhibitor. C12·C13·C14 The authors tentatively attribute this to the strength of drug–polymer interactions and to water disrupting them in PVP, with moisture-induced amorphous–amorphous phase separation explaining why PVP content stops mattering at high RH; these links rest on earlier work and remain to be tested on the films themselves. C44·C42·C43·X12