How the HPMCAS grade and drug ionization decide the release of weakly basic drugs from amorphous solid dispersions

Background and motivation

An amorphous solid dispersion (ASD) disperses a poorly soluble drug in a polymer so that the drug can reach the higher apparent solubility of its amorphous form while the polymer protects it from crystallizing. The polymer also sets the pace of release. Because the polymer is far more soluble than the drug, drug release is fastest when it is tied to the polymer, so that drug and polymer leave the dissolving surface together; this is called congruent release (Section 1; Indulkar 2019, ref 15; Saboo 2019, ref 27). C1 It follows that anything that slows the polymer will slow the drug.

Hydroxypropyl methylcellulose acetate succinate (HPMCAS) is one of the most used ASD polymers, especially for spray-dried products. It is an enteric polymer: its succinoyl substituents carry carboxylic acid groups (COOH) that must ionize before it dissolves, so it is insoluble at gastric pH and dissolves in near-neutral media. The authors use the critical ionization fraction model for this behaviour: the polymer starts to dissolve only after a critical fraction of its acid groups is ionized, and its rate then rises with further ionization, because ionization to carboxylate brings hydration, plasticization and dilution of the chains (Hiew 2022, ref 45; Tsiartas 1997, ref 46). C2 Any component of the ASD that hinders that hydration, for example a hydrophobic drug that binds strongly to the carboxylate, "can impact" the solubility and dissolution rate of the polymer. C3

The group had already seen this in an earlier study with one grade, HPMCAS-MF, and a range of weak acid, neutral and weak base drugs (Bapat 2024, ref 3). Weak bases that were cationic at the dissolution pH released poorly from the MF dispersions, and this was attributed to hydrophobic ion pairs formed by electrostatic attraction between the protonated drug and the polymer carboxylates, which reduce polymer hydration; under such mutual ionization, release fell and insoluble drug–polymer complexes formed. C5 HPMCAS, however, is sold in grades, LF, MF and HF, that differ in the amounts and ratio of the ionizable, hydrophilic succinoyl groups and the hydrophobic acetyl groups: LF has the most succinoyl groups and HF the most acetyl groups (Ueda 2020, ref 13; refs 40, 41, 49, 50, 55–58). C6 Earlier comparisons of the grades looked mainly at crystallization inhibition and at the drug-rich nanodroplets formed on dissolution; how the grade changes release kinetics, and in particular how ionic interactions with a lipophilic cationic drug play out across grades, has been "little explored". C4

The authors therefore set a clear hypothesis: at equal drug loading, a lipophilic cation will impair release from HF dispersions more than from MF dispersions, and LF dispersions will perform best. C7 To test it they chose two lipophilic weak bases that differ in how ionized they are near neutral pH, loratadine (partially cationic at pH 6.8) and cinnarizine (fully cationic), made 25% drug-loaded dispersions with all three grades, measured release at two pH values to change the drug's ionization, titrated the acid content of the grades, and measured in solution how much polymer the drug pulls out as an insoluble complex. C8·A1·A2·A3

Background concepts

Weak base ionization. A weak base is protonated, and so positively charged, at pH values below its pKa and mostly neutral above it. Loratadine (LOR) has a pKa of 5.2 and cinnarizine (CNZ) a pKa of 8.4 (refs 80, 79); both are very lipophilic, with log P 5.2 and 5.8. M1·M2 The authors calculate from the pKa values that LOR is about 14% ionized at pH 6.0, whereas CNZ is printed as more than 99.9% ionized. R13 Note by the AI writer (not discussed in the paper): the usual Henderson–Hasselbalch calculation, which the paper does not print, reproduces the LOR value (13.7%) but gives about 99.6% for CNZ at pH 6.0, not more than 99.9%; the conclusion that CNZ is essentially fully ionized is unaffected. X3 The same kind of calculation gives the authors' 89% ionization for HPMCAS at pH 5.8 with a pKa of 4.9. R13

Acid groups and pKa of a polymer. The number of COOH groups per gram of HPMCAS can be measured by potentiometric titration: the polymer is dissolved with excess sodium hydroxide and back-titrated with hydrochloric acid. The first end point marks neutralization of the excess base, the second the precipitation of all of the polymer; the acid volume between them is proportional to the COOH groups. The pH at half-neutralization is taken as the apparent pKa. P8·C11

Normalized release rate and congruent release. Release was measured from a compact pressed into a die so that only one flat face of known area touches the medium (a rotating-disc, or Wood's, apparatus). The release rate of each component is normalized to the area and to the component's weight fraction in the compact (Equation 1):

In plain terms: $k$ is the slope of the solution concentration of the component against time, $V$ is the medium volume (100 mL), $S$ the exposed die area (0.5 cm²) and $x$ the weight fraction of that component in the compact. Dividing by $x$ puts drug and polymer on the same scale, so equal normalized rates mean that drug and polymer leave the surface in the proportion in which they are mixed, which is congruent release. For a neat polymer, $x=1$ and $R$ is its intrinsic dissolution rate (IDR). E1·P6

Surface pH. The pH at the dissolving surface of an acidic polymer differs from the bulk pH, and in the authors' argument it is the surface pH that sets how ionized drug and polymer are where they meet. C28·C31 The authors use two measures. The slurry pH of each neat grade in water, taken as "the lowest possible surface pH" of the polymer during dissolution, and a value of pH 6.0 at the dissolving ASD surface in 50 mM pH 6.8 phosphate buffer, taken from their earlier media study (Bapat 2023, ref 8). C10·R15

Insoluble complexation. When a protonated lipophilic base meets ionized carboxylates in water, the two can form an ion pair that is too hydrophobic to stay dissolved, and drug and polymer fall out together. The authors quantify this by adding drug and polymer to buffer and measuring how much of the polymer is no longer in solution after centrifugation (Equation 2):

In plain terms: the share of added polymer that has left the solution once the drug is present. Because the same amount of polymer alone dissolves completely, the authors read the lost polymer as polymer bound in an insoluble drug–polymer complex. E2·C9 The amount of drug is expressed as a molar ratio of drug to polymer COOH groups (drug/COOH); at 1:1 there is one drug molecule per acid group. Because LF has twice the acid groups of HF per gram, the same drug/COOH ratio corresponds to a higher drug loading for LF (for CNZ at 1:1, 36% for LF and 22% for HF, by the paper's Table 4). R9.d·R9.j

Methods in detail

The study has three experimental parts, run on the same two drugs and three grades, plus a calculation of ionization from pKa values. The MF grade was partly carried over from earlier work: its release rates and IDR come from the group's media study (Bapat 2023, ref 8), its complexation data at pH 6.0 from the HPMCAS-MF study (Bapat 2024, ref 3) and its acid content from Hiew 2022 (ref 45). X7

Contribution 1: Release of drug and polymer from dispersions of the three grades

Problem. To see whether the grade changes the damage done by a cationic drug, drug and polymer release must be measured separately and on a common scale, under conditions where the drug's ionization differs.

What they did. LOR and CNZ were each dispersed at 25% drug loading (DL; 25:75 w/w drug:polymer) in HPMCAS-LF, -MF and -HF by dissolving drug and polymer in methanol:dichloromethane 1:2 v/v and removing the solvent by rotary evaporation at a 50 °C water bath. The dispersions were kept under vacuum overnight, cryomilled and sieved to 106–250 µm, and were amorphous by powder X-ray diffraction and polarized light microscopy. P5·R19 For each test, 100 mg of ASD (or neat polymer) was compressed at 1500 psi for 1 min in an 8 mm die, giving one exposed face of 0.5 cm², which rotated at 100 rpm in 100 mL of 50 mM phosphate buffer at 37 °C. Samples were taken at 5, 10, 15, 20, 25, 30, 40, 50 and 60 min with replacement by fresh medium, and the medium pH stayed within 0.1 unit of target. P6·R7 The drugs were assayed by reversed-phase HPLC with UV detection (LOR at 210 nm, CNZ at 251 nm) and HPMCAS by HPLC with an evaporative light scattering detector (ELSD), calibrated from 1 to 250 µg/mL. P2·P3 LOR dispersions and neat polymers were tested at pH 6.8 and 7.5, CNZ dispersions at pH 6.8 only; all six dispersions were also tested in pH 1.6 hydrochloric acid medium, for which Table S3 reports the drug only. Each test was run in triplicate. P6·A3 At the end of each test the solution and the compact surface were inspected under polarized light for crystals, and the observation was extended to about 12 h. P7

Why it works. The rotating disc fixes the area and the hydrodynamics, so differences in rate reflect the material. Measuring the polymer as well as the drug shows directly whether the drug follows the polymer (congruent) or not, and whether a slow drug rate comes with a slow polymer rate.

Contribution 2: Acid content, pKa and surface pH of the grades

What they did. HPMCAS-LF and -HF were dried at 105 °C for 30 min and under vacuum overnight, dispersed in ultrapure water and dissolved with 260 µL of 0.5 M sodium hydroxide over 24 h, then back-titrated with 0.1 N hydrochloric acid in small increments while the pH was logged. End points were found by fitting at least five points around each to a polynomial and taking the zero crossing of the second derivative; the COOH content follows from the acid volume between the two end points, and the pKa from the half-neutralization point. Each grade was titrated twice. P8·C11 The polymer mass and the end-point volumes are not printed. For the slurry pH, excess polymer was stirred in water overnight and the final pH recorded; it was stable within 15 min. P10·R18

Contribution 3: A solution test of insoluble complex formation

Problem. What happens at the surface of a dissolving compact cannot easily be sampled. The authors therefore measure, in a simpler system, how readily each drug takes each grade out of solution.

What they did. Three stock solutions were made in tetrahydrofuran (THF): drug alone, polymer alone (52.5 mg/mL) and drug plus polymer (polymer again 52.5 mg/mL). A small aliquot (10–15 µL of THF per mL) was added to stirred phosphate buffer to give 0.5 mg/mL polymer, well below its solubility of 4–8 mg/mL in pH 6.0 phosphate buffer (ref 43), and a drug concentration above its amorphous solubility. P9 In the controls, the drug alone precipitated and gave a pellet of drug, and the polymer alone dissolved completely and gave no pellet; drug plus polymer turned turbid and gave a pellet that contained polymer. R17 After centrifugation at 14800 rpm, the polymer left in the supernatant was measured with the phenol–sulfuric acid colorimetric method at 490 nm, and the lost fraction calculated with Equation 2. P4·E2 CNZ was tested at drug/COOH ratios from 0.25:1 to 2:1 (and 3:1 in Figure 6) at pH 6.0 with LF and HF, and at pH 5.8 with LF and MF; LOR was tested at 1:1 between pH 5.8 and 7.5 with all three grades. Each point is a triplicate. P9·A4·A5

Why pH 6.0. In their earlier media study the authors found that the dissolving ASD surface sits near pH 6.0 when the bulk is 50 mM phosphate at pH 6.8, so pH 6.0 stands for the surface conditions of the release test. C28·R15 pH 5.8 was added because there the polymer is somewhat less ionized (89% at pKa 4.9) while still soluble, and CNZ is still fully ionized. R13

Note by the AI writer (not discussed in the paper): the methods section describes the complexation test for the LF and HF grades in pH 6.0 buffer, yet Figure 7 covers pH 5.8–7.5 and Figure 8 includes MF at pH 5.8 without a stated source; the SI lists buffer recipes only for pH 5.8, 6.8 and 7.5, so the media for pH 6.0, 6.2, 6.3 and 6.5 are not described. X5

Experimental analysis

Neat polymers dissolve in the order LF > MF > HF

Without any drug, the three grades already differ. In pH 6.8 buffer the IDR was about 1.50 mg·min⁻¹·cm⁻² for LF, 0.90 for MF and 0.45 for HF, and at pH 7.5 about 1.73, 1.09 and 0.63 (read from Figure 2, approximate; MF from the earlier study). R1 Each grade dissolved faster at pH 7.5 than at 6.8, and LF was about 3-fold faster than HF at both pH. C12 This is the baseline the drug acts on: the grade with the most acid groups and the fewest acetyl groups dissolves fastest.

Loratadine dispersions: the grade matters greatly at pH 6.8, much less at pH 7.5

Table A collects the drug and polymer release rates of the 25% DL dispersions. For LOR at pH 6.8, the drug was released at about 0.25 mg·min⁻¹·cm⁻² from LF, 0.10 from MF and 0.01 from HF, so LF released more than 10 times faster than HF. R2·C14 At pH 7.5 the rates were about 0.73, 0.62 and 0.44, a ratio between LF and HF of less than 2. R2·C14 Moving from pH 6.8 to 7.5 raised the HF dispersion's rate by more than an order of magnitude, the MF dispersion's by a factor of about 5 and the LF dispersion's by about 3. C14 In every case the polymer left at a similar normalized rate to the drug, which the authors read as release controlled by the polymer. C13

Table A. Normalized release rates of drug and polymer from 25% DL dispersions, n = 3 (read from Figures 3 and 4, approximate; MF values cited by the authors from Bapat 2023). R2·R3

Note by the AI writer (not discussed in the paper): "approximately the same" holds well for the faster systems, where drug and polymer rates differ by about 10% or less, but in the two slowest systems the plotted drug rate is about half the polymer rate (LOR–HF at pH 6.8, about 0.01 against 0.02; CNZ–MF, about 0.014 against 0.025). Release there is closer to polymer-first than to strictly congruent. X6

Cinnarizine dispersions: LF about nine-fold faster than MF, and HF does not release

CNZ, fully ionized at the test pH, showed the same order more sharply. At pH 6.8 the LF dispersion released drug at about 0.13 mg·min⁻¹·cm⁻² and polymer at about 0.14, against about 0.014 and 0.025 for MF (Figure 4); the authors give a factor of about 9 between the two grades. R3·C15 The release rate of the HF dispersion was "too slow to be determined", and after 1 h no CNZ was found in solution (0.0 ± 0.0 µg/mL). R3·R5.i At equal grade, CNZ also released more slowly than LOR: from LF, about 0.13 against 0.25 mg·min⁻¹·cm⁻². R2.c·R3.b Together with the LOR results, this confirms the hypothesized order, HF worst and LF best, for both drugs at pH 6.8, with the caveat that the MF values come from earlier batches. C24·X7

How much drug reached solution, and no crystallization

After 1 h of release, LOR reached 17.6, 7.0 and 2.4 µg/mL from LF, MF and HF at pH 6.8 and 59.1, 39.9 and 29.2 µg/mL at pH 7.5; CNZ reached 5.3, 2.7 and 0.0 µg/mL at pH 6.8 (Table 2). R5 Table 2 also lists cited crystalline and amorphous solubilities: for LOR 1.8 and 9.0 µg/mL at pH 6.8 and 1.4 and 7.6 µg/mL at pH 7.5, for CNZ 2.1 and 12.0 µg/mL at pH 6.8. M1·M2

No crystals were seen under polarized light, in solution or on the compact surface, during any 1 h test; no crystallization was seen for up to 12 h after release from the MF and HF dispersions of either drug, and small crystals appeared in solution only after about 10 h for the LOR–LF dispersion. R6 The authors conclude that the slow release of some systems "cannot be attributed to crystallization". C17 In pH 1.6 hydrochloric acid, where HPMCAS does not dissolve, 2.1–5.5% of the drug was released in 1 h from all six dispersions, without crystallization (Table S3). R4

Note by the AI writer (not discussed in the paper): three details of these data need care. First, the text says acid-medium release was "less than 5%" for every dispersion, but Table S3 gives 5.5 ± 0.07% for CNZ–LF. X2 Second, several LOR concentrations exceed the cited amorphous solubility, up to about 8-fold at pH 7.5 (59.1 against 7.6 µg/mL); the paper does not say whether samples were filtered or centrifuged before analysis, so these values may include drug-rich colloids, and the paper does not discuss them. X9 Third, the 1-h concentrations can be compared with the plotted rates through Equation 1: assuming linear accumulation from time zero and ignoring sample replacement, the expected concentration is about 75 times the drug rate in mg·min⁻¹·cm⁻². This matches five of the six LOR cases within about 20%, but not the CNZ dispersions (LF: about 9.8 expected against 5.3 µg/mL measured; MF: about 1.1 against 2.7), so for CNZ the plotted rates and the tabulated concentrations cannot both be read at face value. X8

LF carries twice the acid groups of HF, at nearly the same pKa

The titration curves differ in shape (Figure 5): LF shows a flat buffering region near pH 4.9 followed by a sharp drop between about 1300 and 1600 µL of 0.1 N acid, whereas HF falls gradually without a clear plateau. R8 From the end points, LF carries 1.52 ± 0.03 mmol/g COOH and HF 0.76 ± 0.02 mmol/g, exactly twice as many acid groups per gram for LF; the MF value of 1.19 ± 0.01 mmol/g is taken from Hiew 2022 (Table 3). M3·M4·M5 The apparent pKa values are close: 4.91 ± 0.01 (LF), 4.97 ± 0.02 (MF) and 5.06 ± 0.02 (HF). M3·M4·M5 The slurry pH of the neat grades in water is 4.52 ± 0.04 (LF), 4.61 ± 0.02 (MF) and 4.73 ± 0.03 (HF) (Table 5), which the authors read as "reflecting their different solubilities"; the source of the MF value is not stated. M3·M4·M5·C20·X4

HF loses more polymer to the insoluble complex than LF

Table B shows how much polymer CNZ took out of pH 6.0 buffer. For HF the loss was already 70 ± 2% at a drug/COOH ratio of 0.25:1 and reached 98–99% from 0.75:1 upward. For LF it rose gradually, from 30 ± 5% at 0.25:1 to 70 ± 6% at 1:1 and 95 ± 3% at 2:1. MF, cited from the earlier study, lay between the two, and all three grades reached about 100% at 3:1 (Figure 6). R9·R10 The authors' summary: at 1:1, "nearly 100%" of HF forms an insoluble complex with CNZ while "only ∼70%" of LF is precipitated, "suggesting" that LF is more soluble in the presence of the drug at a given stoichiometry. C27·C18

Table B. Percent HPMCAS complexed with CNZ in pH 6.0 phosphate buffer, mean ± SD, n = 3 (Table 4; MF read from Figure 6, approximate, cited by the authors from Bapat 2024). The equivalent drug loading is the paper's own conversion. R9·R10·X1

Note by the AI writer (not discussed in the paper): the printed equivalent drug loading of 88% for LF at 2:1 does not fit the rest of Table 4. From the LF 1:1 row (36%), the same conversion gives about 46% at 1.5:1, as printed, and about 53% at 2:1; every LF row should also equal the HF row at twice the ratio, which holds for the printed values except this one. The 88% is most likely a misprint. X1

Complex formation follows drug ionization

Two observations tie complexation to the charge of the drug. The first compares the two drugs with LF at 1:1 and pH 6.0: LOR, about 14% ionized, took out 50 ± 6% of the polymer, whereas CNZ, essentially fully ionized, took out 70 ± 6%. R9.m·R9.d·R13 With HF the difference almost disappears (95 ± 3% for LOR, 99 ± 3% for CNZ). R9.n·R9.j

The second follows LOR at 1:1 across pH (Table C, Figure 7 and Figure S1). As the pH rose from 6.0 to 6.8, the polymer lost to the complex fell from about 95% to about 40% for HF, from about 50% to about 2% for LF, and (cited) from about 70% to about 5% for MF; at pH 7.5 it was about 15%, 0% and 2%. R11 Over the same range the calculated LOR ionization fell from about 14% to about 2.5%, and to about 0.5% at pH 7.5. R13 The authors describe complex formation as falling off "steeply" with pH as LOR ionization is reduced, for all three grades, and present this as illustrating the importance of ionization. C32

Table C. Percent HPMCAS complexed with LOR at a 1:1 LOR/COOH ratio, n = 3, with the authors' calculated LOR ionization from pKa 5.2 (pH 6.0 values from Table 4; other values read from Figures 7 and S1, approximate; MF cited from Bapat 2024). R9.m·R9.n·R11·R13

The CNZ experiment at pH 5.8 approaches ionization from the polymer side (Figure 8). CNZ stays fully ionized at both pH values, but lowering the pH from 6.0 to 5.8 raised complexation at every ratio: for LF from 30–70% to about 40–81%, and for MF from about 41–90% to about 50–100% between 0.25:1 and 1:1. R12·R9.a–d·R10 The authors explain it ("This is because") by the polymer becoming less soluble as its ionization falls, to about 89% at pH 5.8, while the ratio of drug cations to polymer anions rises slightly. C19·R13 So the extent of complexation depends on the charge state of both partners, not of the drug alone.

Putting release next to complexation

Note by the AI writer (not discussed in the paper): the paper never plots release against complexation, but its numbers can be set side by side. Ordered by the polymer lost at 1:1 and pH 6.0, the systems run LOR–LF (50%; drug release about 0.25 mg·min⁻¹·cm⁻² at pH 6.8), LOR–MF (about 70%; about 0.10), CNZ–LF (70%; about 0.13), CNZ–MF (about 90%; about 0.014), LOR–HF (95%; about 0.01) and CNZ–HF (99%; too slow to determine). R2·R3·R9·R10·R11 More complexation goes with slower release, with one near-tie in the middle (LOR–MF and CNZ–LF). Two points limit this comparison. Complexation was measured at a fixed 1:1 ratio, whereas a 25% DL dispersion has a different drug/COOH ratio in each grade; by the paper's own Table 4 conversion, 25% DL of CNZ lies between 0.5:1 and 0.75:1 for LF but between 1:1 and 1.5:1 for HF, so in the HF dispersion the drug can outnumber the acid groups. And complexation was measured from THF stocks in buffer, not at the dissolving surface. R9·X10

Discussion

The authors' explanation, step by step

The authors connect the findings in the chain below. Each step is their reading; the boxes in the two branches summarize the mechanism they propose, which the paper does not measure directly. C2·C28·C34·C25·C26

For the difference between the grades the authors borrow an argument from Heller and co-workers, who studied partial esters of vinyl acetate–maleic anhydride copolymers that dissolve by ionization of their acid groups (ref 78). There, the ionization needed to dissolve the polymer rose with the size of the hydrophobic ester group, from 3% of the carboxyls for an ethyl ester to 91% for a C9H19 ester, while the pKa stayed the same. R14 HPMCAS grades likewise have nearly equal pKa values, and "A similar explanation can be applied": LF, with relatively more hydrophilic succinoyl than hydrophobic acetyl groups, should need less ionization to dissolve and so has a lower dissolution pH threshold. C25 In addition, at the same drug loading LF offers twice as many carboxylates per gram as HF, leaving "spare" carboxylates, not paired with the drug, to keep the polymer hydrated; the authors call the lower impact of the drug on LF release "fairly straightforward to rationalize". C26

The last link, from complexation to release, goes through the interface. Polymer solubility at the interface sets how fast polymer diffuses through the aqueous boundary layer into the bulk, and since the drug leaves with the polymer, a complex that lowers interfacial polymer solubility lowers drug release as well. C29 The authors test this link on the two drugs with LF. Both drugs are lipophilic (log P 5.8 for CNZ, 5.2 for LOR), so lipophilicity does not separate them; but LOR complexes less with LF than CNZ does (about 50% against 70% at 1:1), and the LOR dispersion releases faster. The difference in release "can be most likely attributed" to that lower complexation, and the lower complexation in turn "at least partially" to LOR being only about 14% ionized at pH 6.0. C30·C31 The pH series of Figure 7 and the pH 7.5 release data point the same way: as LOR loses its charge, complexation collapses and the grade gap in release narrows. C32·C14

Where this sits in the wider literature

Ionic interactions between weakly basic drugs and acidic polymers are usually discussed for their benefits in the solid state. Weuts and co-workers found that poly(acrylic acid) salts of loperamide gave a higher glass transition temperature, better stability against crystallization on storage and rapid release, and similar interactions have been reported for lumefantrine, ciprofloxacin, clofazimine, gefitinib, lapatinib and ketoconazole (refs 48, 69–75). C21 The release results are less consistent: lumefantrine and clozapine released better as salts with poly(acrylic acid) (ref 76), whereas lumefantrine released more slowly from enteric polymers as drug loading rose, and quickly from a neutral polymer (Hiew 2022, ref 45). R16 The authors suggest that this conflict "may be related" to the hydrophilicity of the polymer: poly(acrylic acid) dissolves even when un-ionized, whereas an enteric polymer dissolves only above its critical ionization, so drug molecules that block the hydration of a critical number of ionizable groups are "likely to be detrimental" to release. C22 The present paper adds that, within one enteric polymer family, the grade decides how large that penalty is.

What it suggests for formulation

The authors themselves draw two practical messages. With lipophilic weak bases at a pH where drug and polymer are both ionized, LF gave better release than MF or HF. C24 And the stability benefit that makes drug–polymer salts attractive should "Ideally" be balanced against the reduced release seen under mutual ionization. C23 They call the results "a baseline" and ask for studies with the solubilizing components of biorelevant intestinal fluids, which might change complex formation. C35·C36

Note by the AI writer (not discussed in the paper): three further readings follow from the data. First, at a fixed drug loading, the drug/COOH ratio in HF is about twice that in LF, so the choice of grade changes how much of the polymer's acid capacity a cationic drug occupies; for a 25% DL CNZ dispersion the paper's own conversion places HF above 1:1 and LF below 0.75:1. R9·M3·M5 Second, the solution complexation test is quick and ranked the grades and the two drugs in the same order as release, which makes it a candidate screen before release testing, although it was checked here on two drugs only. R9·R11·R2·R3 Third, for a base with a low pKa such as LOR the penalty shrinks at higher intestinal pH, whereas a high-pKa base such as CNZ stays ionized across that range; CNZ was not tested at pH 7.5. R13·A3

Limitations

The authors present the study as a baseline under simple conditions: release was measured in phosphate buffers, and "Additional studies are needed" with solubilizing components such as those of simulated intestinal fluids. C36 They also describe release from acidic polymer–basic drug dispersions as "convoluted". C35

The AI writer's own judgement. The rankings are solid: release of drug and polymer was measured separately in triplicate, the neat polymers rank in the same order as the dispersions, and complexation ranks the grades the same way at every CNZ ratio and across the LOR pH series, with drug-only and polymer-only controls. R1·R2·R3·R9·R11·R17 The causal reading is weaker, for five reasons.

• The MF release, IDR and pH 6.0 complexation data, and the MF acid content, come from earlier studies and batches; the direct comparison within this paper is between LF and HF. X7
• Complexation was measured from THF stocks in buffer at 0.5 mg/mL polymer and a fixed drug/COOH ratio, not at the surface of a dissolving compact, and the surface pH of 6.0 is taken from earlier work; the link from complexation to release rests on two drugs and is a correlation. X10
• The grade mechanism, a lower critical ionization extent for LF and "spare" carboxylates that hydrate, was not measured: substituent contents of the lots, critical ionization extents and hydration were not determined, and the supporting analogy comes from another polymer family. X11
• Raising the pH from 6.8 to 7.5 also speeds up the neat polymers, so the narrowing of the grade gap for LOR mixes the effect of less drug ionization with that of more polymer ionization. R1·R2
• Only one drug loading (25%) was studied, CNZ only at pH 6.8, and several printed values are inconsistent: the 88% drug loading in Table 4, the 5.5% acid-medium release against "less than 5%", the CNZ ionization of more than 99.9%, 1-h CNZ concentrations that do not match the plotted rates, and LOR concentrations above the cited amorphous solubility without comment. A1·X1·X2·X3·X8·X9

Conclusion

For two lipophilic weak bases dispersed at 25% in HPMCAS, the grade changed release by more than an order of magnitude at pH 6.8: LF released drug and polymer fastest, HF slowest, and the cinnarizine–HF dispersion did not release measurably. R2·R3 In solution, HF lost far more polymer than LF to an insoluble complex with the drug at the same drug/COOH ratio, and the less ionized drug, or the same drug at a higher pH, complexed less. R9·R11 The authors conclude that electrostatic complex formation between the cationic drug and the polymer carboxylates compromised release, more for HF than for LF, and that the extent depends on drug ionization and so on pH. C34 Their explanation of the grade effect, more acid groups and lower hydrophobicity for LF, is plausible and fits the titration data, but it was not measured directly. C25·C26·X11