Lumefantrine amorphous solid dispersions made by spraying into cold acidified water: which acidic polymer survives the water, and how much drug it releases

Background and motivation

Lumefantrine is an antimalarial used in combination with artemether. It is a basic drug with an aqueous solubility below 80 ng/mL, and its oral bioavailability of 4–11% leads to high doses, higher treatment cost, poorer compliance and reported treatment failures (Patel et al., ref 32; refs 33, 34) C12·M1. Raising its solubility is therefore a direct way to improve therapy, and turning the drug into an amorphous solid dispersion (ASD) is one of the most widely used routes: the amorphous form dissolves faster and can generate a supersaturated solution, which raises the flux across the intestinal membrane (Frank et al., ref 3) C7. The price is physical instability; the amorphous drug can revert to the crystal, which is why it is dispersed in a polymer that slows crystallization through antiplasticization, drug–polymer interactions, lower molecular mobility or a higher nucleation barrier (refs 6–13) C8.

The industrial routes to ASDs are spray drying and hot-melt extrusion. The authors argue that both are poorly suited to the setting in which malaria is treated. Spray drying uses large volumes of organic solvent, leaves residual solvent and gives low-density, electrostatically charged powder; extrusion needs components that tolerate high temperatures; both need expensive equipment and skilled staff (refs 18–20) C9. Anti-solvent co-precipitation is an older alternative. A drug–polymer solution, usually in a non-volatile solvent, is added to a cold aqueous anti-solvent, and the precipitate is washed and dried; the method has been used both at bench scale and industrially (refs 21–29, including the microprecipitated bulk powder process of Shah et al., ref 29) C10. It needs less organic solvent than spray drying and no inert atmosphere, and the authors see a strong potential for use in developing countries C10.

The weak point of anti-solvent precipitation is water. The precipitate sits in an aqueous medium far longer than in any other ASD process, and absorbed moisture is a known accelerator of crystallization of amorphous drugs (Rumondor et al., refs 30, 31); the residual moisture must therefore be removed quickly C11. For lumefantrine there was a further reason to expect polymer-specific behaviour. Acidic polymers protonate its tertiary amine, which X-ray photoelectron spectroscopy (XPS) sees as a second nitrogen peak (Song et al., ref 38; Trasi et al., ref 39) C21. In lumefantrine ASDs made by plain solvent evaporation, without any water, the degree of protonation correlated strongly with physical stability, and ASDs with EL100 or CAP were stable for more than 30 days at accelerated conditions (Trasi et al., ref 39) R20·C49.

The study therefore had three aims: to make lumefantrine ASDs by spray anti-solvent precipitation, to characterize their physicochemical properties, drug release and solid-state stability, and to find the best polymer and drug:polymer ratio for this process A1·A2. All four polymers tested carry carboxylic acid groups: HPMCP and CAP through phthalyl substituents, HPMCAS through succinoyl substituents and EL100 through methacrylic acid units (Figure 1) M2·M3·M4·M5.

Background concepts

Anti-solvent precipitation of an ASD. The drug and the polymer are dissolved together in a solvent that dissolves both (here dimethylacetamide, DMA), and this solution is introduced into a large excess of a liquid in which the drug–polymer system is insoluble but which mixes with the solvent (here ice-cold 0.02 N hydrochloric acid); the precipitate is then collected and dried P1·C10. The authors note that the lumefantrine ASDs are insoluble in 0.02 N HCl C14. In this paper the solution was sprayed through an ultrasonic nozzle into the stirred anti-solvent, hence the name spray anti-solvent precipitation P1. Note by the AI writer (not discussed in the paper): the usual picture is that the solvent diffuses away into the anti-solvent so fast that drug and polymer lose solubility together and precipitate as one amorphous phase; how much water the wet precipitate then holds, and for how long, is the variable this paper ends up discussing.

Protonation of a basic drug by an acidic polymer, seen by XPS. When the carboxylic acid of a polymer transfers its proton to the tertiary amine of lumefantrine, the drug becomes a cation and the polymer an anion: an ionic, acid–base interaction rather than a simple hydrogen bond. XPS measures the binding energy of core electrons in roughly the top 10 nm of a solid P6. A neutral amine nitrogen gives an N1s peak at about 399 eV; a protonated nitrogen holds its electrons more tightly because of the positive charge and gives a second peak at about 402 eV (refs 38, 39) C21·P6. Fitting the N1s envelope with the two components gives the fraction of drug molecules that are protonated, reported here as % protonation P6·R8.

Glass transition and water. The glass transition temperature (Tg) marks where an amorphous solid turns from a rigid glass into a mobile, rubbery material; the further a sample sits below its Tg, the slower its molecules move and the slower it crystallizes. For a miscible ASD the Tg is expected to lie between the Tg values of the pure drug and the pure polymer, and absorbed water, with a very low Tg of its own, lowers it (refs 40–43) C47. Lumefantrine has a low Tg of 18 °C, whereas the four polymers lie far higher (HPMCP 150 °C, HPMCAS 117 °C, CAP 133 °C; EL100 not reported) M1·M2·M3·M4.

Non-sink dissolution and amorphous solubility. In a sink test the medium can dissolve all of the dose. In a non-sink test it cannot, so the test shows how far above the equilibrium solubility an ASD drives the concentration and how long that supersaturation lasts (refs 35, 36) C13. The relevant ceiling for a supersaturating ASD is the amorphous solubility: the highest concentration of molecularly dissolved drug the amorphous form can sustain, above which a drug-rich phase separates. The authors give the amorphous solubility of lumefantrine as 2 µg/mL in one place and 1.8 µg/mL in another, without a method or reference M1·X5.

Powder descriptors. Laser diffraction gives the volume-based percentiles d10, d50 and d90 (the sizes below which 10, 50 and 90% of the particle volume lies) and the span, a measure of the width of the distribution P10·E4. Carr's compressibility index and Hausner's ratio compare the bulk and tapped density of a powder; low values mean that a powder packs little on tapping and flows well P13·E2·E3. The paper names these indices but does not print their formulas or the densities behind them E2·E3.

Methods in detail

The study is a formulation screen in three stages: make the ASDs, read their solid state and interactions, then test stability and release. Two polymers were carried forward after the first PXRD screen, and only the most promising compositions were tested for powder flow. The flow below summarizes the work.

Preparation by spray anti-solvent precipitation

For a 300 mg batch, lumefantrine and polymer were dissolved in 5 mL DMA and sprayed through an ultrasonic nozzle (Büchi) into 200 mL of ice-cold 0.02 N HCl under stirring P1. The suspension was vacuum-filtered, and the solid was dried in an open Petri dish inside a vacuum desiccator at room temperature for 12 to 100 h, depending on the sample P1·R11. The Discussion specifies that the EL100 ASDs needed 4 days to dry, the others about 12 h R11. All four polymers were first made at a drug:polymer ratio of 4:6 (40% drug loading); combinations "exhibiting satisfactory stability and dissolution properties" were then also made at 2:8 and 3:7 P1·C16. In practice HPMCP and HPMCAS were extended to all three ratios; EL100 was also made at 2:8 and 3:7 but only examined by PXRD, and CAP was made at 4:6 only A2·R2. Pure amorphous lumefantrine for reference was made by melting 50 mg at 150 °C in an aluminium pan, quenching in liquid nitrogen and grinding P2.

Note by the AI writer (not discussed in the paper): the solids concentration in DMA was about 60 mg/mL (300 mg in 5 mL), and the dispersion was sprayed into 40 times its volume of anti-solvent. The drug content of the dried ASDs was not assayed, so the nominal composition is assumed throughout.

Solid state, interactions and thermal analysis

Crystallinity was read by powder X-ray diffraction (Cu Kα, 2θ 4–40°, 4°/min) right after drying and after storage at 40 °C/75% RH for up to 3 months P3·P4. The 4:6 CAP ASD was also stored under vacuum for 3 months as a dry control P4·R5. The paper does not state the storage container, the sample amount or any intermediate time point other than the first day for CAP P4·X15.

Drug–polymer interaction was probed in two ways. Attenuated total reflectance FTIR (64 scans, dry-air purge) recorded the carbonyl region of amorphous lumefantrine, the neat polymers and the ASDs P5. XPS (monochromatic Al Kα, 20 eV pass energy for high-resolution spectra, charge neutralizer, C 1s C–C set to 284.6 eV) recorded the N1s region, which was fitted with the neutral (399 eV) and protonated (402 eV) components to give % protonation, in triplicate P6. Modulated DSC (cooled to −25 °C, heated to 200 °C at 5 °C/min with ±2 °C modulation every 60 s, non-hermetic pans) gave Tg from the reversing heat flow, and thermogravimetry (to 170 °C at 10 °C/min under nitrogen) gave the water content as the weight lost between 35 and 170 °C P7·P8·E1.

Particles, flow and dissolution

Particle morphology was imaged by SEM after platinum coating P9. Particle size was measured by laser diffraction in 0.02 N HCl (pH 1.7), in which the ASDs do not dissolve, at two stages: on the suspension straight after precipitation and on the dried powder that was later used for dissolution P10·C14. Each measurement was made in triplicate P10. Carr's index and Hausner's ratio were measured only for the 2:8 and 4:6 HPMCP powders, chosen because 2:8 released almost all of its drug and 4:6 gave the highest release at a high (40% w/w) loading P13·C40.

Dissolution was run under non-sink conditions: ASD powder containing 20 mg of lumefantrine was added to 100 mL of 50 mM pH 6.8 phosphate buffer in a jacketed beaker at 37 °C, stirred at 300 rpm P11. About 1 mL was withdrawn at each time point up to 2 h, replaced with fresh buffer, passed through a 0.22-µm nylon filter and analysed by HPLC (C18 column, acetonitrile and pH 2.5 sodium sulfate buffer 75:25, detection at 380 nm, calibration 1–500 µg/mL) P11·P12. Each profile is the mean of three runs P11. The marketed product Coartem was tested alongside the 4:6 ASDs, but the Methods do not describe how much of it was added or in what form M9·X17. Groups were compared by one-way ANOVA with Tukey's test at α = 5% P14.

Note by the AI writer (not discussed in the paper): if all 20 mg dissolved in 100 mL the concentration would be 200 µg/mL. This ceiling is useful for reading the release percentages below, and it shows that the test was strongly non-sink against an amorphous solubility of 1.8 µg/mL: the dose was about a hundred times what the medium could hold as dissolved amorphous drug P11·M1·X4.

Experimental analysis

Result 1: HPMCP and HPMCAS keep lumefantrine amorphous; CAP and EL100 do not

Right after drying, the 4:6 ASDs with HPMCP, HPMCAS and CAP gave only broad amorphous halos, centred around 8 and 21° 2θ (Figure 2) R1. The 4:6 EL100 ASD showed small sharp peaks, at about 24.8 and 26.5° 2θ as read from Figure 2, which the authors match to crystalline lumefantrine (Figure S1, not available here) R1·C15. The 2:8 and 3:7 EL100 ASDs also showed lumefantrine peaks immediately after preparation (data not shown), whereas the 2:8 and 3:7 HPMCP and HPMCAS ASDs were amorphous (Figure 3) R2·R3. DSC agreed: no melting peak was seen for any HPMCP, HPMCAS or CAP ASD, which the authors take as "indicating" that the drug was amorphous at all loadings R9·C24.

Table A. Crystallinity of the ASDs by PXRD (from Figures 2–4 and the text; peak positions read from Figure 4, approximate) R1·R2·R3·R4·R5.

Under accelerated storage the 4:6 CAP ASD "began to crystallize within the first day", but the diffraction peak areas did not increase over 3 months, and the same ASD kept under vacuum stayed amorphous for 3 months R4·R5·C17. All six HPMCP and HPMCAS ASDs showed no reported crystallinity after 3 months at 40 °C/75% RH, which the authors call "indicating prolonged retention of the amorphous state" and, in the Discussion, "surprising" for a process that uses water as anti-solvent R4·C1·C42. EL100 was described as "the least stable of the tested formulations" and dropped C30.

Note by the AI writer (not discussed in the paper): two of the 3-month traces in Figure 4 carry a single narrow feature, a spike at about 31.8° 2θ on the 2:8 HPMCAS trace and a small step at about 27.6° 2θ on the 4:6 HPMCP trace R4·X16. Single features of this kind may be noise rather than crystal reflections, and the authors report no crystallinity for these ASDs; a reader who needs certainty should look at the original figure X16. The caption of Figure 4 says "up to 3 months" without labelling the storage time of each trace, so the CAP pattern is presumably the early one described in the text X15.

Result 2: the failing ASDs held more water or dried more slowly

Table 2 of the paper lists the Tg and water content of the ASDs that were not dropped. The HPMCP and HPMCAS ASDs held 4–6% water after drying, the 4:6 CAP ASD 9% R10. The text gives the CAP value as "approximately 10%"; the table value is used here R10·X20. The EL100 ASDs and the lower CAP loadings were not tested "due to the poor stability profile exhibited in the earlier studies" R10·C16.

Table B. Glass transition temperature and water content of the ASDs (from Table 2, n = 3; amorphous lumefantrine Tg 18 °C; EL100 ASDs not tested) R10·M1·M2·M3·M4.

The authors connect the water content to the failures. For CAP, the higher water content "could also be responsible for rapid crystallization under the accelerated storage conditions", and in the Discussion it "could be a reason" C27. The vacuum control points the same way: the CAP ASD crystallized at 40 °C/75% RH but not under vacuum R5. For EL100 the argument is different. The Discussion states that the EL100 ASDs "did not have a high moisture content after drying" but took 4 days to dry against about 12 h for the others, and "such prolonged contact of the amorphous drug with water is known to enhance the risk for crystallization" (Qian et al., ref 46) C43·R11. No water content for EL100 is reported anywhere in the paper, so this statement cannot be checked against data X2.

The Tg data hold two further observations. First, the Tg did not fall as drug loading rose. With both polymers the 2:8 ASD, which has the most polymer, had the lowest Tg (88 °C with HPMCP, 45 °C with HPMCAS), while the 3:7 and 4:6 ASDs lay at 87–112 °C R10. The authors find this "interesting", consider the similar water contents "unlikely" to explain it, and suggest that ionization "may also affect the Tg"; they leave it for further studies C46·C48. The Discussion refers to the higher-loaded ASDs as "3:7 Lum:polymer and 4:8 Lum:polymer"; no 4:8 ratio was made, and 4:6 is meant X12. Second, HPMCP gave the highest Tg at every ratio, and the CAP ASD's Tg (82 °C) sat close to that of the HPMCAS ASD (87 °C) although neat CAP has the higher Tg; the authors attribute the latter, with "could be due to", to the higher water content of the CAP ASD R10·C26·C28. At 3:7 the HPMCP and HPMCAS values (112 ± 19 and 101 ± 22 °C) overlap within one standard deviation X11.

Result 3: all ASDs are partly protonated, but protonation does not rank stability

Fitting the XPS N1s spectra showed that 14–42% of the lumefantrine was protonated in the ASDs, against 3 ± 1% in the amorphous drug alone (Table 1) R8. The authors call the protonation "significant" in all ASDs and note "some correlation" with polymer loading: protonation rose from 37 to 42% across HPMCP 4:6 to 2:8 and from 14 to 37% across HPMCAS 4:6 to 2:8 R8·C22·C23.

Table C. Percent protonation of the lumefantrine tertiary nitrogen (from Table 1) next to the stability outcome (from Figures 2–4) R8·R1·R4.

FTIR gave the first hint of an interaction (Figure 5). On forming the dispersion, a polymer carbonyl band moved to higher wavenumber: for CAP the dominant carbonyl band moved from 1675 cm⁻¹ in the neat polymer to 1733 cm⁻¹ in the 4:6 ASD, for EL100 from 1707 to 1725 cm⁻¹, while HPMCP and HPMCAS showed only minor changes at every loading R6·R7·C18. Neat CAP shows a shoulder that the authors assign to its second type of carbonyl (ester versus acid), and the changed band ratio in the ASD "further suggests an interaction of the polymer with the drug" C19. Because the IR changes could not show whether the drug is ionized, XPS was used C20. Note by the AI writer (not discussed in the paper): the 4:6 EL100 ASD trace in Figure 5 also carries a band at about 1600 cm⁻¹ nearly as strong as its carbonyl band, which the other ASDs do not show at that strength R6; the paper does not assign it.

The key comparison is in the Discussion. At 4:6 the authors rank protonation as HPMCP > EL100 > CAP > HPMCAS > amorphous lumefantrine and stability as HPMCP > HPMCAS > CAP > EL100 > amorphous lumefantrine, and conclude that "a correlation is not observed" between the two, unlike in the earlier solvent-evaporated ASDs (Trasi et al., ref 39) C50·C49·R20. Their reconciliation is that the two polymers that broke the correlation, CAP and EL100, were the ones with more water or longer exposure to water; without water in the process they had been stable for more than 30 days C51·R20. Hence: "a higher degree of tertiary nitrogen protonation in lumefantrine may confer greater stability to the ASD while the presence of high moisture content negatively affects stability" C52·C56.

The AI writer's reading of Table C: within the table, HPMCP (37 ± 3%) and EL100 (38 ± 4%) at 4:6 cannot be ordered, and HPMCAS is the least protonated only at 4:6 (its 2:8 value equals 4:6 HPMCP) X7. The least protonated ASD of all, 4:6 HPMCAS at 14%, was among the stable ones, and the stability of neat amorphous lumefantrine, which ends both rankings, is not reported in this paper R8·X7. The data therefore support the authors' negative statement (protonation does not order stability here) more firmly than their positive one ("may confer greater stability").

Result 4: more polymer gives more release; HPMCP leads at 2 h

At 4:6 (Figure 7) both the HPMCP and HPMCAS ASDs exceeded the stated amorphous solubility (2 µg/mL) within 30 min R15·C34. They behaved differently over time. The HPMCAS ASD dissolved fast and levelled off after about 1 h at about 18–19 µg/mL; the HPMCP ASD rose steadily and reached 23.3 µg/mL at 2 h against 18.4 µg/mL for HPMCAS R15·C34. No drug was detected from the 4:6 CAP ASD over 2 h, and Coartem stayed at about zero throughout R15·C34·C35.

Lowering the drug loading raised the release sharply with both polymers (Figures 8 and 9) R16·R17·C36. The 2:8 HPMCP ASD climbed to about 190 µg/mL within the first hour and reached 208.4 µg/mL at 2 h; the 2:8 HPMCAS ASD reached about 110 µg/mL at 1 h and 142.8 µg/mL at 2 h R16·R17·C38. At 3:7 the HPMCP ASD released at a steady rate to 132.6 µg/mL, the HPMCAS ASD rapidly in the first hour and then slowly to 84.8 µg/mL R16·R17·C38. The authors state that HPMCP gave "significantly higher drug release than HPMCAS for each drug-polymer ratio" C37.

Table D. Drug in solution after 2 h (from the text; 20 mg drug in 100 mL, 0.22-µm filtered, n = 3). The last column is computed by the AI writer from the printed values and the stated amorphous solubility of 1.8 µg/mL R15·R16·R17·R18·R19·X3.

Three numbers in this part of the paper do not line up, and the reader should know which to use. First, the abstract gives the 2:8 HPMCP ASD "a maximum drug release of 140 µg/mL", while the results give 208 µg/mL and the Discussion 208.4 µg/mL; 140 µg/mL matches the 2:8 HPMCAS ASD instead (143 and 142.8 µg/mL), and Figure 8 agrees with 208 C3·X1. Second, the authors describe the 2:8 values as "about 100- and 60-times" the amorphous solubility of 1.8 µg/mL; the quotients are about 116 and 79 R19·X3. Third, the printed release percentages equal the 2-h concentration divided by 200 µg/mL (142.8 → 71.4%, 132.6 → 66.3%, 84.8 → 42.4%), so the 2:8 HPMCP value of 208.4 µg/mL corresponds to 104% of the nominal dose and is called "almost 100%"; no correction for the drug removed in the 1 mL samples, and no assay of ASD drug content, is described R18·X4.

The ranking "HPMCP above HPMCAS at each ratio" holds at the 2-h point. At 4:6 the HPMCAS curve is higher from 15 to 90 min (for example about 18.0 against 8.9 µg/mL at 60 min, as read from Figure 7) and HPMCP overtakes it only at 2 h; the outputs of the ANOVA and Tukey tests are not shown X9·R15.

Result 5: drying enlarges the particles, most for HPMCAS

SEM showed irregular particles that tend to aggregate (Figure 6); in the AI writer's reading of the images, most ASDs are rough aggregates while the 4:6 HPMCP particles appear as smooth-faced fragments R12·C31. Laser diffraction quantified how drying changed the size (Table 3).

Table E. Median and 90th-percentile particle size before and after vacuum drying (excerpt from Table 3, laser diffraction in 0.02 N HCl, n = 3; d10 and span omitted) R13.

In suspension the HPMCAS particles were the smallest (d50 56–65 µm) and the HPMCP particles the largest (99–139 µm) R13. Drying reversed this. All HPMCAS and CAP ASDs grew markedly, HPMCAS up to a d50 of 1180 µm at 2:8, while the HPMCP ASDs changed much less and the 3:7 HPMCP ASD even became smaller (99.2 to 71.5 µm) R13·C33. The authors state that for 2:8 and 4:6 HPMCP "the d90 values were similar for both forms", which holds for 2:8 (335 and 367 µm) but not for 4:6 (378 ± 29.5 and 563 ± 161.3 µm) C33·X18. They then suggest that "the smaller particle sizes of HPMCP ASDs relative to the corresponding HPMCAS ASDs may explain their better dissolution" C39.

Note by the AI writer (not discussed in the paper): the 4:6 pair argues against particle size as the main driver. The dried 4:6 HPMCAS ASD (d50 366 µm) released faster than the dried 4:6 HPMCP ASD (d50 147 µm) for the first 90 min R13·R15·X19. Size and polymer change together in every pair, so the data cannot separate them X19.

Powder flow. The 2:8 and 4:6 HPMCP powders gave Carr indices of 7.1 ± 1.2% and 6.3 ± 3.7% and a Hausner ratio of 1.1 for both (printed with a % sign, although the ratio has no unit), which the authors read as "excellent flow behavior and compressibility properties" of powders made by this route R14·C41·X21. Only these two powders were tested P13.

Discussion

The authors' causal picture. The paper's explanation runs through water. Anti-solvent precipitation keeps the amorphous drug wet for longer than other processes, so the polymer must protect it from moisture during manufacture as well as in storage C11·C42. The flow below lays out the chain as the authors give it; the hedges are theirs.

The authors' chain from process water to outcome R10·R11·R1·R4·R15·R16·C27·C43·C51.

In this picture, protonation of lumefantrine by the acidic polymers is real in every ASD but is not what decides stability in a water-based process; the authors keep the idea that protonation "may confer greater stability", outweighed here by moisture R8·C50·C52. Within the two stable polymers, more polymer gives more release, and HPMCP outperforms HPMCAS at 2 h, which the authors link to its smaller dried particles ("may explain") C36·C37·C39. For CAP they offer a second, release-specific explanation: its poor release "could be related to the stronger drug-polymer interactions as seen from the high degree of protonation", in line with the incongruent release of 40%-loaded CAP ASDs reported earlier (Trasi et al., ref 39) C54·R21·C53.

Where the conclusions apply. All results concern one drug, lumefantrine, made by one bench process at 300 mg scale, dried in a vacuum desiccator at room temperature, stored at one accelerated condition and dissolved in one medium (pH 6.8 phosphate buffer, non-sink, 2 h) P1·P4·P11. The stability read-out is PXRD, which detects crystallinity once it exceeds the method's detection limit, at what is effectively one time point per ASD P3·X15. Within these limits, the screen is clear: HPMCP and HPMCAS ASDs survived the aqueous process and storage, CAP and EL100 ASDs did not R1·R4.

What remains open. The authors list several questions themselves: why the Tg of the 2:8 ASDs is so low and whether ionization lowers Tg ("further studies are warranted"), why CAP releases so poorly ("more studies are warranted"), the exact nature of the interaction behind the FTIR changes, and whether a more efficient drying system would shorten the contact with water C46·C48·C54·C20·C45. They chose not to pursue an all-organic anti-solvent route because it would add cost and was not needed for HPMCP and HPMCAS C44.

Implications for formulation (AI writer's reading, not stated in the paper). For an aqueous anti-solvent process, the screen suggests checking two process properties early: how much water each polymer's precipitate keeps after drying, and how long it takes to dry. In this data set these separated the failures from the successes better than the drug–polymer protonation measured by XPS R10·R11·R8. The release data also show how steeply loading trades against supersaturation for lumefantrine: going from 40 to 20% drug raised the 2-h concentration about 9-fold with HPMCP and about 8-fold with HPMCAS, so the best-performing composition carries four parts polymer per part drug R15·R16·R17. Finally, concentrations of a hundred times the amorphous solubility, measured after a 0.22-µm filter, are best read as "drug passing the filter", which may include drug-rich colloids or polymer-associated drug, rather than as molecularly dissolved drug X6.

Limitations

Limitations stated by the authors. The authors acknowledge that water is a risk inherent to the process and recommend "a more efficient drying system" in future studies C11·C45. They describe the Tg behaviour of the 2:8 ASDs as a "surprising phenomenon" that needs further study, and call for more work on the poor release of CAP ASDs C46·C48·C54. Their explanations are hedged: moisture "could be a reason" or "could have been a cause" of instability, protonation "may confer" stability and particle size "may explain" the release differences C27·C51·C52·C39.

The AI writer's own judgement (1): what the stability data can and cannot show. The central stability result rests on PXRD patterns, one per ASD, after 3 months at 40 °C/75% RH; no intermediate time points, no second humidity and no more sensitive crystallinity method were used R4·X15. This is enough to separate the failures (CAP, EL100) from the successes, but not to rank the successes. The authors nonetheless describe HPMCP ASDs as "more physically stable than HPMCAS" from Tg data and conclude that 2:8 HPMCP showed "the best stability", although no HPMCP or HPMCAS ASD crystallized and 2:8 HPMCP had the lowest Tg among the HPMCP ASDs C29·C57·X10. The moisture explanation is plausible and supported by the CAP vacuum control, but water content, drying time and polymer chemistry change together, no ASD was dried to different water contents, and the EL100 water content and DSC result are not reported (the thermal section names EL100 as the only ASD with a melting peak, while the Table 2 footnote says EL100 ASDs were not tested) R5·C56·X2·X14.

The AI writer's own judgement (2): numbers to read with care. Several stated numbers disagree with the data: the abstract's 140 µg/mL for the 2:8 HPMCP ASD (results: 208.4 µg/mL), the "about 100- and 60-times" supersaturation (quotients about 116 and 79), the 2:8 HPMCP release of 104% of the nominal dose called "almost 100%", and the "approximately 10%" water of the CAP ASD (Table 2: 9%) X1·X3·X4·X20. The amorphous solubility used as the reference (1.8 or 2 µg/mL) has no stated source X5. Some ranking statements go beyond the data: HPMCP versus HPMCAS release at 4:6 depends on the time point, HPMCP and EL100 protonation are equal within error, and the CAP ASD, described as highly protonated, is less protonated than HPMCP X9·X7·X8. The Coartem comparison is qualitative because the amount and form of Coartem tested are not described, and the abstract's "<80 ng/mL" for Coartem is the cited aqueous solubility of lumefantrine, not a measured release X17. It is also not stated whether the CAP powder used for dissolution was still amorphous at the time of the test, given that it crystallizes within a day at 40 °C/75% RH X22. Finally, Eudragit L100 is named as a methacrylic acid–ethyl acrylate copolymer in the text but drawn with a methyl methacrylate unit in Figure 1 X13.

Conclusion

The paper shows that spray anti-solvent precipitation into ice-cold 0.02 N HCl can produce amorphous lumefantrine dispersions with HPMCP and HPMCAS that show no reported crystallinity after 3 months at 40 °C/75% RH and reach 208.4 and 142.8 µg/mL of lumefantrine in 2 h at a 2:8 ratio, far above the 18.4–23.3 µg/mL of the 4:6 ASDs and the near-zero release of Coartem R4·R15·R16·R17. The same process failed with CAP, which held 9% water and crystallized within a day of storage, and with EL100, which dried for 4 days and was crystalline as made R10·R4·R11·R1. All ASDs were partly protonated (14–42%), but protonation did not order their stability R8·C50. The authors' explanations, moisture for the failures, protonation as a possible stabilizer and particle size for the release differences, are hedged and were not tested by varying one factor at a time C51·C52·C39. The authors conclude that the 2:8 HPMCP ASD showed the best stability and drug release, that ASD stability "was reduced by the presence of moisture during manufacturing or storage", and that faster drying should be considered; for a water-based ASD process this puts the water history of the precipitate next to the drug–polymer chemistry as a design variable C57·C56·C45.