Making an amorphous lopinavir/ritonavir combination by solvent granulation, and why the two drugs are better kept apart

Background and motivation

Many new drugs dissolve poorly in water, and the amorphous form is a common answer: an amorphous drug can dissolve to a concentration well above the crystalline solubility, a supersaturated solution, and it is this supersaturation, rather than the total dissolved amount, that drives flux across a membrane (Hancock and Zografi 1997; Frank et al. 2012b; Raina et al. 2015; Borbás et al. 2016) C1. A polymer is usually added to make a molecular dispersion, the amorphous solid dispersion (ASD), which lowers molecular mobility, raises the glass transition temperature and slows crystallization (Kothari et al. 2015; Prasad et al. 2014) C2.

Treatments for HIV and hepatitis C combine several drugs, either for synergy or because one drug inhibits the efflux transporters or metabolic enzymes that limit the other (Zhang et al. 2000; Drewe et al. 1999) C3. Putting the drugs in one fixed dose combination (FDC) tablet can help patients take them, and several FDCs are marketed: Kaletra (lopinavir/ritonavir), Atripla and Harvoni. In some of these, such as lopinavir and ritonavir, the drugs are formulated as amorphous solids for better bioavailability C3. Separately, a body of work on drug–drug "co-amorphous" systems reports stability gains, from interactions that raise the glass transition temperature (Martinez et al. 2016) or from interference with nucleation (Knapik-Kowalczuk et al. 2018); but dissolution is usually compared only with the crystalline drugs, and real gains over the pure amorphous drug appear mainly when an acid and a base form a salt or change the local pH (Fung et al. 2018), or for kinetic reasons (Allesø et al. 2009) C4.

The Taylor group had shown earlier that mixing is not free. When two miscible, uncharged drugs form one amorphous phase, the amorphous solubility of each in the mixture is lower than that of the pure amorphous drug, because mixing lowers each drug's chemical potential (Alhalaweh et al. 2016; Trasi and Taylor 2015a, 2015b) C5. An amorphous FDC therefore risks giving less supersaturation than its single-drug parts.

The second motivation is cost. Commercial ASDs are made by spray drying or hot melt extrusion. Hot melt extrusion needs a drug that survives processing above about 120 °C with a workable melt; spray drying needs organic solvents, control of residual solvent and solvent recycling, and can give powders of low bulk density and high electrostatic charge; both need expensive equipment (Paudel et al. 2013; Repka et al. 2008; Murtomaa et al. 2004) C6. Since drugs for infectious diseases are used widely in low-income countries, the authors look for an inexpensive route that could run there C6.

Their candidate is wet granulation with an organic solvent, also called solvent impregnation: dissolve drug and polymer in a solvent, pour the solution onto an excipient powder, granulate and dry. The idea has a commercial precedent in the amorphous tacrolimus product Prograf, made by solvent impregnation of a polymer (Honbo et al. 1987; Trasi et al. 2017; Yamashita et al. 2003) C36. The paper tests the route on a lopinavir/ritonavir FDC, using X-ray powder diffraction (XRPD), infrared spectroscopy and release testing, and compares release with the marketed tablet, Aluvia A1·A6·M9.

Background concepts

Amorphous solubility and supersaturation. An amorphous drug can raise the concentration of molecularly dissolved drug up to its amorphous solubility; beyond that point the excess separates as a drug-rich phase rather than staying dissolved. Here the amorphous solubilities at pH 6.8 (10 mM phosphate, 37 °C) were measured by adding a methanol stock to buffer containing 10 µg/mL HPMC, ultracentrifuging and assaying the supernatant: 31 µg/mL for ritonavir (RTV) and 19 µg/mL for lopinavir (LPV) P13·M1·M2. The crystalline solubilities at the same pH are printed as 1.8 and 2.9 µg/mL, without a stated method or source M1·M2·X14. The paper measures supersaturation with the relative supersaturation:

In plain terms: $S$ is the measured dissolved concentration $C$ divided by the crystal solubility $C^{*}$, so $S=1$ is a saturated solution and $S=17$ is seventeen times the crystal solubility E1. With the values above, the amorphous form could give about 17-fold supersaturation for ritonavir and about 7-fold for lopinavir R5·E1.

Co-amorphous mixing and mutual solubility reduction. When two drugs are mixed at the molecular level in one amorphous phase, each is "diluted" by the other, its chemical potential falls, and so does the highest concentration it can reach in water C5·C38. The paper adds a point that matters for FDCs: the fall depends on composition, and the minor component is hit harder than the major one C39. In the Aluvia ratio, 200 mg lopinavir to 50 mg ritonavir (4:1), ritonavir is the minor component M9·C39.

Ionization and the pH shift. Ritonavir is a weak base with pKa values of 1.8 and 2.6 for its two thiazole groups; at pH 1.2 (0.1 N HCl) it is highly, though not completely, ionized, and its crystalline solubility rises to about 0.4 mg/mL (Law et al. 2001), so acid is a sink medium for it M1·C14. Lopinavir is not ionized at either pH M2·C14. When a weak base dissolved in acid moves to near-neutral pH, as from stomach to intestine, it loses its charge and becomes supersaturated; if the un-ionized concentration exceeds the amorphous solubility, the solution undergoes liquid–liquid phase separation (LLPS) into a drug-lean solution and drug-rich nanodroplets (Ilevbare and Taylor 2013; Sugihara and Taylor 2018) C40.

Drug-controlled and polymer-controlled release. Release from an ASD tends to fall into one of two regimes. In polymer-controlled release the hydrophilic polymer, here PVPVA, dissolves and carries the drug with it, so release is fast; in drug-controlled release the slow dissolution of the hydrophobic drug sets the pace. Which regime applies depends on the drug:polymer ratio C43. For ritonavir–PVPVA ASDs, earlier Taylor-group work found nanodroplet release, the hallmark of polymer-controlled release, only below 25% drug loading (Indulkar et al. 2019) R22.

Flow and tablet terms. Carr's index and the Hausner ratio compare the bulk and tapped density of a powder; lower values mean better flow. The angle of repose is the slope of a poured powder heap. Friability is the mass lost when tablets are tumbled; hardness is the force needed to break a tablet.

Methods in detail

The study has three parts: making the granules and checking that they are amorphous and stable; testing release against the marketed tablet and then taking the combination apart to see what controls ritonavir release; and two experiments aimed at the mechanism, a pH shift and an infrared look at a dissolving surface. The flow below follows the paper's order.

Contribution 1: amorphous granules from a spatula, then from a granulator

Problem. Can a solvent simply poured onto excipients, without spray drying or extrusion, leave the drugs amorphous, and can the granules be tableted?

What they did. A 1:1 blend of anhydrous lactose and MCC pH 101 was wetted with a 1:1 dichloromethane (DCM):methanol (MeOH) solution of the drugs and PVPVA and mixed with a spatula. The main batch carried the marketed dose, 200 mg LPV and 50 mg RTV, with 250 mg PVPVA in 2–3 mL solvent on 1.5 g of blend, 2 g per tablet; the drug:polymer ratio is therefore 50:50 P1.a. Single-drug granules used the same 1:1 drug:polymer ratio: 50 mg RTV + 50 mg PVPVA + 400 mg blend, and 200 mg LPV + 200 mg PVPVA + 1000 mg blend P1.d·P1.e. Variants contained 5% w/w Span 20 (the surfactant in the marketed product) or 0.1% w/w Tween 80 at the same 2 g, a 20:80 drug:PVPVA ratio on 3 g of blend (over 4 g per tablet), or other LPV:RTV ratios (125:125 and 50:200) P1.b·P1.c·P1.f·P1.g·P1.h. All granules were dried overnight in a vacuum oven and milled for 10 s in a freezer mill P1. Stability was tested by open storage at 40 °C/75% RH over saturated sodium chloride, with XRPD every 7 days for one month P2·P6.

To see whether the route works with more industrial equipment, 60 g of the lactose/MCC blend was granulated in a custom bench-top high-shear granulator (described in Wikstrom et al. 2008) by pumping in a DCM:MeOH solution of 12 g LPV, 3 g RTV and 15 g PVPVA at 25 mL/min, under 4 min in total, "until complete granulation of the powder occurred" P3. The granules were sieved, vacuum-dried, assayed, and characterised by Carr's index, Hausner ratio, angle of repose and sieve analysis; 150 mg portions were compressed directly on a computer-controlled press with a 6 mm punch to a 500 kg target load at 120 mm/min, and tested for hardness and friability P3·P4·P5.

Why it can work. The authors make it a condition of the method that the drug resists crystallization while the solvent evaporates, which favours slowly crystallizing compounds of higher molecular weight and more complex structure, such as ritonavir and lopinavir C35·C32. The price is excipient: enough powder is needed to take up the solvent without forming a sticky mass C33.

Contribution 2: a dissolution design that takes the combination apart

Problem. Does the granulated combination release like the marketed product, and if not, which part of the design is responsible?

What they did. Granules were mixed with 10 wt% croscarmellose sodium and pressed into compacts (500 psi, 3 s), because the loose powder floated and to make a fairer comparison with a tablet P7. Compacts holding 200 mg LPV and 50 mg RTV, or one Aluvia tablet, were stirred at 150 rpm in 250 mL medium at 37 °C; samples were filtered through 0.45 µm PTFE filters and assayed by HPLC (C18 column, acetonitrile:water 60:40, 215 nm) P7·P8·P9·P15. Two media were used. pH 6.8, 10 mM phosphate buffer is strongly non-sink for both drugs with respect to their crystalline solubilities, and neither drug is ionized there P8·C10. 0.1 N HCl mimics the stomach and is a sink for ionized ritonavir P9·C14. Aluvia could not be measured at pH 6.8: its milky dispersion clogged small-pore filters and passed through larger ones R20.

Against the co-granulate the authors set single-drug granules (pH 6.8), Aluvia (acid), co-granules with less lopinavir per ritonavir (acid, ritonavir fixed at 50 mg per vessel), co-granules with surfactant (acid), and separately granulated drugs combined either in one tablet or as a bilayer tablet (acid) A1·A2·A3·A6. Scaled-up bench-top granules equivalent to 100 mg of drug were tested in 100 mL 0.1 N HCl P10.

How it differs from earlier work. Earlier co-amorphous studies mostly compared the mixture with the crystalline drugs C4. Here the reference for each drug is its own single-drug amorphous granule, and the product reference is a marketed amorphous FDC, so a loss caused by mixing shows up directly A1·A6.

Contribution 3: two experiments aimed at the mechanism

pH shift. One Aluvia tablet was dissolved in 200 mL 0.1 N HCl for 30 min; the medium was then neutralized with 750 mg NaOH in 10 mL water and 40 mL of 100 mM pH 6.8 buffer, a 25% increase in volume P11. To see whether ritonavir crystallizes on the shift, the experiment was repeated with a drug solution free of excipients, since insoluble MCC would hide crystals: a pH 1.2 solution with nominally 100 µg/mL RTV and 400 µg/mL LPV was neutralized with NaOH to pH 6.8; part of the suspension was ultracentrifuged and the supernatant assayed, and part was examined under a polarized light microscope P12. A related test added 0.8 mg/mL lopinavir (dissolved from LPV-only granules) to a 200 µg/mL ritonavir solution in 0.1 N HCl (from RTV-only granules) and stirred for 1 h P14.

Surface infrared. Pure ritonavir, pure lopinavir and a 200:50 LPV:RTV mixture were melted and quenched to glasses and measured by attenuated total reflectance FTIR (ATR-FTIR, 64 scans, 400–4000 cm⁻¹). The blend pellet was then exposed to 0.1 N HCl for 1 h, dried, and the exposed surface was scraped off and measured again P16. ATR is somewhat surface sensitive (Planinsek et al. 2006), so a change in the ratio of drug-specific bands is read as a change in surface composition C29. The pellet contained no PVPVA or excipients P16·X7.

Experimental analysis

The granules are X-ray amorphous and stay so for a month

XRPD of the 200/50 LPV/RTV granules with PVPVA showed no peaks from either drug (Figure 2). The Bragg peaks that remain (near 10.5, 19 and 21° 2θ, among others) are attributed to lactose, which stays crystalline in the granules; a physical mixture of the crystalline drugs and excipients at the same loading showed additional drug peaks, several of them below 10° 2θ R1·C8. After one month of open storage at 40 °C/75% RH the pattern was unchanged, and granules of each drug alone also stayed free of crystals (Figure S2, not available here) R2·R3. The authors read the first result as amorphous drugs ("suggesting") and the second as resistance to crystallization "for this time period" ("indicating") C8·C9.

At pH 6.8, sharing one amorphous phase lowers both drugs, ritonavir most

Figure 3 is the paper's clearest experiment. Each drug granulated alone dissolves to its own amorphous solubility: ritonavir rises from about 14 µg/mL at 15 min to about 30 µg/mL by 60 min, against a measured amorphous solubility of 31 µg/mL, and lopinavir sits at about 18–19 µg/mL throughout, against 19 µg/mL R4.a·R4.b·M1·M2. The authors call these solutions "maximally supersaturated", with relative supersaturations of about 17 (ritonavir) and 7 (lopinavir) C12·R5. When the same doses are granulated together, ritonavir stays between about 2 and 4 µg/mL over 2 h and lopinavir climbs only to about 15 µg/mL R4.c·R4.d. The text describes the ritonavir plateau as 4–5 µg/mL, while the plotted points reach about 4.0 µg/mL at most; the corresponding supersaturation, printed as about 3, is about 2–3 X3·R5.

Table A. Dissolved drug (µg/mL) at pH 6.8, 10 mM phosphate, 250 mL, 37 °C; time points read from Figure 3 (approximate); amorphous solubilities and S values from the text R4·R5·M1·M2.

The authors explain the loss by co-amorphous mixing: in the co-granule ritonavir and lopinavir form one amorphous phase, each lowers the other's chemical potential, and so each has a lower amorphous solubility and a lower driving force for dissolution; they state that the lowered concentration "is due to" this mixing C38. Ritonavir loses more because it is the minor component at LPV:RTV 4:1, and the reduction caused by a second amorphous component grows as a component becomes more dilute C39. Note by the AI writer (not discussed in the paper): the amorphous solubility of each drug in this particular mixture was not measured; the explanation rests on the earlier work of Trasi and Taylor (2015b) and on the pattern of Figure 3 C38.

In acid, co-granulated ritonavir lags the marketed tablet

In 0.1 N HCl ritonavir is ionized and acid is a sink medium for it, so far more dissolves than at pH 6.8 C14. From Aluvia, ritonavir rose steadily from about 40 µg/mL at 15 min to about 169 µg/mL at 2 h (Figure 4) R6.a. Full release of 50 mg in 250 mL would give 200 µg/mL, so even the marketed tablet released incompletely in this time ("indicating incomplete release") C15. The co-granulated compacts started at a similar level, about 46 µg/mL, but reached only about 96 µg/mL at 2 h R6.b. Lopinavir, not ionized in acid, stayed near its amorphous solubility, rising from about 16 to 18–19 µg/mL for both products R6.c·C14. The authors call the ritonavir difference "significantly lower" (abstract: "significantly retarded"); the paper gives no replicate numbers or statistics C17·X12. The text cites Figure 3 for this comparison; the data are in Figure 4 X1. Scaled-up bench-top co-granules also released less ritonavir than Aluvia (Figure S3) R7.

Two further acid results point to lopinavir as the cause. Adding 0.8 mg/mL lopinavir to a 200 µg/mL ritonavir solution lowered free ritonavir by 10%, to about 180 µg/mL, after 1 h (data not shown), which the authors read as a small amount of ritonavir precipitation induced by lopinavir ("indicating") R11·C16. And at a fixed 50 mg of ritonavir per vessel, co-granules with less lopinavir released more ritonavir (Figure 8): about 83 µg/mL at 2 h for 4:1 LPV:RTV, about 97 µg/mL for 1:1 and about 161 µg/mL for 1:4 R13. The authors write that this "confirms" that lopinavir intimately mixed with ritonavir retards ritonavir release C27. The polymer and excipient amounts in the 1:1 and 1:4 granules are not printed, so total drug and possibly the drug:polymer ratio change along with the LPV:RTV ratio X11.

Two attempts to fix the co-granule failed. Lowering the drug loading relative to polymer from 50 to 20 wt% gave "no overall improvement" (data not shown) R19·C25. Surfactants did not raise ritonavir release (Figure 7): with 0.1% Tween 80 ritonavir reached about 83 µg/mL at 2 h, which the authors judge similar to no surfactant; with 5% Span 20, the surfactant in Aluvia, it reached only about 37 µg/mL ("seemed to reduce"); lopinavir stayed at about 16 µg/mL with either R12·C26. Figure 7 itself has no surfactant-free control, and the same 4:1 co-granule gave about 96 µg/mL in Figure 4 and about 83 µg/mL in Figure 8, so the Tween result lies within the spread between runs X10.

What did work was keeping the drugs apart (Figure 9). Separately granulated ritonavir and lopinavir, blended and compressed into one tablet, gave about 134 µg/mL ritonavir at 15 min and about 160 µg/mL at 2 h; compressed as a bilayer tablet they gave about 135 µg/mL at 15 min and about 188 µg/mL from 60 min R14. Both are close to or above Aluvia's about 169 µg/mL, and lopinavir again stayed at about 19 µg/mL R14·R6.a. The abstract states that "comparable release could be achieved" this way C28. In the discussion these data are cited as Figure 8 rather than Figure 9 X1.

Table B. Dissolved ritonavir (µg/mL) in 0.1 N HCl at 37 °C, read from Figures 4, 7, 8 and 9 (approximate); complete release would give 200 µg/mL R6·R12·R13·R14.

When the pH rises, ritonavir falls below its own amorphous solubility

In the two-stage test with Aluvia (Figure 5), ritonavir reached about 84 µg/mL after 30 min in acid; after neutralization it fell to about 4.6 µg/mL at 45 min and stayed at about 4 µg/mL to 2 h, while lopinavir dipped from about 17 to 12 µg/mL and recovered to about 16 µg/mL R8. The authors note that the modest dilution cannot account for the drop, and that the remaining ritonavir is at a level similar to that from the co-granule at pH 6.8 C18. Two details of this experiment are loose: the Figure 5 caption gives the acid stage as 250 mL, whereas the methods give 200 mL, brought to 250 mL by the additions; and the 750 mg of NaOH (18.75 mmol) is slightly less than the 20 mmol of HCl in 200 mL of 0.1 N acid, so the final pH depends on the added buffer and was not reported as measured X4·X5.

"At first glance", the authors write, one might expect high supersaturation followed by rapid crystallization C19. The excipient-free drug solution tests this (Figure 6). On neutralization a precipitate formed; ultracentrifuged ritonavir fell from about 97 to about 16 µg/mL and lopinavir from about 19 to about 8 µg/mL, and the precipitate showed no birefringence under polarized light R9·R10. The authors take this to "support" a non-crystalline ritonavir precipitate C20. Because the remaining ritonavir, about 16 µg/mL, is well below the amorphous solubility of ritonavir alone (31 µg/mL), they conclude that ritonavir and lopinavir are mixed in the precipitate when both are un-ionized ("suggesting"), lowering the maximum supersaturation as in their earlier work C21·M1. One oddity: the solution was made with a nominal 400 µg/mL of lopinavir, yet the "initial" lopinavir bar is only about 19 µg/mL, close to its amorphous solubility; how the solution was prepared, and whether lopinavir had already separated before neutralization, is not stated X6.

Table C. Dissolved drug (µg/mL) before and after the pH 1.2 to 6.8 shift, read from Figures 5 and 6 (approximate); amorphous solubilities measured at pH 6.8 R8·R9·M1·M2.

After acid, the surface of a drug–drug glass looks more like lopinavir

The infrared spectra of the two amorphous drugs are very similar (Figure 10). Ritonavir has an extra carbonyl band at 1704 cm⁻¹ next to the shared bands near 1630 cm⁻¹, and bands at 1229 and 878 cm⁻¹ that lopinavir lacks R15. In the 200:50 melt-quenched blend these ritonavir bands are hard to see, showing only as a small shoulder at 1718 cm⁻¹. After 1 h in 0.1 N HCl the shoulder weakened and the ratio of the 1630 and 1718 cm⁻¹ bands rose from 7.7 to 11.2, with other small changes marked by arrows; the spectrum became slightly more lopinavir-like R16. The authors read this as preferential loss of ritonavir from the surface, "consistent with" lopinavir enrichment of the surface, and add that mass balance on the acid dissolution data, where relatively more ritonavir than lopinavir dissolves, points the same way C29·C30. The text places the arrows in Figure 9; they are in Figure 10 X1.

The bench-top granules flow and compress

Granules from the bench-top granulator had acceptable flow by the usual indices, a narrow size distribution and few fines, and gave intact tablets of low friability (Table D) R17·R18·C31. The assay found 8.4% lopinavir and 2% ritonavir, about 10% drug in total, which the methods describe as the target loading R17. Note by the AI writer (not discussed in the paper): the full batch, 15 g of drug in 90 g of solids, would correspond to about 16.7% drug; since the solution was added only until the powder was fully granulated, the volume actually added is not stated X15.

Table D. Bench-top granules and 150 mg tablets (text, p.6) R17·R18.

Discussion

How the authors put the findings together

The authors offer one explanation per medium, both starting from the same fact: in the co-granule, ritonavir and lopinavir share one amorphous phase C38. The chain below follows their discussion; the branch marked good is their remedy.

Neutral pH. Here the explanation is the mutual reduction of amorphous solubility described above, which the authors state without hedging and which, in their words, explains the data of Figure 3 C38·C39.

The pH shift. The situation is more complex when the pH starts low and then rises. Ritonavir dissolves to a high concentration while ionized; when the pH rises it loses its charge and becomes supersaturated, and once the un-ionized concentration exceeds the amorphous solubility, liquid–liquid phase separation forms amorphous nanodroplets. With both drugs in solution, ritonavir and lopinavir mix in these nanodroplets, which "will result in" the same suppression of amorphous solubility C40. The fall of free ritonavir "can be attributed to three factors": the 25% dilution, the change in ionization that takes ritonavir over its amorphous solubility and forms a drug-rich phase, and the mixing with lopinavir in that phase C41.

Acid. Why should ritonavir lag in acid, where it is ionized and the medium is a sink? The authors first point to the drug:polymer ratio: differences in this ratio "most likely" play a part, together with the solubility suppression C42. Their reasoning runs as follows. Aluvia, which turned milky at pH 6.8, "appears" to release in the polymer-controlled regime, with nanodroplets C44. Since ritonavir–PVPVA ASDs form nanodroplets only below 25% drug loading (Indulkar et al. 2019), Aluvia's total drug loading is "likely" in that range or lower C45·R22. The co-granules, by contrast, carry 50% drug relative to polymer, chosen to keep the tablet weight reasonable, and are expected to release in the drug-controlled regime C46. Amorphous ritonavir alone would still release fast and completely in acid, they argue, because it ionizes; but with lopinavir present the polymer is "expected" to leave first, enriching the surface in drug; ritonavir, the more soluble drug, then dissolves preferentially, the surface becomes rich in lopinavir, which does not ionize and dissolves poorly in acid, and the remaining ritonavir must diffuse through this lopinavir-rich layer (Figure 11, the "Proposed mechanism") C46·C47. As support they cite the higher ritonavir release from ritonavir-rich co-granules (the text names Figure 7, but the data are in Figure 8), the infrared surface data and mass balance C48·C29·C30·X1.

Note by the AI writer (not discussed in the paper): two of the authors' own results weigh on this chain. First, the 20 wt% co-granule, below the 25% limit at which ritonavir–PVPVA nanodroplets form, did not improve release (data not shown), which does not fit the drug:polymer explanation X9·R19. Second, the ratio trend of Figure 8 is also what the composition-dependent solubility reduction would predict, so it does not by itself separate that mechanism from the surface barrier X8.

The remedy and where the method applies

Since lopinavir mixed with ritonavir slowed ritonavir release "even under favorable release conditions", the authors conclude that "the best approach appears to be" granulating the two drugs separately and then either mixing with a disintegrant before compaction or making bilayer tablets; "clearly", they write, this lets each drug dissolve independently and better exploits ritonavir's ionization in acid C49. In their conclusions they state that this "largely resolved" the depressed ritonavir release C51.

On the method itself, the authors call organic-solvent wet granulation "a possible alternative" to hot melt extrusion and spray drying for slowly crystallizing compounds such as ritonavir and lopinavir, "albeit with certain caveats" C32. The main caveat is tablet size: a relatively large excipient mass is needed to granulate the drug–polymer solution without forming a sticky mass, so the approach is "likely to be best suited for low dose formulations", or for drugs and polymers very soluble in the solvent (conclusions: "appears best suited for low or moderately low dose drugs") C33. The burden can "potentially, at least partially" be reduced with excipients that take up more solvent (in preliminary work, more MCC allowed a higher drug loading), and fluid-bed granulation, where rapid evaporation tolerates more solvent, is "a possible alternative" C34·R21. The drug must also resist crystallization while the solvent evaporates, so slowly crystallizing compounds of higher molecular weight and more complex structure suit best, and adequate drying is needed to remove the organic solvent C35. Because granules can simply be mixed with a disintegrant and compressed, the route "can potentially" reduce the number of processing steps C37. The abstract sums this up: solvent granulation "may be a viable method" to make ASDs of low-dose drugs with low crystallization tendencies C53. The closing statement is broader: "Clearly the phase behavior of fixed dose combinations of amorphous formulations is highly complex" and needs to be considered in formulation design C52.

What it suggests for formulation

Note by the AI writer (not discussed in the paper): for an amorphous combination product, the useful question is whether the drugs end up in one amorphous phase. If they do, the minor drug's free concentration can fall sharply at neutral pH, and even in acid, where a weak base like ritonavir should dissolve freely, release can slow. Here, copying the reference product's surfactant or lowering the drug loading did not fix this, whereas putting the drugs in separate granules or separate tablet layers did R12·R19·R14. The method itself trades equipment for excipient: at the assayed 10% drug, the combined 250 mg dose would need granules weighing about 2.5 g R17.

Limitations

Stated by the authors

The authors present the method "albeit with certain caveats": the large excipient mass and tablet size, the need for drug and polymer to dissolve well in the solvent, the need for drugs that resist crystallization during solvent evaporation, and adequate drying to remove the organic solvent C32·C33·C35. They describe the stability result as resistance to crystallization "for this time period", one month C9.

The AI writer's own judgement

• The surface barrier was not seen on the granules. The infrared evidence comes from a melt-quenched lopinavir/ritonavir glass without PVPVA or excipients, not from the granules whose release is explained X7·C47.
• Precipitate composition was inferred. Mixing of the two drugs in the precipitate or nanodroplets after the pH shift is inferred from concentrations; the precipitate was checked only for birefringence C21·C40·R10.
• Replicates are not stated. No figure states the number of replicates or what the error bars show, and "significantly" is not backed by a test X12.
• Figure cross-references are wrong in four places. The acid comparison is cited as Figure 3 (data in Figure 4), the infrared arrows as Figure 9 (Figure 10), the ritonavir-rich co-granules as Figure 7 (Figure 8) and the separate-granule results as Figure 8 (Figure 9) X1.
• Small numerical inconsistencies. Ritonavir's amorphous solubility is given as 31 and as about 32 µg/mL; the co-granulated ritonavir plateau is 4–5 µg/mL in the text and about 4 in Figure 3; the pH-shift volumes differ between caption and methods, and the final pH was not measured X2·X3·X4·X5.
• Unsourced inputs and units. The crystalline solubilities behind all supersaturation values have no stated source; the granulator tip speed (94 m/sec) cannot be checked, and tablet hardness is printed in kPa X14·X16.
• Scope. One drug pair, one polymer, in vitro dissolution only, one month of open-dish stability by XRPD, and no residual-solvent data; the supporting information was not available for this reading R2·C35.

Conclusion

The paper shows that a plain organic-solvent wet granulation, drug and PVPVA dissolved in DCM:MeOH and poured onto lactose and MCC, gives X-ray amorphous ritonavir and lopinavir that stay free of crystals for a month at 40 °C/75% RH, with granules that flow and compress into robust tablets R1·R2·R17·R18·C50. Its more lasting lesson concerns combination products: when the two drugs share one amorphous phase, ritonavir, the minor component, dissolves to only about 4 µg/mL at pH 6.8 instead of about 30, and to about 96 instead of about 169 µg/mL from the marketed tablet in acid, while granulating each drug separately restores release to about 160–188 µg/mL R4·R6·R14·C51. The authors explain the neutral-pH loss by mutual reduction of amorphous solubility and propose a lopinavir-rich surface barrier for the acid case; the first rests on their earlier thermodynamic work, the second on a polymer-free model glass C38·C47·X7.