Where commercial amorphous tacrolimus capsules stop dissolving: the amorphous solubility as a ceiling for six marketed products

Background and motivation

Many new drug candidates dissolve poorly in water, and formulators have several ways to raise the amount that reaches solution: salts, cocrystals, micellar solubilization, cyclodextrin complexation and amorphous solid dispersions (ASDs) (refs 1–11). The introduction separates these approaches into two kinds. Salts and cocrystals can produce supersaturated solutions that later crystallize to the free form. Micelles and cyclodextrins instead raise the equilibrium solubility of the crystal without supersaturating the solution, and because passive absorption depends on the free drug concentration, a higher total concentration of this kind need not raise the rate at which drug crosses a membrane. C1 Amorphization works differently: it raises the chemical potential of the drug, so under non-sink conditions, and in the absence of crystallization, the dissolving solid produces a solution with a higher free-drug concentration than the crystal can give. For this reason the authors argue that "it is important to test the in vitro dissolution behavior under non-sink conditions" for ASDs, citing Sun, Wen and Taylor (2016, ref 19). C2 Polymers are added to keep the amorphous solid physically stable and, in solution, to slow crystal nucleation and growth so that supersaturation lasts. C3

Earlier work from the same group set the frame for this paper. With various drugs, the group had found that there is a maximum concentration of molecularly dissolved drug in a supersaturated solution and that this maximum is the amorphous solubility. Above it, if the drug does not crystallize, the solution separates into two phases (liquid–liquid or glass–liquid phase separation, LLPS): a water-saturated, colloidal, drug-rich amorphous phase that is a supercooled liquid or a glass depending on its glass transition temperature, and a water-rich phase (Almeida e Sousa et al. 2015, ref 32). C5 That behaviour had been documented mostly for solutions made by adding a concentrated drug solution in an organic solvent to buffer. Few studies had examined model ASDs during dissolution, and, to the authors' knowledge, none had examined commercial products, which contain ASD particles together with other excipients. C6

Tacrolimus (also called FK-506) is a macrolide immunosuppressant with a molecular weight of 804 g/mol, used to prevent organ rejection. M1 One of the earliest commercial ASD products is a tacrolimus capsule introduced by Fujisawa Pharmaceuticals (now Astellas Pharma); the 0.5, 1 and 5 mg capsules were approved by the FDA in 1994 under the trade name Prograf, and several generic products followed. M9 For a generic, the authors write that "it is essential that the physicochemical characteristics", such as the maximum achievable supersaturation and product stability, match the brand to ensure bioequivalence over the shelf life (Petan et al. 2008, ref 23). C4 Crystalline tacrolimus is a monohydrate with a reported aqueous solubility of 0.7–1.3 µg/mL (refs 15, 24). R5 Its absorption in transplant patients is highly variable, with peak blood levels at 0.5–6 h and a mean bioavailability of 25% (ref 25), and permeability, intestinal and first-pass metabolism, P-glycoprotein efflux and food also affect it (refs 26–30). In beagle dogs the ASD improved the area under the plasma concentration curve by an order of magnitude over the crystal (Yamashita et al. 2003, ref 31).

The paper therefore asks whether the supersaturation that a commercial ASD product generates can be predicted from properties of the drug alone, chiefly its amorphous solubility and its crystallization kinetics in simple solution. The stated hypothesis is that the maximum free-drug concentration produced by dissolving commercial tacrolimus formulations is dictated by the amorphous solubility of tacrolimus. C7 Four questions frame the experiments: is crystallization detected during non-sink dissolution; how similar is the dissolution of products made by different companies; do the products dissolve to reach the amorphous solubility; and do they exceed it, forming a new colloidal phase? C7 The study used one innovator product (Prograf) and five generics, all 5-mg capsules. M6–M11

Background concepts

Amorphous solubility and liquid–liquid phase separation. A crystal in contact with water sets the familiar equilibrium solubility. An amorphous solid has a higher free energy, so it can produce a higher concentration of dissolved drug, but only up to a limit: the amorphous solubility. Beyond it, extra drug does not stay dissolved as single molecules; the solution splits into a water-rich phase, whose drug concentration stays at the amorphous solubility, and a drug-rich phase that appears as small droplets or particles. C5·C25 The paper does not compute the amorphous solubility from theory; it notes that it can be calculated from the crystal solubility, the free-energy difference between crystal and amorphous forms (from melting temperature, enthalpy and heat capacity) and the water content of the amorphous material at 100% relative humidity (ref 32), or measured directly. C26·E5 It measures it directly.

The solubility advantage. The size of the gain over the crystal is expressed as a ratio, stated in words in the paper. E1

In plain terms: the ratio says how many times more drug the amorphous form can hold in solution than the crystal. For tacrolimus the paper uses the ultracentrifugation value of the amorphous solubility, 49 µg/mL, and the crystalline solubility, 1.4 µg/mL, and reports a ratio of approximately 35; the division reproduces this (49/1.4 = 35.0). E1·R4.a·R4.e·R16 Other drugs span about 5 to 65 (ref 32). R17

Sink and non-sink conditions. A dissolution test is "sink" when the medium could dissolve all of the drug while staying well below the solubility limit, so the limit never matters. It is "non-sink" when it cannot, so the solution reaches the limit. One 5-mg capsule in 40 mL gives a total of 125 µg/mL if everything dissolved, about two and a half times the amorphous solubility; the authors note that 100 mL would be needed for exact sink conditions with respect to the amorphous solubility. P11·C29

Membrane flux as a measure of free drug. Concentration alone cannot tell free drug from drug held in micelles or in a colloidal drug-rich phase. The paper uses a second read-out, the rate at which drug crosses a dialysis membrane into a sink, and relies on a relation it states in words (Raina et al. 2014, ref 36). E2·C36

In plain terms: the driving force for crossing the membrane is the thermodynamic activity of the free drug on the donor side, not the total amount of drug in the donor. Drug held in micelles adds to the concentration but not to the flux. Above the amorphous solubility, adding more drug only makes more drug-rich phase, the free drug stays at the amorphous solubility, and the flux stops rising. C34·C36·C37 A flux plateau therefore marks the highest activity the drug can reach in that medium, and a product solution that gives the plateau flux contains drug at that maximum activity.

Detecting the new phase. The five methods used here each detect the onset or the boundary of the drug-rich phase in a different way: light scattering at a wavelength where the drug does not absorb rises when particles form; the fluorescence probe pyrene reports a less polar environment once a hydrophobic drug-rich phase appears (its I3/I1 emission peak ratio rises); a fine filter or an ultracentrifuge separates the drug-rich phase so that the remaining solution can be assayed; and a dissolving amorphous film levels off at the amorphous solubility if it does not crystallize. P4·P7·P6·P5·P8 Nanoparticle tracking analysis (NTA) sizes individual particles from their Brownian motion through the Stokes–Einstein equation, which the paper names but does not print. E4·P13

Methods in detail

The design has two halves. The first fixes a reference: the crystalline and amorphous solubility of pure tacrolimus, the time a solution at the amorphous solubility survives before crystallizing, and the membrane flux of pure-drug solutions across a range of concentrations. The second dissolves the products under non-sink conditions and compares their plateau concentration, their particles and their flux with that reference. All solution work used 5 mM sodium phosphate buffer at pH 4.5 and 37 °C. A1·A2·A3·M12

The six products

Six brands of 5-mg tacrolimus capsules were bought from the Purdue Pharmacy and analysed within their expiry dates. M6–M11 Their inactive ingredients were taken from the package inserts and are summarised in Table A. All contain lactose, croscarmellose sodium (CCS) and magnesium stearate; HPMC is listed for five of them and is "most likely present" in the sixth (Dr. Reddy's), whose insert does not mention it. M6–M11·C9 Only the Mylan product lists a surfactant, sodium lauryl sulfate. M10 The average powder weight per capsule differs more than twofold, from 99.8 mg (Mylan) to 242 mg (Sandoz), for the same 5 mg of drug. M6–M11 The drug loading of the ASD inside each product is not given; the authors later cite "a report from Fujisawa" that the tacrolimus ASD has a 50% drug loading. R19

Table A. The six products (from Table 1; MH monohydrate, mag. st. magnesium stearate). Each capsule contains 5 mg tacrolimus. M6–M11·C9

Two details of Table 1 deserve a flag. Prograf and Sandoz are printed with the same lot number, 047354, both in the text and in the table, so one of the two is presumably a misprint. X1 And the claim later in the discussion that HPMC is present in all the studied products rests, for Dr. Reddy's, on the footnote's inference rather than on the insert. X3

Solid-state characterization

Capsule contents and the pure excipients (tacrolimus in amorphous and crystalline form, CCS-Na, HPMC, lactose monohydrate and magnesium stearate) were examined by attenuated total reflectance FTIR (128 scans, 600–4000 cm⁻¹) and by X-ray powder diffraction (Cu-Kα, 5–40° 2θ, step 0.02°). The lactose form in each product was identified by comparison with reference patterns for β-lactose anhydrous and α-lactose monohydrate from the Cambridge Crystallographic Database. P2·P12 About 5 mg of crystalline tacrolimus monohydrate was heated in a DSC to 160 °C until it melted, cooled at 15 °C/min to 0 °C and reheated at 10 °C/min to see the glass transition and whether the glass recrystallized. P1

Crystalline and amorphous solubility

The crystalline solubility was measured by stirring excess monohydrate in buffer at 37 °C for 48 h, filtering through a 0.45-µm glass syringe filter (no drug adsorption was observed on it) and assaying by HPLC. P3 The amorphous solubility was measured five ways, all at 37 °C in the same buffer. A2

• UV extinction. A tacrolimus solution in acetonitrile was added by syringe pump to 10 mL of stirred buffer while the scattering at 350 nm, where tacrolimus does not absorb, was recorded; the concentration at which scattering rose was taken as the onset of the new phase. P4
• Pyrene fluorescence. Small aliquots of 10 mg/mL tacrolimus in acetonitrile were added to 10 mL of buffer containing 1 µM pyrene; the ratio of the emission peaks at 383 nm (I3) and 373 nm (I1) was plotted against tacrolimus concentration, and its inflection taken as the onset of LLPS. The methods state that the ratio is around 0.51 without a tacrolimus-rich phase and changes to around 0.61 once it forms. P7
• Filtration. A solution containing the drug-rich phase was passed through a 0.22-µm glass fibre filter, which retained the phase-separated particles, and the filtrate assayed. P6·R8
• Ultracentrifugation. A 50 µL portion of 20 mg/mL tacrolimus in acetonitrile was added to 15 mL of buffer containing 10 µg/mL HPMC, stirred for 30 min and ultracentrifuged at 210053 g; the supernatant was assayed. P5
• Film dissolution. Five milligrams of tacrolimus were melted at 140 °C between a glass slide and a cover slip, cooled, and the exposed film dissolved in 40 mL of buffer at 300 rpm while a UV probe followed the absorbance at 204 nm (corrected for scattering at 350 nm). The experiment was repeated until a run reached a plateau without crystallizing; the plateau was taken as the amorphous solubility. P8

Crystallization induction time

A 50 µg/mL tacrolimus solution, about the amorphous solubility, was made by adding 125 µL of 20 mg/mL stock in acetonitrile to 50 mL of buffer. The absorbance at 204 nm was followed with and without 50 µg/mL HPMC, in triplicate, to find when crystallization began to lower the concentration. P10

Non-sink dissolution of the products

The entire contents of one capsule (5 mg tacrolimus) were added to 40 mL of buffer at 37 °C, stirred at 300 rpm, in triplicate. pH 4.5 was chosen because it is the pH of the USP monograph for tacrolimus capsules. P11·C8 The 40 mL volume was chosen to represent the baseline gastric volume: Mudie et al. (ref 38) found that after a 240 mL drink of water the mean gastric volume falls within minutes and reaches a baseline of about 40 mL in approximately 30 min. R18·C29 Samples of 300 µL were filtered through 4-mm, 0.45-µm PTFE syringe filters and assayed by HPLC; the medium removed was replaced. P11 The HPLC method (C-18 column, acetonitrile:water 70:30, 60 °C, 210 nm) was linear from 0.1 to 100 µg/mL with a limit of quantification of 0.2 µg/mL. P15

Particles and mass balance

To look for drug-rich particles, a 60 µg/mL pure-drug solution, just above the amorphous solubility, was made by solvent shift, filtered through a 0.7-µm glass filter and examined by NTA; the 0.7-µm filter was used because the 0.22-µm filter retained the phase-separated particles. P13·R8 Filtrates of capsule contents dissolved for 20 min in 40 mL, and in 100 mL, were examined the same way. P13 For a mass balance, the contents of a capsule were dissolved for 2 h in 40 mL, 2 mL was filtered through a 0.7-µm glass filter and assayed, and 2 mL of acetonitrile was then passed through the same filter to recover any drug retained on it. P14 The expected values follow from the dose: 5 mg in 40 mL is 125 µg/mL in total, of which 50 µg/mL should be dissolved if the system stops at the amorphous solubility and 75 µg/mL should remain undissolved. E3

Membrane flux

A side-by-side diffusion cell held two compartments of 30 mL each, separated by a regenerated cellulose dialysis membrane with a molecular-weight cut-off of 12–14 kDa (pores of about 2–3 nm) and an exposed area of 7.065 cm², at 37 °C, in triplicate. P9·M15 Pure-drug donor solutions were made from a 20 mg/mL acetonitrile stock to give 20–100 µg/mL; for the products, the capsule contents were dissolved for an hour in 30 mL of buffer and then added to the donor compartment. The rise in receiver concentration was followed with a 2-cm UV probe. P9·A3 The paper does not say whether the undissolved solid was transferred with the product solution, nor what the donor concentration was. X7

Experimental analysis

The products contain amorphous tacrolimus in a lactose matrix

None of the six XRPD patterns shows the peaks of tacrolimus monohydrate (Figure 1), which the authors describe as "consistent with the expectation" that the drug is amorphous in all products. R1·C11 Five products share the same set of peaks with different intensity ratios, which the authors attribute to different excipient amounts; the Panacea pattern is different and matches β-lactose anhydrous, and the Mylan pattern has a small β-lactose peak near 10.5°. R1·C10 The FTIR spectra of the six products are almost identical except Panacea's, and the tacrolimus bands are barely visible (Figure 2), so the authors conclude that lactose is the main component of the capsules. R2·C12 At 5 mg of drug in 99.8–242 mg of powder, this is expected. M6–M11

The DSC trace of the monohydrate (Figure 3) shows a broad dehydration endotherm between 60 and 120 °C, followed by melting of the anhydrate with an onset of 121 °C (the trace is annotated 121.62 °C, 37.14 J/g, peak 129.25 °C). R3·M1 The cooled melt was amorphous and on reheating showed a glass transition with an onset of 75 °C (75.53 °C on the trace) and no crystallization. R3·M1 The authors read the high Tg and the lack of recrystallization as a sign that tacrolimus "has good glass-forming ability and stability" ("suggests"). C13

The ceiling: an amorphous solubility of about 50 µg/mL

Crystalline tacrolimus monohydrate dissolved to 1.4 (0.1) µg/mL, which the authors find "in good agreement" with the 0.67–5 µg/mL reported in similar media. R4.a·R5·C14 The five amorphous-solubility methods agreed closely (Table B, from Table 2). R4

Table B. Crystalline and amorphous solubility of tacrolimus in 5 mM phosphate buffer, pH 4.5, 37 °C (from Table 2; the right column summarizes the methods section). R4·P3·P4·P5·P6·P7·P8

The raw signals behind two of these values are shown in Figure 4. The scattering stayed near zero up to about 35 µg/mL and rose steadily from about 40 µg/mL, reaching about 0.91 at about 118 µg/mL; the text places the inflection at around 46 µg/mL. R6 The pyrene I3/I1 ratio was about 0.518–0.521 from 10 to 40 µg/mL, then rose to about 0.527 at 50, 0.540 at 70 and 0.588 at 100 µg/mL (all read from Figure 4b, approximate). R7 The methods text says the ratio changes "to around 0.61" once the drug-rich phase forms, but the highest plotted value is about 0.59. X2 The authors interpret both curves as the formation of a tacrolimus-rich phase between 40 and 50 µg/mL. C15

Using the ultracentrifugation value, the amorphous-to-crystalline solubility ratio is approximately 35, near the middle of the 5–65 range reported for other drugs, which the authors call a "significant water solubility advantage". R16·R17·C16 Note by the AI writer (not discussed in the paper): the ultracentrifugation medium alone contained 10 µg/mL HPMC; with the buffer-only values of 46–50 µg/mL the ratio would be about 33–36, so the conclusion does not change. X9

How long does a solution at the ceiling survive?

A 50 µg/mL solution made by solvent shift began to crystallize within 3 h (Figure 5a): read from the figure, the concentration held at about 50 µg/mL for about 150 min, fell to about 32 µg/mL at about 240 min and to about 12 µg/mL at about 350 min, and was about 8 µg/mL at the last point, near 680 min. R9 With 50 µg/mL HPMC (Figure 5b) the onset moved to 6–8 h: the concentration held at about 50 µg/mL until about 400 min and was about 40 µg/mL at 800 min, with a wide spread between replicates (standard deviation band about 27–52 µg/mL). R9 The authors say these results "suggest that this supersaturation can be maintained for a biologically relevant period of time", in particular with HPMC, which they describe as present in all the products. C17 The test used pure drug in buffer; the product solutions themselves were followed only for 2 h. R10

All six products dissolve to the ceiling

Figure 6 holds the central result. In 40 mL, the products differed at first: at 5 min the concentrations ranged from about 26 µg/mL (Prograf) to about 41 µg/mL (Accord). R10.a·R10.e The authors describe Prograf as dissolving slowest and Accord as appearing to dissolve fastest, but all products reached their maximum concentration, equivalent to the amorphous solubility, within approximately 45 min. C18 Table C gives the readings for the first 30 min.

Table C. Tacrolimus in solution (µg/mL) after adding one 5-mg capsule to 40 mL of pH 4.5 buffer at 37 °C, mean of n = 3, read from Figure 6 (approximate). "Hidden": the Prograf markers lie under the others after 20 min. R10

From 45 to 120 min every readable point lies between about 46 and 50 µg/mL. R10 The profiles showed a plateau after 60 min with no fall in concentration, which the authors read as complete dissolution and no crystallization within the test, "in agreement with" the induction-time results. C30 The maximum concentration was "remarkably consistent" at around 50 µg/mL for all six products, the same as the independently measured amorphous solubility. C31 From this the authors infer that the formulation components do not affect the amorphous solubility, "at least not at the concentrations present following dissolution". C31 They place this against earlier findings: polymers used in ASDs have had little effect on the amorphous solubility (Jackson et al. 2016, ref 39), small molecules that mix into the drug-rich phase lower it (Trasi and Taylor 2015, ref 40), and micellar surfactants raise it by solubilization (Raina et al. 2015, ref 35). C32 Lactose, the main excipient, is "not anticipated" to solubilize tacrolimus or to mix with its amorphous phase. C33

Do the products overshoot the ceiling?

A product could in principle exceed the amorphous solubility in the vessel, either by forming drug-rich particles through LLPS or by shedding undissolved fragments. The pure-drug reference shows what such particles look like: a 60 µg/mL solution, filtered through 0.7 µm, contained a second scattering phase with a mean size of 220 nm (Figure 7), which the authors take as "confirming" LLPS. R11·C19 Filtrates of capsule contents dissolved for 20 min in 40 mL also contained particles, with a mean size of 210 nm, but particles were present too after dissolution in 100 mL, which the authors treat as sink with respect to the amorphous solubility. They therefore attribute the particles to CCS-Na passing the filter, CCS-Na being the only other insoluble material in the solution ("This suggests"). R12·C20 As a result, it "was not possible to conclusively determine" by NTA whether the products formed drug nanoparticles. C39

The filtration test gave a clearer answer. After 2 h in 40 mL, the 0.7-µm filtrates held between 45 and 50 µg/mL "for all filtrates", and the acetonitrile rinse of the filter recovered about 70 µg/mL against an expected 75 µg/mL (Table D). R13·E3 The authors conclude that no drug nanoparticles formed during dissolution passed the filter and that the system is close to mass balance, the small gap being "mostly due to" sampling errors and medium left in the syringe. C21·C22

Table D. Mass balance after dissolving one capsule for 2 h in 40 mL (from the text, p.6). The product(s) and number of replicates are not stated. R13·E3·X6

Because the 0.7-µm filter lets the drug-rich particles through, an overshoot would have raised the filtrate above 50 µg/mL; it did not. On this basis it "appears" that the products dissolve to the amorphous solubility rather than beyond it, a "supposition" in the authors' word. C40

The product solutions reach the maximal membrane flux

For pure-drug solutions, the mass flow rate across the membrane rose with donor concentration and then levelled off (Figure 8, Table E). R14 The text describes the rise as linear and the plateau as starting above 50 µg/mL; the figure caption says "at or above 50 µg/mL". R14 The authors read the plateau as the chemical potential of the drug reaching its maximum, consistent with reaching the amorphous solubility and forming a new phase above 50 µg/mL. C23

Table E. Mass flow rate of pure-drug solutions made by solvent shift, mean of n = 3, all values about, read from Figure 8 (approximate); standard deviations up to about ±0.0007 µg/min. R14

The solutions from the dissolved products gave mass flow rates between 0.0048 and 0.0054 µg/min, shown as the band in Figure 8, which the authors find "in good agreement" with the pure-drug plateau. R15·C24 Their conclusion: the driving force for membrane transport reached its maximum value, corresponding to the amorphous solubility, so "the amorphous solubility advantage is achieved", and no solubilization of the drug by the formulation components is observed, "based on the mass flow rate measurements". C38 This also covers the Mylan product with its surfactant, although the paper reports the product flux only as a single range for all six. M10·R15

Note by the AI writer (not discussed in the paper): read from Figure 8, the pure-drug flux at 20 µg/mL (about 0.0027 µg/min) is about 57% of the flux at 50 µg/mL (about 0.0047 µg/min), not 40% as strict proportionality would give, and a straight line through the points from 20 to 50 µg/mL meets the axis near 0.0013 µg/min. The rise is linear, as the text says, but not proportional to concentration, so the plateau, rather than the slope below it, is the reliable read-out of activity here. X8

Discussion

How the authors connect the results

The argument runs as a short chain. The amorphous drug dissolves until the solution reaches the amorphous solubility; crystallization is slow enough for that level to hold through the test; and at that level the free drug has its highest activity, so membrane flux levels off. C27·C30·C37 What the excipients could change is where the ceiling sits and what the measured concentration means. The flow below sets out the cases the authors discuss.

Two alternatives would have shown up as a mismatch between concentration and flux. If an excipient, such as the surfactant in the Mylan product, held drug in micelles, the total concentration could rise above the amorphous solubility while the flux stayed at the plateau; if a small-molecule excipient dissolved in the drug-rich phase, the amorphous solubility, and with it the plateau and the flux, would fall. C34·C35·C32 Neither happened: all six products reached about the same concentration as the pure drug and gave the pure-drug flux. R10·R15·C38 The authors therefore conclude that, "at least in the case of tacrolimus", the maximum concentration reached during dissolution and the resulting solution thermodynamics of marketed multicomponent amorphous formulations "can be predicted from a knowledge of the amorphous solubility"; the abstract states that the study "demonstrates a link" between amorphous solubility and in vitro formulation performance. C45

Why the products stop at, and not above, the ceiling

Some ASDs, especially at low drug loading, release drug as nanoparticles or droplets that can carry the total drug in the vessel above the amorphous solubility. The authors connect the absence of such an overshoot here to the drug loading. Alonzo et al. (2011, ref 41) reported that higher drug-loading ASDs may not generate nanospecies on dissolution; a report from Fujisawa gives the tacrolimus ASD a 50% drug loading; and so the authors "believe" that the dissolution of the commercial tacrolimus formulations is drug controlled rather than polymer controlled, in the sense of Simonelli et al. (1969, ref 42). C41·R19 This is an explanation, not a measurement: the loading is not referenced in the reference list, it applies to the originator's product, and neither drug loading nor polymer release was measured for any of the six products. X4

What it suggests for formulation and quality testing

The authors' own recommendation is to characterize not only the crystalline solubility but also the amorphous solubility and crystallization kinetics of a drug in the medium of interest, and to use them as reference performance metrics against which a formulated product is compared: "This highlights the importance of characterizing not only the crystalline solubility but also the amorphous solubility". C46 In this view, a product that reaches the amorphous solubility and holds it for longer than the test is performing at the thermodynamic limit, and a product that falls short, or whose flux falls short, points to a formulation problem.

Note by the AI writer (not discussed in the paper): for these six tacrolimus products the ceiling test does not separate brand from generics, since all end at the same concentration and flux; the differences lie in the first 20 minutes, where Prograf, at about 26 µg/mL at 5 min, trails Accord at about 41 µg/mL. R10.a·R10.e A filtered concentration alone also cannot show whether the solubility limit was exceeded, because about 200-nm drug-rich particles may pass a 0.45-µm filter; in this paper the 0.7-µm filtration with mass balance and the flux measurement are what carry that inference. X10·R13·R15

Limitations

Stated by the authors. NTA could not settle whether the products form drug nanoparticles in 40 mL, by LLPS or by incomplete dissolution, because croscarmellose particles also appeared in the filtrates ("Unfortunately, it was not possible to conclusively determine"). C39 At least one product contains a surfactant, and the interaction of tacrolimus with the other formulation ingredients is unknown. C35 The conclusions are framed for tacrolimus ("at least in the case of tacrolimus") and for the conditions evaluated. C45·C42

The AI writer's own judgement.

• The central claim is well supported: five methods agree on the amorphous solubility, all six products plateau at that value, and the flux test reads free-drug activity directly. R4·R10·R14·R15
• The product dissolution data and the pure-drug flux are read from plots, and the product flux is reported only as one range for all six, so product-to-product differences in flux cannot be judged. R10·R15
• The 100 mL control used to blame croscarmellose for the particles corresponds to a nominal 50 µg/mL, at the amorphous solubility rather than below it, and no croscarmellose-only or placebo control is reported, so the identity of the 210-nm particles remains open. X5
• The product(s) and replicates behind the NTA and mass-balance tests are not stated, and the donor composition for the product flux tests is not described. X6·X7
• Smaller points: Prograf and Sandoz share a printed lot number; the pyrene end value in the methods (about 0.61) does not match the plot (about 0.59); and the ratio of about 35 uses an HPMC-containing medium. X1·X2·X9
• All work used 5 mM phosphate buffer at pH 4.5. Media with bile salts and phospholipids could solubilize tacrolimus and change both the total concentration and the free-drug activity, and the in vivo relevance of the 40-mL test was not examined here.

Conclusion

Six marketed 5-mg tacrolimus capsules, the innovator and five generics with different excipient lists and fill weights, all dissolved under non-sink conditions to about 46–50 µg/mL, the amorphous solubility of the pure drug measured five ways, and held it for 2 hours. R10·R4 Their solutions crossed a membrane at the same maximal rate as pure drug at or above that concentration, so the excipients neither solubilized the drug nor changed its amorphous solubility at these levels. R15·R14·C38·C31 Filtration and mass balance indicate that the products stop at the amorphous solubility rather than exceed it, while the drug-loading explanation for this remains the authors' belief. R13·C40·C41 For tacrolimus, then, the dissolution ceiling of a multicomponent amorphous product can be read from a property of the drug alone. C45