WO2010148288A2 - Pharmaceutical formulations with low aqueous levels of free unbound drug - Google Patents

Abstract

The complexation properties of cyclodextrins are combined with the partitioning properties of reversed lyotropic liquid crystalline phase microparticles to prepare pharmaceutical formulation dispersions with reduced levels of free unbound aqueous drug. Safe to inject formulations including, for example, propofol, include dispersions of reversed lyotropic liquid crystalline phase microparticles with propofol in an aqueous carrier together with cyclodextrins.

Description

PHARMACEUTICAL FORMULATIONS WITH LOW AQUEOUS LEVELS OF FREE UNBOUND DRUG

BACKGROUND

Aqueous dispersions comprising microparticles of reversed lyotropic liquid crystalline phase material and drug are known as pharmaceutical formulations, for example, in US 7,713,440 and US 6,482,517 to Anderson, and US Patent 5,531,925 to Landh et al. The drug is strongly partitioned in the reversed hexagonal and reversed cubic phase material, as opposed to the aqueous phase, of the formulation, and thus sequestered and protected. Nevertheless, in the case of certain drugs, at equilibrium, not insignificant amounts of drug remain in the aqueous phase of the formulation. This can be problematic for drugs or other actives which are subject to various forms of water- mediated degradation, such as hydrolysis and oxidation, and for drugs or other actives which in unbound free form are associated with undesired effects, such as pain on injection or local toxicity.

Cyclodextrins are cyclic oligosaccharides often used, in particular, in pharmaceutics to increase the aqueous solubility of drugs or Active Pharmaceutical Ingredients (APIs). In such an approach, a cyclodextrin is dissolved in aqueous solution, always at molar excess to the API and usually at high cyclodextrin concentration, very often 10-30% or more. Molar ratio of cyclodextrin to drug is typically on the order of 5: 1 CD:API (cyclodextrin to API). Several cyclodextrins are approved for intravenous injection, and injectable products containing a cyclodextrin compound are marketed at the time of this disclosure, all of which are clear aqueous solutions. When a drug, or other compound, is solubilized by complexation with a cyclodextrin as per the prior art, the complex is — in terms of strict thermodynamic criteria — truly solubilized in the water; that is, it is not to be considered as dispersed, and in the absence of other, dispersed material such a formulation is a true aqueous solution, not an emulsion or dispersion of any type. A number of publications have reported drug formulations in which drugs are complexed with cyclodextrins, and the complex in tum is encapsulated in the aqueous interior of liposomes. See, e.g., Loukas et al. [Int. J. Pharm. 162: 137 (1998)].

U.S. 5,571,534 to Jalonen et al. describes several formulations containing toremifene or tamoxifen or their desmethyl derivatives. In one formulation, POPC/cholesterol/DMPG liposomes were prepared incorporating a complex of toremifene citrate and 2-hydroxypropyl-β-cyclodextrin. An emulsion preparation of toremifene that did not contain any cyclodextrin was also prepared in the Jalonen patent disclosure, in accordance with the well-known solubility properties of cyclodextrins, which neither partition into emulsion droplets nor have sufficient solubility therein to transfer APIs into emulsion droplets.

Indeed, by virtue of cyclodextrins' defining property of increasing aqueous-phase solubility of hydrophobic or amphiphiϋc compounds, the addition of a cyclodextrin to a lipid-based formulation would be antithetical to the goal of increasing lipid-phase sequestration of any such compound. Phrased otherwise, in a lipid dispersion or emulsion containing a hydrophobic or amphiphilic API, the addition of cyclodextrin will tend to increase the fraction of API that is in the aqueous phase, measured at any instant in time.

U.S. Patent Application 10/585,31 1 to Takeda et al. describes fat emulsions comprising propofol and 2 to 20 w/v% of an oil and at least one compound selected from the group consisting of cyclodextrins, cyclodextrin derivatives and pharmacologically acceptable salts thereof. The oil is in most embodiments soybean oil.

Emulsions containing triglyceride fats such as soybean oil or medium-chain/long- chain triglyceride mixtures are known to cause a number of fat-related problems, not only in propofol formulations but more generally in pharmaceutical emulsion products. The fats, together with the water, provide the nutrients that promote microbial growth, and it is well documented that, for example, sedation with emulsion-based propofol over several days results in nosocomial infections in a significant proportion of cases. In addition, the administration of fat emulsions to pediatrics, in particular, can cause "propofol infusion syndrome", and in any case, the triglyceride load delivered during the course of treatment confounds efforts to maintain proper nutritional balance. Recently, Anderson in U.S. 7,713,440 has reported an injectable lipidic formulation of propofol that contains no triglycerides (nor diglycerides or monoglycerides). The formulation comprises particles of reversed liquid crystalline phase material containing propofol, wherein the particular liquid crystalline material used is that known in the art as a "reversed cubic phase", also known as an "inverted" cubic phase, or "Type H" cubic phase.

U.S. 7,138,387 to Pai et al. and U.S. 7,034,013 to Thompson et al. describe clear aqueous solution formulations of propofol solubilized with, respectively, hydroxypropyl- β-cyclodextrin and a sulfoalkylether cyclodextrin. The cyclodextrin to drug ratio (CD:API) in 7,034,013 lies between 1 : 1 and 5:1 , with exemplified compositions having CD:AP1 molar ratios near 3:1. In 7,138,387 the CD:API molar ratio is restricted to between 4:1 and 8:1. Bielen et al. [Anesth. Analg. (1996) 82:920] reported that an aqueous solution formulation of propofol using 20% hydroxypropyl-β-cyclodextrin (HPCD, or HPβCD) caused severe bradycardia when injected according to standard protocol. Trapani ct al. [Trapani G, Latrofa A, Franco M, Lopedota, A, Sanna E And Liso, G (1998) J. Pharm. Sci. 87(4):514] have solubilized propofol with a molar excess of hydroxypropyl-β-cyclodextrin and determined the complexation constant.

U.S. 5,814,330 to Puttcman describes a mucoadhesive, water-in-oil emulsion composition comprising a drug selected from the group consisting of antibacterial, antiviral, contraceptive, and antifungal agents, and a cyclodextrin or a derivative thereof in an amount from 10% to 70% by weight based on the total weight of the composition. The CD: API molar ratio in the case of itraconazole lies in the range of approximately 1 :1 to 350: 1.

Propofol, a widely used intravenous anesthetic agent, is well known to cause pain at the site of injection. In lipid formulations of propofol, and indeed in propofol formulations more generally, it is widely believed that the pain on injection is due not to the propofol that is partitioned into lipid particles, but rather to the free propofol in the aqueous phase of the emulsion or dispersion.

Brazeau et al. [Journal of Pharmaceutical Sciences 1998, 87(6):667] have published a review of drugs (and other injected medical compounds) that cause pain on injection (POI) in a significant fraction of patients. These include propofol, amoxicillin, penicillin G, sodium sulbactam, cephalosporins, amikacin, gentamicin, kanamycin, neomycin, streptomycin, tetracycline, pentamidine, oxarnniquine, clarithromycin, bleomycin, methotrexate, diazepam, midazolam, chlorpromazine, promethazine, etomidatc, methohexitone, botulinum toxin A, gallamine, methocarbamol, myochrysine, epinephrine, bradykinin, erythropoietin, follicle stimulating hormone, heparin, diclofenac, ketorolac, diatrizoate (contrast agent), edetate (EDTA), haemophilus influenza type B vaccine, phenytoin, polymyxin B sulfate, progesterone, testosterone, theotepa, trimethobenzamide, and vitamins A, D and K. In many, though not all, of these formulations, it is the aqueous-phase drug that is responsible, at least in part, for the POl. Pain on injection is most consequential in pediatric and geriatric patients, but even in healthy adults can lead to thrombophlebitis, nerve end neuromas, and other complications, in some cases so severe as to prevent the use of particular drug formulations. In the case of propofol, for example, several attempted formulations that had the advantage of low or no triglycerides — thus potentially reducing lipid load complications and infection rates particularly in use of propofol for sedation — were not acceptable pharmaceutically because of severe POI.

Indeed, the cyclodextrin formulations of propofol cited above, described in U.S. 7,138,387 to Pai et al. and U.S. 7,034,013 to Thompson ct al., which employ high concentrations of cyclodextrin, nevertheless still suffer from unacceptable pain on injection, in addition to problems with bradycardia. For example, taking the complexation constant between propofol and 2-hydroxypropyl-β-cyclodextrin (HPCD) from Trapani et al. (cited above), and even a very high HPCD concentration of 30% (a near-saturation level), with 1% propofol in a clear aqueous solution, one can calculate an estimated concentration of unbound propofol of about 40 mcg/mL. This is more than double the concentration of aqueous-phase propofol in the marketed emulsion formulations such as Diprivan, which is approximately 15-20 mcg/mL, already high enough to cause significant POI. Furthermore, at this concentration of HPCD, the amount of cyclodextrin that would be administered in the course of a one-day sedation treatment with this cyclodextrin-propofol solution would be far greater than the maximum allowable dose under current regulatory guidelines. Moyano et al., [Moyano JR, Arias-Bianco MJ, Gines JM₁ Rabasco AM and Perez- Martinez JI (1997) J. Pharm. ScL 86:72] and Djedaini and Perly [Djedaini F and Perly B, in New Trends in Cyclodextrins and Derivatives, Duchene D, ed., Editions de Sante, Paris, 1991, pp. 217-246] have described NMR-based methods for determining the complexation constant and fraction of bound drug in a cyclodextrin-drug solution. Other methods, including UV-Vis spectroscopy-based methods, are also known. .

Emulsions and other lipid-based vehicles are one proven way to reduce pain on injection. By sequestering a significant, often dominant, fraction of the drug in lipidic particles, pain on injection is reduced, essentially in proportion to the reduction in the concentration of drug in the aqueous phase — that is, the drug that is free in aqueous solution, outside of the lipidic particles or droplets. With most such vehicles, the equilibrium between lipid-phase drug and aqueous-phase drug is a "dynamic equilibrium", meaning that while any given drug molecule can quickly exchange between the two phases, the amount of drug in, say, the lipid phase at any instant is very nearly equal to that at any other instant (once the formulation has had time to equilibrate, which is typically very fast). Thus it makes sense to speak of the "aqueous-phase concentration" of drug. Mathematically, the equilibrium between the lipid-phase and aqueous-phase drug concentrations can be described by the appropriate partition coefficient, which is the ratio between the lipidic and aqueous concentrations.

Some methods that have been proposed, including some actually used, for reducing this pain on injection arc cited in the Brazeau et al. publication cited above, including the addition of a local anesthetic such as lidocaine, formulation as an emulsion, changing the pH, and changing the site or frequency of injection. Addition of a local anesthetic to an injectable formulation can lead to dangerous precipitation of drug and resulting emboli, introduces another drug into the patient, and can further complicate and delay already intricate and/or time-sensitive procedures. Changing the site or increasing the frequency (with reduced dose per needle stick) of injections can at the least lead to suboptimal pharmacodynamics, and can increase the risk of infections and other problems associated with frequent needle sticks (both to the patient and the clinician).

In addition to potentially causing pain on injection, aqueous-phase drug is more prone to hydrolytic degradation. Such degradation reactions arc numerous and well known, and in pharmaceutics commonly include ester hydrolysis, amide hydrolysis, base- catalyzed oxidation, lactone ring-opening, asparaginyl deamidation, and glycoside hydrolysis.

SUMMARY OF THE INVENTION

The inventor has found surprisingly that, although addition of low concentrations of 2-hydroxypropyl-β-cyclodextrin ("HPCD") in dispersions of reversed cubic phase liquid crystalline microparticles comprising propofol increases the total aqueous concentration of propofol, the concentration of free, unbound aqueous propofol is strongly reduced. The method as taught herein can be applied to formulations of certain drugs in which the aqueous-phase fraction of the drug is problematic, such as causing pain on injection, local toxicity, or being subject to hydrolysis, oxidation, lactone ring- opening, counterion-induced precipitation (such as, e.g., by the common ion effect), or other water-mediated degradation reaction. Such drugs include injectable propofol. The formulations employ only compounds that are safe for injection, and satisfy other criteria known in the art to be essential for safety in injectable and other pharmaceutical formulations, such as absence of large particles, identity of excipients, ability to be sterilized. It has been found that low concentrations of cyclodextrin prevent de- stabilization of the lipid-based particles, as well as other complications including cost, toxicities, regulatory barriers, and most significantly in some cases, interference with drug pharmacology — for example, it is known that certain cyclodextrins actually reverse the pharmacologic effects of rocuronium. Total cyclodextrin concentrations in the practice of this invention should be less than 75 mg/mL (7.5%), more preferably below about 50 mg/mL (5.0%), and most preferably equal to or less than about 20 mg/mL (2.0%).

An exemplary embodiment of the invention combines the complexation properties of cyclodextrins with the partitioning properties of certain lipid-based, surfactant-containing formulations to reduce the level of free unbound aqueous drug in pharmaceutical formulations, in order to reduce such problems as pain on injection, irritation, local toxicity, or drug hydrolysis or other water-mediated drug degradation. In particular the formulations described are acceptable for injection and other routes of pharmaceutical and veterinary administration. The reduction in unbound, aqueous-phase drug, compared with the lipidic vehicle in the absence of cyclodexrrin, is at least 35%, more preferably 50% or greater, and most preferably about 80% or greater. The drug is preferably selected from the group consisting of propofol, amoxicillin, penicillin G, cephalosporins, rocuronium, tetracycline, pentamidine, oxamniquine, methotrexate, diazepam, midazolam, chlorpromazine, promethazine, etomidate, methohexitone, gallamine, methocarbamol, myochrysine, epinephrine, diclofenac, ketorolac, progesterone, testosterone, olanzapine, mupirocin, remifenlanyl, hydralazine, phenoxybenzamine, levothyroxine, amlodipine, bendamustine, naloxone, cocaine, ibafloxacin, clofoctol, xibornol, tuberin, elaiomycin, virginiamycin, camptothecin and its derivatives, and 2-(sulfonamido)methyl-carbapenem antibiotics. Drugs containing phenolic groups, such as propofol, and cytotoxic drugs, such as anticancer compounds, are also preferred. Anticancer drugs known to cause irritation at the site of injection, most of which are good candidates for formulation in lipidic vehicles, include carmustine, doxorubicin, etoposide, ifosfamide, teniposide, and vinorelbine, and these are generally good candidates for the invention as well, not only because they are hydrophobic or amphiphilic (and/or of low solubility in water), but also because they are prone to causing local toxicity when administered, being cytotoxic. Prodrugs, which are designed to undergo some sort of degradation which is most commonly hydrolysis, can be good candidates for the invention. Most preferably, the drug is selected from the group consisting of propofol, olanzapine, mupirocin, remifentanyl, tetracycline, amoxicillin, and oxamniquine. The invention can also be applied to drug formulations where free aqueous phase concentrations of one or more drug degradants must be minimized. The water-mediated drug degradation reactions which are retarded by the invention include but are not limited to hydrolysis and base-catalyzed oxidation.

Another exemplary embodiment of the invention provides a pharmaceutical formulation of a drug in a dispersion of reversed lyotropic liquid crystalline phase material particles with relatively low concentrations of cyclodextrin and significantly reduced amounts of free unbound drug. A further exemplary embodiment of the invention provides a method of combining the complexation properties of cyclodextrins with the partitioning properties of dispersions of reversed lyotropic liquid crystalline microparticle formulations to reduce the level of free unbound aqueous propofol in pharmaceutical formulations.

Still another exemplary embodiment of the invention provides a pharmaceutical formulation of propofol in a lipid dispersion of reversed liquid crystalline phase material particles with relatively low concentrations of cyclodextrin and greatly reduced amounts of free unbound propofol, with a reduction of most preferably about 80% or greater.

DESCRIPTION OF THE DRAWINGS

Figure 1 is a graph of the change in chemical shift (in ppm) of the meta (triangles) and para (squares) protons as a function of the molar ratio of HPCD to Propofol. A monotonic upfield movement is observed.

Figure 2 is a graph plotting the upfield peak movement increments as a function of the molar ratio of HPCD to propofol. Squares are for the peaks at the para position, triangles for the peaks at the meta position. The dotted line shows the calculated asymptotic value for the para position, and the dashed line the asymptotic value calculated for the meta position.

Figure 3 is a plot of the resulting concentration of unbound aqueous propofol, calculated using the dialysis data and the computed complexation constants across three levels of aqueous-phase deoxycholate levels spanning the possible range, namely from 0 to 1 mg/mL (where the overall concentration of deoxycholate in the dispersion is 1.3 mg/mL). The plot uses "deoxy" to indicate sodium deoxycholate.

DETAILED DESCRIPTION OF THE INVENTION

This invention is based on the surprising discover}' that cyclodextrins can be combined with drug formulations comprising dispersions of reversed lyotropic liquid crystalline microparticles in such a way that, first of all, the concentration of unbound drug in the aqueous phase of the dispersion is significantly reduced below the level achievable either with cyclodextrin alone or with the reversed liquid crystalline phase microparticles alone, and second, that the level of cyclodextrin in this formulation is low enough that the inevitable complexation of stabilizing surfactants in the reversed liquid crystalline material does not lead to flocculation of dispersed particles. In the course of the inventor's work with propofol as a representative drug, it was found that the level of aqueous propofol increased strongly with addition of even small amounts of cyclodextrin over that in the absence of cyclodextrin — however, a deeper analysis shows that there was a strong reduction in the amount and concentration of unbound drug. In this way, the numerous advantages of reversed liquid crystal dispersions — such as triglyceride-free status, little or no microbial support, intimate interactions with biomembranes, and others — over emulsions and other lipidic vehicles can be applied to drugs that partition preferably into the lipidic particles, but not so overwhelmingly as to completely circumvent the problems associated with unbound aqueous-phase drug.

The invention can be applied to a number of drugs and medicinal compounds for which free unbound aqueous concentrations of drug are problematic, for example, cause pain on injection, or are subject to water-mediated degradation, such as oxidation or hydrolysis, which can often be base- and/or acid-catalyzed, with base-catalyzed oxidation being particularly common in addition to hydrolysis reactions such as de-esteri fication, de-amidation, lactone and other ring-opening reactions, etc. Drugs that are subject to water-mediated degradation include, for example, olanzapine, mupirocin, remifentanyl, tetracycline, hydralazine, phenoxybenzamine, levothyroxine, amlodipine, bendamustine, naloxone, cocaine, ibafloxacin, clofoctol, xibomol, tuberin, elaiomycin, virginiamycin, camptothecin and its derivatives, and 2-(sulfonamido)methyl-carbapcnem antibiotics. Drugs that are currently in formulations associated with pain on injection, which may also benefit from the invention, include propofol, amoxicillin, penicillin G, cephalosporins, rocuronium, tetracycline, pentamidine, oxamniquine, methotrexate, diazepam, midazolam, chlorpromazine, promethazine, etomidate, methohexitone, gallamine, methocarbamol, myochrysine, epinephrine, diclofenac, ketorolac, progesterone, and testosterone. Drugs containing phenolic groups, such as propofol, and cytotoxic drugs, such as anticancer compounds, particularly benefit from this invention. Anticancer drugs known to cause irritation at the site of injection, most of which are good candidates for formulation in lipidic vehicles, include carmustine, doxorubicin, etoposide, ifosfamide, teniposide, and vinorelbine, and these are generally good candidates for the invention as well. Prodrugs, which are designed to undergo some sort of degradation which is most commonly hydrolysis, are good candidates for the invention. Most preferably, the drug is selected from the group consisting of propofol, olanzapine, mupirocin, remifentanyl, tetracycline, amoxicillin, and oxamniquine. The invention can also be applied to drug formulations where free aqueous phase concentrations of one or more drug degradants must be minimized.

Furthermore, the drug may be one in which solubilization within cyclodextrins, without the use of a lipidic vehicle, leads to problematic pharmacological issues. For example, far more serious than pain on injection is the fact that cyclodextrin solutions of propofol have been shown to cause severe bradycardia, sinus arrest and atrioventricular block upon normal administration. In this case, replacing a cyclodextrin solution formulation, having greater than or equal to about 1 : 1 molar ratio of cyclodextrin to drug, with the present invention having lipidic particles present may prove highly ad%'antageous, irrespective of any pain on injection issues.

Throughout this disclosure, it is to be understood that when referring to a drug being solubilized in the lipid phase of a lipidic vehicle (or "lipid-based vehicle"), it is meant that the drug is solubilized by the lipid phase of that vehicle, and not simply surrounded by or encapsulated by a lipid. It implies a true thermodynamic solubilization within lipid domains, and without simultaneous complexation with a cyclodextrin. Thus, in particular, the case of drug being carried inside a liposome does not satisfy this requirement, since in this case the drug will be in aqueous solution in the water-filled interior of the liposome, in all prior art cases of relevance herein. By contrast, in the case of a dispersion of particles comprising reversed liquid crystalline phase material, it means that the drug will reside substantially inside the lipid-rich, hydrophobic domains of the liquid crystalline material. Clearly this invention, which relies on small quantities of cyclodextrins in the continuous phase of the dispersion to sequester drug stranded in the continuous phase external to the particles, with the vast majority of drug being solubilized inside lipid domains of the particles, is sharply distinguished from cases wherein the majority of the drug is in aqueous domains, e.g., inside a liposome, or encapsulated in cyclodextrins which are in turn encapsulated in liposomes.

As stated, the majority of drug in this invention is dissolved in the reversed lyotropic liquid crystalline particles, wherein the term "dissolved" rules out the case where the drug is dispersed in the lipid. Thus, for example, it is not true, nor is it to be taken to be true in this invention, that the drug in a dispersion of solid lipid particles is dissolved in the lipid; rather, the drug is dispersed. A solution means, for the puiposes of this disclosure, that the lipid domains containing the drug are in a fluid, noncrystalline state. The lipid-rich phase is either a liquid or a liquid crystal, or a combination thereof. The distinction is an important one because, among other things, solid lipid particles frequently do not need the delicately balanced surfactant stabilization conditions that dispersed fluid particles require.

Also throughout this disclosure, it should be remembered that surfactant solutions, micellar solutions, and microemulsions are, both by definition and by strict thermodynamic criteria, not dispersed systems, but are rather true solutions, single-phase liquids. These are thus in sharp contrast with the dispersions that are at the heart of the current invention. That is, every embodiment of this invention comprises a multiphase lipid dispersion that has water as the continuous phase, and a plurality of lipid-rich particles as the dispersed phase. The cyclodextrin(s) reside in the continuous aqueous phase.

In the exemplary case of propofol, the invention uses low concentrations of a cyclodextrin — far below equimolar to the overall propofol concentration, which in a commercial 10 lng/mL propofol formulation is about 56 mM, or approximately 75 mg/mL of cyclodextrin — to bind the propofol that is not partitioned into the reversed lyotropic liquid crystalline particles. Indeed, in the practice of this invention, concentrations of cyclodextrin must be kept low in order to avoid de-stabilizing the

I l dispersion of reversed lyotropic liquid crystalline particles via complexation of the stabilizing surfactants, in addition to other complications caused by the high cyclodextrin concentrations of the prior art. In the case of reversed cubic phase dispersion formulations ofpropofol at 10 mg/mL propofol overall (the standard concentration of all injectable propofol formulations currently on the market), only on the order of 0.1 mg/mL, or about 1%, of the propofol is outside of the particles in the absence of any cyclodextrin, as 99% of it is partitioned into the lipid-based particles. Thus, the requirements for the amount of cyclodextrin are greatly reduced compared to formulations, such as that in U.S. 7,138,387 to Pai et al., in which the cyclodextrin is the only means for reducing free propofol. Thus, the present invention not only reduces the costs and toxicities associated with the cyclodextrin, but more importantly, by employing a judiciously chosen combination of lipid vehicle and cyclodextrin, reduces the amount of free, unbound drug beyond what a pharmaceutically acceptable amount of cyclodextrin can accomplish by itself. The formulations are thus highly turbid, milk-like to the eye, due to the fact that a relatively, high concentration of reversed cubic phase particles are present in dispersed form in the aqueous liquid.

Thus, the invention creates a powerful combination of reversed lyotropic liquid crystalline particles on the one hand, into which the vast majority of the drug is partitioned, and the cyclodextrin on the other hand, which then binds a high percentage of the remaining minority of drug. This is demonstrated herein in the case of a propofol- laden dispersions of reversed liquid crystalline phase material, particularly those . described in U.S. 7,713,440 to Anderson (the complete contents of which is herein incorporated by reference). The methodology could also apply to dispersions according to U.S. 5,531,925 to Landh and Larsson (the complete contents of which is herein incorporated by reference), and to formulations comprising a combination of these two types of particles.

As taught in 7,713,440 to Anderson, typical triglyceride oils such as soybean oil do not form reversed cubic phases in equilibrium with excess water (aqueous phase), and thus tend to form emulsions rather than the reversed cubic phase-based particles of that patent and of 5,531 ,925. Also as taught in 7,713,440 to Anderson, particles comprising reversed cubic phase and reversed hexagonal phase liquid crystalline material require cither a coating phase, as in 5,531,925 or an electrostatic charge greater than or equal to about 25 mV in magnitude, in order to stabilize the particles in dispersion (and quite generally, in order to create them in the first place). This is because of the fusogenic properties of these phases, properties which are not present in general in lamellar phase materials (or remnants thereof), such as those coating emulsions. In Anderson, the charge is created by a bilayer-associated charged compound, such as an ionic surfactant, typically a bile salt surfactant. As shown herein in the preferred case of a bile salt, these compounds tend to complex strongly with cyclodextrins. Similarly, the particles of 5,531 ,925 are often stabilized in dispersion by coating phases induced by the incorporation of surfactant. Hence, many dispersions of reversed liquid crystalline phase particles have a strong tendency to be de-stabilized by the incorporation of higher levels of cyclodextrin, leading to flocculation, coalescence, or ripening. The high concentrations of cyclodextrins typified in prior art pharmaceutical preparations, usually 10% or higher, will have a strong tendency to de-stabilize dispersions of reversed cubic and reversed hexagonal phase particles. In contrast, the lack of fusogenic properties in emulsion droplet coatings, in addition to the stabilizing effect of emulsifϊer film viscoelasticity [see, e.g., C. Wabel, Ph.D. thesis, Universit&t Erlangen-Nϋrnberg, July 39, 1998], makes emulsions inherently less prone to the de-stabilizing effect of cyclodextrins. However, dispersions of reversed liquid crystalline phases, particularly reversed cubic phases, can in many cases often have very substantial advantages over pharmaceutical emulsions, including much more efficient solubilization of drug, much more intimate interactions with cell biomembranes, production at ambient temperatures, etc.

The dispersions of particles comprising reversed cubic and reversed hexagonal phase liquid crystalline materials that are used in preferred embodiments of this invention contain both lipids of low solubility in water, such as phospholipids or low-HLB poloxamers, as well as bilayer-associated compounds that act as stabilizers, which are most preferably surfactants. A significant surface charge, such as that from an ionic surfactant, and/or other stabilizing effects attributable to water-soluble surfactants, can be highly effective stabilizers of fusion-prone reversed liquid crystal materials, as taught in 7,713,440. A surfactant, or other stabilizer, is considered of low solubility in water if it is soluble in water to less than about 3%, and/or of high octanol-water partition coefficient, i.e., if Kow is greater than or equal to about 10, more preferably greater than about 100. Low HLB (hydrophilic-lipophilic balance) is considered low when less than about or equal to about 7. It is the sensitivity of water-soluble surfactants (and other stabilizers) to cyclodextrins — which results in significant complexation of the surfactants as the cyclodextrin concentration is raised — that is prevented in the invention by reducing the cyclodextrin concentration down to very low levels, this in turn being made possible by the sequestration of the vast majority of drug by the lipidic particles. According to thermodynamics, the fraction of stabilizer bound by cyclodextrin is governed by the following equation, where K_sc is the complexation constant of the particular stabilizer- cyclodextrin combination:

[complex] = K_*c * [free stabilizer] • [free HPCD]

This is studied in much greater depth in the Examples section below. Indeed, the Examples teach the complexing effect of cyclodextrin on both drug and stabilizer is accounted for, thus providing for the deteπnination of the concentrations of stabilizer and cyclodextrin needed for a stable formulation satisfying the goals of the formulation design.

The vast majority of the drug will lie in the reversed lyotropic liquid crystalline particles that comprise the dispersed phase. At equilibrium — meaning at any snapshot in time — preferably over 75% of the drug resides inside the particles (or inside any surface phase, which must still be considered to be inside the particle), more preferably over 90%, and most preferably over 95%. In the Example 3 herein, for example, at 2% cyclodextrin one can compute that approximately 97.5% of the propofol lies inside the reversed lyotropic liquid crystalline particles.

A surfactant is an amphiphile that possesses two defining properties. First, it significantly modifies the interfacial physics of the aqueous phase (at not only the air- water but also the oil-water and solid-water interfaces) at unusually low concentrations compared to non-surfactants. Second, surfactant molecules associate reversibly with each other (and with numerous other molecules) to a highly exaggerated degree to form thermodynamically stable, macroscopically one-phase, solutions of aggregates or micelles. Micelles are typically composed of many surfactant molecules (lO's to 1000's) and possess colloidal dimensions. [See R. Laughlin, Advances in liquid crystals, vol. 3, p. 41 , 1978]. Some of the surfactants that are used in injectable pharmaceutical (and more generally, medicinal) formulations include soaps (salts of fatty acids), bile salts, phospholipids, benzalkonium chloride, benzethonium chloride, and myristyl-γ-picolinium chloride. Of these surfactants listed above, all of those which are water-soluble are able to complex strongly with cyclodextrins, with, say, a complexation constant greater than or equal to about 300M^"1 with a β-cyclodextrin.

High concentrations of cyclodextrin — viz., higher than those employed in the invention described herein — can complex a significant fraction of the surfactant(s) that are responsible for stabilizing the dispersed lipidic particles against flocculation and even coalescence, and that this can lead to a significant decrease in the stability of the dispersion. Bile salts in particular are complexed by common pharmaceutical cyclodextrins, and at cyclodextrin concentrations on the order of 10% or higher will be greatly inhibited in their ability to stabilize a lipidic dispersion, such as a cubic phase dispersion. Complexation of bile salts by 2-hydroxypropyl-β-cyclodextrin (HPCD) is in fact quantified in the course of the work described herein. Also, addition of 2% HPCD to a dispersion of propofol-laden reversed cubic phase particles, produced according to U.S. 7,713,440, was found to induce particle size growth instability at 50⁰C storage condition over several months, wherein said instability does not occur over this same time period in the absence of cyclodextrin. Thus the present invention relied fundamentally on recognition of the need for, and on the successful discovery of, compositions in which the cyclodextrin concentration, lipidic particle composition, and especially the stabilizer selection and concentration allowed for dispersion stability on the one hand, and a high degree of propofol sequestration on the other hand. The dispersions of the invention are preferably stable against particle size increase for at least 18 months at 25⁰C, such that over this time span the D90 remains less than 5 microns, and more preferably less than 2 microns, and most preferably the D95 remains less than 2 microns for 18 months.

It is important that the formulation does not contain high concentrations (greater than about 15 milliMolar) of water-soluble or micellar components that have complexation constants with the cyclodextrin greater than that of propofol (about 1500 M^"1 for hydroxypropyl-beta-cyclodextrin), otherwise the propofol will be displaced from the cyclodextrin, leading to higher free aqueous propofol levels. Most lipid formulations, in particular, are deliberately or at least tacitly designed to minimize the concentration of lipid or surfactant species in the aqueous phase, since aqueous phase lipid or surfactant does little more than increase the irritation of the formulation and amount of free propofol. Indeed, micellar formulations of propofol have been plagued with extreme pain on injection.

If a reversed lyotropic liquid crystalline particle dispersion based propofol formulation is such that the total aqueous phase concentration of components with complexation constants greater than about 1500 M^"1 is less than about 15 mM, then addition of cyclodextrin as per the instant invention will significantly reduce the free, unbound aqueous propofol without requiring pharmaceutically unacceptable, limiting, or de-stabilizing amounts of cyclodextrin. In particular, the level of cyclodextrin should preferably be less than about 20 mg/mL. In part this is because at higher concentrations than 20 mg/mL (or more generally, higher than twice the propofol concentration), the total mass of cyclodextrin injected into a patient during the course of Total Intravenous Anesthesia (TIVA), or sedation on a 24-hour or more basis, could be more than the 16,000 mg currently approved for use in an intravenous product. Conversely, if the aqueous phase concentration of strongly complexing (K greater than about 1500 M^"1) component or groups of components is greater than about 15 mM, then this will strongly interfere with complexation of the aqueous propofol with the cyclodextrin. This result can be demonstrated by applying the mathematical analysis described herein to the case where the concentration of the competing species (herein specifically sodium deoxycholate, but more generally any strongly-binding aqueous-phase component(s)) is greater than 15 mM.

For propofol formulations that are designed mainly for anesthetic induction, the formulation should be such that the total aqueous phase concentration of components with complexation constants greater than about 1500 M^'1 is less than that of the molar concentration of the cyclodextrin. In case a different cyclodextrin or cyclodextrin derivative besides HPCD is used, then the formulation should be such that the total aqueous phase concentration of components, with complexation constants greater than that of propofol with the particular cyclodextrin used, is less than that of the molar concentration of the cyclodextrin.

Aqueous-phase concentrations of potentially interfering species, and their complexation constants, can be determined according to methods demonstrated and used herein for, viz., the sodium deoxycholate in one of the main formulations of focus. In the practice of this invention for propofol formulation, it is preferred to maintain the molarity of the cyclodextrin at less than 25 milliMolar (25 mM), and more preferably less than or equal to about 15.5 mM (which corresponds to approximately 20 nig/mL in the case of HPCD). It should be noted that since a propofol concentration of 10 mg/mL (1%) corresponds to approximately 56 mM, these cyclodextrin concentrations are far too small to encapsulate (or "complex") the majority of the overall propofol in the formulation, and are intended only to encapsulate the aqueous-phase propofol, recognizing that this will in general lead to higher levels of total aqueous-phase propofol (i.e., less in the lipid phase), but when practiced as taught herein will nonetheless lead to much lower levels of free, unbound aqueous-phase propofol. Thus, the invention provides pharmaceutically- acceptable for injection compositions containing propofol, preferably at a concentration of approximately 10 mg/mL, a carrier for the propofol which is a dispersion comprising hpid-based liquid crystalline particles such as reversed cubic phase particles, and one or more cyclodextrin compounds, wherein the total concentration of said cyclodextrin compounds is less than 25 mM, and more preferably less than about 15.5 mM. The composition exhibits a concentration of free, unbound aqueous-phase propofol that is significantly less than the concentration of free, unbound aqueous-phase propofol that would exist in the absence of said cyclodextrin compound(s). Preferably the concentration of free, unbound aqueous-phase propofol is less than 30 micrograms per milliliter (mcg/mL), more preferably less than 20 mcg/mL, and most preferably less than or equal to about 15 mcg/mL. It is well known to those skilled in the art that it is problematic to directly measure bound or unbound cyclodextrin by traditional methods in a turbid dispersion, thus indirect means are required. This unbound aqueous-phase propofol concentration is determined as described herein by first dialyzing the dispersion against an aqueous solution containing the buffer and cyclodextrin at concentrations matching those in the dispersion, measuring the concentration of total aqueous propofol in the dialysate, then multiplying this by the calculated (or otherwise determined) fraction of unbound propofol in an aqueous solution containing said total aqueous propofol concentration, buffer, cyclodextrin, and dispersion stabilizer(s), all at concentrations approximating those in the aqueous phase of the dispersion. It is understood that the volume ratio of dialysate to dispersion in this dialysis measurement is no larger than 30: 1.

Cyclodextrin and cyclodextrin derivatives which may be useful in the present invention include α-cyclodextrin, β-cyclodextrin, γ-cyclodextrin, hydroxypropyl- β- cyclodextrin, dimethyl-β-cyclodextrin, sulfobutylether cyclodextrin, 2,6-dimethyl 14-β cyclodextrin, 2, 3, 6- tri methyl 21-β cyclodextrin. In the case of propofol, most preferred cyclodextrins are 2-hydroxyopropyl-b-cyclodextrin and sulfobutylether cyclodextrin.

Injectable routes of administration for which the invention can be applied include, but are not limited to, intravenous, intraocular, intramuscular, intraarterial, intraarticular, subcutaneous, intrathecal, intraperitoneal, periocular, intraocular, conjunctival, subconjunctival, transconjunctival, peribulbar, retrobulbar, subtenons, transscleral, intraorbital, intrascleral, intravitreal, subretinal, transretinal, choroidal, uveal, intracameral, intracorneal, intralenticular (including phakia and psuedophakia), and in or adjacent to the optic nerve. Other routes of administration for which the invention can be applied include topical, ocular, ophthalmic, oral, intranasal, sublingual, endotracheal, intraductal, intragastric, intralumenal (for duidenal, jejunal and colonic), intrademal, intraosseus, intrathoracic, as well as via the lymphatics, urethra, bladder, ureter and vagina, and also extracorporeal for administration via cardiopulmonary bypass, extracoφoreal membrane oxygenation, dialysis and plasmapheresis..

EXAMPLES

Example 1. A dispersion of reversed cubic phase lyotropic crystalline material, in the form of charge-stabilized particles comprising propofol, phosphatidylcholine, sodium glycocholate and HFCD, was prepared as follows, with overall concentration of 10 mg/ml propofol. Preparation of Oil/Phospholipid Mixture. Into a 25OmL screw top Erlenmeyer flask, 29.2gm propofol and 5.81gni vitamin E were added. The mixture was blanketed with nitrogen and capped. The mixture was swirled by hand with mild heating to mix.

Preparation of Precursor Material. Into a stainless steel mixing container, 2.91 gm deoxycholic acid, sodium salt and 41.70gm deionized water for injection were added. The solution was mixed using a planetary mixer on the lowest speed for one minute. The speed was increased and the components mixed for an additional 10 minutes. Into the tared mixing container, 35.65gm phosphatidylcholine, Lipoid "90G", was added. The mixture was mixed using a planetary mixer on the lowest speed for one minute. The speed was increased to speed 4 and mixed for an additional 30 minutes. Into the tared mixing container, 1 1.80gm of the premixed oil/phospholipid mixture were added. The entire mixture was mixed using the planetary mixer on the lowest speed for one minute. The speed was increased and mixed for an additional 30 minutes.

Preparation of the Final Product. Into a glass container, 504.6gm sterile water for injection were added. This container was placed in an ice bath containing a cooling mixture of water and ice. The temperature was adjusted to 23°C. The container was placed under a Silverson AX60 homogenizer and homogenization was started. Into a tared Wilton cookie press, 82.26gm of precursor material were added. Into the glass container, the partial cubic phase was added slowly over 15 minutes while continuously homogenizing. The dispersion was homogenized for 58 minutes at 5400RPM's after the addition of the partial cubic phase. The temperature of dispersion was kept between 18- 26⁰C by adding more ice to the ice bath when the temperature of the dispersion reached 24⁰C. Mixing was halting after 53 minutes and a ImL aliquot was removed for an in- process microscopic evaluation in a Differential Interference Contrast microscope.

To the dispersion, 13.37gm of additional oil/phospholipid mixture were added over 5 minutes while continuously homogenizing. The dispersion was homogenized for an additional 143 minutes. The container was removed from the homogenizer and replaced with a 3L container of a solution of 3.1% glycine, as tonicity adjuster, and 3.3% FfPCD, enough to create a final cyclodextrin concentration of 2.0%. This equates to a molar ration of HPCD : propofol approximately 1 :4. The dispersion was quantitatively transferred into the 3L container. The glycine solution previously held back was used to rinse the dispersion from its container, with the rinses being added to the final 3 L container. The dispersion was homogenized with the Silverson AX60 homogenizer for an additional 6 minutes at 5400RMS. A ImL sample was taken for in-process microscopic evaluation. pH was measured to be 7.62. To 20OmL of this product was added 0.01% EDTA. The final bulk product was sparged with sterile-filtered nitrogen prior to filling vials. The drug product was placed into vials and flushed with sterile- filtered nitrogen. The vials were stoppered, sealed and autoclaved at 121⁰C for 15 minutes.

Example 2. The dispersion prepared similarly as in Example 1, but without cyclodextrin, was used as a starting point to study impact on aqueous propofol levels of various amounts of HPCD. The aqueous phase of the dispersion was separated. The aqueous phase contains approximately 0.1 mg/mL propofol, and on the order of 0.1 - 0.5 mg/mL sodium glycocholate. To this aqueous phase were added various amounts of hydroxylpropyl-beta-cyclodextrin, and after 24 hours of equilibration, to use ¹H NMR to determine the amount of unbound propofol in that aqueous mixture. The results are shown in the following table, showing that the amount of total aqueous propofol increases as the concentration of HPCD increases:

Example 3. The interactions between Propofol and 2-hydroxypropyl-β- cyclodextrin (HPCD), and the effect of sodium deoxycholate, were determined by ¹H NMR spectroscopy on a JEOL 400 MHz NMR 687047-156. Each complex was dissolved in D₂O, Aldrich Chemical Co. (Lot #01817BE). Spectra were recorded under the following conditions: number of scans, 8; acquisition time, 2.73 s; pulse width, 6.25 μs; spectral width, 6006 Hz. The temperature ranged between 24.7°C and 25.8°C. Propofol was dissolved in D₂O, in all cases herein at 0.1 mg/niL (0.56 mM). To test for any effect of deuterium exchange, ¹H NMR spectra of the simple solution were recorded at time periods of 0, 3, 6, and 24 hours.

Propofol (at 0.1 mg/mL) and HPCD (5 mg/mL, 3.6 mM), from Cyclodextrin Technologies Development, lnc (Lot #0807101C7), were dissolved in D₂O. ¹H NMR spectra were recorded at time periods of 0, 3. 6, 24, and 48 hours. Conditions were identical to that of the Propofol solution, other than the 48 hour time period. In this case, the number of scans was increased to 256.

Deoxycholate-HPCD was prepared using varying concentrations of sodium Deoxycholate, Marcor Development Corporation (Lot #DSN0205001 ). Three concentrations were investigated, using 0.18 mg/mL, 0.3 mg/mL, and 1 mg/mL of sodium deoxycholate, respectively, with HP-β-CD concentrations at a constant of 5 mg/mL in D₂O. ¹H NMR spectra were recorded at time periods of 3, 6, and 24 hours.

Propofol-Deoxycholate-HPCD was created from 4 mL of the Propofol and D₂O complex. To this, sodium deoxycholate (0.45 mg/mL, 0.48 mM) was added, followed by HPCD (5.1 mg/mL, 3.7 mM). ¹H NMR spectra were recorded at time periods of 0 and 24 hours. Spectra were recorded under the following conditions: number of scans, 256; acquisition time, 2.73 s; pulse width, 6.25 μs; spectral width, 6006 Hz. Initially, spectra were obtained at 25°C. At the 24 hour time point, spectra were recorded at 25°, 50°, and 75°C.

To determine the effect of the molar ratio of HPCD to Propofol on the interactions, samples were made using ratios of 6.6: 1 , 3: 1 , 1.4: 1 , and 0.8: 1. The samples v^ere run under the same standard conditions and 8 scans for the ¹H NMR.

In all of the runs, the HOD peak was seen at approximately 4.8 ppm. This peak then became the reference peak when determining if a chemical shift made a significant move upfield or down field. Upon inspection of the NMR data, the significant peaks for the Propofol sample were seen near 1.2 ppm (doublet - protons at ends of isopropyl branches), 3.3 ppm (triplet - middle of isopropyl), 7.0 ppm (triplet - para position to phenolic group), and 7.2 ppm (doublet - meta position).

Initial scans of the individual components in D₂O, at concentrations most relevant to the complexation experiments to follow, showed the following: 1. The deuterium exchange with protons on the Propofol over 24 and even 48 hours was insignificant, as spectra taken at 0, 3, 6 and 24 hours after dissolution of the Propofol were virtually identical, except for the phenolic proton which could not be assigned at any time point.

2. Neither HPCD nor sodium deoxycholate exhibited peaks farther downfield than 5.3 ppm. This allowed monitoring of Propofol peaks near 7.0 and 7.2 without confounding interferences.

3. While a Propofol peak near 1.2 ppm was relatively free of interference, peaks nearby from both Deoxycholate and HPCD were potential confounders, and therefore the results from analysis of this peak were performed but taken to be less reliable than those near 7.0 and 7.2.

4. The Propofol peak near 3.3 ppm was in a region far too crowded with interferences from the cyclodextrin and bile salt, and therefore did not provide useful information.

5. Deoxycholate and HPCD solutions were also run at the initial time point, 3 hours, 6 hours, and 24 hours, and throughout these times the peaks remained at the same points with the same strength they had at the initial time point.

Comparison of scans of the same sample at 25°, 50° and 75°C showed a linear increase in chemical shift of the peaks near 7.2 of 0.008 ppm per degree. The results below show that the small temperature variations recorded between samples were thus insignificant.

Inspection of NMR chemical shifts of the Propofol-HPCD solutions at the four ratios investigated are summarized in Table 1. It is seen that as HPCD is added at a constant 0.1 mg/mL Propofol concentration, the chemical shifts of the meta and para position protons steadily march upfield.

Table 1. Proton chemical shifts for Propofol peak positions that provide information on possible complexation with HPCD, at 4 molar ratios of HPCD:propofol. The change in chemical shift from the simple Propofol solution (without HPCD) is shown in parentheses. The meta and para rows are to be considered more reliable than the "branches" row due to confounding peaks. TABLE I - PROTON CHEMICAL SHlI-TS of SAMPLES (no Deoxycholale)

The following chart, and Figure 1 , show the magnitude of the upfield change in chemical shifts, averaged for the para and meta positions, as the molar ratio is changed: IfPCD:propofol δave (meta) δave (para)

0 0 0

0.8 0.0355 0.054

1.4 0.0515 0.0765

3 0.074 0.1 145

6.6 0.099 0.155

.With reference to the graph of Figure 1, where the change in chemical shift (in ppm) of the meta (triangles) and para (squares) protons as a function of the molar ratio of HPCD to Propofol are presented, it can be seen that a monotonic upfield movement is observed. The upfield shift is a clear indication that the Propofol is being complexed increasingly as KPCD is added. From the study of Trapani et al. [ 1 ], it is believed that when Propofol is complexed by HPCD, the aromatic ring lies entirely within the cyclodextrin, whereas the isopropyl groups lie primarily outside the HPCD. This is yet another reason why the data for the meta and para positions is more reliable than those of the isopropyl group protons, and the data here show a monotonic upfield shift with increasing cyclodextrin for the former positions, and a non-monotonic behavior for the latter. NMR spectra for samples with added sodium Deoxycholate (0.45 mg/mL) at the 6.6: 1 , 3: 1 , and 1.4: 1 molar ratios (0.8: 1 data not available) are summarized in Table 2.

Table 2. Proton chemical shifts for Propofol peak positions that provide information on possible complexation with HPCD in the presence of competition and other effects from sodium Deoxycholate, at 4 molar ratios of HPCD:propofol. The change in chemical shift from the simple Propofol solution (without HPCD and without deoxycholate) is shown in parentheses.

TAUl-E 2 - PROTON CHEMICAL SHUTS of SAMPLES WITH DEOXYCHOLATE*

The Propofol samples deliberately did not have deoxycholate, since in the absence of cyclodextrin there is no competition for binding, making this a better reference material.

Table 2 shows the exact same trend as seen in Table I , except that the upfield movement in the chemical shifts with added HPCD are slightly smaller in magnitude. This could due to competition for HPCD binding, a possibility that tends to be supported by the fact that the effect of Deoxycholate is greater as the molar ratio decreases. Thus, the effect of Deoxycholate is approximately 0.030 ppm in the 1.4: 1 ratio case (corresponding to approximately 40% of the δ=-0.076 in the Deoxycholate-free case), and only about 0.018 ppm in the 6.6:1 case, corresponding to about 12% of the δ in that case. With fewer cyclodextrin molecules around, the effect of competition is more pronounced. Nevertheless, other effects could be playing a role, even a dominant one, as, for example, the presence of the Deoxycholate salt could be having a non-specific ionic strength effect, and/or the Deoxycholate, being surface-active, may be interacting with the complex.

Complexation analysis. The magnitude of the upfield movement in the peaks near 7.2 (averaged over the two doublet peaks) and near 7.0 (averaged over the three triplet peaks) was plotted. These data were then analyzed using the standard stability constant formalism:

K = [complex]/ {[free propofol]«[free HPCD]} where K is the stability constant, analogous to the equilibrium constant for the reaction

(free propofol) + (free HPCD) •*-» complex .

Several references [1-3] have noted that in NMR experiments on HPCD-propofol complexes, the fast exchange between bound and unbound states gives rise to a single peak at each peak position, which is a weighted average of the two peak positions, unbound and (fixed) bound. Since the unbound positions are known accurately from the CD- free spectrum, we have confidence in setting the movement of the peak position zero, for the unbound propofol, and thus the linearly weighted mean peak movement δ is given simply by: δ = b*δ_max where b is the fraction of bound propofol. According to the literature, complexes formed from propofol and HPCD are each formed from one molecule of each, and thus the molar concentration of bound HPCD is equal to the molar concentration of bound propofol. For a given peak and a given total HPCD concentration [CD], with the total propofol concentration fixed at [P] (=0.1 mg/mL, or 0.56 niM), the expression for the stability constant — as calculated from that given peak position — is then given by:

Kcalc = [P]^«(δ/δ_nBX)/{([P]^»(l-δ/δ_σβχ))^«([CD]-[P]-δ/δ_lnβ)}

The parameters S_n^_x for the 7.0 and 7.2 ppm peaks giving the most nearly constant value of Kcaic are 1.22 and 1.87, respectively. The dotted lines in Figure 2 show these asymptotic values calculated from this analysis. The RMS deviation of the Kcalc values from the mean (1322M ¹) was 1 16.7, or 8.8% of the mean. The following Table 3 shows the input data, namely the HPCD:propofol molar ratio, observed upfield movements for the meta and para position peaks, and the calculated stability constants Kcalc. The stability constant differs several fold from the value reported in Trapani [1], which are in the range of 2,200 to 3,50OM ¹. This could be due to the method of preparation of the complexes. In [1], complexes were prepared in solution (generally at much higher concentrations than those used here), equilibrated for 5 days, then dried, and resolubilized before measurement. Such a regimen could well yield more strongly-bound complexes than the passive method used here, which was intended for a different purpose. Another difference was the use of phosphate buffer in Trapani [I].

Table 3. Results of the calculated estimates of the stability constant K based on the NMR upfield peak movements, with no deoxycholate present. The best fit values yielding this fit are δmax=0.187 ppm for the para position, and δmax=0.122 ppm for the meta position. The RMS deviation amongst the Kcalc values was 8.8% of the mean, which was 1 ,322 M^"'. δave δave Kcalc (para) Kcalc

HPCDφropofol (meta) (para) [HPCD] M-¹ (meta) M^"1

0 0 0 0 ND ND

0.8 0.0355 0.054 0.000448 1416 1437

1.4 0.0515 0.0765 0.000785 1246 1332

3 0.074 0.1 145 0.001682 1 180 1 149

6.6 0.099 0.155 0.0037 1497 1326

Based on the Kcalc values, and using the relation δ = b*δ_max from above, we can estimate that at the molar ratio of 6.6: 1, which corresponds to a HPCD concentration of about 5 mg/mL at this value of the propofol concentration, the bound fraction of propofol b is estimated to be 0.099/0.122 = 81.1 % based on the meta position, and 0.155/0.187 = 82.9% based on the para position. The good agreement between these values lends support to the analysis. The unbound propofol concentration would then be the remaining 18% of 100 mcg/mL, or about 18 mcg/mL. The effect of stabilizer (sodium deoxycholate). The complexation constant for deoxycholate-cyclodextrin complexation was also determined, and found to be approximately 2300 M^"1. Because the concentration of aqueous-phase deoxcholate in the lipid dispersion is difficult to determine due to its low overall concentration (1.3 mg/mL), the NMR analysis was performed at deoxycholate concentrations of 0.04, 0.12, 0.18 and 0.32 mg/mL. In all four of these series, the span of the movement of the meta-position propofol resonance, δ_ma_*, from the complexed to uncomplexed position was δ_m^O.πδ. The RMS deviation between experimental and calculated peak positions remained at approximately 3% of this δ_n,_ax for all four Series, as did that of the deoxycholate peak. Application of the complexation constants yielded a bound (complexed) fraction of 84% at only 5 mg/mL cyclodcxtrin, decreasing mildly with increasing deoxycholate. Indeed, only at quite low HPCD concentrations was the propofol-deoxycholate competition for binding significant. This was also supported by UV- Vis measurements at the deoxycholate level of 0.24 mg/mL.

Example 4. Final result: unbound aqueous propofol. After full analysis of the UV-Vis and NMR data both with and without the presence of deoxycholate, which yielded a more accurate value for the propofol-HPCD complexation constant of 1535M^"1, the resulting concentration of unbound aqueous propofol, calculated using the dialysis data and the computed complexation constants, was plotted in Figure 3 across three levels of aqueous-phase deoxycholate levels spanning the possible range, namely from 0 to 1 mg/mL (recalling that the overall concentration of deoxycholate in the dispersion is 1.3 mg/mL). The plot uses "deoxy" to indicate sodium deoxycholate.

It can be seen from the resulting plot of Figure 3 that at 2% HPCD, the result is nearly independent of the value of aqueous-phase deoxycholate, and is between about 10 and 12 mcg/mL (i.e., 0.010 and 0.012 mg/mL). This is lower than the concentration of free propofol in the currently marketed Diprivan® formulation. As stated above, the molar ratio of cyclodextrin to drug at 2% HPCD and 1 % propofol is about 0.26:1 (i.e., CD:API=0.26). Over the range of 0.5% to 2.0% HPCD, the fraction of the total aqueous- phase propofol that was cyclodextrin-bound was calculated to increase from about 0.80 (80%) at 0.5% HPCD, to about 0.90 (90%) at 1%, to just over 0.95 (95%) at 2% (20 mg/mL) of 2-hydroxypropyl-beta-cyclodextrin.. In short, while the total aqueous-phase propofol jumps from about 90 mcg/mL without HPCD to over 200 mcg/mL with the addition of 2% HPCD, the unbound aqueous propofol gets dramatically reduced to about Il mcg/mL. This represents a reduction of about 88% in the amount of unbound aqueous propofol.

Example 5. The formulation prepared in Example 1 with 2% HPCD and 0.01% EDTA was analyzed after several months, and showed excellent stability over 12 weeks (at the time of writing) at 25" and 40⁰C storage temperatures, but with 2-5 micron particles visible in the sample stored at 5O⁰C. As stated above, formulations at this composition without cyclodextrin have been found in the inventor's laboratory to be stable even at 50⁰C for 3 months within any substantial increase in particle size (viz., the mean particle size does not increase by more than 40%).

25°C stability data:

Total

Time Zeta potential Propofol

(Weeks) Particle size (mV) pH (mg/ml)

0 314.6 -57.5 7.56 1 1.38

8 352.5 -56.7 7.57 11.28

12 404.1 -62.7 7.35 10.94

40⁰C data:

Time Total

(Weeks) Particle size Zeta potential pH Propofol

0 314.6 -57.5 7.56 1 1.38

8 398.1 -55.0 7.54 10.1 1

12 369.3 -56.4 7.30 10.68

The level of total aqueous propofol was measured at both temperatures to be stable at approximately 0.25 mg/mL over that same time period. This yields a calculated determination of the unbound aqueous propofol of just under 0.013 mg/mL, or 13 mcg/mL.

Claims

CLAIMSWe claim:

1. A pharmaceutical composition comprising: stabilized microparticles of reversed lyotropic liquid crystalline phase material comprising a lipid, surfactant and a drug; an aqueous carrier; and cyclodextrin, wherein i. the concentration of said cyclodexrrin is less than or equal to about

7.5%; and, ii. the molar ratio of said cyclodextrin to said drug is less than or equal to about 1 :2.

2. The composition of claim 1 wherein said concentration of cyclodextrin is less than or equal to about 5%.

3. The composition of claim 1 wherein said concentration of cyclodextrin is less than or equal to about 2%.

4. The composition of claim 1 wherein said molar ratio of cyclodextrin to said drug is less than or equal to about 1 :3.

5. The composition of claim 1 wherein said molar ratio of cyclodextrin to said drug is less than or equal to about 1 :4.

6. The composition of claim 1 wherein the amount of free unbound drug in the aqueous phase of said composition is reduced by about 35% or more compared to the amount of free unbound drug in the aqueous phase of the same composition without cyclodextrin.

7. The composition of claim 1 wherein the amount of free unbound drug in the aqueous phase of said composition is reduced by about 50% or more compared to the amount of free unbound drug in the aqueous phase of same composition without cyclodextrin.

8. The composition of claim 1 wherein the amount of free unbound drug in the aqueous phase of said composition is reduced by about 80% or more compared to the amount of free unbound drug in the aqueous phase of the same composition without cyclodextrin.

9. The composition of claim 1 wherein the drug is selected from the group consisting of propofol, olanzapine, mupirocin. remifentanyl, tetracycline, amoxicillin, and oxamniquine.

10. The composition of claim 10 wherein the drug is propofol.

1 1. The composition of claim 1 wherein the drug is selected from the group consisting of hydralazine, phenoxybenzamine, levothyroxine, amlodipine, bendamustine, naloxone, cocaine, ibafloxacin, clofoctol, xibornol, tuberin, elaiomycin, virginiamycin, camptothecin and its derivatives, and 2-(sulfonamido)methy]-carbapenem antibiotics.

12. The composition of claim 1 wherein the composition is pharmaceutically suitable for IV injection.

13. The composition of claim 1 wherein said reversed lyotropic liquid crystalline phase microparticles are charged, uncoated, and are selected from a reversed cubic phase or a reversed hexagonal phase.

14. A method of reducing a level of free unbound aqueous drug in an aqueous pharmaceutical stabilized dispersion of reversed lyotropic liquid crystalline phase microparticles comprising the step of incorporating cyclodextrin in the dispersion in a concentration equal to or less than about 7.5%.

15. The method of claim 14 wherein said reversed lyotropic liquid crystalline phase microparticles are charged, uncoated, and are selected from a reversed cubic phase or a reversed hexagonal phase.

16. The method of claim 14 wherein said drug is propofol.

17. A method of retarding water mediated degradation reactions of drugs in pharmaceutical dispersions of reversed lyotropic liquid crystalline phase microparticles comprising the step of incorporating cyclodextrin in the dispersion in a concentration equal to or less than about 7.5%.

18. The method of claim 17 wherein said reversed lyotropic liquid crystalline phase microparticles are charged, uncoated, and are selected from a reversed cubic phase or a reversed hexagonal phase.

19. The method of claim 14 wherein said drug is propofol.