Lipsome/anthraquinone drug composition and method

Description Submit all comments and votes

FIELD OF THE INVENTION

The present invention relates to an improved liposome/anthraquinone therapeutic composition having reduced lipid oxidation and free-radical damage, and to methods for preparing and using the composition.

REFERENCES

Aubel-Sadron, G., et al, Biochemie, 66: 333 (1984).

Augustin, M. A., J Am Oil Chem, 60: 105 (1983).

Forssen, E. A., et al, Proc Nat Acad Sci, USA, 78(3) 1873 (1981).

Gabizon, A., et al, Cancer Res, 42: 4734 (1982).

Gabizon, A., et al, Cancer Res, 43: 4730 (1983).

Gabizon, A., et al, Brit J Cancer, 51: 681 (1985).

Goormaghtigh, E., et al, Res Commun Chem Path and Pharm, 42(1): 149 (1983).

Goormaghtigh, E., et al, Biochem Biophys Acta, 779: 271 (1984).

Gutteridge, J. M. C. Biochem Pharm, 33(11): 1725 (1984a).

Gutteridge, J. M. C., FEBS, 172(2): 245.

Juliano, R. L., et al, Biochem Pharmacol, 27: 21 (1978).

Klein, R. A., Biochem Biophys Acta, 210: 486 (1970).

Mayhew, E., et al., Cancer Treat Repts, 63 (11-12): 1923 (1979).

Myers, C. E., et al, Cancer Treat Repts, 60(7): 961 (1976).

Olson, F., et al, Eur J Clin Oncol, 18(2): 167 (1982).

Poznansky, M. L., et al, Pharm Revs 36(4): 277 (1984).

Rahman, A., et al, Cancer Res, 40: 1532 (1980).

Samuni, A., et al, Cancer Res, in press.

Sikic, B. I., et al, Science, 228: 1544 (1985).

Sonneveld, P., Cancer Treat Repts, 62 (7): 1033 (1978).

Stuart-Harris, R. C., et al, Biochem Pharmacol, 12: 1 (1984).

Sunamoto, J., et al, Biochem Biophys Acta, 833: 144 (1985).

Szoka, F., et al, Ann Rev Biophys Bioeng, 9: 467 (1980).

Tomasz, M., Chem-Biol Interact, 13: 89 (1976).

Wang, Y. M., et al, Cancer Res, 40: 1022 (1980).

Young, R. C., et al, N Eng J Med, 305: 139 (1981).

BACKGROUND OF THE INVENTION

Liposome delivery systems have been proposed for a variety of drugs, particularly those which are administered parenterally. Liposomes have the potential of providing a controlled "depot" release of the administered drug over an extended time period, and of reducing toxic side effects of the drug, by limiting the concentration of free drug in the bloodstream. Liposomes can also alter the tissue distribution and uptake of drugs, and the altered tissue distribution, combined with other advantages just mentioned, can significantly increase the therapeutic effectiveness of the drug. Liposome/drug compositions can also increase the convenience of therapy by allowing higher drug dosage and less frequent drug administration. Liposome drug delivery systems are reviewed generally in Poznansky et al.

One group of drugs whose use in liposome delivery systems has been widely studied is the class of anthracene quinones, including, particularly, the anthracycline glycoside antibiotics, exemplified by the anti-tumor drug doxorubicin or "Adriamycin", doxorubicinol, daunorubicin, and daunorubicinol, and their cyanomorpholinyl derivatives. Doxorubicin (DXR) is a potent chemotherapeutic agent effective against a broad spectrum of neoplasms (Aubel-Sadron et al and Young). However, use of the drug in soluble form is limited by serious side effects. Its acute toxicity includes malaise, nausea, vomiting myelosuppression, and severe alopecia. In addition, cumulative and irreversible cardiac damage occurs with repeated administration, which seriously limits the use of the drug in protracted treatment (Young). When administered in liposome form, the drug retains its therapeutic effectiveness against animal tumors, but is significantly less toxic, as judged by reduced mortality (Forssen, Gabizon 1985). The drug-protective effect of liposomes is due, at least in part, to a marked alteration in tissue disposition and drug-release rate of the injected drug (Gabizon 1982; Gabizon 1983; Juliano).

The cardiomyopathy observed in doxorubicin treatment is similar to the cardiac muscle lesions seen in experimental animals under conditions of alpha-tocopherol (.alpha.-T) deficiency (Tomasz), suggesting that the drug-induced lesions are caused by increased free-radical reactions involving membrane lipids. DXR and other anthraquinones, which have the general structure shown in FIG. 1, contain both quinone and hydroquinone groups, and thus might be expected to promote peroxidation reactions involving electron transfer to or from the quinones or hydroquinones. In addition, binding of the drug to lipids through the anthracene moiety would be expected to facilitate lipid involvement in peroxidations reactions. It is known, for example, that DXR binds tightly to cardiolipin, a major lipid component in mitochondria, and enzyme-catalyzed electron transport results in formation of covalent linkages between the drug and lipid (Goormaghtigh).

The free radical mechanism of cardiac toxicity proposed for DXR suggests that a lipophilic free-radical quencher such as vitamin E would be effective in reducing drug toxicity and this, in fact has been found (Myers; Wang; Sonneveld). The referenced studies have shown that vitamin E is effective in reducing cardiotoxicity when administered prior to or concurrent with DXR administration. More recent studies have shown that a liposome drug system with coentraped DXR and vitamin E is less toxic in animals, and produces less cardiomyopathy, than either vitamin E/DXR or liposome/DXR combinations alone (Olson). The reduced toxicity of liposomes with coentrapped DXR and vitamin E apparently results from a combination of the altered drug distribution and/or lower free drug levels--due to liposomal entrapment of the drug--and from reduced free-radical damage--due to the free-radical quenching activity of vitamin E.

Despite the reduced toxicity of a liposome/DXR/vitamin E formulation, it has been found, in studies conducted in support of the present invention, that the drug and lipids in such formulation can undergo substantial chemical modification, even under anoxic during storage conditions. Such damage increases the toxicity of the drug formulation and appears to compromise the therapeutic action of the drug.

SUMMARY OF THE INVENTION

The present invention includes a liposome/anthraquinone composition having anti-oxidant properties designed to reduce lipid oxidation and free radical damage to both the lipid and drug components of the composition.

According to the invention, it has been discovered that drug and lipid damage related to oxidative and free-radical mechanisms is substantially reduced by including in the liposome formulation, a lipophilic free-radical scavenger, such as .alpha.-T, and a water-soluble trihydroxamic acid chelating agent, such as ferrioxamine, having a high and selective binding affinity for ferric iron. The extent of protection against lipid-peroxidation and free-radical damage is much greater than that afforded by free-radical quenchers alone or by chelating agents, such as ethylenediamine tetraacetic acid (EDTA), which have been used heretofore. The chelating agent is present in a molar excess of the ferric iron in the suspension.

One preferred liposome composition contains DXR (or a pharmacologically acceptable analog, derivative, or salt thereof) at a molar ratio of at least about 2.5 mole percent in liposomes composed 20-50 mole percent cholesterol, 10-40 mole percent negatively charged phospholipid, such as phosphatidylglycerol (PG), phosphatidylserine (PS) or phosphatidylinositol (PI), and phosphatidylcholine (PC). The liposomes are predominantly in the size range of 0.05 to 0.5 microns, and the amount of free drug in the suspension--that is, drug not associated with liposomes--is preferably less than about 15% of the total drug in the suspension. Ferrioxamine is contained in the suspension at about 50 .mu.M, and .alpha.-T in the liposomes, at a concentration of at least about 0.2 mole percent.

Also included in the invention is a method for treating human neoplasms with a DXR/liposome composition formed as above. Clinical trials on human cancer patients in the frame of a Phase I study indicate significant reduction of many side effects such as discomfort (malaise), headaches, nausea, vomiting, local pain at the site of injection, and alopecia, frequently associated with treatment by free DXR. In addition, anti-tumor activity has been shown on hepatocellular carcinoma.

It is one general object of the invention to provide a liposome composition containing an entrapped anthraquinone drug in which in vitro peroxidative damage to the lipid and drug components of the composition is substantially reduced.

Another object of the invention is to provide a general method for reducing, in a liposome/anthraquinone composition, lipid and drug toxicity which are related to oxidation and free-radical reactions.

A specific object is to provide such a composition for treatment of human neoplasm, wherein the drug is DXR or a pharmacologically accepted analogue, derivative, or salt thereof.

A related object of the invention is to provide a method of treating human neoplasms, and, particularly, primary and metastatic liver tumors, hematopoietic proliferative disorders, and leukemias with liposome-entrapped DXR with significant amelioration of normal drug side effects.

These and other objects and features of the invention will become more fully apparent when the following detailed description of the invention is read in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIGS. 1A-1C show the general structures of class I and II anthracycline aminoglycosides (1A and 1B, respectively), and mitoxanthrone (1C);

FIG. 2 shows the fluorescent emission spectra of DXR in soluble (upper dashed line) and liposome-bound (upper solid line) form, and a DXR-Fe (III) complex in soluble (lower dashed line) and freshly prepared liposome-bound (lower solid line) form;

FIG. 3 illustrates oxidative and free radical mechanisms which may contribute to drug and lipid modifications in the presence of ferric iron, oxygen, and unsaturated lipids;

FIG. 4 illustrates ferric iron chelation by a trihydroxamic acid chelating agent;

FIG. 5 shows UV spectra of lipids isolated from liposomes incubated in the presence (spectrum A) or absence (spectrum B) of ferrioxamine;

FIG. 6 plots the effects on survival in mice of graded doses of free DXR (F-DXR) and liposome-entrapped DXR (L-DXR) after a single injection; and

FIGS. 7 and 8 are plots of the survival curves for male and female mice, respectively, after repeated administration of graded doses of F-DXR and L-DXR.

DETAILED DESCRIPTION OF THE INVENTION

I. Preparing the Liposome/Drug Composition

A. The Anthraquinone Drug

The anthraquinone drug used in preparing the composition of the invention is an anti-neoplastic anthraquinone drug having an anthracene ring structure and quinone, semi-quinone, or hydroquinone functionalities carried on the ring structure. More precisely, and as the term is defined herein, anthraquinone drugs include those anthraquinone structures having one quinone and hydroquinone group on adjacent rings of the anthracene ring structure. Anti-neoplastic anthraquinones having these features can be classed into three groups which are illustrated in FIGS. 1A-1C.

The first group is the class I anthracene glycosides which are illustrated in FIG. 1A. Included in this group are a number of clinically important anti-neoplastic drugs, such as doxorubicin (DXR, commonly known as Adriamycin), daunomycin, carcinomycin, N-acetyladriamycin, N-acetyldaunomycin, rubidazone, and 5-imidodaunomycin. Table I below gives the structural variations of the these several class I drugs, in terms of the R1, R2, and R3 groups in FIG. 1A. More recently, a cyanomorpholino derivative of DXR has been reported (Sikic). Drugs in this class are known to have anti-neoplastic effects against a variety of cancers, including acute leukemias, breast cancer, Hodgkin disease, non-Hodgkin lymphomas, and sarcomas. The primary mechanism of the drugs appears to be intercalation into DNA and DNA damage (Young, Goormaghtigh 1984), although membrane binding (Goormaghtigh 1984) and enzyme-catalyzed free-radical formation (Aubel-Sadron) have also been suggested as possible contributing mechanisms of drug action.

TABLE I __________________________________________________________________________ R.sub.1 R.sub.2 R.sub.3 __________________________________________________________________________ Adriamycin .dbd.O --CO--CH.sub.2 OH --NH.sub.2 Daunomycin .dbd.O --CO--CH.sub.3 --NH.sub.2 N--Acetyladriamycin .dbd.O --CO--CH.sub.2 OH --NH--CO--CH.sub.3 N--Acetyldaunomycin .dbd.O --CO--CH.sub.3 --NH--CO--CH.sub.3 Rubidazone .dbd.O --C--N--NH--C-- --NH.sub.2 CH.sub.3 O 5-Imidodaunomycin .dbd.NH --CO--CH.sub.3 --NH.sub.2 __________________________________________________________________________

The second group includes the class II anthracene glycosides, which are distinguished from the class I compounds by more complex (multimeric) aminoglycoside residues, as seen in FIG. 1B. These compounds share the same general therapeutic and toxicity properties of their class I counterparts. Representative class II anthracene aminoglycosides are listed in Table II, with reference to the R1, R2, and R3 groups shown in FIG. 1B.

TABLE II ______________________________________ Anthracycline R.sub.1 R.sub.2 R.sub.3 ______________________________________ Musettamycin OH COOCH.sub.3 H Rudolfomycin OH COOCH.sub.3 Rednosamine Aclacinomycin H COOCH.sub.3 Cinerulose Marcellomycin OH COOCH.sub.3 2-Deoxyfucose Descarbomethoxy- OH H 2-Deoxyfucose marcellomycin Descarbomethoxy- OH H Rednosamine rudolfomycin ______________________________________

The third group of anthraquinones includes those which have the general ring structure shown in FIG. 1A, i.e., with quinone and hydroquinone groups on adjacent anthracene rings, but which lack the glycoside groups characteristic of the anthracycline antibiotics. One important drug in this group is mitoxanthrone, an anti-neoplastic drug which is reported to be effective against a variety of cancers, including breast cancer, but which may have reduced side effects, including cardiotoxicity, when compared with free DXR (Stuart-Harris).

Studies on the binding of ferric iron to DXR which have been carried out in collaboration with one of the inventors indicate that the anthraquinone drugs, and particularly the anthracyclic aminoglycosides, form stable metal complexes with ferric iron (Samuni). The studies examined the binding of DXR to ferric iron, both in the presence and absence of small unilamellar vesicles (SUVs) formed from the saturated phospholipid dimyristylphosphatidylcholine (DMPC). With reference to FIG. 2, the fluorescent emission spectrum of DXR at 480 nm excitation wavelength (upper dashed line) shows peaks at about 590 and 560 nm, and this spectrum is changed very little when the drug is associated with SUVs (upper solid line). When DXR is complexed with ferric iron in the presence of ADP, a fluorescence emission spectrum having nearly the same shape but significantly reduced intensity was observed (lower dashed line). The reduced intensity presumably results from the quenching effect of iron. When the DXR-Fe(III) complex was added to DMPC SUVs, less quenching effect was observed, as evidenced by a fluorescence emission spectrum which was intermediate in intensity between that of the free drug and the DXR-Fe(III) complex in solution (lower solid line). The same emission spectra for DXR-Fe(III) complex in association with liposomes was observed whether iron was added before or after drug binding to the liposomes, indicating that the drug-iron complex forms readily with lipid-bound drug.

Of interest with respect to the present invention was the finding that DXR undergoes relatively rapid and irreversible chemical modification when complexed with ferric iron in the presence of SUVs formed with saturated and/or unsaturated phospholipids. One measure of this modification is a change in the 590/560 nm ratio of the fluorescence spectra, which would result from modification of the antracene ring. The 590/560 nm ratio of DSR-Fe(III) in association with DMPC is initially about 1.9. In the presence of ferric iron, at an iron:drug ratio of 2:1, this ratio falls to about 0.9 over a five hour period at 30.degree. C. With a large molar excess of ferric iron, the decline in 590/560 is significantly faster, decreasing to less that 0.8 in two hours. The modified DXR, when extracted from the liposomes, has a 590/560 ratio of about 0.6. The modified drug also shows increased hydrophobicity, as evidenced by its partitioning characteristics in a standard biphasic partition system. This change may be due to cleavage of the more hydrophilic sugar moiety from the antracene ring structure. It is noted that the drug modification reactions occur either in the presence or absence of oxygen, in liposomes composed either of saturated or unsaturated phospholipids. That is, the events leading to drug modification do not appear to require oxygen dependent electron transfer or free radical formation involving unsaturated lipids. However, as will be seen below, both oxygen and unsaturated lipids do contribute to drug modification in the drug/liposome formulation, and under these conditions, significant lipid peroxidative damage may also occur. One of the important aspects of the formulation of the invention, as will be seen below, is a significant reduction in such drug and lipid modifications.

B. Lipid Components

The liposomes in the composition are formed from standard vesicle-forming lipids, which generally include neutral and negatively charged phospholipids and a sterol, such as cholesterol. The selection of lipids is guided by considerations of (a) drug-release rate in serum, (b) drug-entrapment efficiency, (c) liposome toxicity, and (d) biodistribution and targeting properties.

Considering the effect of lipid components on in vivo drug-release rates, the most important compositional factors are chain length, degree of unsaturation, and head group charge and side groups in the phospholipids. The dependence of drug release rate on lipid composition is due in part to differences in the rate of exchange of amphipathic drug with the outer liposome bilayer, and in part to the differing stability of liposomes having different lipid compositions. From studies below, it will be seen that negatively charged phospholipids, such as phosphatidylglycerol (PG) and phosphatidylserine (PS), tend to enhance drug liposome stability as measured by DXR release in 50% plasma, whereas cardiolipin (CL) produces a marked destabilizing effect on the liposomes. The latter effect may be related to the apparently strong interaction between DXR and cardiolipin which is reflected by drug and lipid cross-linking which also occur in vivo (Goormaghtigh 1983). Neutral phospholipids, particularly phosphatidylcholine (PC), and negatively charged phospholipids, such as PG, PS, and phosphatidylinositol (PI), having acyl chain components of selected chain lengths and degree of saturation are available from commercial sources, or can be prepared by known techniques.

Another lipid component which is important to liposome stability, and therefore to drug-release rate, is cholesterol. In one study conducted in support of the present invention, the in vitro release of DXR from liposomes whose lipid components contained varying amounts of cholesterol, PC, and PG, PS, or CL, was examined. The liposomes were prepared substantially as described in Example I below, but contained the mole percentages of PC, PG, and cholesterol indicated in Table III below. Prior to each a test, the liposome suspensions were freed of free (soluble) drug by molecular sieve chromatography. The DXR/liposome compositions were each incubated at 37.degree. C. for 1 hour in the presence of 50% human plasma, and the liposome-associated drug then separated from released, soluble drug, either by molecular sieve chromatography or ultracentrifugation. The amount of DXR retained in the liposomes was calculated as a percentage of the total original drug. The results are shown in Table III:

TABLE III ______________________________________ % Liposome-Retained Liposome Composition Mole Ratio DXR ______________________________________ PC:CHOL (DXR) 4:4 72 PC:CHOL (DXR) 4:2 62 PC:CHOL (DXR) 4:1 76 PC:CHOL (DXR) 4:0 29 PG:PC:CHOL (DXR) 3:7:5 99 PG:PC:CHOL (DXR) 3:7:4 96 PG:PC:CHOL (DXR) 3:7:2.5 81 PG:PC:CHOL (DXR) 3:7:0 70 PS:PC:CHOL (DXR) 3:7:10 98 PS:PC:CHOL (DXR) 3:7:0 62 CL:PC:CHOL (DXR) 1:4:5 39 CL:PC:CHOL (DXR) 1:4:0 28 ______________________________________

As seen from the table, 20 mole percent cholesterol produces a 2-fold to 3-fold enhancement in drug retention in PC liposomes, although no additional improvement is seen up to a cholesterol mole ratio of 50%. In liposomes formed with 30 mole percent PG, good drug retention achieved in the absence of cholesterol, but progressively greater stability is observed with increasing amounts of cholesterol. PS, another negatively charged phospholipid, gives substantially the same result as found with PG. Additional studies not reported here indicate that the increased drug retention is seen over a range of negatively charged phospholipid of between about 10-40 mole percent. Interestingly, the presence of only 10 mole percent cardiolipin (diphosphatidylglycerol), which contains a double negatively charged head group, substantially eliminated the cholesterol effect, giving poor drug retention even at 50 mole percent cholesterol.

Drug-entrapment efficiency, another factor to be considered in selecting a lipid composition, refers to the total amount of drug which can be loaded into liposomes, expressed as a ratio of drug per mole per liposome lipid. High entrapment efficiency is desirable both in terms of preparation costs and for maximizing the amount of drug which can be delivered in liposomal form in a given volume of liposomes. Experiments conducted in support of the present invention have examined the efficiency of DXR entrapment in both large and smaller, sonicated liposomes, and with a number of different lipid compositions. The liposome and lipid composition variables are shown in Table IV below. In this table, MLV refers to large multilamellar vesicles, which typically contain heterogeneous liposome sizes averaging between about 0.5-5 microns, and SUVs, to sonicated but unfractionated vesicles. The final entry is for vesicles which have been sonicated and fractionated to remove larger vesicles, yielding predominantly small liposomes which are designated SUV(F). DPPG and DPPC are abbreviations for dipalmitoyl PG and dipalmitoyl PC, respectively, as distinguished from PC and PG, which are derived from egg lipids and which contain a mixture of both saturated and unsaturated phospholipid components.

TABLE IV __________________________________________________________________________ DXR Entrapment DXR/Phospholipids Liposome Composition Mole Ratio % Total m mole/mole __________________________________________________________________________ MLV-PC:CHOL:DXR 4:4:1 14 35 MLV-PC:DXR 4:1 10 26 MLV-CL:PC:CHOL:DXR 1:4:5:1 64 128 MLV-CL:PC:DXR 1:4:1 58 116 MLV-PS:PC:CHOL:DXR 3:7:10:2 59 129 SUV-PC:CHOL:DXR 10:10:1 15 16 SUV-CL:PC:CHOL:DXR 1:4:2:1: 47 94 SUV-CL:PC:CHOL:DXR 1:4:5:1 45 90 SUV-CL:CHOL:DXR 10:5:2 90 180 SUV-PS:PC:CHOL:DXR 3:7:4:2 50 100 SUV-PG:PC:CHOL:DXR 3:7:4:2 61 128 SUV-DPPG:PC:CHOL:DXR 3:7:4:2 39 78 SUV-DPPG:DPPC:CHOL:DXR 3:7:4:2 28 56 SUV(F)-PS:PC:CHOL:DXR 3:7:4:2 25 50 __________________________________________________________________________

The data relating to MLVs suggest that a negatively charged phospholipid--either CL or PS--is required for good entrapment efficiency, but that cholesterol has only a minor effect, if any, on drug entrapment. With SUVs, a similar increase in entrapment efficiency was observed with either CL or PG, or PS. Also, entrapment efficiency was relatively poor when saturated phospholipids (DPPG alone or with DPPC) were substituted for the more unsaturated egg lipids, and in the predominantly very small SUV(F) liposomes. The data in Tables III and IV, taken together, suggest that the optimal liposome composition, for obtaining both high entrapment efficiency and good drug retention in vitro (in serum plasma), contains between 10-40 mole percent negatively charged phospholipid, particularly PG, but excluding cardiolipin (CL); between 20-50 more percent cholesterol; and at least about 40 mole percent of a natural phospholipid, such as PC. Within this range of lipid components, the degree of acyl chain saturation and chain length may be varied to achieve desired drug-release characteristics in vivo, although it would be expected that a composition containing a high percentage of saturated lipids would have a reduced drug entrapment efficiency.

Another important consideration in the choice of lipid components is the toxicity of the drug/liposome composition which is produced. Since the soluble drug is generally more toxic than liposome-entrapped drug, it is desirable that the liposomes have both good entrapment efficiency and good drug retention in vivo. The lipid composition factors which effect drug retention and entrapment are discussed above. In addition, the acyl chain composition of the lipids may produce toxicity effects which are unrelated to the amount of drug present in the liposomes. Experiment conducted in support of the present invention to asses the toxicity of empty liposomes (not containing entrapped drug) on laboratory animals gave the following results: (1) Liposomes composed predominantly of saturated lipids, such as DPPG and DPPC, were more toxic than liposomes formed from corresponding unsaturated phospholipids. (2) Although unsaturated lipids are less toxic, they are also much more susceptible to lipid peroxidation damage on storage, and liposomes with substantial lipid oxidation damage are considerably more toxic, in terms of LD.sub.50 values, than corresponding fresh liposomes. Also, it would be expected that lipid oxidation damage would increase toxicity in a drug/liposome composition, through decreased drug retention. As will be seen, an important aspect of the invention involves reducing peroxidative damage to lipid and drug components of the anthraquinone/liposome composition by a combination of protective agents, and therefore the susceptibility of unsaturated lipids to oxidative damage is less of a concern. That is, the composition of the invention contributes to reduced toxicity, in part, by allowing the use of unsaturated lipids, which are both less toxic and show greater drug entrapment, without concomitant lipid peroxidative damage, which would increase toxicity effects.

Biodistribution and liposome targeting may be affected by liposome size, surface charge, and the presence of specific surface-bound molecules which act to target the liposomes to specific sites in the body. Of particular interest to the success of the present invention is the enhanced accumulation of drug/liposomes in certain target organs, such as liver and spleen, and the reduced accumulation in non-target organs, such as kidney and heart, where drug toxicity is largely localized. In accordance with one aspect of the invention, optimal biodistribution is achieved within a selected size range of liposomes, as discussed below with reference to liposome sizing.

C. Protective Agents

The interaction of an anthraquinone drug with ferric iron in the presence of lipid, and the chemical modification of the drug which can occur in vitro have been described in Section A. As noted there, the drug modification reaction occurred under anoxic conditions, and in the presence of saturated lipids only. In Section B it was seen that, for a number of reasons, natural phospholipids (containing both saturated and unsaturated phospholipids) are generally advantageous in the drug/lipid composition of the invention. Further, it is reasonable to expect that the preparation and handling of the composition will involve some exposure to molecular oxygen, so that oxidative and free radical reactions which involve unsaturated lipids and oxyen can also be expected to produce lipid and drug modifications in the composition. The present section examines the oxidative and free radical mechanisms which can lead to such drug and lipid modification and demonstrates how such modification reactions can be controlled to a great extent by including in the composition a combination of lipophilic and water-soluble protective agent which act at different points in peroxidation/free radical pathways.

A scheme showing likely oxidative and free radical reactions in a composition containing an anthraquinone drug, lipid, and ferric iron is given in FIG. 3. Here, as in Section A, the liposome/anthraquinone formulation is indicated by Lip/AnQ; the acyl-chain lipid components forming the liposomes may be saturated, unsaturated, or a mixture of both. The upper left portion of the Figure shows formation of the Lip/AnQ-Fe(III) complex, described in Section A. In the presence or absence of oxygen and either saturated or unsaturated lipid, the events leading to drug modification are likely to involve, first, possible formation of a semi-quinone-Fe(II) complex, and inter- or intramolecular free radical damage to the drug. Since the modified drug (DXR) is considerably more hydrophobic than the original molecule, it is possible that modification involves cleavage of the hydrophilic aminoglycoside residue from the anthracene ring. The reaction may also involve electron transfer from the semi-quinone to other functionalities in the ring structure, particularly to the hydroquinone groups, acting to delocalize the radical on the anthracene ring. The conversion of the AnQ-Fe(III) complex to modified drug is shown at the left in the Figure.

Also as shown in the Figure, the AnQ-Fe(II) complex can combine with molecular oxygen, leading to hydrogen peroxide formation, and ultimately hydroxyl radicals, which can then propagate free radical reactions. This oxygen-mediated pathway is suggested by earlier studies on the possible mechanisms of DXR-stimulated membrane damage in vivo where initial semi-quinone formation may occur by enzyme-catalyzed electron transfer (Goormaghtigh 1984), and, in vitro, where direct oxygen participation was involved (Gutteridge 1984a).

Where the liposomes contain an unsaturated lipid (UL), a separate pathway involving oxidation and free radical damage may also be involved, as illustrated at the right in FIG. 3. Initial reaction of the lipid with molecular oxygen leads to an oxygenated species which can complex with ferric iron, as indicated. The lipid complex, after undergoing a redox reaction to form an iron/lipid-radical complex, can form both peroxide and oxide lipid radicals, which in turn can propagate free radical reactions in the lipid and drug components of the composition, leading to drug and lipid modifications, as indicated. This pathway involving lipid oxidation and free radical propagation is suggested by studies on the mechanism of peroxidative damage in liposomal membranes (see, for example, Gutteridge 1984b and Sunamoto).

Heretofore, attempts to limit peroxidative damage in liposomes have focused on free radical scavenging, typically by including in the liposomes a lipophilic free radical scavenger, such as .alpha.-T. In theory, such free radical scavengers have the capacity to block free radical propagation in lipids and lipid-associated drugs, such as anthraquinones, and therefore to limit peroxidation-relative damage to that produced by "early" radical formation reactions. Assuming that the radical-forming events are relatively benign in comparison to free radical propagation, it would be expected that little additional protection, above that provided by .alpha.-T, could be obtained. Therefore an important aspect of the present invention is the discovery that significantly greater reduction in lipid and drug modification in an anthraquinone/liposome composition can be achieved by a combination of lipophilic free radical quencher and a water-soluble protective agent which acts at the level of free radical formation.

The lipophilic free radical scavenger used in the composition of the invention is preferably .alpha.-T, or a pharmacologically acceptable analog or ester thereof, such as .alpha.-T succinate. Other suitable free radical scavengers include butylated hydroxytoluene (BHT), propyl gallate (Augustin), and their pharmacologically acceptable salts and analogs. Additional lipophilic free radical quenchers which are acceptable for parenteral administration in humans, at an effective level in liposomes, may also be used. The free radical quencher is typically included in the lipid components used in preparing the liposomes, according to conventional procedures. Preferred concentrations of the protective compound are between about 0.2 and 2 mole percent of the total lipid components making up the liposomes; however, higher levels of the compound, particularly .alpha.-T or its succinate analog, are compatible with liposome stability and are pharmacologically acceptable.

The water soluble protective agent is an iron-specific chelating agent selected from the class of natural and synthetic trihydroxamic acids and characterized by a very high binding constant for ferric iron (on the order of 10.sup.30) and a relatively low binding constant for 2-valence cations, such as calcium and magnesium. A variety of trihydroxamic acids of natural origin have been described, including compounds in the ferrichrome class, such as ferrichrome, ferrichrome A, and albomycin; compounds in the ferrioxamine class, including the ferrioxamines and ferrimycines; and compounds in the fusaramine class. The structure and iron coordination properties of these compounds have been reviewed (Emery).

One preferred chelator is ferrioxamine B, also known variously as ferrioxamine, deferoxamine, desferrioxamine B, and Desferal. This compound shows exceptional iron binding affinity and has been proven safe for parenteral use in humans in treating iron-storage disease and iron-poisoning (Keberle). The structure of iron-coordinated deferoxamine is shown in FIG. 4. As seen, the compound has three hydroxamic acid groups (--C(O)N(O)--) which can arrange octahedrally about the chelated iron, to form six symmetrical Fe-oxygen ligand bonds. The binding constants of ferrioxamine for iron and several 2-valence metals, including Fe.sup.2+, are given in Table V.

TABLE V ______________________________________ Metallic Ion Binding Affinity ______________________________________ Fe.sup.3+ 10.sup.31 Ca.sup.2+ 10.sup.2 Mg.sup.2+ 10.sup.4 Sr.sup.2+ 10.sup. Zn.sup.2+ 10.sup.11 Ni.sup.2+ 10.sup.10 Co.sup.2+ 10.sup.11 Cu.sup.2+ 10.sup.14 Fe.sup.2+ 10.sup.10 ______________________________________

The chelating agent is present in the composition at a concentration which is in molar excess of the ferric iron in the liposome suspension. Typically, aqueous media used in liposome preparation contains at least about 1-2 .mu.M ferric iron, and may contain up to 100 .mu.M or more ferric iron. For aqueous medium containing up to about 20 .mu.M iron, concentrations of chelating agent of about 50 .mu.M are preferred.

The chelating agent is preferably added to vesicle-forming lipids at the time of liposome formation, so that the lipids are protected against drug-promoted lipid oxidation damage during liposome preparation. Methods for preparing liposomes by addition of an aqueous solution of chelating agent are described below. Here it is noted only that the liposome suspension formed by this method contains chelating agent both in the bulk aqueous phase and in the encapsulated form, i.e., within the aqueous internal liposome region. Alternatively the chelating agent may be included in the suspension after liposome formation.

Experiments conducted in support of the present invention have examined the degree of protection of lipid and drug components in a composition containing both .alpha.-T and ferrioxamine. In one study, liposomes containing DXR and .alpha.-T were prepared substantially as in Example 1, either in the presence or absence of 50 .mu.m ferrioxamine. The drug/liposome compositions was then stored under anoxic conditions for 1 day at 4.degree. C., and the lipids were extracted conventionally by chloroform/methanol. The extracted lipids were dissolved in heptane and the UV spectra (190-300 nm) were monitored with a double-beam spectrophotometer. FIG. 5 shows the spectra obtained for lipids incubated in the presence (spectrum A) or absence (spectrum B) of chelator. The FIG. 5B spectrum shows strong peaks at about 233 nm, corresponding to conjugated diene formation and at about 275, corresponding to conjugated triene formation. The solid line spectrum in FIG. 5A shows little or no evidence of diene or triene formation, and spectral subtraction shows that diene and triene formation was inhibited more than about 99% in the ferrioxamine-containing composition.

In a second study, DXR/liposomes were prepared substantially as described in Example I, under nitrogen atmosphere or air, and in the presence or absence of .alpha.-T and ferrioxamine, as indicated in Table VI below. As in the experiment described above, the liposomes were composed of PG:PC:cholesterol, in a mole ratio of 3:7;4. .alpha.-T, when included, was present at a concentration of about 1.5 mole percent, and ferrioxamine, when included, was present at a concentration of 50 .mu.M. The liposomes were stored under anoxic conditions for 1 day at 4.degree. C., as above. Chemical modification of the drug was detected both by fluorescence emission spectroscopy and by assaying the change in drug toward a more lipophilic species, as described in Section A above.

TABLE VI __________________________________________________________________________ Hydrophilic Fluorescence Composition .alpha.T Desferal DXR Intensity Ratio PG PC CHOL (mole ratio) (50 .mu.m) Air % (590/560 nm) __________________________________________________________________________ 3 7 4 - - - 67 .+-. 2 1.4 .+-. 0.1 3 7 4 + - - 68.1 .+-. 2 1.5 .+-. 0.1 3 7 4 - + - 67.1 .+-. 2 1.6 .+-. 0.1 3 7 4 + + - 86 .+-. 2 1.9 .+-. 0.1 3 7 4 - - + 47 .+-. 3 1.3 .+-. 0.1 __________________________________________________________________________

The righthand column in Table VI shows the 590/560 nm fluorescence intensity ratios of the hydrophobic derivative of DXR for the five compositions. The data indicates two types of drug modification, as discussed above: (1) possible cleavage of the sugar moiety to give a more hydrophobic DXR product, and (2) damage to the anthracene ring which affects the 590/560 ratio of the drug. As will be recalled from section A, the 590/560 ratio is largest in unmodified DXR, and decreases with anthracene ring modification. The ratios observed show that .alpha.-T alone and ferrioxamine along both gave moderate protections against drug modification, with ferrioxamine having a greater protective effect. Significantly, the combination of .alpha.-T and ferrioxamine provided much greater protection than either protective agent alone, or what would be predicted from a sum of the two individual agent effects. The 1.9 value observed was substantially the same as that observed immediately after liposome preparations under nitrogen, indicating nearly complete protection of the drug through the storage period.

The studies just reported suggest that the protective mechanism of ferrioxamine is qualitatively different than that of .alpha.-T. It is like that ferrioxamine acts primarily to inhibit peroxide generation and other "early" free-radical generating events, whereas .alpha.-T acts to quench free-radical reactions being propagated within the lipid bilayer. The likely points of inhibition by trihydroxamic acid chelator are indicated by double lines in FIG. 3.

The inhibitory effect of trihydroxamic acid chelation on peroxidation damage in liposomes is also in contrast to the peroxidation-stimulating effect observed when a conventional tetraacetic acid-type chelator, such as ethylenediaminetetraacetic acid (EDTA), is added to liposomes. The observed difference between trihydroxamic acid and tetraacetic acid chelators, in their ability to protect against peroxidative damage in liposomes may be due to the difference in number of iron coordination sites available in the two species. Whereas a trihydroxamic chelator is able to form octahedral ligand bonding to all six ferric iron coordination sites, tetraacetic acid chelators are limited to four-site binding, leaving two iron coordinations sites available for participation in redox reactions. In this regard, it is noted that DTPA, a pentaacetic acid chelator, also shows a protective effect on anthraquinone modification in vitro, in the presen