Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

It is well established that the therapeutic activity of anticancer agents can be improved through application of liposomal drug carrier technology (Fielding, 1991; Sugarman and Perez-Soler, 1992; Kim, 1993). In general, liposomes engender pharmacokinetic and biodistribution characteristics that lead to increases in therapeutic activity and/or reductions in drug-related toxicities (Fielding, 1991; Mayer et al., 1994). Although the mechanism of therapeutic activity for liposomal anticancer drugs is not well understood, studies have suggested that increased drug exposure at the site of tumor growth is important (Gabizon and Papahadjopoulos, 1988; Wu et al., 1993; Mayer et al., 1994; Bally et al., 1994). These increases in tumor drug levels result from preferential accumulation of the liposome carrier within tumors (Gabizon, 1992; Wu et al., 1993; Bally et al., 1994; Ogihara-Umeda et al., 1994; Uchiyama et al., 1995). It is important to note, however, that there is no evidence suggesting that the encapsulated form of the drug is therapeutically active. It is postulated, therefore, that antitumor activity is mediated by drug released from regionally localized liposomes (Mayer et al., 1994).

The emphasis of investigators developing liposomal anticancer agents has been, for the reasons cited above, on the use of liposomal lipid compositions that are less permeable to the encapsulated agent and exhibit increased circulation lifetimes. Liposomes that are retained in the plasma compartment for extended periods of time exhibit a greater tendency to accumulate in regions of tumor growth (Gabizon and Papahadjopoulos, 1988; Gabizon, 1992; Wu et al., 1993). However, the kinetics of this extravasation process, where liposomes leave the blood compartment and enter an extravascular site, are slow (Nagy et al., 1989; Bally et al., 1994). Efficient drug delivery can, therefore, be achieved only with liposomes that effectively retain the drug after systemic administration. The problem that arises through applications of liposomal carriers that are optimized for enhanced drug retention concerns evidence from studies with liposomal doxorubicin that demonstrate reduced therapeutic activity, despite efficient delivery of drug to tumors (Mayer et al., 1994). A balance between doxorubicin retention (to maximize drug accumulation in a site of tumor growth) and release (to effect therapy) has not been established.

Attempts to improve the therapeutic properties of liposomal doxorubicin formulations through changes in drug release characteristics have been unsuccessful due to specific adverse effects of free doxorubicin, including cardiotoxicity (Minow et al., 1975) and drug-mediated free radical damage (Rajagopalan et al., 1988). More specifically, effective modulation of doxorubicin release rates has been achieved with relatively simple changes in liposomal lipid composition (Mayer et al., 1989; Bally, et al., 1990); however, liposomal formulations of doxorubicin that release drug after i.v. administration exhibit enhanced toxicity and increased doxorubicin accumulation in cardiac tissue. This effect is most dramatic for doxorubicin formulations prepared using DMPC/Chol liposomes, which release >90% of the encapsulated contents in the blood compartment within 24 hr after i.v. administration and are 3 times more toxic than free drug (Mayer et al., 1994).

We examined the influence of liposome drug release properties on the biological activity of mitoxantrone. The rationale for selecting mitoxantrone is based on the fact that this drug is less cardiotoxic than doxorubicin (Weiss, 1989) and is not capable of generating free radical damage in nonproliferating cells (Durr, 1984). It is demonstrated that the in vivo rate of mitoxantrone release from DMPC/Chol liposomes is at least 68-fold greater than that from DSPC/Chol liposomes. The pharmacodynamic characteristics of these formulations have been characterized using murine tumor models where the primary site of tumor progression is in the liver. The data illustrate how a balance between drug release characteristics and liposome-mediated drug delivery to sites of tumor progression is required for optimal therapeutic activity.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. Novantrone (mitoxantrone hydrochloride) was obtained from the British Columbia Cancer Agency and is a product of Lederle Laboratories Division, Cyanamid Canada (Montreal, Canada). DSPC and DMPC were purchased from Avanti Polar Lipids (Alabaster, AL). HEPES, citric acid, Chol, nigericin and Sephadex G50 (medium) were purchased from Sigma Chemical Co. (St. Louis, MO). Dibasic sodium phosphate was obtained from Fisher Scientific (Fair Lawn, NJ). Bio-Gel A15m gel was purchased from Bio-Rad (Mississauga, Ontario, Canada). Fetal bovine serum was obtained from Gibco Laboratories (Grand Island, NY). Solvable was obtained from NEN Research Products (DuPont Canada, Mississauga, Ontario, Canada). [¹⁴C]Mitoxantrone, used as a tracer, was generously donated by the American Cyanamid Company (Montreal, Canada). [³H]CDE, a lipid marker that is not exchanged or metabolized in vivo (Stein et al., 1980), was purchased from Amersham (Oakville, Ontario, Canada). Aquacide II was purchased from Terochem Laboratories Ltd. (Edmonton, Canada). L1210 and P388 tumor cell lines were originally purchased from the National Cancer Institute Tumor Repository (Bethesda, MD), and cells were obtained from ascites fluid generated weekly by passage in BDF1 mice. Cells were used for experiments after the third passage and before the 20th. We, therefore, revert back to the original tumor stock every 4 to 6 months. Female CD1 and BDF1 mice (8-10 weeks old) were purchased from Charles River Laboratories (St. Constant, Quebec, Canada).

Preparation of liposomes. DSPC/Chol (55:45, mol/mol) and DMPC/Chol (55:45, mol/mol) liposomes were prepared using well-established extrusion technology (Hope et al., 1985). The indicated phospholipid and Chol molar ratios were dissolved in chloroform and dried to a homogeneous lipid film under a stream of nitrogen gas. This lipid film was dried under vacuum to remove any residual chloroform. The lipid film was then hydrated in a 300 mM citric acid buffer (pH 4.0) at a concentration of 100 mg of lipid/1 ml of buffer. The resulting multilamellar vesicle mixture was frozen and thawed five times (Mayer et al., 1986) and extruded through three stacked 100-nm polycarbonate filters (Nuclepore, Pleasanton, CA), using an extrusion device (Lipex Biomembranes Inc., Vancouver, Canada). The resulting large unilamellar vesicles were sized by quasielastic light scattering using a Nicomp 270 submicron particle sizer (Pacific Scientific, Santa Barbara, CA) operating at 632.8 nm. The mean diameter of these liposomes was 100 to 120 nm.

Transmembrane pH gradient loading of mitoxantrone. Mitoxantrone was encapsulated using a transmembrane pH gradient-driven loading procedure (Mayer et al., 1985; Madden et al., 1990). The procedure used was analogous to that used for vincristine (Boman et al., 1993) and consisted of incubation of liposomes at 65°C for 10 min before addition of sufficient mitoxantrone to achieve a final drug to lipid weight ratio of 0.1. The pH of this mixture was then increased from pH 4.0 to 7.2 by the addition of 350 µl of 0.5 M Na₂HPO₄ buffer to 1.0 ml of the drug/liposome mixture. The resulting mixture was incubated at 65°C for an additional 15 min. Encapsulation efficiency for mitoxantrone was determined at three different temperatures (37°C, 50°C and 65°C) using size-exclusion chromatography on mini-spin columns made of Sephadex G-50 (Madden et al., 1990). Aliquots of the sample (100 µl) were taken at intervals over a 2-hr time period and assayed for drug encapsulation. Drug and lipid concentrations in the samples collected in the void volume of these columns were determined by measuring [³H]CDE and [¹⁴C]mitoxantrone. Radioactivity was assessed by mixing the samples with 5 ml of Pico-Fluor 40 scintillation cocktail (Packard, Meriden, CT) and counting them with a Packard 1900 scintillation counter (Packard).

In vitro characteristics of liposomal mitoxantrone. For release studies, liposomal mitoxantrone formulations were prepared as outlined above. The resulting drug-loaded liposomes were transferred into 25-mm-diameter Spectrapor dialysis tubing (10,000-12,000 molecular weight cut-off; Spectrum Medical Industries, Los Angeles, CA), and the samples (3 ml) were dialyzed against 1 liter of HEPES-buffered saline at 37°C. At the indicated time points, 100-µl samples were taken from the dialysis bag and assayed for drug and lipid using the mini-spin columns, as described above. The experiment was then repeated with addition of nigericin (an ionophore that collapses the pH gradient) to the sample and external buffer to a concentration of 120 nM.

Plasma elimination and distribution studies. Female CD1 mice (20-25 g, 4/group) were injected with a 10 mg/kg drug dose via the lateral tail vein. At 1, 4, 24 and 48 hr, animals were sacrificed by CO₂ asphyxiation, and whole blood was collected via cardiac puncture and placed into EDTA-coated tubes (Microtainers; Becton Dickinson). Plasma was isolated after centrifugation of whole blood at 500 × g for 10 min. Aliquoted plasma samples (100 µl) were mixed with 5 ml of Pico-Fluor 40 and measured for ³H and ¹⁴C.

Establishment of the MTD and efficacy studies. The MTD was determined in limited dose-ranging studies, approved by the local Animal Care Committee (University of British Columbia). The work was completed in accordance with the guidelines of the Canadian Council on Animal Care. Briefly, female BDF1 mice in groups of two were given drug by a single i.v. injection. Weight loss and signs of stress/toxicity were monitored twice daily for 30 days. If individual animals lost >25% of the original body weight, they were sacrificed. The MTD was estimated as the dose where tumor-free animals survived for a period of 30 days after drug administration. At the end of the 30-day period, animals were sacrificed by CO₂ asphyxiation and necropsies were completed to identify any additional toxicities. The exact LD₁₀ dose of the different mitoxantrone formulations was not determined. Such toxicity studies are not approved by the Canadian Council on Animal Care or the institutional Animal Care Committee.

View larger version (41K): [in this window] [in a new window]   Fig. 4.   Liver and spleen weights of untreated BDF1 mice () and BDF1 mice previously (7 days) injected i.v. with 104 L1210 cells (). At day 7, livers and spleens were taken from the mice and weighed. The results were obtained from four animals and error bars indicate the S.D.

View larger version (171K): [in this window] [in a new window]   Fig. 5.   Hematoxylin and eosin staining of paraffin-embedded livers of untreated BDF1 mice (A) and BDF1 mice injected i.v. with 105 P388 cells 1 day (B), 3 days (C) and 7 days (D) earlier. Structural features are as follows: H, hepatocytes; S, sinusoid; V, blood vessel; rbc, red blood cell; K, Kuffer cell. Arrowheads, inflammatory infiltrate; arrows, disorganization and lack of hepatocytes during tumor cell invasion. Bar in C, 30 µm. E, endothelial cell.

Statistical analysis. Analysis of variance was performed on the results obtained after administration of the two liposomal formulations and free mitoxantrone. Common time points were compared using the post hoc comparison of means and Scheffé test. Differences were considered significant at P < .05. For the efficacy studies, survival times (in days) were ranked and statistically analyzed using a Cox's F test. Comparisons indicated as having statistical significance had P values from the Cox's F test of <.05.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

In vitro mitoxantrone uptake and release characteristics. Studies evaluating in vitro drug accumulation in liposomes prepared from DMPC (C₁₄)/Chol and DSPC (C₁₈)/Chol at 37°C, 50°C and 65°C are shown in figure 1. At 37°C, <15% of the drug was encapsulated in either liposome formulation over the 2-hr time course. In contrast, >98% of the drug was efficiently entrapped when the incubation temperature was increased above 50°C. The time required to achieve maximum uptake was 45 min and <5 min when the incubation temperature was 50°C and 65°C, respectively. Uptake rate was enhanced slightly at 50°C for the DMPC/Chol system, compared with the DSPC/Chol system. The results suggest that the phase transition temperature of the phospholipid species does not markedly affect mitoxantrone loading characteristics. This result is consistent with the in vitro drug release studies (fig. 2) that demonstrate no difference in drug release from either liposomal formulation. The in vitro release assay used is based on dialysis against a large volume (1 liter) of buffer with (results not shown) and without 10% fetal bovine serum. Under these conditions, free mitoxantrone equilibrates across the dialysis membrane in <8 hr. In contrast, <2% drug release was observed from the liposomal formulations over a 72-hr incubation period at 37°C. Figure 2 also includes data obtained for mitoxantrone-loaded liposomes incubated with nigericin, a H⁺/monovalent cation exchanger. In the presence of nigericin, the liposomes exhibited a transmembrane pH gradient of <0.5 unit (results obtained using the pH gradient probe [¹⁴C]methylamine are not shown). Although drug release rates were increased in the presence of nigericin, there were minor differences in release rates observed for the two liposomal systems studied. After the 48-hr incubation period, the DMPC/Chol liposomes released <30% of the encapsulated drug, in comparison with 20% drug release observed for the DSPC/Chol systems.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Effect of temperature on pH gradient loading of mitoxantrone into DSPC/Chol (A) and DMPC/Chol (B) liposomes, at 37°C (), 50°C () and 65°C (). At time 0, mitoxantrone and the liposomes were mixed together at a drug to lipid ratio of 0.1 (w/w). Encapsulated drug was determined by the mini-spin column procedure described in "Materials and Methods." Duplicate samples were taken and [3H]CDE and [14C]mitoxantrone were measured as described in "Materials and Methods." Data represent the average values ± S.D. of four measurements.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Release of mitoxantrone from DSPC/Chol () and DMPC/Chol () liposomes in HEPES-buffered saline at 37°C. Solid lines, absence of nigericin. Dashed lines, addition of nigericin at time 0. Samples (100 µl) were taken from the dialysis bags, applied to Sephadex G-50 mini-spin columns in duplicate and centrifuged at 500 × g for 2 min. Duplicate samples were taken from the resulting mixture, and 3H and 14C were measured as described in "Materials and Methods." Data represent the average values ± S.D. of at least four measurements for studies in the presence of nigericin.

In vivo plasma elimination of liposomal lipid and mitoxantrone. Results in figure 3 show that the plasma elimination of liposomal lipid is similar after i.v. administration of mitoxantrone-loaded DMPC/Chol and DSPC/Chol liposomes (fig. 3A). An estimation of the amount of mitoxantrone retained in the liposomes remaining in the circulation can be made by determining the ratio of mitoxantrone to lipid at the indicated time points, an estimation that assumes that the level of free drug in the plasma of animals given liposomal mitoxantrone is negligible. The results shown in figure 3B demonstrated greater release of mitoxantrone from DMPC/Chol liposomes than DSPC/Chol liposomes (P < .05 for 24- and 48-hr time points). For DMPC/Chol liposomes, 73% of the mitoxantrone originally associated with the carrier was released within 48 hr. In contrast, <5% of the drug was released from DSPC/Chol liposomes. Between the 4-hr and 48-hr time points, the rate of mitoxantrone release was estimated to be 1.7 and <0.025 µg drug/µg lipid/hr for DMPC/Chol and DSPC/Chol liposomes, respectively. These results are consistent with those obtained using entrapped doxorubicin (Mayer et al., 1994) and clearly demonstrate that control of in vivo mitoxantrone release rates can be achieved through simple changes in liposomal lipid composition. It should be noted that plasma drug levels obtained after administration of free drug are significantly less than those obtained with the liposomal formulations. This is indicated in figure 3C, a plot of plasma drug levels measured after i.v. administration of the indicated formulation. Trapezoidal AUC analysis of these plasma drug levels, from 1 to 48 hr, indicated plasma AUCs of 0.01, 167.86 and 229.86 µg drug/100 µl plasma/hr after administration of free mitoxantrone, DMPC/Chol/mitoxantrone and DSPC/Chol/mitoxantrone, respectively.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   In vivo plasma elimination of mitoxantrone from DSPC/Chol () or DMPC/Chol () liposomes or free mitoxantrone (). Liposomes were loaded with mitoxantrone at a drug to lipid ratio of 0.1 (w/w). Female CD1 mice were injected at a 10 mg/kg drug dose i.v. via the lateral tail vein. A, elimination of lipid from the plasma compartment over 48 hr. B, change in the drug to lipid ratio over the 48-hr time period. C, elimination of the free drug from the plasma compartment over 48 hr. Data represent the means ± S.D. obtained from at least four animals. *P < .05.

Acute toxicity of free and liposomal mitoxantrone. Formal LD₁₀ and LD₅₀ studies are not sanctioned by the Canadian Council on Animal Care; therefore, toxic dose range-finding studies in tumor-free female BDF1 mice were conducted using only 2 mice/dose. These limited dose escalation studies suggested that the MTD of free drug was approximately 10 mg/kg. When drug was encapsulated in DSPC/Chol or DMPC/Chol, the MTD increased to approximately 30 mg/kg. At this dose, 100% of the animals treated survived for >30 days. Necropsies suggested no gross abnormalities in any of the tissues examined. An evaluation of drug-induced weight loss, however, suggested that the DMPC/Chol liposomal formulation was more toxic than the DSPC/Chol system. This result was confirmed in efficacy experiments, where changes in weight were measured over 14 days after initiation of treatment. For animals given 10⁴ L1210 cells i.v. and treated 24 hr later with mitoxantrone, the maximum therapeutic dose of free and liposomal mitoxantrone was 10 and 20 mg/kg, respectively. The nadir in weight loss after treatment of tumor-bearing animals occurred between day 12 and 13; at this time point animals treated with free drug (10 mg/kg) lost <25% of their original body weight and had to be killed. In contrast, animals treated with DSPC/Chol and DMPC/Chol mitoxantrone (20 mg/kg) exhibited a body weight loss of 8% and 25%, respectively.

L1210 and P388 antitumor activity of free and liposomal mitoxantrone. The murine tumor models used for evaluating the antitumor activity of liposomal mitoxantrone were based on i.v. injection of L1210 or P388 cells. Although these cells are typically used to initiate ascitic tumors after i.p. inoculation, the cells can also be given by alternate routes of injection. When given i.v., primary sites of cell seeding include the liver and spleen. Evidence to support this is provided in figures 4 and 5. Seven days after i.v. inoculation of 10⁴ L1210 cells, the liver and spleen of the recipient animals showed a 2 and 3 fold increase in weight, respectively (fig. 4). Untreated animals had to be sacrificed as a result of significant tumor-related disease within 10 days. Histological studies (not shown) indicated the presence of massive, diffuse, cell infiltration throughout the liver. There were no other gross abnormalities in any other organs or tissues derived from these animals. For mice injected with P388 cells, liver and spleen weight increases were also observed. The histopathological analysis, however, revealed discrete foci of tumor cells that progressively became larger over a 7-day time course (fig. 5). In our laboratory these i.v. tumor models were typically not responsive to chemotherapy with doxorubicin or vincristine (free or liposomally encapsulated drug), and thus these models were used as a stringent measure of mitoxantrone antitumor activity.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Survival times of BDF1 mice injected with 104 L1210 cells (A) or 105 P388 cells (B) i.v. via the lateral tail vein and treated with mitoxantrone. Twenty four hours after tumor cell inoculation, the mice were treated with a 10 mg/kg dose of free mitoxantrone () or DSPC/Chol () or DMPC/Chol () liposomal formulations. Untreated (saline) animals served as controls ().

Drug and liposomal lipid uptake in liver. The results presented to this point demonstrate that 1) the rate of mitoxantrone release from DMPC/Chol liposomes after i.v. administration was significantly greater than that measured for DSPC/Chol liposomes and 2) DMPC/Chol liposomal mitoxantrone was significantly more efficacious than free drug or DSPC/Chol liposomal mitoxantrone when tested against a tumor model where the primary site of disease progression is in the liver and spleen. It has been proposed that differences in the therapeutic activity of encapsulated anticancer drugs would be a consequence of liposomal characteristics that regulate the drug exposure within sites of disease progression. Therefore, in addition to assessing drug release from liposomes in the plasma compartment, it is also important to correlate antitumor activity with drug levels at the site of tumor progression. For this reason, drug delivery to the liver, a primary site of disease progression for the i.v. tumor models used, was evaluated. Results, shown in figure 7, were obtained in tumor-free CD1 mice. It should be noted that drug/liposome plasma elimination and biodistribution data were similar in tumor-free CD1 and tumor-bearing BDF1 mice. As shown in figure 7A, liposomal lipid accumulation in the liver was similar for both DSPC/Chol and DMPC/Chol liposomal mitoxantrone formulations over 48 hr. Unlike doxorubicin (Bally et al., 1990), the presence of entrapped mitoxantrone did not cause significant reductions in liposomal lipid accumulation in the liver. Empty DMPC/Chol liposomal lipid uptake in the liver, for example, was not significantly different from that of mitoxantrone-loaded DMPC/Chol liposomes. Figure 7B demonstrates that the level of mitoxantrone achieved in the liver after i.v. administration of DMPC/Chol liposomal mitoxantrone is less than that observed for DSPC/Chol liposomal mitoxantrone (P < .01 for the 48-hr time point). AUC analysis of liver drug levels, from 1 to 48 hr, indicate liver AUCs of 2564, 1810 and 1070 µg drug/g liver/hr after i.v. administration of DSPC/Chol/mitoxantrone, DMPC/Chol/mitoxantrone and free mitoxantrone, respectively. Notably, the liposomal formulation that engenders the greatest level of drug exposure in the liver (DSPC/Chol) did not provide the greatest therapeutic benefit.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Lipid and drug levels in the liver of mice after injection of DSPC/Chol/mitoxantrone (), DMPC/Chol/mitoxantrone (), empty DMPC/Chol liposomes () and free mitoxantrone (). The liposomal lipid dose was 100 mg/kg, and the drug dose was 10 mg/kg. A, amount of lipid per gram of liver; B, amount of mitoxantrone per gram of liver measured over 48 hr. Drug and lipid levels were determined as described in "Materials and Methods." The data represent the mean ± S.D. from at least three animals. **P < .01.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

The therapeutic index of most anticancer drugs is narrow, with severe toxic side effects occurring within the dose range required to mediate effective therapy. Although a variety of experimental strategies have been developed to improve the therapeutic index of anticancer drugs, these strategies have a common aim, i.e., to improve drug specificity. The principle benefit postulated for the use of liposomes as carriers of anticancer drugs involves liposome-mediated increases in drug delivery to the disease site and decreases in drug delivery to healthy tissues and organs (Sugarman and Perez-Soler, 1992; Mayer et al., 1994). Using this as a rationale, we and others have emphasized the importance of designing liposomes that have a greater propensity to accumulate within disease sites (Gabizon and Papahadjopoulos, 1988; Gabizon, 1992; Ogihara-Umeda et al., 1994; Uchiyama et al., 1995). In this regard, liposome carriers have been optimized with respect to maximizing the amount of drug contained per liposome (Mayer et al., 1989, 1994), increasing drug retention characteristics (Mayer et al., 1989; Boman et al., 1994) and augmenting the circulation lifetime of the drug-loaded carrier (Gabizon, 1992; Wu et al., 1993). However, it can be suggested that the therapeutically active component of a liposomal anticancer drug formulation is the free drug. We believe that the primary source of free drug is from regionally localized liposomes (Mayer et al., 1994). Therefore, our research has attempted to establish a balance between efficient liposome delivery to the disease site and controlled drug release. The latter can be achieved for certain drugs by changing the liposomal lipid composition (Mayer et al., 1994; Boman et al., 1994). This study illustrates how controlled drug release can engender significant improvements in therapeutic activity of the anticancer drug mitoxantrone.

It was surprising that differences in drug accumulation and leakage rates for DSPC/Chol and DMPC/Chol liposomes were not substantial when evaluated in vitro, even when the liposomes were incubated in the presence of nigericin. The phase transition temperatures for DSPC and DMPC are 55.3°C and 23.9°C, respectively (Lewis et al., 1987), and it was anticipated that differences in the gel-to-liquid crystalline phase transition of these phospholipids would be reflected by changes in permeability characteristics. This was evident for liposomal formulations of vincristine, where a good correlation between phospholipid phase transition temperatures and drug leakage in vitro was observed (Boman et al., 1993). Collapse of the transmembrane pH gradient did increase drug release from the liposomal formulations; however, no substantial differences in the rate of drug release from the DSPC/Chol and DMPC/Chol liposomes were noted. After pH gradient-mediated uptake, drugs such as mitoxantrone can form insoluble precipitates within the liposome (Madden et al., 1990; Bolotin et al., 1994). If this is the case, then permeability characteristics of the drug in a precipitated form may be less dependent on membrane characteristics or the presence of a residual transmembrane pH gradient. It is not understood, however, why differences in drug permeability become apparent in vivo.

Mitoxantrone was selected as a model drug for these studies for two reasons. First, the drug loading and release characteristics of mitoxantrone are comparable to those of doxorubicin (Madden et al., 1990). Second, mitoxantrone is less cardiotoxic than doxorubicin, and mitoxantrone is not capable of generating free radical-mediated toxicities in nondividing cell populations (Durr, 1984). Liposome-mediated increases in mitoxantrone MTD observed in this report are comparable to those reported for a liposomal mitoxantrone formulation prepared using an anionic lipid-drug complex (Schwendener et al., 1991, 1994). The liposomal formulations evaluated here, however, exhibit significantly better drug retention characteristics than do those formulations described by Schwendener et al. This is reflected in higher blood levels and improved circulation lifetimes for mitoxantrone encapsulated in the phosphocholine/Chol-based liposomal carriers. Differences in drug release characteristics may be a consequence of the use of anionic lipids. Anionic lipids increase liposome elimination rates (Hwang, 1987) and have been shown to enhance release of the anthracycline doxorubicin even when the drug is encapsulated using the transmembrane pH gradient-loading procedure (Mayer et al., 1989). Clearly, when the rate of drug dissociation from the liposomal carrier is very rapid, carrier-mediated changes in drug pharmacokinetics and biodistribution are not significant and changes in biological activity (relative to drug administered in free form) are minimal.

Studies evaluating the therapeutic activity of DSPC/Chol and DMPC/Chol liposomal mitoxantrone (fig. 6; table 1) establish that both drug delivery and drug release are important attributes of an optimal liposomal anticancer drug formulation. The i.v. L1210 tumor model was selected for these studies in part because L1210 cells seed primarily in the liver and spleen after i.v. administration. It is well established that these tissues are primary sites of liposome accumulation (Hwang, 1987; Sugarman and Perez-Soler, 1992). Furthermore, other investigators have shown, using experimental models of liver cancer, that the therapeutic activity of liposomal formulations of a novel platinum compound and doxorubicin analog is enhanced, compared with free drug (Perez-Soler, 1989; Gabizon, 1992). It is perplexing, therefore, that models of liver cancer have not been used more frequently to characterize the pharmacodynamic behavior of liposomal anticancer drugs. We have shown that mitoxantrone delivery to the liver is enhanced when DSPC/Chol liposomes are used, in comparison with DMPC/Chol liposomes (fig. 7B). Increased liposomal drug exposure in this tissue, however, does not result in improved therapeutic activity. In fact, the DMPC/Chol liposomal formulation, which exhibits controlled release characteristics and a reduced capacity to deliver drug to the liver, was significantly more effective. Thus, it is not sufficient to develop drug carriers that accumulate at the disease site in high levels; one must also engineer appropriate drug release rates.

Unpublished studies completed in this laboratory have demonstrated, using the i.v. L1210 tumor model, that egg phosphatidylcholine/Chol liposomal doxorubicin, DSPC/Chol liposomal doxorubicin and DSPC/Chol liposomal vincristine are relatively ineffective in treating this model, typically producing ILS of <50% at the maximum therapeutic doses. A possible explanation for the effectiveness of liposomal mitoxantrone may be related to the fact that this encapsulated drug does not appear to affect the liver Kupffer cells. We have shown, for example, that empty and mitoxantrone-loaded liposomes exhibit comparable plasma elimination profiles and comparable levels of uptake in liver (fig. 7). This is contrary to effects observed with vincristine-loaded (Boman et al., 1994) or doxorubicin-loaded (Bally et al., 1990) liposomes, where encapsulated drug significantly increases the circulation lifetime of the liposomal carrier. This effect is due, in part, to drug-mediated blockade of phagocytic cells in the liver. It can be suggested that the blockade effect may adversely affect the therapeutic activity of liposomal anticancer drugs in treating tumors that are progressing in the liver and that phagocytic cells in the liver may have a significant role in defining the antitumor activity of liposomal mitoxantrone.

In conclusion, we have developed a liposomal mitoxantrone formulation with significant therapeutic activity. The plasma elimination curves and biodistribution data demonstrate that effective control of both drug release characteristics and target site delivery can work synergistically to achieve optimal therapy. Our laboratory will continue to study liposomal formulations of mitoxantrone, with the following objectives: 1) to further improve the therapeutic index of the drug; 2) to target the liposomal drug for use in treatment of specific cancers, such as hepatocellular carcinomas, and/or 3) to develop novel formulations that effect delivery of the drug-loaded carrier to tumor cells, thereby bypassing normal cellular drug uptake mechanisms. The DMPC/Chol formulation described here meets the first objective.