The use of glucocorticoids (GCs) in rheumatoid arthritis is limited by side effects related to unfavorable pharmacokinetics and biodistribution. Liposomal GC formulations have been studied since the 1970s in an attempt to overcome this obstacle, but none has entered clinical use. We undertook this study to determine whether a novel approach could overcome the limitations that have thus far prevented the clinical use of these formulations: low drug:lipid ratio, low encapsulation efficiency, and lack of controlled release.
We used ∼80-nm sterically stabilized (pegylated) nanoliposomes (NSSLs), which were remote-loaded with an amphipathic weak acid GC (such as methyl prednisolone hemisuccinate) utilizing an intraliposome (aqueous compartment)–high/extraliposome (bulk medium)–low transmembrane calcium acetate gradient. This unique method actually “traps” the GC in the liposomal aqueous phase as a calcium–GC precipitate.
Our liposome formulation exhibited high encapsulation efficiency (94%) and a high drug:lipid mole ratio (0.41) and demonstrated controlled release of the encapsulated GC during systemic circulation and in inflamed paws in rats with adjuvant-induced arthritis. In addition, both in arthritic rats and in a Beagle dog, we showed the pharmacokinetic advantage of using liposomes as GC carriers. Finally, we demonstrated the superior therapeutic efficacy of our liposome formulation over that of free GCs in arthritic rats, both in early and in peak disease stages.
Amphipathic weak acid GCs remote-loaded into ∼80-nm NSSLs overcome past limitations of liposomal GC formulations. The unique loading method, which also leads to controlled release, improves the therapeutic effect both systemically and locally. Such a development has great potential for improving GC therapy.
Glucocorticoid (GC) treatment is widely used in the treatment of rheumatoid arthritis (RA) as well as other inflammatory joint diseases because of its antiinflammatory and immunosuppressive features (1–3). GCs have a beneficial role in both short-term and long-term treatment of RA. In short-term use, GCs are more effective antiinflammatory agents than are nonsteroidal antiinflammatory drugs, and their long-term use has been shown to stop progression of bone erosions caused by RA, similar to the other disease-modifying antirheumatic drugs (DMARDs) (1). However, the systemic application of GCs is often accompanied by substantial side effects, especially when long-term treatment is used. Due to the fast clearance of GCs from the body, frequent high-dose intravenous (IV) and/or oral administration is required to achieve efficacy, but this affects many tissues and causes serious side effects. Therefore, improvement of the therapeutic efficacy would be beneficial.
Intraarticular (IA) administration of GCs serves as an alternative to systemic treatment. It enables the use of lower doses of GC, since it achieves sufficient levels of the drug at the lesion with a lower level of side effects compared with systemic administration. However, the use of IA injection of GCs in RA is limited to monarthritis or oligoarthritis of the large joints; it cannot be used for patients with polyarthritis, which involves many large and small joints.
A systemic “targeted” administration that will bring the GC selectively to the inflamed joints can increase the therapeutic index of GCs and enable their use for both short-term treatment and prolonged therapy for polyarthritis while minimizing their dose-dependent side effects. This can be achieved by GC delivery via liposomes, and especially via sterically stabilized nanoliposomes (NSSLs) (4, 5) (also referred to as pegylated liposomes). The NSSLs are unique in terms of their prolonged plasma circulation time and also their ability to extravasate at tumor and inflammation sites (6–11). Although GC delivery via liposomes is one of the earliest attempts to use liposomes as drug carriers, until now no liposomal GC formulation has reached clinical application.
A systemic liposome formulation needs to fulfill several key requirements in order to be efficacious and attractive enough to justify development as a drug: 1) it should accumulate at the inflammation sites as a result of prolonged circulation due to reduction in uptake by the reticuloendothelial system; 2) during the prolonged circulation, the drug leakage rate from the NSSLs should be slow; 3) the liposomal formulation should be chemically and physically stable during storage; 4) liposomes should have a high drug:lipid mole ratio in order to avoid problems associated with toxicity induced by high liposome dose; and 5) high efficiency of drug encapsulation is required (>70%). None of the GC–liposome formulations developed so far meets all these requirements.
In this report we describe the successful development of an NSSL formulation encapsulating GC prodrugs (NSSLs-GC) in which these 5 criteria are achieved, and we demonstrate its superiority to the nonliposomal (free) GC. Our formulation is based on a novel approach, using water-soluble amphipathic weak acid GC prodrugs, such as methylprednisolone hemisuccinate (MPS) or betamethasone hemisuccinate (BMS), which are efficiently remote-loaded into NSSLs (∼80 nm) utilizing an intraliposome (aqueous compartment)–high/extraliposome (bulk medium)–low transmembrane calcium acetate gradient (12). (See Materials and Methods for a more detailed explanation.)
The therapeutic efficacy of the NSSLs encapsulating MPS (NSSLs-MPS) and NSSLs encapsulating BMS (NSSLs-BMS) was evaluated in Lewis rats with adjuvant-induced arthritis (AIA) both at an early disease stage (before clinical signs appear) and at the peak of the disease. In both cases, superior efficacy of the NSSLs-GC over the free (nonencapsulated) drug was clearly demonstrated.
MATERIALS AND METHODS
Hydrogenated soybean phosphatidylcholine was obtained from Lipoid (Ludwigshafen, Germany). The hydrogenated soybean phosphatidylcholine has an iodine value of 3.0, an acyl chain composition of 86% stearic acid (C18:0), 13% palmitic acid (C16:0), and <1% other acyl chains, and a gel-to–liquid crystalline phase transition temperature (Tm) of 52.5°C (13). Cholesterol (>99% pure) was obtained from Sigma (St. Louis, MO). PEG-DSPE-2000 was obtained from Genzyme Pharmaceuticals (Liestal, Switzerland). The prodrug MPS sodium salt was obtained from Pharmacia (Puurs, Belgium). BMS was obtained from Steraloids (Newport, RI). All other chemicals, including buffers, were of analytic grade or better and were obtained from Sigma. Highly pure water was obtained using the WaterPro PS HPLC/Ultrafilter Hybrid System (Labconco, Kansas City, MO), which provides low levels of total carbon and inorganic ions in sterile pyrogen-free water of 18.2 MΩ electrical resistance.
Female inbred Lewis rats and a Beagle dog were obtained from Harlan (Jerusalem, Israel). The animal studies were reviewed and approved by the Ethics Committee and the Institutional Animal Care and Use Committee.
Remote loading of amphipathic weak acid GC prodrugs.
Remote loading was performed using our approach as described elsewhere (12) and is briefly explained here. [Ca acetate]liposome >> [Ca acetate]medium transmembrane gradient was used as the driving force for the remote loading of the amphipathic weak acid GC prodrugs. Such a gradient is characterized by an excess of the membrane-impermeable Ca2+ ions over the membrane-permeable acetic acid; the leakage of protonated acetic acid of the liposomes also leads to creation of a pH gradient in which the pHliposome > pHmedium. In such liposomes, the amphipathic weak acid, such as MPS, in its protonated (uncharged) form is “pumped” into the slightly alkaline intraliposomal aqueous phase by diffusing through the lipid bilayer. There it loses its proton, becomes negatively charged, and forms a poorly soluble MPS (or BMS)–calcium salt, which precipitates in the intraliposomal aqueous phase. This precipitation stabilizes loading and helps avoid problems associated with osmotic effects, and it also supports a slow release of the drug at 37°C.
The loading process continues as long as there is an excess of intraliposomal acetate. However, for reasons of encapsulation stability, the process is stopped before all acetate gradient is utilized.
Based on our previous experience with remote loading of amphipathic weak acids via a transmembrane calcium acetate gradient (12) and optimization of MPS remote loading described elsewhere (Avnir Y and Barenholz Y: manuscript in preparation), we selected a liposome lipid composition of hydrogenated soybean phosphatidylcholine/cholesterol/PEG-DSPE-2000 of mole ratio 55:40:5. Briefly, lipids were dissolved in ethanol and then were hydrated with 200 mM calcium acetate at 70°C, reaching (during the hydration step) a final lipid concentration of 10% (weight/volume). The large multilamellar vesicles formed upon lipid hydration were downsized by sequential extrusion (using an extruder device [Northern Lipids, Vancouver, British Columbia, Canada]) at 70°C through polycarbonate filters of decreasing defined pore size, starting with a 400-nm pore size filter and ending with a 50-nm pore size filter, under increasing nitrogen pressure (up to 200 pounds per square inch). This procedure results in 83 ± 15–nm liposomes (mean ± SD), defined here as NSSLs. To create the transmembrane calcium acetate gradient, the calcium acetate of the external liposome medium was replaced by either 0.9% NaCl (w/v) or 5% dextrose (w/v) at 4°C in 4 dialysis steps.
Remote loading of MPS into NSSLs exhibiting transmembrane calcium acetate gradient to form NSSLs-MPS.
MPS was dissolved in 5% dextrose to a concentration of ∼16.9 mM (8 mg/ml) and was mixed with preformed NSSL dispersion (∼38.6 mM phospholipids), in which a transmembrane calcium acetate gradient was created. The remote loading was achieved by incubation of the above NSSLs for 20 minutes at 60–65°C (above the 52.5°C Tm of hydrogenated soybean phosphatidylcholine), and then liposomes were cooled to 4°C and dialyzed against 5% dextrose to remove acetate released during loading and residual unloaded MPS. Alternatively, in some cases unloaded drug and released acetate were removed by the Dowex anion exchanger, 1×4 400 mesh Cl− form (Sigma). This anion exchanger binds the negatively charged MPS and acetate anion but almost no NSSLs. After dialysis, MPS and phospholipid concentrations were 13.4 mM and 32.5 mM, respectively (drug:lipid mole ratio of 0.41). The identical method was used to obtain NSSLs loaded with BMS.
Chemical and physicochemical characterization of NSSLs-MPS.
The total and intraliposome (after treatment with Dowex anion exchanger) concentrations of MPS and BMS and their hydrolysis products were quantified using high-performance liquid chromatography (14). The concentration of NSSL phospholipid was determined as the concentration of organic phosphorus using the modified Bartlett procedure (15). The concentration of intraliposome calcium was determined using atomic absorption spectrometry after exhaustive dialysis, Dowex anion exchanger, or gel permeation chromatography using Sepharose CL-4B (see below). Size distribution analysis of NSSLs and NSSLs-MPS was made by dynamic light scattering, using the particle size analyzer ALV-NIBS/HPPS equipped with an ALV-5000/EPP multiple digital correlator (ALV-LaserVertriebsgesellschaft, Langen, Germany).
Cryo–transmission electron microscopy (TEM) analysis of NSSLs and NSSLs-MPS after loading.
Cryo-TEM was used to confirm liposome size distribution measured by dynamic light scattering and to characterize the detailed structure of the NSSLs and NSSLs-MPS, as described by Schroeder et al (16).
Kinetics of MPS release from NSSLs-MPS in biologic fluids.
The kinetics of MPS release in human plasma were quantified by separating liposomes from bulk medium using gel permeation chromatography on a Sepharose CL-4B column equilibrated with 0.9% NaCl. The NSSLs-MPS were incubated in 80% (v/v) human plasma at 37°C for a period of 96 hours. At the desired time points, aliquots were taken and chromatographed, and fractions of ∼1 ml were collected and analyzed for particle size distribution (by dynamic light scattering) and for levels of phospholipids, MPS, and Ca2+ ions.
The NSSLs and their load (NSSL-associated MPS and NSSL-associated MPS derivatives ) were eluted at fractions 4–7, while free MPS and free MPS derivatives released from NSSLs were eluted at fractions 9–16. Analysis of size distribution and [Ca2+]:[phospholipid] ratio were used to assess physical stability of NSSLs. The change in [MPS]:[phospholipid] ratio was used to assess the kinetics of MPS release from NSSLs during the incubation.
For studying the release of MPS in human inflamed synovial fluid, NSSLs-MPS were incubated in 80% (v/v) human inflamed synovial fluid at 37°C. At the desired time points, aliquots were removed and mixed with the Dowex anion exchanger to remove released MPS.
Pharmacokinetics and biodistribution in rats.
For pharmacokinetics and biodistribution studies, Lewis rats were injected with either free MPS (10 mg/kg) or NSSLs-MPS (10 mg/kg) radiolabeled with the nontransferable, nonmetabolizable lipid marker 3H-cholesteryl hexadecyl ether (17). At defined time points after IV injection, serum was obtained, rats were killed, and organs were harvested. Organs were homogenized using a homogenizer (Polytron; Kinematica, Littau, Switzerland), and the homogenates were extracted as described elsewhere (18), followed by quantification of MPS (as described above). The level of the liposome 3H-cholesteryl hexadecyl ether was determined by burning samples in a Sample Oxidizer (Model 307; Packard, Meriden, CT), after which the [MPS]:[3H-cholesteryl hexadecyl ether] ratio was calculated and used to determine the rate of release of MPS from NSSLs-MPS in vivo. The small blood volume of rats allowed for only 4 time points in the pharmacokinetic study; therefore, a more detailed study was conducted on a Beagle dog as described below.
Pharmacokinetics of free MPS and NSSLs-MPS in a Beagle dog.
We studied the pharmacokinetics of MPS administered to a 15-kg, 14-year-old Beagle dog in both the free (nonliposomal) and liposomal (NSSLs-MPS) forms. In both cases the dog received MPS at 6.3 mg/kg (total of ∼94.5 mg). The drug was administered via an 18-gauge Venflon catheter (Becton Dickinson, Helsingborg, Sweden) inserted into a superficial limb vein. The free MPS was infused at a rate of 3 ml/minute, and the total time of infusion was 5 minutes. For a period of 53 minutes, 17.2 ml of NSSLs-MPS containing ∼600 μmoles phospholipids was infused using an injection pump (Type MS200; Graseby Medical, Watford, UK) to prevent side effects that may be induced by IV injection of particles. Blood was withdrawn after free MPS or NSSLs-MPS infusion by rinsing the catheter with saline and then with 1 ml diluted heparin (250 units/ml) to prevent clotting.
To evaluate the levels of MPS after infusion of free MPS and NSSLs-MPS at each bleeding time, sera were prepared and analyzed in triplicates. Results are presented as the mean ± SEM. The large number of data points allowed for a detailed data analysis using WinNolin software (Pharsight, Mountain View, CA).
Induction and clinical assessment of AIA.
Six-week-old female Lewis rats were injected subcutaneously at the base of the tail with 1 mg of Mycobacterium tuberculosis (Mt) H37Ra (Difco, Detroit, MI) in Freund's complete adjuvant (Difco). Severity of arthritis (the arthritis index) was assessed every other day by a blinded observer, as follows: 0 = no arthritis; 1 = redness of the joint; 2 = redness and swelling of the joint. The ankle and tarsometatarsal joints of each paw were scored. A maximal score of 16 could be obtained. The collective data were subjected to t-tests using Prism 4 software (GraphPad Software, San Diego, CA).
Treatment protocols for AIA.
Treatment at early disease stage.
Rats (n = 8 per group) were injected IV on days 10 and 14 after induction of AIA with 5% dextrose (control), free MPS (50 mg/kg), or NSSLs-MPS (0.4, 2, and 10 mg/kg). Another group of rats were treated IV with 3 injections of NSSLs-MPS (10 mg/kg) on days 10, 14, and 18.
Treatment at late disease stage.
Nineteen days after immunization with Mt, when the animals had an average arthritis score of 9, they were divided into groups of 6 and injected IV with 5% dextrose, free MPS (50 mg/kg), free BMS (10 mg/kg), NSSLs-MPS (2 and 10 mg/kg), or NSSLs-BMS (1 and 5 mg/kg). The treatment was repeated on day 23.
Evaluation of liver and kidney toxicity.
Rats were treated twice with phosphate buffered saline or with NSSLs-MPS (10 mg/ml) on days 10 and 14 after induction of AIA. After 40 days, rats were killed, and blood samples and liver tissue were analyzed. Sera were tested for glucose, globulin, and albumin levels, liver enzymes (alanine aminotransferase [ALT], aspartate aminotransferase [AST], lactate dehydrogenase [LDH], and alkaline phosphatase [AP]), cholesterol, and renal function (urea and creatinine levels). Liver specimens were stained with hematoxylin and eosin and examined under light microscopy.
Comparison of different methods for encapsulating GC into NSSLs for systemic administration.
Table 1 demonstrates a comparison between our results of NSSLs remote-loaded with the amphipathic weak acid MPS (logD at pH 7.0 of 0.02 ) and the results of 2 recent studies of similar NSSLs loaded by the conventional passive loading method with either the water-soluble nonamphipathic (highly hydrophilic, logD at pH 7.0 of −4.02 ) prednisolone phosphate (4, 5) or the liposome membrane–associated (highly hydrophobic, logD at pH 7.0 of 10.02 ) prednisolone palmitate (20). Table 1 shows that remote loading of MPS via the transmembrane calcium acetate gradient achieved a drug:lipid mole ratio of 0.41 and encapsulation efficiency of 94%, while a much lower drug:lipid mole ratio (≤0.19) was reported for the 2 other GC prodrugs. Thus, their use will require a much higher dose (compared with liposomes remote-loaded with MPS), which may lead to liposome-induced toxicity (21, 22).
Table 1. Comparison of different methods of encapsulating various GCs into sterically stabilized nanoliposomes
Figures 1A and B are cryo-TEM images of NSSLs before and after remote loading of MPS. It is clear that after remote loading, MPS precipitates inside the NSSLs. In another study we proved that this precipitation is due to formation of calcium–MPS salt, and that the precipitate is amorphous rather than crystalline in nature (Avnir Y, et al: manuscript in preparation).
Unique mechanism of release of MPS from NSSLs-MPS and its relevance to pharmacokinetics and biodistribution.
The kinetics of MPS release in biologic fluids was studied by incubating NSSLs-MPS in human inflamed synovial fluid and in human plasma at 37°C. Figure 2A shows that MPS release was linear in both fluids, having similar kinetics (a half-life [T1/2] of ∼50 hours). In contrast to MPS release, intraliposome calcium concentrations, as well as NSSL size (83 nm, not shown), remained unchanged throughout the experiment, suggesting that the release of MPS is related to its amphipathic nature and can be explained by its distribution coefficient (Table 1) and not by physical damage to the NSSL membrane, since, in the latter case, change in size and/or Ca2+:phospholipid ratio should occur.
We also verified this unique release mechanism in vivo by monitoring the plasma pharmacokinetics and time-dependent biodistribution of NSSLs-MPS (Figures 2B and C). First, these studies show the superiority of using liposomes as GC carriers, since the free MPS is rapidly cleared from the plasma in a time frame of a few minutes (T1/2 <10 minutes), while encapsulated MPS has an estimated half-life of 11 hours. Second, the in vivo controlled release is apparent by monitoring the clearance of the MPS carriers, the NSSLs, which have a half-life of ∼34 hours (Figure 2B). Thus, the faster clearance of the encapsulated MPS compared with its NSSL carriers indicates that MPS is released from them during circulation. This behavior is also observed in most organs in which NSSLs-MPS accumulate (except the spleen) and is described in Figure 2C as the injected dose ratio of MPS:NSSLs; that is, as the NSSLs become empty, the ratio is decreased.
It is important to note that such behavior is observed in the inflamed paw, whereas in accordance with findings of other studies of nanoliposomes (4, 23–26), our NSSLs accumulated in a significantly higher concentration in inflamed paws than in healthy paws (>3-fold at the 24-hour time point) (inset in Figure 2C). Therefore, not only do NSSLs bring the GC content to the inflamed site, but they also slowly release it there.
Comparison of pharmacokinetics of free MPS and NSSLs-MPS in a Beagle dog.
The application of pharmacokinetic studies of liposome-associated drugs in rats to clinical use in humans may be misleading due to the fact that rat total plasma volume is small (∼5.6 ml for a 150-gram rat) (i.e., only ∼0.2% that of humans and ∼1% that of dogs). Thus, the level of dilution upon infusion on a body weight basis is much smaller in rats than in dogs or humans. Driven by the drug liposome/medium partition coefficient, the larger the dilution is, the higher may be the drug release of liposomes (27). Therefore, in large animals having large blood volume, drug release may be much faster than in small animals (28). To determine the extent to which this dilution issue was relevant to the present study, we studied the pharmacokinetics of NSSLs-MPS in a 15-kg Beagle dog.
Figure 3 and Table 2 show the comparison between the dog serum pharmacokinetics of MPS in its free form and as NSSLs-MPS. Results indicate that 5 minutes after infusion of free MPS, only 52.4% of the injected dose was detected (82.3 μg/ml), while 15 minutes and 30 minutes after infusion, only 0.36% (0.57 μg/ml) and 0.19% (0.303 μg/ml), respectively, of the free MPS–injected dose remained in serum. Ten minutes after infusion of NSSLs-MPS, almost all (94%) of the infused MPS dose was detected in the dog serum (∼148 μg/ml); 6 hours after infusion, a small decrease to 70% of the infused dose was observed (108.2 μg/ml); and 9 hours after injection, the MPS serum level was 78.03 μg/ml, which was 50% of the total infused dose, which compared well with the mean residence time distribution value described in Table 2. Even 24 hours after infusion, the MPS serum level was still ∼15% of the injected dose, which was similar to that in the rat (∼20% of the injected dose). Therefore, comparison between the rat (Figure 2B) and the dog (Figure 3) revealed that the effect of in vivo dilution in plasma was small.
Table 2. Serum pharmacokinetic parameters in a Beagle dog, comparing free MPS and MPS originating from NSSLs-MPS*
NSSLs-MPS, mean ± SEM
Free MPS, mean ± SEM
NSSLs-MPS:free MPS ratio
NSSLs = sterically stabilized nanoliposomes; AUC = area under the concentration-time curve; AUMC = area under the first moment-time curve; MRT = mean residence time distribution.
The elimination half-life, which is the time required for the plasma concentration to be reduced by 50%.
The maximum observed concentration in the concentrations data range; the expected methylprednisolone hemisuccinate (MPS) serum concentration based on Beagle serum volume of 600 ml (4% of body weight) is ∼160 μg/ml.
Comparison of free and NSSL-loaded MPS treatment at early disease stage.
We compared the therapeutic effects of IV administration of NSSLs-MPS and free MPS at various doses at an early stage of the disease. Figure 4A shows that a dose of 10 mg/kg MPS delivered as NSSLs-MPS, injected on days 10 and 14 after arthritis induction, resulted in a clinical (arthritis) score close to 0 for a period of 9 days (from day 14 to day 23), while free MPS at a 5-fold higher dose (50 mg/kg), injected at the same times, showed almost no therapeutic benefit. Furthermore, administration of a third injection, of 10 mg/kg MPS as NSSLs-MPS on day 18, extended the duration of the therapeutic effect for 10 more days. No significant effect was observed with empty liposomes administered at the equivalent NSSL concentration (data not shown). The therapeutic effect was dose related in the range of 0.4–10 mg/kg NSSLs-MPS.
Comparison of free and NSSL-loaded MPS and BMS treatments at late disease stage.
In this experiment we continued to test the therapeutic effect of free MPS and NSSLs-MPS, this time starting treatment at the peak of the disease (IV injection on days 19 and 23). We also studied the therapeutic effect of BMS (an amphipathic weak acid prodrug 5-fold more potent than MPS ) in its free and remote-loaded forms.
Figures 4B and C show the therapeutic effects of the different treatments. As can be seen in Figure 4B, the antiinflammatory therapeutic effect of NSSLs-loaded MPS or BMS was superior to that of the free GCs. Although free BMS exhibited a reasonably good therapeutic effect on day 25, with a mean ± SEM clinical score of 2.17 ± 0.47 (7.67 ± 0.84 in the control group), this effect was short lived, and the rats redeveloped arthritis with a score of 4.2 after 2 more days. In contrast, treatment with 5 mg/kg NSSLs-BMS resulted in a score of 0.33 ± 0.2 on day 25, and the score increased to 4 only after 15 days. Hence, treatment with 5 mg/kg NSSLs-BMS resulted in not only a better immediate therapeutic effect but also a significant prolongation of this effect.
Using the same basis of comparison, the results for MPS were even more dramatic. While 50 mg/kg free MPS showed only minimal efficacy, decreasing the mean clinical score only slightly, 10 mg/kg NSSLs-MPS exhibited a therapeutic effect similar to that of 5 mg/kg NSSLs-BMS. Figure 4C shows the dose-response relationship of NSSLs-MPS, revealing that even a dose of 2 mg/kg NSSLs-MPS was more efficacious than 50 mg/kg free MPS. Comparison of the early and late treatments revealed that free MPS (50 mg/kg) and NSSLs-MPS (2 mg/kg), which were ineffective in the early phase, were effective at the peak of disease, whereas treatment with 10 mg/kg NSSLs-MPS resulted in similar long-term therapeutic effects at both early and late stages of disease.
We tested the effect of treatment with 10 mg/kg NSSLs-MPS on liver and kidney functions. Liver function (evaluated by blood levels of albumin, globulin, cholesterol, and liver enzymes [AP, AST, LDH, and ALT]), renal function (evaluated by serum urea and creatinine levels), and blood glucose levels were in the normal range. There was no evidence of any histopathologic changes in liver specimens from any of the liposome-treated rats.
The synthesis of GCs and the discovery of liposomes both emerged at about the same time (40–50 years ago). Since then both methodologies have been constantly and extensively studied and developed. Today, we have in clinical use water-soluble synthetic GC prodrugs that are 25 times more potent than hydrocortisone (29) (as exemplified by BMS), and liposome technology has developed to yield doxorubicin, the first nanomedicine (10) (approved by the Food and Drug Administration in 1995), as well as more than 10 other approved liposome-based drugs and vaccines.
The development of liposomal GCs for treating inflammatory diseases has been a long-sought goal, as evident from numerous studies starting in the 1970s and continuing to this day. However, although much progress has been made, some major limitations and restrictions still prevent the pharmaceutical development of liposomal GC formulations for clinical use.
From a historical perspective, in the early stage of research related to liposomal GCs, the main idea was to utilize their lipophilic nature and to encapsulate them in the liposome membrane. This form of encapsulation has encountered a major limitation of stability, since most GCs are not lipophilic enough, as is obvious from their logD and mass aqueous solubility. For instance, cholesterol, one of the liposome membrane components, has a logD at pH 7.0 of 9.85 and mass solubility of 6 × 10−9 gm/liter, while hydrocortisone has a much lower logD at pH 7.0 of 1.43 and mass solubility of 0.16 gm/liter (19). Therefore, such GCs are not suitable for “encapsulation” in the liposome bilayer, since upon dilution they will desorb to the aqueous medium. A good example of such behavior was described by Mishina et al (30), who demonstrated that >50% of encapsulated methylprednisolone was released to the bulk medium during 10 days of storage.
Two opposite approaches were used to solve this major obstacle. First, an old approach was revisited in which a GC is chemically modified to form a prodrug that is highly lipophilic, as demonstrated long ago by Shaw et al (31), who used the cortisol palmitate ester, or recently by Teshima et al (20), using the more potent prednisolone palmitate ester. However, this approach has a major drawback imposed by the fact that the mole% of GC palmitate in the lipid bilayer is limited, since this molecule is not a “liposome-forming lipid” (13, 32). Due to its large “packing parameter,” there is a rather low limit to its membrane mole%, above which formation of liposomes is prevented (32). Thus, the liposome lipid bilayer cannot accommodate a high level of the GC–palmitate without losing its integrity and/or being transferred to a nonlamellar phase (32).
The second approach is well presented by Metselaar et al (4). Those investigators used the highly hydrophilic, nonamphipathic, acidic GC prednisolone phosphate, which is encapsulated passively in the intraliposome aqueous phase. However, the second approach also has the same drawback as the first (i.e., low drug:lipid ratio), although for totally different reasons (the limitation of the passive loading process and the small volume of NSSLs , which result in an additional disadvantage of having a very poor encapsulation efficiency of 5%) (see Table 1). The low drug:lipid ratio obtained with the classic passive loading is imposed by the paradox related to the absolute need for nanoliposomes, since only they can reach efficiently into the inflammation sites, and to the extremely small volume of NSSLs resulting in a low drug:lipid ratio, even for drugs that are highly soluble (33, 34). Thus, for such a formulation to reach a therapeutically sufficient drug level at the inflammation site, a large amount of the NSSLs has to be administered. This may cause various types of toxicity (21, 22).
In the present study we have tried to overcome all of the above-mentioned major obstacles by replacing the passive loading approach with active remote loading of GCs. Table 1 shows that, indeed, our NSSLs-GCs achieved this, reaching a novel drug:lipid mole ratio of 0.41, under conditions of 94% drug encapsulation efficiency.
Another common limitation for most previously described liposomes encapsulating GCs is lack of a controlled slow release of the encapsulated GC. This issue did not receive enough attention and remained unaddressed in many of the liposome-based formulations (33). Lack of GC controlled release dictates that drug availability is dependent on the intra- or extracellular destruction of liposome membrane. Therefore, liposomes that show lack of GC release may have inferior efficacy. Alternatively, a fast release will result in losing the advantage of liposomes as a means of improving drug biodistribution and bioavailability. Our loading approach is also aimed at filling this gap, and indeed it exhibits a unique mechanism of releasing encapsulated GC, as shown in vitro (in human plasma and inflamed synovial fluid) and in vivo (in AIA rats) (Figures 2A–C).
We suggest that the mechanism of release is a result of intraliposome events associated with the rise in temperature. Although there is evident release of MPS at 37°C, under long-term storage conditions at 4°C (14 months), there is a minimal (<6%) release of MPS (not shown).
The profile of MPS release resembles that of other remotely loaded substances (35–37). Therefore, the release profile of MPS is best described by the Arrhenius equation, which correlates chemical reactions to temperature (i.e., MPS release rate is a function of the reciprocal of Kelvin temperature [1/T]). Thus, we propose that at 37°C a larger fraction of the calcium–MPS intraliposome precipitate dissolves, which leads to partial MPS protonation and subsequent release from the liposomes, as demonstrated in Figure 2A.
As mentioned, this controlled release also operates in vivo, as seen by monitoring NSSLs-MPS in the plasma and in selected organs of AIA rats. In rat plasma (Figure 2B), it was demonstrated that NSSLs have an estimated half-life of 34 hours, while their encapsulated MPS has a half-life of 11 hours, and monitoring NSSLs-MPS time-dependent accumulation in organs, as well as in the inflamed paw, revealed a similar release profile (Figure 2C). Hence, in regard to therapeutic efficacy, while administration of free MPS results in rapid plasma clearance with a half-life of minutes, administration of NSSLs-MPS results in a major reduction of clearance rates concomitant with the slow release of MPS at a constant rate in the blood and at the organs and inflamed joint where it accumulates (Figure 2C). Therefore, our novel NSSLs-GCs can exert their therapeutic effect both systemically and locally.
Recently, Metselaar et al (38) demonstrated localization of the NSSLs containing gold particles in the inflamed synovial lining of mice with collagen-induced arthritis, suggesting uptake by phagocytic cells such as monocyte/macrophages. This may be an important antiinflammatory activity, but by itself it is probably not sufficient, since the GC must also affect the nonphagocytic lymphoid cells that play a major role in the synovial inflammatory process as well as activated lymphocytes outside the joint.
These requirements serve as the basis of our working hypothesis for the need for, and advantages of, local and systemic slow and controlled release. Based on the prolonged circulation time (Figures 2B and C and Table 2) and on the localization in the inflamed joint (Figure 2C), it is expected that our NSSLs-MPS, in addition to being taken up by macrophages, will also “spray” their MPS load in the synovial joint, thus also affecting the nonphagocytic immunoactive cells (e.g., T cells and B cells) there. In parallel, the systemic slow release may affect the lymph nodes. Although the superior therapeutic efficacy demonstrated in Figure 4 supports our working hypothesis, further studies are required to achieve direct proof of it.
Plasma pharmacokinetics of NSSLs-MPS and free MPS in 2 mammalian species, the small (<200-gm) rodent (the rat) and the much larger (∼15-kg) Beagle dog, demonstrate high similarity (compare Figures 2B and 3). In both animal species, MPS released from NSSLs has a much longer circulation time, with a mean residence time of ∼11 hours in the rat and ∼9 hours in the dog, compared with mean residence times of <15 minutes for free MPS in both animals. The pharmacokinetic similarity between the 2 species and the pharmacokinetic superiority of NSSLs-MPS over free MPS in all other pharmacokinetic parameters (Table 2) are prerequisite for therapeutic efficacy and suggest that the observed pharmacokinetic advantages will also be retained in humans, similar to what was demonstrated for doxorubicin (39).
GCs might have an important role in the treatment of RA. During short-term use, GCs are frequently added to minimize disease activity while awaiting the effect of the slower-acting antirheumatic drugs. With prolonged use, they have been shown to act like disease-modifying drugs and to stop progression of bone erosion (1). Patients treated with prednisone in addition to the traditional DMARDs are also more likely to achieve remission (1). Most of those studies used prednisone in dosages >7.5 mg/day, and, unfortunately, long-term use of prednisone in dosages >5 mg/day is associated with multiple adverse events that preclude its routine use. In the present study we have shown that we can increase the therapeutic efficacy of GCs significantly, as evaluated by their antiinflammatory effect (compared with that of free GCs) when they are delivered as NSSLs-GCs. If similar effects are observed in human clinical trials, we will be able to use NSSLs-GCs as effective remitting agents in doses that minimize their side effects. We will therefore be able to use modern nanomedical technology to change one of the oldest antiinflammatory agents to become a new, effective, disease-modifying drug for the treatment of RA.
Drs. Naparstek and Barenholz had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Avnir, Ulmansky, Barenholz, Naparstek.
Acquisition of data. Avnir, Ulmansky, Wasserman, Even-Chen, Broyer, Barenholz.
Analysis and interpretation of data. Avnir, Ulmansky, Broyer, Barenholz, Naparstek.