Abstract

AR-67 is a lipophilic camptothecin analogue currently under clinical investigation using a Cremophor EL based formulation. However, as potential toxicity limitations exist in the clinical use of Cremophor, an alternative cyclodextrin (SBE-β-CD) based formulation has been proposed. Here we conducted pharmacokinetic (PK) studies in mice and compared the SBE-βCD based formulation to the Cremophor EL formulation. PK studies were conducted following intravenous or oral administration of AR-67 in either Cremophor or SBE-β-CD formulation in mice. Noncompartmental analysis was used to determine the plasma and tissue drug distribution. A non-linear mixed effects (population) PK model was developed to fit both the oral and intravenous data and estimate key PK parameters. The effect of formulation was explored as a covariate in the PK model. AR-67 in SBE-β-CD formulation had similar plasma PK and biodistribution as that in the Cremophor EL formulation. The proposed 2-compartment model described the plasma PK of AR-67 in both formulations adequately. AR-67 in the SBEβ-CD formulation exhibited dose linearity following both oral and intravenous administration. Our studies indicate the SBE-β-CD is a viable alternative to Cremophor EL as a pharmaceutical excipient for formulating AR-67.

Keywords: Camptothecin, SBE-β-CD, Cremophor EL, pharmacokinetic modeling

Introduction

AR-67 (7-t-butyl-dimethylsilyl10-hydroxycamptothecin, also known as DB-67) is a lipophilic camptothecin analog exhibiting potent anticancer activity in tumor cell lines and
xenograft models (A. Y. Chen et al; 2005;Lopez-Barcons, Zhang, Siriwitayawan, Burke, & Perez-Soler, 2004;Pollack et al; 1999). AR-67 is currently under clinical investigation for the treatment of solid tumors. Like all camptothecin analogs, AR-67 exerts its anticancer effect by “poisoning” the nuclear enzyme topoisomerase-I, which is responsible for unwinding supercoiled DNA through cutting and then ligating phosphate backbone during DNA replication and transcription (Champoux, 2001). At nanomolar concentrations, camptothecins form the DNA-enzyme-drug complex that prevents topoisomerase-I mediated DNA ligation.

Subsequently, collision of the advancing DNA-replication fork with this complex may cause double strand breaks, which ultimately lead to apoptosis (Hsiang, Hertzberg, Hecht, & Liu, 1985;Hsiang & Liu, 1988). The pharmacological use of camptothecins presents a challenge because they undergo pH-dependent reversible hydrolysis from the closed-ring lactone moiety to the open-ring carboxylate form with equilibrium favoring the charged carboxylate species at pH 7 and higher.

Additionally, avid binding of the camptothecin carboxylate to albumin further shifts the equilibrium towards lactone hydrolysis in plasma (Bom et al; 2000;Burke & Mi, 1993). Initial enthusiasm about camptothecin waned until mechanistic studies demonstrated that the lactone form was primarily responsible for the anticancer effect, while the carboxylate form was
primarily responsible for systemic toxicity (Creaven, Allen, & Muggia, 1972;Hertzberg et al; 1989;Muggia, Creaven, Hansen, Cohen, & Selawry, 1972). In the past four decades, several analogs have been developed in the effort to maintain the lactone form stability (Bom et al; 1999;Pratesi, Beretta, & Zunino, 2004). Compared to the clinically available camptothecins (irinotecan, topotecan), AR-67 has superior lactone stability in rodent and human blood (Adane, Liu, Xiang, Anderson, & Leggas, 2012;Arnold et al; 2010;Bom et al; 2000). This property maybe attributed to the lipophilicity of AR-67, which allows it to rapidly associate with lipid membranes and thereby stabilize the lactone form (Bom et al; 2001; Ziomkowska, Cyrankiewicz, & Kruszewski, 2007). In addition, the 10-hydroxyl substitution appears to abrogate its interaction with albumin and further inhibit lactone hydrolysis (Burke & Mi, 1994; Mi, Malak, & Burke, 1995). In vivo, the lactone form exceeds 95% of the total plasma concentration partly due to the higher carboxylate clearance (Adane, Liu, Xiang, Anderson, & Leggas, 2010;Arnold et al; 2010).

However, the enhanced lactone stability of this highly lipophilic molecule limits its aqueous solubility (0.11 µg/mlatpH 5.2) (Xiang & Anderson, 2002), which poses challenges in formulating it for intravenous or oral administration. Following preclinical evaluation through the National Cancer Institute’s Rapid Access to Interventional Development (RAID) program, AR-67 is currently under clinical investigation using a Cremophor EL/ethanol (1:1 v/v) diluent formulation. While the current treatment schedule calls for daily intravenous administration over the first five days of a 21-day cycle, future clinical trials will evaluate an alternative schedule of weekly administration of a higher dose, which has been shown to be efficacious with another camptothecin analogue, irinotecan (Friedman et al; 1999;Rothenberg et al; 1993). However, Cremophor EL may not be the optimal excipient for higher clinical dosages of AR-67. Cremophor EL-based paclitaxel has been associated with side effects following high doses, including severe anaphylaxis, hyperlipidemia, erythrocyte aggregation
and peripheral neuropathy (Gelderblom, Verweij, Nooter, & Sparreboom, 2001;Kloover, den Bakker, Gelderblom, & van Meerbeeck, 2004). Dose-dependent decrease in paclitaxel clearance and oral absorption has also been observed (Bardelmeijer et al; 2002;Sparreboom, van Tellingen, Nooijen, & Beijnen, 1996) most probably due to Cremophor EL mediated inhibition of the efflux transporter P-glycoprotein and entrapment of paclitaxel within micelles, respectively (Bardelmeijer et al; 2002; Nerurkar, Burton, & Borchardt, 1996). Given the
potential limitations of Cremophor EL, we sought an alternative pharmaceutical excipient for preparing intravenous and oral formulations that allow flexibility in dosing and limit drug-related toxicity.

A chemically modified β-cyclodextrin, sulfobutylether-β-cyclodextrin (SBE-β-CD), was previously used to improve AR-67 solubility to 1 mg/mL (Xiang & Anderson, 2002). The results of preliminary pharmacokinetic studies of AR-67 in rats indicated that SBE-β-CD could be used as an alternative formulation for AR-67 for both oral and intravenous administration (Adane et al; 2012). However, studies comparing the pharmacokinetic profiles of AR-67 in SBE-β-CD and the currently used Cremophor EL based formulation have not been conducted.

In this study, we compared the pharmacokinetics of the SBE-β-CD and Cremophor EL based formulations of AR-67 and assessed dose linearity of AR-67 following intravenous and oral administration of the SBE-β-CD based formulation.

Materials and Methods

Chemical Reagents

All reagents and chemicals were of the highest purity available. Ammonium acetate (Mallinckrodt Baker, Phillipsburg, NJ), HPLC grade acetonitrile and methanol (Burdick and Jackson, Muskegon, MI) were purchased from VWR (West Chester, PA). Tetrabutylammonium dihydrogen phosphate (TBAP; 1.0 M aq. solution) was obtained from Sigma-Aldrich (St. Louis, MO). Dimethylsulfoxide (≥99.7%), ethanol, dextrose, hydrochloric acid and acetic acid came from Fisher Scientific (Fair Lawn, NJ). Blank mouse plasma used in the preparation of calibrators and quality control solutions was from Abacell Corp. (San Mateo, CA). Consumables were treated with AquaSilTM siliconizing reagent (Pierce, Rockford, IL). Siliconized pipette tips were obtained from Cole-Parmer and amber siliconized microcentrifuge tubes from Crystalgen Inc. (Plainview, NY). Magnesiumand calcium-free Dulbecco’s phosphate buffered saline (PBS) was from Gibco (Invitrogen Carlsbad, CA). AR-67 (≥ 98 % purity) was obtained from Novartis Pharmaceuticals Corporation. The compound was stored at 4°C, and handled as if it were a known chemical carcinogen. Sulfobutylether-βcyclodextrin (Captisol®) was received as a gift from CyDex, Inc. (Overland Park, KS). Cremophor EL was purchased from Sigma-Aldrich (St. Louis, MO).

AR-67 Formulations

Cremophor EL-based AR-67 solutions were prepared by diluting a 5.13 mg/mL stock solution inCremophor EL:ethanol (50:50; v/v) to 0.2 mg/mL with 5% dextrose in water (Xiang & Anderson, 2002). The cyclodextrin-based AR-67 solutions were prepared by saturating AR-67 with 22.2% sulfobutylether-β-cyclodextrin (SBE-β-CD) containing 2 mM acetic acid and 8.37 mM HCl. The solution was analyzed using a validated HPLC bioanalytical method for AR-67 concentration and diluted to the desired concentration with 5% dextrose in water (Horn et al; 2006).

Animals

C57BL/6 male mice (31 to 36 day-old) were obtained from Harlan (Indianapolis, IN). Mice were housed in a 12 h/12 h (light/dark) cycle and provided with free access to food pellets and water. All experimental procedures were approved by the Institutional Animal Care and Use Committee at the University of Kentucky, which is under the purview of the US Public Health Service’s Policy on Humane Care and Use of Laboratory Animals.

Plasma Pharmacokinetics, Tissue Distribution and Dose Linearity Studies Cremophor ELand SBEβ-CD based AR-67 doses were administered to mice as 1 mg/kg bolus dose intravenously or orally to assess the plasma pharmacokinetics and tissue distribution. 37 and 32 mice were given intravenous and oral SBE-β-CD based AR-67, respectively; 36 and 33 mice received intravenous and oral Cremophor EL based AR-67, respectively. Preliminary studies showed zebrafish bacterial infection no food-effect, and all oral gavage studies were conducted in mice which had free access to food. Blood samples were collected by cardiac puncture and were transferred into heparinized containers; heart, kidney, lung, liver, testes and spleen were collected from these animals at predetermined time points between 5 minutes and 8 hours after drug administration (Supplemental Table 1). Excised tissues were snap-frozen in liquid nitrogen and stored at -80°C until further processing.

Another cohort of mice received 5 mg/kgAR-67 intravenously for brain biodistribution studies. 19 mice received Cremophor EL based and 30 received SBE-β-CD based AR-67,respectively. Brain tissues were collected at predetermined time points between 5 minutes and 6 hours after drug administration (Supplemental Table 2). Tissues were snap-frozen in liquid nitrogen after collection and stored at -80°C until analysis.

To assess the dose linearity of the SBE-β-CD formulation, mice received a bolus dose of 0.25, 1, 5, and 12 mg/kg intravenously and 1, 5, 12 mg/kg orally. Two saphenous vein blood samples per mouse were collected before a terminal sample was taken by cardiac puncture. Plasma samples were analyzed for total DB-67 concentrations in dose linearity analysis.

Sample Processing

Siliconized pipette tips and micro centrifuge tubes were used during sample processing. Immediately after blood collection, plasma was separated from blood cells by centrifuging at 10,000xg, 4°C for 2 min. Plasma was then extracted with methanol on dry-ice as previously described (Horn et al; 2006). Briefly, one volume of plasma was extracted by adding four volumes of cold methanol. The mixture was vortexed for 5 sec and centrifuged at 13,000xg, 4°C for 2 min. Tissues were homogenized using Tissuemiser (Fisher Scientific, Pittsburgh, PA) for 2 min in PBS (1:1, w/v) on ice. Extraction was performed following a procedure similar to that for plasma, with the exception of the centrifugation condition (23,000xg, 4°C for 20 min). After centrifugation, supernatants were collected into siliconized amber tubes and placed on dry ice or stored at -80°C until analysis.

HPLC Analysis

Plasma and tissue extracts were analyzed using a validated HPLC method for the simultaneous quantitation of lactone and carboxylate AR-67 (Horn et al; 2006). Experimental samples, calibrators, and quality control solutions were diluted with one volume of ice-cold mobile phase buffer prior to injection. Samples (50 µL) injected within a 3 h window following dilution had minimal carboxylate to lactone conversion (<5%). Using fluorescence detection minimized the likelihood of interference and there were no interfering peaks observed from endogenous substances or blank samples spiked with formulation vehicles. Concentrations of AR-67 carboxylate and lactone in plasma (ng/mL) and tissue (ng/g) samples were determined by interpolation from external calibration curves. Calibration curves were prepared in plasma or appropriate tissue matrix. Plasma calibration curves were linear in the range of 1-300 ng/mL for carboxylate and 2.5-300 ng/mL for lactone analytes. Calibration curves for liver and heart were linear in the range of 2-200 ng/g for carboxylate and 5-200 ng/g for lactone. Calibration curves for all other tissues were linear in the range of 21000 ng/g for carboxylate and 51000ng/g for lactone analysis. Quality control samples were also analyzed to ensure system suitability, which required samples to be within 15% of their nominal values. Experimental sample extracts with concentrations above the calibration range were diluted with appropriate methanol extracted matrix, andreanalyzed alongside a similarly treated quality control sample extract. Samples with concentrations beneath the lower limit of quantification (LLOQ) were reported as containing no analyte.

Non-compartmental Analysis

The area under the plasma and tissue concentration versus time curve of AR-67 was determined using the PK package (Version 1.3-2) under R environment (version 3.1.2). The calculation of AUC was based on the trapezoidal rule. The variance of the AUC was calculated using Bailer’s method (Bailer, 1988). The confidence interval of AUC was constructed with modification as described by Nedelman (Nedelman, Gibiansky, & Lau, 1995). The tissue to plasma AUC ratio was calculated based on AUC determined as aforementioned. The construction of AUC ratio
confidence intervals and statistical analysis were performed using bootstrap approach under R environment (version 3.1.2) (Jaki, Wolfsegger, & Ploner, 2009). Noncompartmental analysis (NCA) was performed to estimate other pharmacokinetic parameters using NCA module in Phoenix (Phoenix X64, Build 6.3.0.395). Bioavailability was calculated as the percentage of AUC following oral administration relative to that following intravenous administration.

Nonlinear Mixed Effects Modeling and Simulation

A simultaneous oral/intravenous two-compartmental model was fit to plasma concentrations of AR-67 calculated from the sum oflactone and carboxylate forms using the nonlinear mixed effects (NLME) module in Phoenix software package. The first order conditional estimation (FOCE) algorithm was implemented with maximum iterations of 1000. Plasma concentrations below the LLOQ were input as ½ of the LLOQ (Hing, Woolfrey, Greenslade, & Wright, 2001; Hughes, 2000). As only one PK sample was taken from each mouse, a fixed proportional error model was used, with standard deviation of the error term fixed to the maximum coefficient of variation (CV) of the HPLC assay (Horn et al; 2006). Interindividual variability on the population parameter was modeled as an exponential term. After a structural model was selected, formulation as a categorical covariate was evaluated as a potential covariate on each PK parameter in Phoenix. The predictive performance of the model was assessed by visual predictive check in Phoenix using 1000 replicates per time point.

Statistics

Two-sample t-test was used to compare AUCs with a significance level “=0.05 following the Bailer-Satterthwaite method (Nedelman et al; 1995). Bootstrapping two-sample test was used to compare AUC ratios with a significance level “=0.05. Likelihood ratio test was conducted to compare the objective function values of the covariate and base models with a significance
level “=0.01 in 2 distribution. Linear regression analysis was conducted using GraphPad Prism to assess the dose linearity. R2 was calculated as a reference for AUC-dose correlation. The slope was analyzed using t-test to assess non-zero significance and considered significant if pvalue is less than 0.05.

Results

Noncompartmental Analysis of AR-67

The plasma pharmacokineticsofAR-67 in Cremophor EL and SBE-β-CD formulations were assessed in male C57BL/6 mice following an intravenous or oral bolus dose of 1 mg/kg. Non-compartmental analyses (NCA) were conducted to compare exposure following administration of the two formulations. Figure 1 shows the comparison of plasma profiles of the two formulations following intravenous and oral dosing. The corresponding AUC and associated statistics for lactone and carboxylate forms and the total amount are listed in Table 1. As previously published (Adane et al; 2012), the majority of AR-67 in plasma was in the lactone form following intravenous or oral administration. There was a trend toward elevation in the AUC of lactone and total drug in the SBE-β-CD formulation compared with Cremophor EL after intravenous and oral dosing, but this increase did not reach a significant level based on the two-sample t-test (Table 1). Clearance, bioavailability and additional PK parameters are presented in Table 2. The bioavailability was in the order of 23-25% with either formulation and the estimated clearance was similar in both formulations.

Biodistribution of AR-67

The AUC in several tissues and their ratios to plasma AUC with associated statistics arepresented in Table 3. For both formulations, irrespective of administration route, the highest exposure was observed in the liver, with tissue to plasma AUC ratio of approximately 10. The second highest exposure was observed in the kidney, with tissue to plasma AUC ratio of approximately 2. In other organs, the tissue distribution was much lower. The ratios in testes and brain were especially low compared to others, presumably due to the blood-tissue barriers present in these organs. Compared with Cremophor EL, SBE-β-CD based formulation had a significantly lower AUC and AUC ratio in the heart after intravenous and oral dosing, and in the liver after oral dosing. The distribution of SBE-β-CD based AR-67 in other tissues was statistically undistinguishable from that of Cremophor EL formulation.

Nonlinear Mixed Effects Pharmacokinetic Modeling of AR-67

To further describe the pharmacokinetics of AR-67 and evaluate the effect of formulation on PK parameters, we developed a two-compartment pharmacokinetic model to simultaneously fit 1 mg/kg oral and intravenous bolus dosing of Cremophor EL or SBE-β-CD based formulations (Figure 2). From the total of 128 plasma observations used for model development, 62 and 66 samples were from Cremophor ELand SBE-β-CD based formulations, respectively; 68 and 60 samples were from intravenous and oral dosing, respectively. Initial model development was performed by pooling data from both formulations. Interindividual variability terms were fixed to zero if they were not estimated with adequate accuracy (i.e; with CV>50%), and the resulting structural model contained interindividual variability terms for clearance (CL) and central volume (V1). All PK parameters were estimated with good accuracy, as shown in Table 4. The goodness offit of the structural model was assessed with routine diagnostic plots (Figure 3 and Supplemental Figure 1), which demonstrate the model describes PK observations reasonably well in the absence of misspecification.

Formulation was parameterized into the structural model to study its effect on each PK parameters, but no significant decrease in the objective function value (OFV) was observed as assessed by the 2 test at a significance level of 0.01. This suggested that the SBE-β-CD based AR-67 formulation could not be statistically discriminated from the Cremophor EL formulation, as assessed by our pharmacokinetic model. As shown in Figure 4, visual predictive check demonstrated good agreement between predicted and observed plasma concentrations, with the majority of observations centered around median predicted values and contained within the predicted 90% intervals.

Assessment of Dose find more Linearity

Analysis of the dose linearity of AR-67 is depicted in Figure 5, and additional associated statistics calculated from NCA are summarized in Table 5. A significant linear relationship was observed between AUC and dose after both intravenous and oral dosing (P<0.05 andR2 >0.98). Additionally, no significant correlation or nonzero slope was found between dose-normalized AUC and dose, indicating linear pharmacokinetics.

Discussion

In this study, we demonstrated that AR-67 formulated in SBE-β-CD had similar pharmacokinetic profiles with that of Cremophor EL based formulation. NCA showed that AR67 in both formulations exhibited comparable plasma AUC after both intravenous and oral administration. The lactone form and total drug tended to be slightly higher in SBE-β-CD than in Cremophor EL, although the difference did not reach statistical significance. We also demonstrated that both formulations had similar biodistribution with moderate exposure differences observed in the heartand the liver. Through nonlinear mixed effects pharmacokinetic modeling, we established a two-compartmental model that described the plasma AR-67 concentration profiles with good accuracy. Using this model, we demonstrated that the estimated pharmacokinetic parameters of AR-67 were statistically undistinguishable between SBE-β-CD and Cremophor EL based formulations. Lastly, we showed that the AR-67 formulated in SBE-β-CD exhibited significant dose linearity after both intravenous and oral administration, suggesting it as a potential alternative excipient to the currently used Cremophor EL.

Despite the potent anticancer activity of AR-67, its poor water solubility is still a formulation challenge. A Cremophor EL based formulation was developed earlier and is currently under clinical trials following the intravenous route. However, due to the undesirable properties of Cremophor EL, an alternative excipient with equivalent solubilizing ability and more biologically inert properties is desirable. In the current study, we demonstrated that the SBE-β-CD based formulation can serve as an alternative to Cremophor EL based formulation.

Our findings indicated that there was no significant difference in the plasma pharmacokinetics between SBE-β-CD and Cremophor EL based formulations. Tissue exposure of AR-67 was lower in heart and liver when formulated with SBE-β-CD but comparable to when formulated with Cremophor EL in other tissues. Bioavailability estimates of AR-67 in both formulations were comparable to those of irinotecan (25%) in mice and 9-nitrocamptothecin in rats (22%), but were lower than that of topotecan in mice (48%) (J. Chen, Ping, Guo, Chu, & Song, 2006; Jonker et al; 2002; Stewart et al; 2004). One notable finding from our study was that the percentage of AR-67 lactone AUC in the plasma was more than 80% in both formulations following intravenous and oral administration. This is consistent with previous observations of high AR-67 lactone ratio in the plasma of rats and patients (Adane et al; 2010,2012;Arnold et al; 2010). Such a high percentage of lactone form is in contrast to that of irinotecan, topotecan and other camptothecin analogs. Plasma lactone AUC percentage ranged between 49-60% for 9-nitrocamptothecin in rats, between 43-50% for topotecan in dogs and 56% for irinotecan in rats (J. Chen et al; 2006;Davies, Minthorn, Dennis, Rosing, & Beijnen, 1997; Kaneda, Hosokawa, Yokokura, & Awazu, 1997). Given that the relatively lipophilic lactone form of camptothecins is responsible for the antitumor effect, the nutritional immunity maintenance of such a high plasma lactone AUC by SBE-β-CD as well as by Cremophor EL may contribute to better antitumor efficacy of AR-67.

SBE-β-CD based formulation showed similar tissue distribution of AR-67 compared with Cremophor EL based formulation. The liver had the highest exposure after both oral and intravenous administration, which was several folds higher than in other examined tissues. Such high liver extraction as well as efflux by ABC transporters located in the liver and gut might explain the modest oral bioavailability of AR-67 (~25%). This is consistent with previous findings showing that AR-67 is a substrate of P-gp and BCRP (Adane et al; 2010). In addition to its modest oral bioavailability, its brain penetration might also be affected by efflux transporters. Our studies showed that brain to plasma ratio was approximately 0.1 for both Cremophor EL and SBE-β-CD based formulations following 5 mg/kg intravenous dose.

Inhibition of efflux transporters located at the blood brain barrier has been shown to enhance brain penetration of camptothecin analogs (de Vries et al; 2007;Kruijtzer et al; 2002;Leggas et al; 2006;Stewart et al; 2004). Cremophor EL has also been shown to inhibit P-gp in vitro (Hugger, Novak, Burton, Audus, & Borchardt, 2002;Nerurkar et al; 1996). However, at the excipient concentration of Cremophor EL used in our formulation we did not see evidence of such inhibition, since the SBE-β-CD formulation produced similar brain exposure.

Nonlinear mixed effects PK modeling provided an alternative quantitative method to compare the pharmacokinetics of the SBE-β-CD and Cremophor EL formulations. The initial model development applied the two-compartmental model (Figure 2) to plasma concentrations from both formulations. Using the estimated parameters, our structural model was able to explain the plasma observations very well. In further model development, we evaluated the influence of formulation on PK parameters, but incorporation of this covariate did not significantly improve the model fit, suggesting that both formulations had undistinguishable PK behaviors. Additionally, the bi-exponential character of the pharmacokinetics with either formulation is consistent with the pharmacokinetics observed during the phase-I study, which used the Cremophor EL based formulation (Arnold et al; 2010).

It is recognized that Cremophor EL contributes to nonlinear drug disposition (Sparreboom et al; 1996). Therefore, an alternative excipient facilitating linear drug pharmacokinetics following both oral and intravenous administration is desirable. Our study demonstrated that following both oral and intravenous administration of various doses of SBEβ-CD based AR-67, there was a linear relationship between doses and plasma total AUC. In addition, dose normalized AUC, which corresponds to clearance, was not significantly correlated with doses. This further confirmed the linear plasma pharmacokinetics of SBE-βCD based AR-67 and highlighted the potential advantage of SBE-β-CD as an alternative formulation excipient.

Conclusion

Our current study indicates that SBE-β-CD is a viable alternative to Cremophor EL as a pharmaceutical excipient for AR-67 formulations, with equivalent bioavailability and plasma AUC, similar biodistribution and dose linearity. However, SBE-β-CD based AR-67 exhibited different exposure than Cremophor EL based AR-67 in some examined tissues, such as heart and liver. Such difference in exposure may be caused by the difference in tissue partition of AR-67 in the two different formulations. Future studies will examine the whole-body PK of AR-67 in both formulations using physiologically-based pharmacokinetic (PBPK) modeling and further investigate potential causes for differences in tissue distribution between SBE-βCD and Cremophor EL based formulations.