- Email: [email protected]
In vitro and in vivo entrapment of bupivacaine by lipid dispersions
Journal of Chromatography A, 1254 (2012) 125–131
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
In vitro and in vivo entrapment of bupivacaine by lipid dispersions Erik Litonius a,∗∗ , Jana Lokajova b , Gebrenegus Yohannes b , Pertti J. Neuvonen c , Juha M. Holopainen d , Per H. Rosenberg a , Susanne K. Wiedmer b,∗ a
Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital and University of Helsinki, Finland Laboratory of Analytical Chemistry, Department of Chemistry, University of Helsinki, Finland Department of Clinical Pharmacology, University of Helsinki and HUSLAB, Helsinki, Finland d Helsinki Eye Lab, Department of Ophthalmology, University of Helsinki, Helsinki, Finland b c
a r t i c l e
i n f o
Article history: Received 26 April 2012 Received in revised form 4 July 2012 Accepted 5 July 2012 Available online 11 July 2012 Keywords: Bupivacaine Electrokinetic capillary chromatography Intralipid In vivo Liposome Pig model
a b s t r a c t Intravenous lipid emulsion is recommended as treatment for local anesthetic intoxication based on the hypothesis that the lipophilic drug is entrapped by the lipid phase created in plasma. We compared a 15.6 mM 80/20 mol% phosphatidyl choline (PC)/phosphatidyl glycerol (PG)-based liposome dispersion with the commercially available Intralipid® emulsion in a pig model of local anesthetic intoxication. Bupivacaine–lipid interactions were studied by electrokinetic capillary chromatography. Multilamellar vesicles were used in the first in vivo experiment series. This series was interrupted when the liposome dispersion was discovered to cause cardiovascular collapse. The toxicity was decreased by an optimized sonication of the 50% diluted liposome dispersion (7.8 mM). Twenty anesthetized pigs were then infused with either sonicated PC/PG liposome dispersion or Intralipid® , following infusion of a toxic dose of bupivacaine which decreased the mean arterial pressure by 50% from baseline. Bupivacaine concentrations were quantified in blood samples using liquid chromatography/mass spectrometry. No significant difference in the context-sensitive plasma half-life of bupivacaine was detected (p = 0.932). After 30 min of lipid infusion, the bupivacaine concentration was 8.2 ± 1.5 mg/L in the PC/PG group and 7.8 ± 1.8 mg/L in the Intralipid® group, with no difference between groups (p = 0.591). No difference in hemodynamic recovery was detected between groups (p > 0.05). © 2012 Elsevier B.V. All rights reserved.
1. Introduction Treatment of severe local anesthetic intoxication using intravenous lipid emulsion (ILE) has been widely adopted following the recommendations of Weinberg et al. [1]. It is thought that ILE entraps lipophilic substances in circulation by creating a substantial lipid phase in plasma, popularly called a “lipid sink” [2]. The lipid sink theory is supported by clinical experience; most intoxications successfully treated with ILE have been caused by lipophilic drugs [3,4]. Also, in vitro measurements confirm a strong interaction between lipid emulsion and the more lipophilic local anesthetics [5].
∗ Corresponding author at: Laboratory of Analytical Chemistry, Department of Chemistry, POB 55, 00014 University of Helsinki, Finland. Tel.: +358 9 191 50264; fax: +358 9 191 50253. ∗∗ Corresponding author at: Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital and University of Helsinki, POB 340, 00029 HUS, Finland. Tel.: +358 50 582 6268; fax: +358 9 471 74017. E-mail addresses: erik.litonius@hus.fi (E. Litonius), susanne.wiedmer@helsinki.fi (S.K. Wiedmer). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.07.013
Assuming that lipophilic entrapment is the mechanism behind the apparent beneficial effect of ILE on intoxications, any factor increasing the affinity of drugs to lipid emulsion would be expected to improve the therapeutic life-saving effect of ILE [6]. Previous studies on the in vitro evaluation showed that a liposome dispersion consisting of 80 mol% 1-palmitoyl-2-oleyl-sn-glycero3-phosphocholine (POPC) and 20 mol% 1-palmitoyl-2-oleyl-snglycero-3-[phospho-rac-(1-glycerol)] (POPG) had the strongest interaction with the local anesthetics bupivacaine, lidocaine and prilocaine [7]. Compared to the commercially available lipid emulsion, Intralipid® , currently the lipid emulsion of choice for the treatment of intoxications, the POPC/POPG dispersion had a much stronger interaction with the most lipophilic local anesthetic bupivacaine [7]. Previous work to increase the entrapment of drugs has involved the use of nano-sized vesicles [8], PEGylated anionic liposomes [9], and varying the triglyceride chain length in the lipid emulsion [10]. The entrapment of drugs by lipid dispersion is following the dynamic partitioning of drugs between the lipid and plasma environment. Recently we have studied drug to liposome interactions using a number of techniques, e.g. electrokinetic capillary chromatography (EKC), capillary electrochromatography (CEC),
126
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
immobilized intralipid chromatography, atomic force microscopy, and polarization fluorescence [5,7,11,12]. The previously developed liposome electrokinetic chromatography (LEKC) method for the interaction studies, which resembles micellar EKC [13], could not be applied here, because the therapeutic concentration of the lipid dispersion in the background electrolyte solution exhibit too high UV-absorbance leading to a very noisy baseline. In the current work we develop a method based on partial filling electrokinetic capillary chromatography (PF-EKC) that allows using high lipid concentration to mimic the real therapeutic doses [14–17]. We tested the entrapment efficiency of the lipid dispersions prepared in various ways suitable for up scaling using PF-EKC. The in vivo study aimed to determine whether the 80/20 mol% POPC/POPG dispersion would outperform the commercially available lipid emulsion Intralipid® when used to treat bupivacaine intoxication in a pig model. When the POPC/POPG liposomes unexpectedly proved to be toxic, we developed the liposome preparation to reduce the toxicity using the asymmetrical field flow fractionation (AsFlFFF). In our previous experiments we tested the entrapment efficiency of 1 mM extruded POPC/POPG liposomes [7]. Here we compare the entrapment of bupivacaine by therapeutic doses of Intralipid® and PC/PG dispersion of the same phospholipid concentration. 2. Experimental
neutral marker to measure the electroosmotic flow. All buffer and sample solutions were stored in a refrigerator. 2.3. Preparation of liposomes Lipid vesicles were prepared from POPC, POPG, and DMPG either by the lipid film method or by direct hydration of lipid powder by 0.9% NaCl solution. In the lipid film method POPC (20 mM), POPG (13 mM), and DMPG (10 mM) in chloroform were stored in a freezer. The appropriate amounts of the lipid stock solutions in chloroform were mixed to obtain the desired compositions. The resulting mixture was evaporated to dryness under a stream of pressurized air, and traces of solvent were removed by evacuation under reduced pressure for at least 16 h. The lipid residues were hydrated in 0.9% NaCl solution at 60 ◦ C for 60 min with shaking to yield multilamellar vesicles (MLVs). The vesicle-containing dispersion was vortexed 4 times during the hydration process. The resulting dispersion of MLVs was used either crude or it was further processed by sonication or filtration. In case of filtration the dispersion of MLVs was filtered through syringe filters with 450 nm pore size (Millipore, Billerica, MA, USA). The sonication procedure consisted of sonication for different times using a microtip Branson Sonifier 250 sonicator (Branson Ultrasonic, Danbury, CT, USA). The lipid dispersion was kept in an ice-water bath during sonication. The sonicated lipid dispersion was centrifuged for 5 min at 14 900 RCF at 15 ◦ C to remove metal chips abrupted from the sonication tip.
2.1. Materials and methods 2.4. Determination of dynamic viscosities using a CE apparatus 1-Palmitoyl-2-oleyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (sodium salt) (POPG), 1,2-dimyristoyl-sn-glycero-3-phospho(1 -rac-glycerol) (sodium salt) (DMPG), and 1-palmitoyl-2-oleylsn-glycero-3-phosphocholine (POPC) were purchased from Avanti Lipids (Alabaster, AL, USA). Bupivacaine and thiourea were from Sigma (St. Louis, MO, USA). The Intralipid® emulsion (20%) was from Fresenius Kabi AG (Bad Homburg, Germany). Sodium chloride 9 mg/mL (NaCl, 0.9%) was purchased from Baxter (Chicago, IL, USA). Hydrogen sodium phosphate was purchased from Sigma (Darmstadt, Germany), dihydrogen sodium phosphate monohydrate, and HPLC-grade methanol was from Mallinckrodt Baker (Deventer, The Netherlands). pH solutions (7 and 10) used for calibrating the pH meter were purchased from Merck (Darmstadt, Germany). Sodium hydroxide (1.0 M) was from FF-Chemicals (Yli-Ii, Finland) and chloroform from Rathburn (Walkerburn, UK). Distilled water was further purified with a Millipore water-purification system (Millipore, Molsheim, France). 2.2. Buffer and sample preparation Sodium phosphate buffer was prepared by mixing of hydrogen sodium phosphate and dihydrogen sodium phosphate monohydrate to yield a total concentration of all phosphate anion forms equaling to 8.2 mM and sodium cations equaling to 14.1 mM in the prepared buffer. The ionic strength of the sodium phosphate buffer at pH 7.4 was 20 mM. Before use, the sodium phosphate buffer solution was filtered through a 0.45-m syringe filter (Gelman Sciences, Ann Arbor, MI, USA). The sodium phosphate buffer was used as bupivacaine solvent and as a running buffer. A 0.9% NaCl solution was used to dilute the lipid dispersion. Bupivacaine for the capillary electromigration study was prepared from stock solution (1 mg/mL in water). The concentration of all drugs in the injected sample was 30 g/mL in sodium phosphate buffer at pH 7.4. During method optimization we tested different concentrations of bupivacaine. The concentration of 30 g/mL was well visible using UV detection and the peak of bupivacaine was gaussian. Thiourea in the concentration of 0.5 mM was used as a
A Hewlett Packard 3D Capillary Electrophoresis system (Agilent, Waldbronn, Germany) equipped with a diode array detector (wavelength 200 nm) was used to determine the dynamic viscosities. We measured the time of the flow of lipid dispersions through the capillary using a pressure of 30 mbar. The temperature of the capillary and the CE carusel with samples was kept at 37 ◦ C. The total length of the capillary was 38.5 cm, the length of the capillary to the detector was 30 cm and the inner dimension of the capillary was 50 m. The dynamic viscosities were determined using the Poisseuille equation. 2.5. Partial filling electrokinetic capillary chromatography For studies on bupivacaine–liposome interactions we employed a PF-EKC method using a Hewlett Packard 3D Capillary Electrophoresis system (Agilent, Waldbronn, Germany). Uncoated fused-silica capillaries were from Polymicro Technologies (Phoenix, AZ, USA). Dimensions of the capillaries were 50 m I.D. (375 m O.D.) with a length of 30/38.5 cm (length to the detector/total length). A new capillary was employed for electrophoretic measurements using Intralipid® and POPC/DMPG pseudostationary phases. The new capillary was preconditioned by rinsing at a pressure of 940 mbar for 15 min with 0.1 M sodium hydroxide, for 10 min with water, and for 5 min with lipid dispersion. A plug of lipid dispersion at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions were as follows: voltage 20 kV; temperature 37 ◦ C; sample injection 10 s at 15 mbar, UV-detection at 200 nm. The electrophoretic runs were repeated 3–6 times. 2.6. Determination of particle sizes The particles sizes were determined by dynamic light scattering (DLS) with Malvern Nano Zetasizer (Worcestershire, UK) using the software (DTS v6.01) or by AsFlFFF. The AsFlFFF experiments were conducted with AF2000 system (Postnova Analytics, Landsberg,
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Germany). The instrument was equipped with UV, LS, and RI detectors. Differential refractive index (DRI) was used at 125 RIU/V. Processing of the measured data was done with the Postnova software (AF2000 Control, version 1.1.026). The AsFlFFF separation channel had a tip-to-tip length of 27.5 cm, a thickness of 350-m, and an initial width of 2.0 cm with a final width of 1.0 cm in a trapezoidal shape. Both ends of the spacer were cut in a triangular shape. Regenerated cellulose membrane having a cutoff of 10 kDa (Z-MEM-AQU-627, Postnova) was placed at the accumulation wall of the AsFlFFF channel. Sample was injected using an autosampler. Application of constant cross flow was used during the analysis time. A pre-filtered 0.9% NaCl was the eluent. After flow equilibration, the sample (100 L) was injected with a flow rate of 0.2 mL/min, followed by a 4 min focusing with a cross-flow rate at 1.5 mL/min and a detector flow rate of 0.2 mL/min. Following a 1 min transition, the cross flow rate was decreased to 1.0 mL/min, and the detector flow rate at 1.0 mL/min for 180 min. The detector flow rate was kept at 1 mL/min throughout. 2.7. Experimental animals The Finnish National Animal Experiment Board approved the experimental protocol (ESAVI-2010-08544/Ym-23), and the experiments were performed in accordance with the European Community guidelines for the care and use of experimental animals at the animal research laboratory at Helsinki University Central Hospital. For the study, a total of 36 pigs were used. Ten of these were used for a first series, during which we observed serious hemodynamic side effects of the MLVs of POPC/POPG dispersion. Following this, changes were made to the POPC/POPG dispersion and tested in pilot experiments to determine whether the toxicity could be attenuated. Upon producing a dispersion that did not cause toxic side effects in pilot experiments, we performed a full series using a total of twenty pigs. 2.8. In vivo experiment protocol The pigs were anesthetized with isoflurane, a volatile general anesthetic, and placed on a ventilator (inspiratory O2 21%) following tracheal intubation. The pigs were monitored as previously described, with hemodynamic measurements performed by arterial cannulation and pulmonary artery catheter [18]. ECG and peripheral oxygen saturation were continuously recorded. Cardiac output was measured with the thermodilution method (mean of three measurements) just before each blood sampling for bupivacaine assays. Following a stabilization period, the pig was infused a toxic dose of bupivacaine (2 mg/kg/min) at a constant rate until its mean arterial blood pressure (MAP) had dropped to half of its baseline value (defined as the Toxic Point). At this point, the bupivacaine infusion and isoflurane were discontinued, the inspiratory O2 was increased to 100%, and an immediate infusion of lipid commenced. The pigs had been randomized to receive either POPC/POPG or Intralipid® , first a 1-min loading dose of 1.5 mL/kg, followed by a continuous infusion at 0.25 mL/kg/min for 29 min according to current clinical recommendations. The solution bags and tubings were covered with aluminum foil and tape in order to maintain blinding of the investigators. Hemodynamic parameters were recorded and arterial blood samples for bupivacaine concentration measurements taken at baseline, Toxic Point and 5, 10, 20 and 30 min later. Arterial blood gas samples were taken at baseline, Toxic Point, and at the end of the experiment to evaluate the blood pH of the pig. Thirty minutes after the Toxic Point, the experiment was ended and the pig euthanized by a rapid injection of potassium chloride. Since bupivacaine is in itself a potent general anesthetic at the dose used
127
in this study we only restarted isoflurane anesthesia in case the pig showed signs of imminent emergence from anesthesia. If MAP dropped below 25 mmHg we administered 0.25 mg intravenous epinephrine. To avoid bias, we set the criteria for preemptive epinephrine to MAP < 25 mmHg or heart rate < 50 bpm. In case of actual cardiac arrest, we would commence chest compressions at 100 compressions per minute, administering 0.25 mg epinephrine and defibrillating (150 J × 2 in case of ventricular fibrillation or hemodynamically compromised ventricular tachycardia) as needed. 2.9. Bupivacaine quantification The blood samples were separated and handled as previously described [18,19]. In short, the plasma was separated from the blood samples by centrifugation for 10 min at 3000 rpm (2500× g) and stored at −22 ◦ C. The total plasma bupivacaine concentration was determined in aliquots of all plasma samples. In addition, of the plasma samples drawn 5 min after the start of the lipid infusions, the free (non-proteinbound) fraction of bupivacaine was separated by ultrafiltration using Centrifree® Ultrafiltration Devices (Millipore Ireland Ltd, Tullagreen, Carrigtwohill, Co. Cork, Ireland) at 1260 g for 30 min at 25 ◦ C. Plasma concentrations of bupivacaine were quantified using solid phase extraction and liquid chromatography–tandem mass spectrometry (LC/MS/MS). The bupivacaine concentrations were measured using a previously described method [20] with some modifications, employing a SCIEX API 2000 Q Trap LC/MS/MS system (Sciex Division of MDS, Toronto, ON, Canada) with ropivacaine as internal standard. Plasma sample aliquots of 200 L were mixed with 200 L of water, 50 L of internal standard solution (100 ng ropivacaine/mL methanol) and 200 L of 4% phosphoric acid. The mixture was then loaded into extraction cartridges (Oasis MCX 1 cc/30 mg, Waters, Milford, MA, USA) that had been pre-conditioned with 1 mL of methanol and 1 mL of water. The cartridges were subsequently washed with 1 mL of 0.1% hydrochloric acid and 1 mL of methanol. The analytes were eluted with two applications of 0.6 mL of 5% ammonium hydroxide in methanol. The eluent was dried at 40 ◦ C under a nitrogen stream. The residue was dissolved in 100 L of the mobile phase, and 5 L of the sample was injected to the LC/MS/MS system. The chromatographic separation was carried out on an AtlantisT3 column (100 mm × 2.1 mm i.d., particle size 3 m; Waters Corp., Milford, MA, USA) protected by an AtlantisT3 precolumn (10 mm × 2.1 mm i.d., 3 m). The mobile phase consisted of (A) 2 mM ammonium acetate (pH 3.65) and (B) methanol. The total running time was 18 min with the mobile phase gradient of 0.5 min at 85% A, 6 min to 5% A, 1 min at 5% A, 0.1 min to 85% A, and 10.4 min at 85% A. The flow rate was 200 L/min. The ion transitions monitored were mass-to-charge ratio (m/z) 289 to m/z 140 for bupivacaine and m/z 275 to m/z 126 for ropivacaine. The lower limit of quantification of plasma bupivacaine was 0.5 ng/mL, and the between-day coefficient of variation (CV) was less than 15% at relevant concentrations (n = 11). 2.10. Statistical analysis We used nonlinear regression to determine the contextsensitive plasma half-life of bupivacaine, comparing the two lipid dispersions. To ensure that we also would note any transient effects, we compared the parameters at each time point using uncorrected unpaired t-tests. Toxicity in the first experiment series was evaluated using Fisher’s exact test. All statistical analysis was performed using Prism 5.0d for OS X (GraphPad Software, Inc., La Jolla, CA, USA).
128
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Table 1 The dynamic viscosity and time of the flow through the capillary with ID of 50 m and the length to the detector of 30 cm. The applied pressure was 50 mbar and the temperature was 37 ◦ C. Intralipid® Intralipid® diluted 50% undiluted Time (s) 126.35 4.27 Dynamic viscosity 10−4 (Pa s)
145.70 4.93
POPC/ POPC/ POPG 8 mM POPG 16 mM 126.18 4.27
130.06 4.40
3. Results and discussion 3.1. Up-scaling the preparation of the liposome dispersion and testing its entrapment efficiency The concentration of egg-yolk phosphatidylcholine (egg PC) in Intralipid® is 1.5%, which corresponds to 15.6 mM egg PC dispersion. Because the preparation of extruded liposomes at such high concentrations and volumes (approx. 270 mL for one animal) was not manageable, we prepared and tested the entrapment efficiency of vesicles suitable for up-scaling, i.e. multilamellar vesicles, sonicated vesicles, and filtered multilamellar vesicles. The tested liposomes consisted of 80 mol% PC and 20 mol% PG. The DMPG lipid was used in in vitro interaction tests instead of POPG lipid. The effect of the liposome acyl chain was tested on the interaction with bupivacaine in our previous experiments, where we compared the retardation of the mobility of bupivacaine driven through the plug of POPC/POPG 80/20 (v/v) or POPC/DMPG 80/20 (v/v) vesicles (both of 1 mM and extruded vesicles) using the plug length of 1250 mbar s. The mobility of bupivacaine decreased to 5.6 and 5.9 × 10−9 m2 V−1 s−1 using the POPC/POPG 80/20 (v/v) and POPC/DMPG 80/20 (v/v) vesicles, respectively. Based on this result and knowing that the overall negative charge of PG is given by the lipid head group and there is only 20 mol% of POPG in the liposome, we concluded that the differences in the acyl chain lengths of POPG and DMPG did not have a critical effect on the results in this work. The entrapment efficiency of bupivacaine by multilamellar, sonicated, and filtered vesicles was studied using PF-EKC. The previously established LEKC method for the interaction studies could not be applied here because liposome dispersions with the concentration higher than approximately 4 mM exhibit too high UV-absorbance of the background electrolyte leading to a very noisy baseline. Applying the PF-EKC mode it is possible to employ very high lipid concentrations. With such a system we can mimic the real therapeutic doses of lipids. In the PF-EKC method the sample plug is driven through the plug of ligand causing the retardation of interacting species. Maintaining the same length of the ligand plug it is possible to compare interactions between analytes and ligands. The main assumption of the identical length of lipid plug is the identical viscosity of the lipid emulsions. Therefore, we determined the dynamic viscosities using the CE apparatus. The time of the flow of the lipid dispersion through the capillary at 50 mbar was measured and the viscosities were calculated (see Table 1). Crude Intralipid® containing 15.6 mM egg PC was more viscous than 15.6 mM POPC/DMPG. We reached more or less the same viscosities of POPC/DMPG and Intralipid® using dispersions diluted to 50%, which were therefore selected for further experiments using PF-EKC. Next we optimized the length of the injected liposome plug to determine bupivacaine–liposome interactions by PF-EKC. The electrophoretic mobility of bupivacaine decreased with an increase in the length of the plug which was varied between 0 and 2.5% of the total length of the capillary (using 0–200 mbar s injections), see Fig. 1. The electrophoretic mobility of bupivacaine using a PC/PG and Intralipid plug length of 1.9% of the total length of the capillary, equaling 150 mbar s (15 mbar for 10 s), was about 46% and
Fig. 1. Electrophoretic mobilites of bupivacaine using various lengths of the plug of Intralipid® diluted 50% and 7.8 mM 80/20 mol% POPC/DMPG. BGE solution: phosphate buffer (20 mM ionic strength), pH 7.4. Injection of a lipid dispersion plug at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions: length of capillary 30/38.5 cm; voltage 20 kV; temperature 37 ◦ C; UV-detection at 200 nm.
58% of the value obtained by capillary electrophoresis (CZE). Using this optimal plug length (injection for 150 mbar s) we could clearly observe the difference in the mobility of bupivacaine caused by the PC/PG or Intralipid® dispersion and this plug length was selected for further experiments. The electrophoretic mobility of bupivacaine was lower using any plug length of 7.8 mM POPC/DMPG when compared to 50% Intralipid® (Fig. 1). This clearly reflects a stronger interaction of bupivacaine with POPC/DMPG than with Intralipid® emulsion. Using the optimal plug length of 150 mbar s we tested the entrapment efficiency of 7.8 mM POPC/DMPG vesicles that were prepared by different procedures. The entrapment efficiency of multilamellar, filtered, sonicated, and filtered/sonicated vesicles was compared with that of Intralipid® . Both multilamellar and sonicated vesicles showed higher interaction with bupivacaine when compared to Intralipid® (Fig. 2). The highest decrease of bupivacaine mobility, hence reflecting the strongest interaction, was observed using a plug of MLVs (Fig. 2). The sonication procedure slightly decreased the interaction with bupivacaine, while filtration fully abolished it. The filtration of multilamellar POPC/DMPG vesicles most probably erased most of the MLVs from the dispersion. Sonication of the filtered MLVs did not significantly change
Fig. 2. Electrophoretic mobility of bupivacaine using the optimal length of the plug of 15 mbar for 10 s. Lipid dispersions: Intralipid® diluted 50% and 7.8 mM 80/20 mol% POPC/DMPG. BGE solution: phosphate buffer (20 mM ionic strength), pH 7.4. The lipid plug at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions: length of capillary 30/38.5 cm; voltage 20 kV; temperature 37 ◦ C; UV-detection at 200 nm. The PC/PG dispersions were sonicated for 30 min using an ice water bath during sonication.
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
129
the interaction with bupivacaine (Fig. 2). The slight decrease of the interaction of bupivacaine to sonicated vesicles when compared to MLVs is in agreement with Howell and Chauhan who showed that MLVs entrap bupivacaine better than unilamellar vesicles [21]. The reason is the reduced dielectric constant, which lowers the energy barrier for the cationic drugs to enter the aliphatic bilayer. The multilamellar dispersion possessed the strongest interaction with bupivacaine when compared to other tested dispersions and was therefore selected for the tests in vivo. 3.2. First series of in vivo experiments During the experiments, we noticed that some pigs experienced cardiovascular collapse rapidly following the start of lipid infusion. Although the researchers were blinded to what lipid dispersion was infused, it was suspected that POPC/POPG caused the violent reaction since nothing similar had occurred previously using the commercially available lipid emulsions. Subsequently, the blinding was broken and we concluded that POPC/POPG had caused cardiovascular collapse in all three pigs that had received it. Of the seven pigs that had been infused Intralipid® after bupivacaine, only one needed resuscitation (p = 0.03). 3.3. Reducing the toxicity of POPC/POPG To determine whether POPC/POPG alone had caused the cardiovascular collapse, we infused an anesthetized pig with the same dose of lipid as in the study series without any prior bupivacaine intoxication. Five minutes after the start of the lipid infusion, MAP had risen from 75 mm Hg to 147 mm Hg, mean pulmonary artery pressure from 15 to 34 mm Hg, and pulmonary capillary wedge pressure from 12 to 35 mmHg. Concurrently, the hemoglobin oxygen saturation dropped from a normal 97% to 84%, indicating a disturbance in the pulmonary gas exchange. Our hypothesis is that POPC/POPG caused global vasoconstriction, leading to left ventricular failure and pulmonary hypertension, rapidly followed by exudation of fluid into the alveoles. One possible mechanism for this is complement activation-related pseudoallergy (CARPA), a hypersensitivity reaction to liposomes occurring both in some humans (5–8%) and pigs [22]. The pig survived for 30 min, and blood pressure returned back to normal during the later stage of the lipid infusion. However, blood oxygen saturation never recovered. Considering that the lipid in itself caused such a severe disturbance of the circulatory system, it is not surprising that the pigs already in severe bupivacaine intoxication perished when administered POPC/POPG. We subsequently tested the commercial lipid emulsion Intralipid® , concluding that it did not have any significant effect on the circulation, although a slight increase in mean arterial pressure could be observed during infusion. Thus, it was obvious that the multilamellar POPC/POPG dispersion had to be altered to reduce its toxicity. 3.4. Development of POPC/POPG dispersion of reduced toxicity Vesicles of high polydispersity and bigger size can increase the risk of in vivo toxicity [23]. One traditional method to decrease the lamellarity and size of MLVs is sonication of vesicles. Another method to decrease the vesicle size is filtration (extrusion) of MLVs. Both these procedures are fast and thus suitable for up-scaling. We tested several different sonication procedures to decrease the polydispersity and size of POPC/POPG based lipid vesicles (Table 2). We altered the sonication volumes (1 mL and 50 mL), sonication times (30 and 60 min), and lipid concentrations (7.8 mM and 15.6 mM). To provide a fast preparation of a large volume of liposome dispersion we also tested two hydration methods of the lipid.
Fig. 3. Particle sizes (hydrodynamic radius, rH) by FFF and DLS of POPC/POPG liposome samples. See Table 2 for description of the samples.
We used the traditional lipid film method and as an alternative we directly hydrated the lipid powders. Table 2 shows the sizes and PDIs by DLS of all the samples. Sample no. 1 is the sample of highest polydispersity. The polydispersity index (PDI) is a measure of size distribution in the sample. Monodispersed particle size distributions give PDI value < 0.1, whereas polydispersed or multi modal particle size distributions give PDI value > 0.1. Sample no. 1 was prepared by 30-min sonication of a 50-mL sample of 15.6 mM POPC/POPG, prepared by direct hydration of lipid powder (see Section 2). We clearly observed that longer sonication times (30 vs 60 min) decreased the polydispersity index (compare samples 4 and 6). Sonication of samples of smaller volumes (1 mL) decreased the PDI but not the size. In general, direct hydration of lipid powder produced bigger liposomes with slightly higher PDIs when compared to liposomes prepared by the traditional lipid film method. Asymmetrical field flow fractionation (AsFlFFF) and dynamic light scattering (DLS) were used to measure the sizes and PDIs of the POPC/POPG based vesicles. The two methods gave similar particle sizes (Fig. 3), but for some of the samples slightly higher standard deviations were obtained using AsFlFFF. The size of Intralipid® particles, as measured by DLS in a previous study, was 303.2 ± 15.6 nm [11]. Based on the results in Table 2 for in vivo applications we prepared liposomes by the lipid film method and used 60 min sonication of 50-mL of dispersion. Next the toxicity of the POPC/POPG dispersion was tested in vivo. 3.5. Main series of in vivo experiments After developing a POPC/POPG dispersion that we were confident would not have a deleterious effect on the pigs’ hemodynamics based on both theoretical reductions in toxic characteristics [23] and pilot experiments indicating no deleterious effect on the circulation (unpublished data), we restarted the study using a total of twenty pigs following the protocol described above. At the Toxic Point, the average plasma bupivacaine concentration was 23.4 ± 3.9 (SD) mg/L in the POPC/POPG group and 22.1 ± 2.3 mg/L in the Intralipid® group, with no difference between groups (p = 0.382, Fig. 4A). At the end of the experiment, after 30 min of lipid infusion, the bupivacaine concentration was 8.2 ± 1.5 mg/L in the POPC/POPG group and 7.8 ± 1.8 mg/L in the Intralipid® group, with no difference between groups (p = 0.591). The total plasma bupivacaine concentrations in the Intralipid® group are similar to those seen in our previous study using the same pig model of local anesthetic intoxication [19]. When estimating the contextsensitive half-life of bupivacaine in the present study, a two-phase model of decay fits the data best (R2 = 0.869). The fast and slow
130
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Table 2 Size and polydispersity indexes of POPC/DMPG vesicles prepared by different methods. #
1 2 3 4 5 6 7 8
Hydration method
Concentration
Water
15.6 mM
Lipid film
x x x x
7.8 mM
x x
Sonication volume
Sonication time
50 mL
30 min
x x x x
x x x x x x
1 mL
x x
x
x x x x x
x x x
x
phase half-lifes were 1.5 and 131 min for POPC/POPG and 1.3 and 78 min for Intralipid® , with no significant difference between groups (p = 0.932). Five minutes after the start of the lipid infusion, the average concentration of free (non-proteinbound) bupivacaine in the plasma samples was 1.2 ± 0.4 mg/L in the POPC/POPG group and 0.9 ± 0.4 mg/L in the Intralipid® group, with no difference between groups (p = 0.199). Thus, POPC/POPG did not affect the pharmacokinetics of bupivacaine when compared to Intralipid® . In accordance with this, we observed no difference between groups in the number of pigs requiring epinephrine due to severe hypotension (7 in
x
x x x
DLS, dH (nm)
PDI
103.3 108.0 102.3 114.0 90.7 88.2 64.1 103.6
0.40 0.24 0.27 0.22 0.29 0.22 0.34 0.23
60 min
the POPC/POPG group, 5 in the Intralipid® group, p = 0.650). Yet, the use of epinephrine may cause bias to the hemodynamic data of the pigs. In the remaining pigs that did not require epinephrine, there was no difference in hemodynamic recovery (Fig. 4B–D) as indicated by heart rate (p = 0.599), MAP (p = 0.892), or cardiac output (p = 0.717). The blood pH measured at baseline, Toxic Point and at 30 min remained between 7.45 and 7.55, with no difference between groups (p > 0.05). The hemodynamic response was quite similar to that observed by us in anesthetized pigs intoxicated with bupivacaine and treated using another clinical nutritional lipid emulsion [19].
Fig. 4. Total plasma bupivacaine concentration (A), heart rate (B), mean arterial pressure (C), and cardiac output (D) during lipid emulsion infusion preceded by infusion of a toxic dose of bupivacaine. No significant differences between lipid emulsions were found (p > 0.05).
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
4. Conclusions We conclude that, although the in vitro interaction between bupivacaine and POPC/DMPG was much stronger than between bupivacaine and Intralipid® , POPC/POPG did not have any significant effect on the in vivo pharmacokinetics of bupivacaine when compared to Intralipid® . The somewhat unexpected equal interaction of bupivacaine with POPC/POPG and Intralipid® may reflect the larger lipid phase formed in the plasma by the excess neutral lipids of Intralipid® . We cannot exclude that the use of an in vitro pharmacodynamic measure of the effect of entrapping bupivacaine, such as the inhibition of methemoglobin production in whole blood [24], would have predicted the similarity in pharmacokinetic effect more accurately. Acknowledgments Funding from the University of Helsinki Research Funds (SKW, JL: University of Helsinki project no. 2105060), Finska läkaresällskapet, Finland (PHR) and Helsinki University Hospital Disctrict EVO research funds (PHR) is acknowledged. We are grateful to Jouko Laitila for bupivacaine assays and to Veikko Huusko and Olli Valtanen for assistance in the pig experiments. References [1] G.L. Weinberg, Reg. Anesth. Pain Med. 35 (2010) 188. [2] G. Weinberg, B. Lin, S. Zheng, G. Di Gregorio, D. Hiller, R. Ripper, L. Edelman, K. Kelly, D. Feinstein, Crit. Care Med. 38 (2010) 2268.
131
[3] C. Jamaty, B. Bailey, A. Larocque, E. Notebaert, K. Sanogo, J.M. Chauny, Clin. Toxicol. 48 (2010) 1. [4] G. Cave, M. Harvey, Acad. Emerg. Med. 16 (2009) 815. [5] J. Laine, J. Lokajova, J. Parshintsev, J.M. Holopainen, S.K. Wiedmer, Anal. Bioanal. Chem. 396 (2010) 2599. [6] D. French, C. Smollin, W. Ruan, A. Wong, K. Drasner, A.H.B. Wu, Clin. Toxicol. 49 (2011) 801. [7] J. Lokajova, J. Laine, E. Puukilainen, M. Ritala, J.M. Holopainen, S.K. Wiedmer, Electrophoresis 31 (2010) 1540. [8] A.B. Dhanikula, D. Lamontagne, J.C. Leroux, Cardiovasc. Res. 74 (2007) 480. [9] B. Howell, A. Chauhan, J. Colloid Interface Sci. 324 (2008) 61. [10] W. Ruan, D. French, A. Wong, K. Drasner, A.H.B. Wu, Anesthesiology 116 (2012) 334. [11] J. Muhonen, J.M. Holopainen, S.K. Wiedmer, J. Chromatogr. A 1216 (2009) 3392. [12] J. Lokajova, J. Pukkila, J.M. Holopainen, S.K. Wiedmer, Eur. J. Pharm. Sci. 41 (2010) 515. [13] S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya, T. Ando, Anal. Chem. 56 (1984) 111. [14] E. Boija, A. Lundquist, M. Nilsson, K. Edwards, R. Isaksson, G. Johansson, Electrophoresis 29 (2008) 3377. [15] K. Vainikka, K. Reijmar, G. Yohannes, J. Samuelsson, K. Edwards, M. Jussila, M.-L. Riekkola, Anal. Biochem. 414 (2011) 117. [16] D. Corradini, G. Mancini, C. Bello, Chroma 60 (2004) S125. [17] G. Manetto, M. Silvana Bellini, Z. Deyl, J. Chromatogr. A 990 (2003) 205. [18] E. Litonius, T. Niiya, P.J. Neuvonen, P.H. Rosenberg, Basic Clin. Pharmacol. Toxicol. 110 (2012) 378. [19] E. Litonius, T. Niiya, P.J. Neuvonen, P.H. Rosenberg, Anesth. Analg. 114 (2012) 901. [20] G. Hoizey, D. Lamiable, A. Robinet, H. Ludot, J.-M. Malinovsky, M.L. Kaltenbach, L. Binet, C. Boulanger, H. Millart, J. Pharm. Biomed. Anal. 39 (2005) 587. [21] B.A. Howell, A. Chauhan, Langmuir 25 (2009) 12056. [22] J. Szebeni, C.R. Alving, L. Rosivall, R. Bunger, L. Baranyi, P. Bedocs, M. Toth, Y. Barenholz, J. Liposome Res. 17 (2007) 107. [23] J. Szebeni, L. Baranyi, S. Savay, M. Bodo, D.S. Morse, M. Basta, G.L. Stahl, R. Bunger, C.R. Alving, Am. J. Physiol. Heart. Circ. Physiol. 279 (2000) 1319. [24] T.L. Samuels, J.W. Willers, D.R. Uncles, R. Monteiro, C. Halloran, H. Dai, Anaesthesia 67 (2011) 23.
Contents lists available at SciVerse ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
In vitro and in vivo entrapment of bupivacaine by lipid dispersions Erik Litonius a,∗∗ , Jana Lokajova b , Gebrenegus Yohannes b , Pertti J. Neuvonen c , Juha M. Holopainen d , Per H. Rosenberg a , Susanne K. Wiedmer b,∗ a
Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital and University of Helsinki, Finland Laboratory of Analytical Chemistry, Department of Chemistry, University of Helsinki, Finland Department of Clinical Pharmacology, University of Helsinki and HUSLAB, Helsinki, Finland d Helsinki Eye Lab, Department of Ophthalmology, University of Helsinki, Helsinki, Finland b c
a r t i c l e
i n f o
Article history: Received 26 April 2012 Received in revised form 4 July 2012 Accepted 5 July 2012 Available online 11 July 2012 Keywords: Bupivacaine Electrokinetic capillary chromatography Intralipid In vivo Liposome Pig model
a b s t r a c t Intravenous lipid emulsion is recommended as treatment for local anesthetic intoxication based on the hypothesis that the lipophilic drug is entrapped by the lipid phase created in plasma. We compared a 15.6 mM 80/20 mol% phosphatidyl choline (PC)/phosphatidyl glycerol (PG)-based liposome dispersion with the commercially available Intralipid® emulsion in a pig model of local anesthetic intoxication. Bupivacaine–lipid interactions were studied by electrokinetic capillary chromatography. Multilamellar vesicles were used in the first in vivo experiment series. This series was interrupted when the liposome dispersion was discovered to cause cardiovascular collapse. The toxicity was decreased by an optimized sonication of the 50% diluted liposome dispersion (7.8 mM). Twenty anesthetized pigs were then infused with either sonicated PC/PG liposome dispersion or Intralipid® , following infusion of a toxic dose of bupivacaine which decreased the mean arterial pressure by 50% from baseline. Bupivacaine concentrations were quantified in blood samples using liquid chromatography/mass spectrometry. No significant difference in the context-sensitive plasma half-life of bupivacaine was detected (p = 0.932). After 30 min of lipid infusion, the bupivacaine concentration was 8.2 ± 1.5 mg/L in the PC/PG group and 7.8 ± 1.8 mg/L in the Intralipid® group, with no difference between groups (p = 0.591). No difference in hemodynamic recovery was detected between groups (p > 0.05). © 2012 Elsevier B.V. All rights reserved.
1. Introduction Treatment of severe local anesthetic intoxication using intravenous lipid emulsion (ILE) has been widely adopted following the recommendations of Weinberg et al. [1]. It is thought that ILE entraps lipophilic substances in circulation by creating a substantial lipid phase in plasma, popularly called a “lipid sink” [2]. The lipid sink theory is supported by clinical experience; most intoxications successfully treated with ILE have been caused by lipophilic drugs [3,4]. Also, in vitro measurements confirm a strong interaction between lipid emulsion and the more lipophilic local anesthetics [5].
∗ Corresponding author at: Laboratory of Analytical Chemistry, Department of Chemistry, POB 55, 00014 University of Helsinki, Finland. Tel.: +358 9 191 50264; fax: +358 9 191 50253. ∗∗ Corresponding author at: Anaesthesiology and Intensive Care Medicine, Helsinki University Central Hospital and University of Helsinki, POB 340, 00029 HUS, Finland. Tel.: +358 50 582 6268; fax: +358 9 471 74017. E-mail addresses: erik.litonius@hus.fi (E. Litonius), susanne.wiedmer@helsinki.fi (S.K. Wiedmer). 0021-9673/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chroma.2012.07.013
Assuming that lipophilic entrapment is the mechanism behind the apparent beneficial effect of ILE on intoxications, any factor increasing the affinity of drugs to lipid emulsion would be expected to improve the therapeutic life-saving effect of ILE [6]. Previous studies on the in vitro evaluation showed that a liposome dispersion consisting of 80 mol% 1-palmitoyl-2-oleyl-sn-glycero3-phosphocholine (POPC) and 20 mol% 1-palmitoyl-2-oleyl-snglycero-3-[phospho-rac-(1-glycerol)] (POPG) had the strongest interaction with the local anesthetics bupivacaine, lidocaine and prilocaine [7]. Compared to the commercially available lipid emulsion, Intralipid® , currently the lipid emulsion of choice for the treatment of intoxications, the POPC/POPG dispersion had a much stronger interaction with the most lipophilic local anesthetic bupivacaine [7]. Previous work to increase the entrapment of drugs has involved the use of nano-sized vesicles [8], PEGylated anionic liposomes [9], and varying the triglyceride chain length in the lipid emulsion [10]. The entrapment of drugs by lipid dispersion is following the dynamic partitioning of drugs between the lipid and plasma environment. Recently we have studied drug to liposome interactions using a number of techniques, e.g. electrokinetic capillary chromatography (EKC), capillary electrochromatography (CEC),
126
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
immobilized intralipid chromatography, atomic force microscopy, and polarization fluorescence [5,7,11,12]. The previously developed liposome electrokinetic chromatography (LEKC) method for the interaction studies, which resembles micellar EKC [13], could not be applied here, because the therapeutic concentration of the lipid dispersion in the background electrolyte solution exhibit too high UV-absorbance leading to a very noisy baseline. In the current work we develop a method based on partial filling electrokinetic capillary chromatography (PF-EKC) that allows using high lipid concentration to mimic the real therapeutic doses [14–17]. We tested the entrapment efficiency of the lipid dispersions prepared in various ways suitable for up scaling using PF-EKC. The in vivo study aimed to determine whether the 80/20 mol% POPC/POPG dispersion would outperform the commercially available lipid emulsion Intralipid® when used to treat bupivacaine intoxication in a pig model. When the POPC/POPG liposomes unexpectedly proved to be toxic, we developed the liposome preparation to reduce the toxicity using the asymmetrical field flow fractionation (AsFlFFF). In our previous experiments we tested the entrapment efficiency of 1 mM extruded POPC/POPG liposomes [7]. Here we compare the entrapment of bupivacaine by therapeutic doses of Intralipid® and PC/PG dispersion of the same phospholipid concentration. 2. Experimental
neutral marker to measure the electroosmotic flow. All buffer and sample solutions were stored in a refrigerator. 2.3. Preparation of liposomes Lipid vesicles were prepared from POPC, POPG, and DMPG either by the lipid film method or by direct hydration of lipid powder by 0.9% NaCl solution. In the lipid film method POPC (20 mM), POPG (13 mM), and DMPG (10 mM) in chloroform were stored in a freezer. The appropriate amounts of the lipid stock solutions in chloroform were mixed to obtain the desired compositions. The resulting mixture was evaporated to dryness under a stream of pressurized air, and traces of solvent were removed by evacuation under reduced pressure for at least 16 h. The lipid residues were hydrated in 0.9% NaCl solution at 60 ◦ C for 60 min with shaking to yield multilamellar vesicles (MLVs). The vesicle-containing dispersion was vortexed 4 times during the hydration process. The resulting dispersion of MLVs was used either crude or it was further processed by sonication or filtration. In case of filtration the dispersion of MLVs was filtered through syringe filters with 450 nm pore size (Millipore, Billerica, MA, USA). The sonication procedure consisted of sonication for different times using a microtip Branson Sonifier 250 sonicator (Branson Ultrasonic, Danbury, CT, USA). The lipid dispersion was kept in an ice-water bath during sonication. The sonicated lipid dispersion was centrifuged for 5 min at 14 900 RCF at 15 ◦ C to remove metal chips abrupted from the sonication tip.
2.1. Materials and methods 2.4. Determination of dynamic viscosities using a CE apparatus 1-Palmitoyl-2-oleyl-sn-glycero-3-[phospho-rac-(1-glycerol)] (sodium salt) (POPG), 1,2-dimyristoyl-sn-glycero-3-phospho(1 -rac-glycerol) (sodium salt) (DMPG), and 1-palmitoyl-2-oleylsn-glycero-3-phosphocholine (POPC) were purchased from Avanti Lipids (Alabaster, AL, USA). Bupivacaine and thiourea were from Sigma (St. Louis, MO, USA). The Intralipid® emulsion (20%) was from Fresenius Kabi AG (Bad Homburg, Germany). Sodium chloride 9 mg/mL (NaCl, 0.9%) was purchased from Baxter (Chicago, IL, USA). Hydrogen sodium phosphate was purchased from Sigma (Darmstadt, Germany), dihydrogen sodium phosphate monohydrate, and HPLC-grade methanol was from Mallinckrodt Baker (Deventer, The Netherlands). pH solutions (7 and 10) used for calibrating the pH meter were purchased from Merck (Darmstadt, Germany). Sodium hydroxide (1.0 M) was from FF-Chemicals (Yli-Ii, Finland) and chloroform from Rathburn (Walkerburn, UK). Distilled water was further purified with a Millipore water-purification system (Millipore, Molsheim, France). 2.2. Buffer and sample preparation Sodium phosphate buffer was prepared by mixing of hydrogen sodium phosphate and dihydrogen sodium phosphate monohydrate to yield a total concentration of all phosphate anion forms equaling to 8.2 mM and sodium cations equaling to 14.1 mM in the prepared buffer. The ionic strength of the sodium phosphate buffer at pH 7.4 was 20 mM. Before use, the sodium phosphate buffer solution was filtered through a 0.45-m syringe filter (Gelman Sciences, Ann Arbor, MI, USA). The sodium phosphate buffer was used as bupivacaine solvent and as a running buffer. A 0.9% NaCl solution was used to dilute the lipid dispersion. Bupivacaine for the capillary electromigration study was prepared from stock solution (1 mg/mL in water). The concentration of all drugs in the injected sample was 30 g/mL in sodium phosphate buffer at pH 7.4. During method optimization we tested different concentrations of bupivacaine. The concentration of 30 g/mL was well visible using UV detection and the peak of bupivacaine was gaussian. Thiourea in the concentration of 0.5 mM was used as a
A Hewlett Packard 3D Capillary Electrophoresis system (Agilent, Waldbronn, Germany) equipped with a diode array detector (wavelength 200 nm) was used to determine the dynamic viscosities. We measured the time of the flow of lipid dispersions through the capillary using a pressure of 30 mbar. The temperature of the capillary and the CE carusel with samples was kept at 37 ◦ C. The total length of the capillary was 38.5 cm, the length of the capillary to the detector was 30 cm and the inner dimension of the capillary was 50 m. The dynamic viscosities were determined using the Poisseuille equation. 2.5. Partial filling electrokinetic capillary chromatography For studies on bupivacaine–liposome interactions we employed a PF-EKC method using a Hewlett Packard 3D Capillary Electrophoresis system (Agilent, Waldbronn, Germany). Uncoated fused-silica capillaries were from Polymicro Technologies (Phoenix, AZ, USA). Dimensions of the capillaries were 50 m I.D. (375 m O.D.) with a length of 30/38.5 cm (length to the detector/total length). A new capillary was employed for electrophoretic measurements using Intralipid® and POPC/DMPG pseudostationary phases. The new capillary was preconditioned by rinsing at a pressure of 940 mbar for 15 min with 0.1 M sodium hydroxide, for 10 min with water, and for 5 min with lipid dispersion. A plug of lipid dispersion at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions were as follows: voltage 20 kV; temperature 37 ◦ C; sample injection 10 s at 15 mbar, UV-detection at 200 nm. The electrophoretic runs were repeated 3–6 times. 2.6. Determination of particle sizes The particles sizes were determined by dynamic light scattering (DLS) with Malvern Nano Zetasizer (Worcestershire, UK) using the software (DTS v6.01) or by AsFlFFF. The AsFlFFF experiments were conducted with AF2000 system (Postnova Analytics, Landsberg,
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Germany). The instrument was equipped with UV, LS, and RI detectors. Differential refractive index (DRI) was used at 125 RIU/V. Processing of the measured data was done with the Postnova software (AF2000 Control, version 1.1.026). The AsFlFFF separation channel had a tip-to-tip length of 27.5 cm, a thickness of 350-m, and an initial width of 2.0 cm with a final width of 1.0 cm in a trapezoidal shape. Both ends of the spacer were cut in a triangular shape. Regenerated cellulose membrane having a cutoff of 10 kDa (Z-MEM-AQU-627, Postnova) was placed at the accumulation wall of the AsFlFFF channel. Sample was injected using an autosampler. Application of constant cross flow was used during the analysis time. A pre-filtered 0.9% NaCl was the eluent. After flow equilibration, the sample (100 L) was injected with a flow rate of 0.2 mL/min, followed by a 4 min focusing with a cross-flow rate at 1.5 mL/min and a detector flow rate of 0.2 mL/min. Following a 1 min transition, the cross flow rate was decreased to 1.0 mL/min, and the detector flow rate at 1.0 mL/min for 180 min. The detector flow rate was kept at 1 mL/min throughout. 2.7. Experimental animals The Finnish National Animal Experiment Board approved the experimental protocol (ESAVI-2010-08544/Ym-23), and the experiments were performed in accordance with the European Community guidelines for the care and use of experimental animals at the animal research laboratory at Helsinki University Central Hospital. For the study, a total of 36 pigs were used. Ten of these were used for a first series, during which we observed serious hemodynamic side effects of the MLVs of POPC/POPG dispersion. Following this, changes were made to the POPC/POPG dispersion and tested in pilot experiments to determine whether the toxicity could be attenuated. Upon producing a dispersion that did not cause toxic side effects in pilot experiments, we performed a full series using a total of twenty pigs. 2.8. In vivo experiment protocol The pigs were anesthetized with isoflurane, a volatile general anesthetic, and placed on a ventilator (inspiratory O2 21%) following tracheal intubation. The pigs were monitored as previously described, with hemodynamic measurements performed by arterial cannulation and pulmonary artery catheter [18]. ECG and peripheral oxygen saturation were continuously recorded. Cardiac output was measured with the thermodilution method (mean of three measurements) just before each blood sampling for bupivacaine assays. Following a stabilization period, the pig was infused a toxic dose of bupivacaine (2 mg/kg/min) at a constant rate until its mean arterial blood pressure (MAP) had dropped to half of its baseline value (defined as the Toxic Point). At this point, the bupivacaine infusion and isoflurane were discontinued, the inspiratory O2 was increased to 100%, and an immediate infusion of lipid commenced. The pigs had been randomized to receive either POPC/POPG or Intralipid® , first a 1-min loading dose of 1.5 mL/kg, followed by a continuous infusion at 0.25 mL/kg/min for 29 min according to current clinical recommendations. The solution bags and tubings were covered with aluminum foil and tape in order to maintain blinding of the investigators. Hemodynamic parameters were recorded and arterial blood samples for bupivacaine concentration measurements taken at baseline, Toxic Point and 5, 10, 20 and 30 min later. Arterial blood gas samples were taken at baseline, Toxic Point, and at the end of the experiment to evaluate the blood pH of the pig. Thirty minutes after the Toxic Point, the experiment was ended and the pig euthanized by a rapid injection of potassium chloride. Since bupivacaine is in itself a potent general anesthetic at the dose used
127
in this study we only restarted isoflurane anesthesia in case the pig showed signs of imminent emergence from anesthesia. If MAP dropped below 25 mmHg we administered 0.25 mg intravenous epinephrine. To avoid bias, we set the criteria for preemptive epinephrine to MAP < 25 mmHg or heart rate < 50 bpm. In case of actual cardiac arrest, we would commence chest compressions at 100 compressions per minute, administering 0.25 mg epinephrine and defibrillating (150 J × 2 in case of ventricular fibrillation or hemodynamically compromised ventricular tachycardia) as needed. 2.9. Bupivacaine quantification The blood samples were separated and handled as previously described [18,19]. In short, the plasma was separated from the blood samples by centrifugation for 10 min at 3000 rpm (2500× g) and stored at −22 ◦ C. The total plasma bupivacaine concentration was determined in aliquots of all plasma samples. In addition, of the plasma samples drawn 5 min after the start of the lipid infusions, the free (non-proteinbound) fraction of bupivacaine was separated by ultrafiltration using Centrifree® Ultrafiltration Devices (Millipore Ireland Ltd, Tullagreen, Carrigtwohill, Co. Cork, Ireland) at 1260 g for 30 min at 25 ◦ C. Plasma concentrations of bupivacaine were quantified using solid phase extraction and liquid chromatography–tandem mass spectrometry (LC/MS/MS). The bupivacaine concentrations were measured using a previously described method [20] with some modifications, employing a SCIEX API 2000 Q Trap LC/MS/MS system (Sciex Division of MDS, Toronto, ON, Canada) with ropivacaine as internal standard. Plasma sample aliquots of 200 L were mixed with 200 L of water, 50 L of internal standard solution (100 ng ropivacaine/mL methanol) and 200 L of 4% phosphoric acid. The mixture was then loaded into extraction cartridges (Oasis MCX 1 cc/30 mg, Waters, Milford, MA, USA) that had been pre-conditioned with 1 mL of methanol and 1 mL of water. The cartridges were subsequently washed with 1 mL of 0.1% hydrochloric acid and 1 mL of methanol. The analytes were eluted with two applications of 0.6 mL of 5% ammonium hydroxide in methanol. The eluent was dried at 40 ◦ C under a nitrogen stream. The residue was dissolved in 100 L of the mobile phase, and 5 L of the sample was injected to the LC/MS/MS system. The chromatographic separation was carried out on an AtlantisT3 column (100 mm × 2.1 mm i.d., particle size 3 m; Waters Corp., Milford, MA, USA) protected by an AtlantisT3 precolumn (10 mm × 2.1 mm i.d., 3 m). The mobile phase consisted of (A) 2 mM ammonium acetate (pH 3.65) and (B) methanol. The total running time was 18 min with the mobile phase gradient of 0.5 min at 85% A, 6 min to 5% A, 1 min at 5% A, 0.1 min to 85% A, and 10.4 min at 85% A. The flow rate was 200 L/min. The ion transitions monitored were mass-to-charge ratio (m/z) 289 to m/z 140 for bupivacaine and m/z 275 to m/z 126 for ropivacaine. The lower limit of quantification of plasma bupivacaine was 0.5 ng/mL, and the between-day coefficient of variation (CV) was less than 15% at relevant concentrations (n = 11). 2.10. Statistical analysis We used nonlinear regression to determine the contextsensitive plasma half-life of bupivacaine, comparing the two lipid dispersions. To ensure that we also would note any transient effects, we compared the parameters at each time point using uncorrected unpaired t-tests. Toxicity in the first experiment series was evaluated using Fisher’s exact test. All statistical analysis was performed using Prism 5.0d for OS X (GraphPad Software, Inc., La Jolla, CA, USA).
128
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Table 1 The dynamic viscosity and time of the flow through the capillary with ID of 50 m and the length to the detector of 30 cm. The applied pressure was 50 mbar and the temperature was 37 ◦ C. Intralipid® Intralipid® diluted 50% undiluted Time (s) 126.35 4.27 Dynamic viscosity 10−4 (Pa s)
145.70 4.93
POPC/ POPC/ POPG 8 mM POPG 16 mM 126.18 4.27
130.06 4.40
3. Results and discussion 3.1. Up-scaling the preparation of the liposome dispersion and testing its entrapment efficiency The concentration of egg-yolk phosphatidylcholine (egg PC) in Intralipid® is 1.5%, which corresponds to 15.6 mM egg PC dispersion. Because the preparation of extruded liposomes at such high concentrations and volumes (approx. 270 mL for one animal) was not manageable, we prepared and tested the entrapment efficiency of vesicles suitable for up-scaling, i.e. multilamellar vesicles, sonicated vesicles, and filtered multilamellar vesicles. The tested liposomes consisted of 80 mol% PC and 20 mol% PG. The DMPG lipid was used in in vitro interaction tests instead of POPG lipid. The effect of the liposome acyl chain was tested on the interaction with bupivacaine in our previous experiments, where we compared the retardation of the mobility of bupivacaine driven through the plug of POPC/POPG 80/20 (v/v) or POPC/DMPG 80/20 (v/v) vesicles (both of 1 mM and extruded vesicles) using the plug length of 1250 mbar s. The mobility of bupivacaine decreased to 5.6 and 5.9 × 10−9 m2 V−1 s−1 using the POPC/POPG 80/20 (v/v) and POPC/DMPG 80/20 (v/v) vesicles, respectively. Based on this result and knowing that the overall negative charge of PG is given by the lipid head group and there is only 20 mol% of POPG in the liposome, we concluded that the differences in the acyl chain lengths of POPG and DMPG did not have a critical effect on the results in this work. The entrapment efficiency of bupivacaine by multilamellar, sonicated, and filtered vesicles was studied using PF-EKC. The previously established LEKC method for the interaction studies could not be applied here because liposome dispersions with the concentration higher than approximately 4 mM exhibit too high UV-absorbance of the background electrolyte leading to a very noisy baseline. Applying the PF-EKC mode it is possible to employ very high lipid concentrations. With such a system we can mimic the real therapeutic doses of lipids. In the PF-EKC method the sample plug is driven through the plug of ligand causing the retardation of interacting species. Maintaining the same length of the ligand plug it is possible to compare interactions between analytes and ligands. The main assumption of the identical length of lipid plug is the identical viscosity of the lipid emulsions. Therefore, we determined the dynamic viscosities using the CE apparatus. The time of the flow of the lipid dispersion through the capillary at 50 mbar was measured and the viscosities were calculated (see Table 1). Crude Intralipid® containing 15.6 mM egg PC was more viscous than 15.6 mM POPC/DMPG. We reached more or less the same viscosities of POPC/DMPG and Intralipid® using dispersions diluted to 50%, which were therefore selected for further experiments using PF-EKC. Next we optimized the length of the injected liposome plug to determine bupivacaine–liposome interactions by PF-EKC. The electrophoretic mobility of bupivacaine decreased with an increase in the length of the plug which was varied between 0 and 2.5% of the total length of the capillary (using 0–200 mbar s injections), see Fig. 1. The electrophoretic mobility of bupivacaine using a PC/PG and Intralipid plug length of 1.9% of the total length of the capillary, equaling 150 mbar s (15 mbar for 10 s), was about 46% and
Fig. 1. Electrophoretic mobilites of bupivacaine using various lengths of the plug of Intralipid® diluted 50% and 7.8 mM 80/20 mol% POPC/DMPG. BGE solution: phosphate buffer (20 mM ionic strength), pH 7.4. Injection of a lipid dispersion plug at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions: length of capillary 30/38.5 cm; voltage 20 kV; temperature 37 ◦ C; UV-detection at 200 nm.
58% of the value obtained by capillary electrophoresis (CZE). Using this optimal plug length (injection for 150 mbar s) we could clearly observe the difference in the mobility of bupivacaine caused by the PC/PG or Intralipid® dispersion and this plug length was selected for further experiments. The electrophoretic mobility of bupivacaine was lower using any plug length of 7.8 mM POPC/DMPG when compared to 50% Intralipid® (Fig. 1). This clearly reflects a stronger interaction of bupivacaine with POPC/DMPG than with Intralipid® emulsion. Using the optimal plug length of 150 mbar s we tested the entrapment efficiency of 7.8 mM POPC/DMPG vesicles that were prepared by different procedures. The entrapment efficiency of multilamellar, filtered, sonicated, and filtered/sonicated vesicles was compared with that of Intralipid® . Both multilamellar and sonicated vesicles showed higher interaction with bupivacaine when compared to Intralipid® (Fig. 2). The highest decrease of bupivacaine mobility, hence reflecting the strongest interaction, was observed using a plug of MLVs (Fig. 2). The sonication procedure slightly decreased the interaction with bupivacaine, while filtration fully abolished it. The filtration of multilamellar POPC/DMPG vesicles most probably erased most of the MLVs from the dispersion. Sonication of the filtered MLVs did not significantly change
Fig. 2. Electrophoretic mobility of bupivacaine using the optimal length of the plug of 15 mbar for 10 s. Lipid dispersions: Intralipid® diluted 50% and 7.8 mM 80/20 mol% POPC/DMPG. BGE solution: phosphate buffer (20 mM ionic strength), pH 7.4. The lipid plug at various lengths was followed by a sample plug of 30 g/mL bupivacaine and 0.5 mM thiourea in BGE (15 s for 10 mbar). Running conditions: length of capillary 30/38.5 cm; voltage 20 kV; temperature 37 ◦ C; UV-detection at 200 nm. The PC/PG dispersions were sonicated for 30 min using an ice water bath during sonication.
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
129
the interaction with bupivacaine (Fig. 2). The slight decrease of the interaction of bupivacaine to sonicated vesicles when compared to MLVs is in agreement with Howell and Chauhan who showed that MLVs entrap bupivacaine better than unilamellar vesicles [21]. The reason is the reduced dielectric constant, which lowers the energy barrier for the cationic drugs to enter the aliphatic bilayer. The multilamellar dispersion possessed the strongest interaction with bupivacaine when compared to other tested dispersions and was therefore selected for the tests in vivo. 3.2. First series of in vivo experiments During the experiments, we noticed that some pigs experienced cardiovascular collapse rapidly following the start of lipid infusion. Although the researchers were blinded to what lipid dispersion was infused, it was suspected that POPC/POPG caused the violent reaction since nothing similar had occurred previously using the commercially available lipid emulsions. Subsequently, the blinding was broken and we concluded that POPC/POPG had caused cardiovascular collapse in all three pigs that had received it. Of the seven pigs that had been infused Intralipid® after bupivacaine, only one needed resuscitation (p = 0.03). 3.3. Reducing the toxicity of POPC/POPG To determine whether POPC/POPG alone had caused the cardiovascular collapse, we infused an anesthetized pig with the same dose of lipid as in the study series without any prior bupivacaine intoxication. Five minutes after the start of the lipid infusion, MAP had risen from 75 mm Hg to 147 mm Hg, mean pulmonary artery pressure from 15 to 34 mm Hg, and pulmonary capillary wedge pressure from 12 to 35 mmHg. Concurrently, the hemoglobin oxygen saturation dropped from a normal 97% to 84%, indicating a disturbance in the pulmonary gas exchange. Our hypothesis is that POPC/POPG caused global vasoconstriction, leading to left ventricular failure and pulmonary hypertension, rapidly followed by exudation of fluid into the alveoles. One possible mechanism for this is complement activation-related pseudoallergy (CARPA), a hypersensitivity reaction to liposomes occurring both in some humans (5–8%) and pigs [22]. The pig survived for 30 min, and blood pressure returned back to normal during the later stage of the lipid infusion. However, blood oxygen saturation never recovered. Considering that the lipid in itself caused such a severe disturbance of the circulatory system, it is not surprising that the pigs already in severe bupivacaine intoxication perished when administered POPC/POPG. We subsequently tested the commercial lipid emulsion Intralipid® , concluding that it did not have any significant effect on the circulation, although a slight increase in mean arterial pressure could be observed during infusion. Thus, it was obvious that the multilamellar POPC/POPG dispersion had to be altered to reduce its toxicity. 3.4. Development of POPC/POPG dispersion of reduced toxicity Vesicles of high polydispersity and bigger size can increase the risk of in vivo toxicity [23]. One traditional method to decrease the lamellarity and size of MLVs is sonication of vesicles. Another method to decrease the vesicle size is filtration (extrusion) of MLVs. Both these procedures are fast and thus suitable for up-scaling. We tested several different sonication procedures to decrease the polydispersity and size of POPC/POPG based lipid vesicles (Table 2). We altered the sonication volumes (1 mL and 50 mL), sonication times (30 and 60 min), and lipid concentrations (7.8 mM and 15.6 mM). To provide a fast preparation of a large volume of liposome dispersion we also tested two hydration methods of the lipid.
Fig. 3. Particle sizes (hydrodynamic radius, rH) by FFF and DLS of POPC/POPG liposome samples. See Table 2 for description of the samples.
We used the traditional lipid film method and as an alternative we directly hydrated the lipid powders. Table 2 shows the sizes and PDIs by DLS of all the samples. Sample no. 1 is the sample of highest polydispersity. The polydispersity index (PDI) is a measure of size distribution in the sample. Monodispersed particle size distributions give PDI value < 0.1, whereas polydispersed or multi modal particle size distributions give PDI value > 0.1. Sample no. 1 was prepared by 30-min sonication of a 50-mL sample of 15.6 mM POPC/POPG, prepared by direct hydration of lipid powder (see Section 2). We clearly observed that longer sonication times (30 vs 60 min) decreased the polydispersity index (compare samples 4 and 6). Sonication of samples of smaller volumes (1 mL) decreased the PDI but not the size. In general, direct hydration of lipid powder produced bigger liposomes with slightly higher PDIs when compared to liposomes prepared by the traditional lipid film method. Asymmetrical field flow fractionation (AsFlFFF) and dynamic light scattering (DLS) were used to measure the sizes and PDIs of the POPC/POPG based vesicles. The two methods gave similar particle sizes (Fig. 3), but for some of the samples slightly higher standard deviations were obtained using AsFlFFF. The size of Intralipid® particles, as measured by DLS in a previous study, was 303.2 ± 15.6 nm [11]. Based on the results in Table 2 for in vivo applications we prepared liposomes by the lipid film method and used 60 min sonication of 50-mL of dispersion. Next the toxicity of the POPC/POPG dispersion was tested in vivo. 3.5. Main series of in vivo experiments After developing a POPC/POPG dispersion that we were confident would not have a deleterious effect on the pigs’ hemodynamics based on both theoretical reductions in toxic characteristics [23] and pilot experiments indicating no deleterious effect on the circulation (unpublished data), we restarted the study using a total of twenty pigs following the protocol described above. At the Toxic Point, the average plasma bupivacaine concentration was 23.4 ± 3.9 (SD) mg/L in the POPC/POPG group and 22.1 ± 2.3 mg/L in the Intralipid® group, with no difference between groups (p = 0.382, Fig. 4A). At the end of the experiment, after 30 min of lipid infusion, the bupivacaine concentration was 8.2 ± 1.5 mg/L in the POPC/POPG group and 7.8 ± 1.8 mg/L in the Intralipid® group, with no difference between groups (p = 0.591). The total plasma bupivacaine concentrations in the Intralipid® group are similar to those seen in our previous study using the same pig model of local anesthetic intoxication [19]. When estimating the contextsensitive half-life of bupivacaine in the present study, a two-phase model of decay fits the data best (R2 = 0.869). The fast and slow
130
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
Table 2 Size and polydispersity indexes of POPC/DMPG vesicles prepared by different methods. #
1 2 3 4 5 6 7 8
Hydration method
Concentration
Water
15.6 mM
Lipid film
x x x x
7.8 mM
x x
Sonication volume
Sonication time
50 mL
30 min
x x x x
x x x x x x
1 mL
x x
x
x x x x x
x x x
x
phase half-lifes were 1.5 and 131 min for POPC/POPG and 1.3 and 78 min for Intralipid® , with no significant difference between groups (p = 0.932). Five minutes after the start of the lipid infusion, the average concentration of free (non-proteinbound) bupivacaine in the plasma samples was 1.2 ± 0.4 mg/L in the POPC/POPG group and 0.9 ± 0.4 mg/L in the Intralipid® group, with no difference between groups (p = 0.199). Thus, POPC/POPG did not affect the pharmacokinetics of bupivacaine when compared to Intralipid® . In accordance with this, we observed no difference between groups in the number of pigs requiring epinephrine due to severe hypotension (7 in
x
x x x
DLS, dH (nm)
PDI
103.3 108.0 102.3 114.0 90.7 88.2 64.1 103.6
0.40 0.24 0.27 0.22 0.29 0.22 0.34 0.23
60 min
the POPC/POPG group, 5 in the Intralipid® group, p = 0.650). Yet, the use of epinephrine may cause bias to the hemodynamic data of the pigs. In the remaining pigs that did not require epinephrine, there was no difference in hemodynamic recovery (Fig. 4B–D) as indicated by heart rate (p = 0.599), MAP (p = 0.892), or cardiac output (p = 0.717). The blood pH measured at baseline, Toxic Point and at 30 min remained between 7.45 and 7.55, with no difference between groups (p > 0.05). The hemodynamic response was quite similar to that observed by us in anesthetized pigs intoxicated with bupivacaine and treated using another clinical nutritional lipid emulsion [19].
Fig. 4. Total plasma bupivacaine concentration (A), heart rate (B), mean arterial pressure (C), and cardiac output (D) during lipid emulsion infusion preceded by infusion of a toxic dose of bupivacaine. No significant differences between lipid emulsions were found (p > 0.05).
E. Litonius et al. / J. Chromatogr. A 1254 (2012) 125–131
4. Conclusions We conclude that, although the in vitro interaction between bupivacaine and POPC/DMPG was much stronger than between bupivacaine and Intralipid® , POPC/POPG did not have any significant effect on the in vivo pharmacokinetics of bupivacaine when compared to Intralipid® . The somewhat unexpected equal interaction of bupivacaine with POPC/POPG and Intralipid® may reflect the larger lipid phase formed in the plasma by the excess neutral lipids of Intralipid® . We cannot exclude that the use of an in vitro pharmacodynamic measure of the effect of entrapping bupivacaine, such as the inhibition of methemoglobin production in whole blood [24], would have predicted the similarity in pharmacokinetic effect more accurately. Acknowledgments Funding from the University of Helsinki Research Funds (SKW, JL: University of Helsinki project no. 2105060), Finska läkaresällskapet, Finland (PHR) and Helsinki University Hospital Disctrict EVO research funds (PHR) is acknowledged. We are grateful to Jouko Laitila for bupivacaine assays and to Veikko Huusko and Olli Valtanen for assistance in the pig experiments. References [1] G.L. Weinberg, Reg. Anesth. Pain Med. 35 (2010) 188. [2] G. Weinberg, B. Lin, S. Zheng, G. Di Gregorio, D. Hiller, R. Ripper, L. Edelman, K. Kelly, D. Feinstein, Crit. Care Med. 38 (2010) 2268.
131
[3] C. Jamaty, B. Bailey, A. Larocque, E. Notebaert, K. Sanogo, J.M. Chauny, Clin. Toxicol. 48 (2010) 1. [4] G. Cave, M. Harvey, Acad. Emerg. Med. 16 (2009) 815. [5] J. Laine, J. Lokajova, J. Parshintsev, J.M. Holopainen, S.K. Wiedmer, Anal. Bioanal. Chem. 396 (2010) 2599. [6] D. French, C. Smollin, W. Ruan, A. Wong, K. Drasner, A.H.B. Wu, Clin. Toxicol. 49 (2011) 801. [7] J. Lokajova, J. Laine, E. Puukilainen, M. Ritala, J.M. Holopainen, S.K. Wiedmer, Electrophoresis 31 (2010) 1540. [8] A.B. Dhanikula, D. Lamontagne, J.C. Leroux, Cardiovasc. Res. 74 (2007) 480. [9] B. Howell, A. Chauhan, J. Colloid Interface Sci. 324 (2008) 61. [10] W. Ruan, D. French, A. Wong, K. Drasner, A.H.B. Wu, Anesthesiology 116 (2012) 334. [11] J. Muhonen, J.M. Holopainen, S.K. Wiedmer, J. Chromatogr. A 1216 (2009) 3392. [12] J. Lokajova, J. Pukkila, J.M. Holopainen, S.K. Wiedmer, Eur. J. Pharm. Sci. 41 (2010) 515. [13] S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya, T. Ando, Anal. Chem. 56 (1984) 111. [14] E. Boija, A. Lundquist, M. Nilsson, K. Edwards, R. Isaksson, G. Johansson, Electrophoresis 29 (2008) 3377. [15] K. Vainikka, K. Reijmar, G. Yohannes, J. Samuelsson, K. Edwards, M. Jussila, M.-L. Riekkola, Anal. Biochem. 414 (2011) 117. [16] D. Corradini, G. Mancini, C. Bello, Chroma 60 (2004) S125. [17] G. Manetto, M. Silvana Bellini, Z. Deyl, J. Chromatogr. A 990 (2003) 205. [18] E. Litonius, T. Niiya, P.J. Neuvonen, P.H. Rosenberg, Basic Clin. Pharmacol. Toxicol. 110 (2012) 378. [19] E. Litonius, T. Niiya, P.J. Neuvonen, P.H. Rosenberg, Anesth. Analg. 114 (2012) 901. [20] G. Hoizey, D. Lamiable, A. Robinet, H. Ludot, J.-M. Malinovsky, M.L. Kaltenbach, L. Binet, C. Boulanger, H. Millart, J. Pharm. Biomed. Anal. 39 (2005) 587. [21] B.A. Howell, A. Chauhan, Langmuir 25 (2009) 12056. [22] J. Szebeni, C.R. Alving, L. Rosivall, R. Bunger, L. Baranyi, P. Bedocs, M. Toth, Y. Barenholz, J. Liposome Res. 17 (2007) 107. [23] J. Szebeni, L. Baranyi, S. Savay, M. Bodo, D.S. Morse, M. Basta, G.L. Stahl, R. Bunger, C.R. Alving, Am. J. Physiol. Heart. Circ. Physiol. 279 (2000) 1319. [24] T.L. Samuels, J.W. Willers, D.R. Uncles, R. Monteiro, C. Halloran, H. Dai, Anaesthesia 67 (2011) 23.