Binding of Long-lasting Local Anesthetics to Lipid Emulsions

The solubility of LAs in two commercially available lipid emulsions was measured using the classic shake-flask method; i.e ., by mixing the solutions by continuous agitation. The validity of this technique was also confirmed by using a method of cobinding with activated charcoal.17 

The LAs used were rac-bupivacaine (Sigma, Saint Quentin-Fallavier, France), ropivacaine (Naropeine; AstraZeneca, Rueil-Malmaison, France), and levobupivacaine (Chirocaine; Abbott, Rungis, France). The lipid emulsions were a long-chain triglyceride emulsion (Intralipid 20%) and a medium- and long-chain triglyceride (50/50) emulsion (Medialipide 20%). Medialipide is also known as Lipofundin in Germany and as Vasolipid in Denmark, Norway, and Sweden. All experiments were conducted using a buffer solution containing (in mm) NaCl 120, NaH²PO⁴10, Na²HPO⁴10, and CaCl²1, pH 7.4. The lipid emulsions were diluted in buffer at the following final concentrations: 0.5, 1, and 2%. A preliminary experiment with Medialipide emulsion diluted to a 10% concentration was also performed with 64, 128, 256, and 512 mg/l rac-bupivacaine.

All experiments were done in triplicate at 20°C. To minimize problems related to emulsion stability with time and dilution,18the emulsion was diluted in buffer immediately before use. The LA was then added and the emulsion was mixed by shaking for a 20-min period. A measure of binding according to the duration of shaking, done with rac-bupivacaine and ropivacaine, showed that steady state was nearly attained after 1–3 min (fig. 1). The aqueous phase was then separated by 2 centrifugation steps (15,000 g, 10 min). LAs were added at the following concentrations (considering the hydrochloride salt for all LAs): 1, 2, 4, 8 16, and 32 mg/l in the 0.5, 1, and 2% Medialipide emulsions; at 8, 16, 32, and 64 mg/l in 1% Medialipide and 1% Intralipid emulsions; and at 2, 4, 8, 15.60, 31.25, 62.50, 125, 250, 500, and 1,000 mg/l in 1% and 2% Medialipide and 1% Intralipid emulsions. The accuracy of the shake-flask method was checked using cobinding with dextran-coated charcoal (Sigma, St. Quentin-Fallavier, France) at a final concentration of 2 mg/ml and a 5% lipid emulsion. In addition, rac-bupivacaine and ropivacaine binding were measured at 2, 4, 8, 15.60, 31.25, 62.50, 125, 250, 500, and 1,000 mg/l in 2% Medialipide emulsion at pH 7.00 and pH 7.40. The effect of temperature on rac-bupivacaine and ropivacaine binding was also tested on a separate set of data at 15.60, 31.25, 62.50, 125, 250, 500, and 1,000 mg/l in 1% Medialipide and 1% Intralipid at 20 and at 37°C, respectively. The data are reported as molar concentrations. LAs were assayed using gas chromatography.19 

The distribution coefficient was calculated as ρ, the ratio of the mole fraction in lipid and in buffer§and as D, the ratio of molar concentration. The relationship between free concentration (concentration in the aqueous phase) and bound concentration (concentration in the lipid phase) was also analyzed using nonlinear regression. This bound (B) versus  free (F) drug concentration was modeled as

, considering a one-site model. Bmax is the maximum binding capacity, Kd is the dissociation constant, and Lip is the concentration of lipid in the emulsion. The whole data set was simultaneously fitted using mixed-effect modeling with the NONMEM software (version 6, level 1; GloboMax, LLC, Hanover, MD).∥We added a random parameter to each of the fixed (structural) parameters (Bmax and Kd) to account for interbatch variability. These random parameters were modeled according to an exponential equation (we made the usual assumption that fixed parameters were log-normally distributed). A combined additive and constant coefficient residual (intraindividual) error model was used. The significance of random parameters and of Bmax and Kd for each LA and each emulsion was tested using the log-likelihood ratio test. Fitting was performed using the first order conditional estimation method with interaction. Because of the asymptotic nature of fitting, P < 0.01 was considered the minimum level of significance for this part of the study. The full (initial) model was based on the hypothesis that each LA and each emulsion had different binding properties. In the final model, Bmax was only dependent on the emulsion, and Kd was only dependent on the anesthetics. Kd was also similar for levobupivacaine and rac-bupivacaine. Random parameters associated with all of these fixed parameters were always significant. The data sets corresponding to the effect of pH and to the effect of temperature were separately fitted using a full model considering different Bmax and different Kd for each emulsion and each LA. The precision (95% CI) and accuracy (bias) of fitting were estimated using a nonparametric bootstrap with 1,000 replicates, and by a visual predictive check. The bootstrap was stratified by batch. In addition to NONMEM, this part of modeling was performed with the software R (version 2.5.1; The R Foundation for Statistical Computing, Vienna, Austria)#, and the normalized prediction distribution errors procedure.20Data are reported as the mean ± SD, or as the fitted value and 95% CI.

The preliminary experiment done with a 20% Medialipide emulsion diluted to 10% showed similar results as those obtained with more diluted emulsions, but the coefficient of variation was exceedingly high. The shake-flask technique was also compared to a technique of cobinding with activated charcoal. The two techniques yielded similar results. It was not possible to differentiate between rac-bupivacaine and levobupivacaine; i.e ., stereospecificity was not observed. Binding was rapid, since steady state was nearly attained after 3 min of mixing.

Binding of LAs was almost linear, from 2 to 125 mg/l in the 0.5, 1 and 2% emulsions; i.e.,  from 6.16 to 385 μm for bupivacaine and from 6.42 to 401 μm for ropivacaine. The apparent distribution coefficients calculated in different ranges of concentrations are reported in table 1. Decreasing the pH of the Medialipide/buffer emulsion from 7.40 to 7.00 led to an average 40–60% decrease in the apparent distribution coefficient (table 1). Increasing the temperature from 20 to 37°C led to an increase in the apparent distribution coefficient (table 1). At higher concentrations, saturation occurred progressively (fig. 2). A one-site binding model was used to fit the free versus  bound concentrations. Fitting was excellent; the visual predictive check did not show any departure of the residuals from normality (the Shapiro-Wilk test was > 0.2). As with the calculation of ρ and the ratio of molar concentration, it was not possible to differentiate between rac-bupivacaine and levobupivacaine. Bmax, the binding capacity, was different for the two emulsions, but similar for the three drugs in each emulsion group (table 2). Conversely, Kd, the dissociation constant, was different between drugs (bupivacaine vs . ropivacaine) but similar for the same drug between the Medialipide and the Intralipid emulsions. (See table, Supplemental Digital Content 1, which shows Bayesian estimates of the parameters, https://links.lww.com/A678.) Decreasing the pH of the 2% Medialipide emulsion from 7.40 to 7.00 led to a decrease in the affinity (1/Kd), but without any significant effect on the capacity: Kd was increased by a factor of 1.68 (95% CI 1.65–1.75) similar for bupivacaine and ropivacaine. Increasing the temperature from 20°C to 37°C led to an increase in affinity. Bmax was similar, whether the experiment was performed at 20°C or at 37°C. Conversely, Kd was significantly lower at 37°C than at 20°C, but the decrease was significantly higher with bupivacaine (rac-bupivacaine or levobupivacaine) than ropivacaine (table 2) (fig. 3).

Since binding was linear in the range 0.5–2%, and because a preliminary study done with the 10% Medialipide emulsion had given approximate distribution coefficients and parameters of binding similar to those measured with the low concentrated emulsions, we may speculate that the binding capacity is approximately 17 mm for the 20% Medialipide emulsion and 44 mm for the 20% Intralipid® emulsion.

The main finding of the study was that the two lipid emulsions were capable of binding each of the three local anesthetics tested in vitro . However, no stereospecific difference between the two enantiomers of bupivacaine was observed. Bupivacaine was more bound than ropivacaine and the difference was in close relation to the octanol/water distribution coefficient of each LA. The binding capacity of Intralipid was 2–3 times greater than that of Medialipide.

Lipid emulsions were diluted in an ionic buffer to accurately measure the concentration of LAs in the aqueous phase. Indeed, dilution in buffer is known to modify the physicochemical properties of the lipid emulsion, but these conditions were close to those encountered in the clinical situation, where dilution of the emulsion occurs rapidly in the blood stream. Both emulsions are made of chylomicron-like particles. The volume of distribution of chylomicrons has been shown to be slightly greater than the plasma volume (≈4,500 ml), with a half-life between 5 and 7 min.21The concentration of intact droplets in plasma during a rapid infusion of lipid emulsion is then expected not to exceed 1%, even if large amounts are infused. The stability of the chylomicron-like droplets depends on their surface charge. The zeta potential (the zeta potential, which stabilizes the emulsion, is the potential surrounding the droplet at its interface with the solvent) of both Intralipid and Medialipide is between -50 and -45 mV, and the use of ionic buffer to dilute the emulsion may have increased the potential to a point where flocculation begins.18However, this may also occur in vivo . Mixing time was limited to 20 min to reduce this phenomenon. A measure of binding according to the duration of shaking showed that steady state was attained after 1–3 min (fig. 1). This expected rapid binding process demonstrates that in clinical situations, an infusion of the emulsion may rapidly trap the circulating anesthetic molecules.

We calculated the distribution coefficient as the ratio of molar concentration in the two solvents, but also as the ratio of the mole fraction: 1 l of water contains ≈55.6 moles, whereas 1 l of the lipid fraction of Intralipid and Medialipide contains ≈4.19 and ≈4.61 moles of the mixture of phospholipids, lecithins, and glycerol, respectively; similarly, 1 l octanol contains ≈7.68 moles. Calculated in this way, the classic octanol/buffer distribution coefficient of ropivacaine (5 mm) and of bupivacaine (1 mm) at 25°C is then approximately 1,000 and 3,000 for ropivacaine and bupivacaine, respectively;22similar to the distribution coefficient of ropivacaine and of bupivacaine in Intralipid at 20°C (approximately 1,200 and 2,000, respectively) (table 1). The ratio of the mole fraction at saturation calculated from the one-site binding model was ≈3 for the Intralipid emulsion and ≈1 for the Medialipide emulsion. However, there is a large confidence interval of the estimated capacity, and these numbers need to be interpreted with care.

The drugs distribute more in Intralipid than in Medialipide. For both drugs, the distribution ratio is 2.4–3.0 greater in Intralipid than in Medialipide, suggesting that Intralipid might be preferable to Medialipide in a clinical situation. Also, bupivacaine distributes 2.0–2.6 times more than ropivacaine in Intralipid and in Medialipide. This is close to the distribution ratio of 3 of both drugs in octanol.22Then, the clearance of the drug exerted by infusion of a lipid emulsion is expected to be lower for ropivacaine than for bupivacaine.

At concentrations greater than 400 μm, saturation becomes significant and a one-site binding model adequately fitted the data. Interestingly, the binding capacity of the emulsion was similar for bupivacaine and ropivacaine, but about 2.5 times greater with Intralipid than with Medialipide. This ratio is similar to the distribution ratio calculated at lower concentrations. The affinity of the emulsion for the drug (1/Kd) was comparable for both emulsions, but about 2.6 times greater for bupivacaine than for ropivacaine. Like the distribution ratio calculated at lower concentrations, this ratio of affinity is not very different from the ratio of 3 (bupivacaine/ropivacaine) observed for the distribution in octanol.22In serum, bupivacaine binds to apha-1 acid glycoprotein and to albumin. The affinity of bupivacaine to apha-1 acid glycoprotein is very high (Kd = 1.62–3.69 μm), depending on experimental conditions and the enantiomer considered.22–24The affinity of ropivacaine to apha-1 acid glycoprotein is similar to that of bupivacaine (Kd = 2.78 μm).25However, because apha-1 acid glycoprotein is not very abundant in the human plasma (≈20 μm), the capacity is low (42–45 μm).23–25Conversely, serum albumin is abundant in plasma (≈640 μm), and its capacity to bind bupivacaine is large (898–1510 μm), but the affinity of bupivacaine for albumin is relatively low (Kd = 680–1200 μm).23,24The affinity of bupivacaine and ropivacaine to Intralipid and Medialipide is close to the affinity of bupivacaine to albumin (table 2). The capacity of lipid emulsions to bind bupivacaine or ropivacaine is important, since the capacity of the 1% diluted emulsion is 1.5 to 4 times the capacity of human serum23,24: 100 ml of a 20% Intralipid emulsion has the same capacity as the amount of albumin contained in 7 to 12 l of serum.

Increasing the temperature from 20 to 37°C led to an important increase in affinity and thermal energy (diffusion) seems to drive the uptake. The capacity of both emulsions remained unchanged (tables 1 and 2). In this case, the decrease in Kd (increase in affinity) was significantly greater for ropivacaine (55%) than for bupivacaine (27%). Both emulsions have a similar negatively charged double-layer interface (the zeta potential is between -45 and -50 mV at pH 7.40). Then, part of the adsorption process of the positively charged tertiary amines may also occur at the shear plane that surrounds the droplets. When the pH of the buffer was decreased from 7.40 to 7.00, the capacity remained also unchanged, but the affinity decreased by a factor of 1.68, similar for both anesthetics. Indeed, when the pH of the buffer decreases, the zeta potential increases (the droplets are less charged), but ionization of the tertiary amines bupivacaine and ropivacaine increases. In addition to the difference in phospholipid composition, the two emulsions differ also by the size of the droplets,**which may possibly explain the difference in capacity. However, the marked decrease in affinity when ionization increases, together with the temperature-dependence of binding and the close relationship between affinity and hydrophobicity of the molecules, tends to favor the hypothesis of a predominant hydrophobic uptake.

In conclusion, the solubility of long-acting LAs in lipid emulsions and the high capacity of binding presented by these emulsions most probably explain the clinical efficacy presented by a rapid infusion in case of LA toxicity. The long-chain triglyceride emulsion Intralipid appears to be about 2.5 times more efficacious than the 50/50 medium-chain/long-chain Medialipide emulsion. Also, because of their higher hydrophobicity, rac-bupivacaine and levobupivacaine seem to be more rapidly cleared than ropivacaine. Acidosis likely lowers the binding capacity of the emulsions, and this fact needs to be kept in mind. Intoxication with other drugs might also be effectively treated if their lipid solubility is in the same range of values, but further studies are needed to ascertain this potential.