Serum pharmacology of amphotericin B applied in lipid emulsions

Application of amphotericin B in lipid emulsions (AmB/L) reduced membrane toxicity in vitro and decreased amphotericin B-associated toxic side effects in vivo when compared to that of amphotericin B applied in 5% glucose (AmB/G). Therefore, a comparative analysis of the pharmacological parameters of AmB/L and AmB/G was performed. Thirteen patients were analyzed, and nine of these patients received a subsequent treatment with AmB/G and AmB/L. In patients in both treatment groups amphotericin B showed a biphasic elimination from serum, with a prolonged terminal half-life of approximately 27 h. Patients treated with AmB/L showed significantly lower peak concentrations (44.2%; P = 0.008) and correspondingly lower area under the drug concentration-time curve (AUC) values (64.3%; P = 0.015) compared to the values for the same patients treated with AmB/G at a dose range of 0.6 to 1.5 mg/kg of body weight. The enhanced clearance of AmB/L may be due to a faster initial elimination of amphotericin B-lipid aggregates by the reticuloendothelial system. Lower peak concentrations and AUC values in serum and a correspondingly faster deposition of AmB/L in tissues may at least partly explain the lower toxicity of AmB/L. A comparative pharmacokinetic analysis with data for a single patient treated with AmB/L demonstrated that hemodialysis did not significantly affect the disposition of amphotericin B.

Amphotericin B is a drug with major importance in the treatment of invasive fungal infections. Determinants of an efficient antifungal treatment are early start of therapy and application of high initial doses (1 to 1.5 mg/kg of body weight per day) of amphotericin B (2,13). Since the bioavailability of amphotericin B in peripheral tissues is rather low, fungicidal concentrations of the drug in tissue may only be provided by early application of high drug doses. This treatment approach is, however, limited by the severe toxicity of amphotericin B. Acute side effects such as fever and chills were observed in up to 60% of patients, while chronic toxicity such as impairment of renal function occurred in up to half of the patients. In a search for an improvement of clinical drug tolerability, several investigators described the lower toxicity of amphotericin B when applied in lipid emulsions. In fact, the maximal tolerated dose of amphotericin B in mice could be raised by more than ninefold when the drug was applied in Intralipid (14). This was supported by a number of recent clinical studies reporting that amphotericin B infused in Intralipid induced a significant reduction of nephrotoxicity and acute side effects compared to amphotericin B applied in dextrose (4)(5)(6)16).
The exact mechanism of the Intralipid-mediated reduction of amphotericin B-associated toxicity is still unclear. However, it has been hypothesized that lipid emulsions like Intralipid decrease the amount of oligomeric amphotericin B and thereby reduce the interaction of amphotericin B with cholesterolcontaining cell membranes. The remaining monomeric amphotericin B, however, retains its potential to bind to the ergosterol of fungal cell membranes (1,8,10,12).
Binding of amphotericin B to phospholipids may also mod-ulate the pharmacology of the drug and may thereby affect the tolerability of amphotericin B. The present study was designed to analyze the impact of lipid emulsions like Intralipid on the pharmacokinetics of amphotericin B.
Patients. The pharmacology of amphotericin B was evaluated in 13 patients (3 females and 10 males). The median age of the patients was 55 years (range, 38 to 76 years). Patients were nonrandomly assigned to treatment with amphotericin B applied in 5% glucose (AmB/G) (patient 10) or amphotericin B in lipid emulsion (AmB/L) (patients 8, 9, and 13) or subsequently with both formulations (patients 1 to 7, 11, and 12). Amphotericin B therapy was initiated for treatment of fever of unknown origin (n ϭ 4) or for presumptive (n ϭ 6) or proven (n ϭ 3) fungal infections (see Table 1). Presumptive fungal infection was defined as antibiotic-resistant fever with microbiological or serological indications for fungal infection and/or typical ultrasound or roentgenographic lesions. A proven fungal infection was diagnosed when histological and/or cultural proof of the fungal infection was obtained from biopsy material of otherwise sterile tissue or from bronchoalveolar lavage specimens.
Treatment regimens. The standard preparation of amphotericin B was established by dissolving the drug in 250 ml of 5% glucose (AmB/G). For application of amphotericin B in the lipid emulsion, 20% Intralipid was added to achieve a final proportion 2 mg of amphotericin B/1 ml of Intralipid (AmB/L). This ratio was chosen in agreement with previous reports stating adequate clinical tolerability and that the formulation is effective (4)(5)(6). Both AmB/G and AmB/L formulations were strictly applied as 1-h infusions.
Hemodialysis. Hemodialysis was performed by using a standard hemodialysis device (MTS 2008; Fa. Fresenius) equipped with a GFS 12 plus dialyzer (Hemophan membrane; Fa. Gambro, Hechingen, Germany) at a blood flow rate of 240 ml/min and a dialysate flow rate of 500 ml/min. The dialysate (SKF 411 plus bicarbonate) contained potassium at a concentration of 4 mmol/liter.
Methods. (i) Sample preparation and extraction. Serum was obtained before drug infusion, at the end of infusion, and for at least five further time points within an interval of 24 h after infusion (2,4,6,20, and 24 h). Having demonstrated that arterial and venous blood samples showed identical concentrations of amphotericin B, blood was generally taken from an arterial access. Serum was immediately separated by centrifugation, and samples were stored at Ϫ20ЊC until analysis. Separate analyses had demonstrated the stability of amphotericin B in serum samples for at least 3 months when they were kept at Ϫ20ЊC. Serum was extracted by the addition of 3 volumes of methanol-EDTA (5 mM) to 1 volume of serum. After centrifugation, the clear supernatant containing the total amount of amphotericin B was separated.
(ii) HPLC analysis. High-pressure liquid chromatographic (HPLC) analysis was performed with a Waters model 490E detector, two Waters model 510 pumps, an automatic sample injector (Waters 717), and Maxima, version 3.10, software, which is an International Business Machines computer-based program designed for pump control and chromatogram evaluation. Peak separation was achieved with a -Bondapak C 18 column (0.39 by 30 cm) and a mobile phase consisting of methanol-5 mM disodium EDTA, which was used at a flow rate of 1 ml/min. Under these conditions, amphotericin B demonstrated a retention time of 14 min and was detected at a wavelength of 405 nm. Quantification of amphotericin B was performed by external standard calculation. The HPLC assay of amphotericin B in serum showed an adequate reproducibility, with a standard deviation of 3.7%. The lower limit of detection was reached at an amphotericin B concentration of 0.05 g/ml. All datum points were based on at least two independent determinations.
Pharmacokinetic evaluation of the data was performed by computer-based analysis with the program TopFIT, version 2.0 (9). Areas under the concentration-time curves (AUCs) were calculated by use of the trapezoidal rule. A two-compartment model of amphotericin B elimination was chosen to provide the best fit of the data.
(iii) Erythrocyte lysis assay. The toxicity of amphotericin B to human cell membranes was analyzed by using an adaptation of the erythrocyte lysis assay described by Butler and Cotlove (3). In short, a 3% suspension of washed human erythrocytes was incubated for 2 h at 4ЊC with the desired amphotericin B concentrations in a final assay volume of 1,000 l. Drug exposure was terminated by centrifugation and pelleting of the erythrocytes. The supernatant was subsequently analyzed for its content of lactate dehydrogenase (LDH) by using an enzyme-based photometric assay (BM/Hitachi 747; Boehringer Mannheim GmbH). The release of LDH from erythrocytes was related to the total LDH content of erythrocytes in the assay, which was determined by a 2-h exposure of erythrocytes to distilled water.
Statistical analysis. Matched pairs analyses were performed by use of the Wilcoxon test. For correlation analyses the Spearman coefficient (R) was determined.

RESULTS
Pharmacokinetics of AmB/G. The pharmacokinetics of amphotericin B were monitored in the 13 patients identified in Table 1. Ten patients (Table 2) received the standard preparation of AmB/G. Amphotericin B was applied at three dose levels: 0.6 mg/kg (n ϭ 2), 1.0 mg/kg (n ϭ 6), and 1.5 mg/kg (n ϭ 2). The respective pharmacokinetic parameters are presented in Table 3 as mean values for each dose. At an amphotericin B dose of 1.0 mg/kg, the mean maximum concentration of drug in serum (C max ) and AUC values amounted to 1.76 g/ml and 18 g ⅐ h/ml, respectively. Between different patients, there was a correlation neither between the dose of amphotericin B and the respective C max values (R ϭ 0.55) nor between the dose and the respective AUC values (R ϭ 0.06).
Pharmacokinetics of AmB/L. Twelve of the patients identified in Table 1 received AmB/L. Four doses were administered ( Table 2): 0.6 mg/kg (n ϭ 2), 1.0 mg/kg (n ϭ 7), 1.5 mg/kg (n ϭ 2), and 2.0 mg/kg (n ϭ 2). Patient 8 received two doses. The respective pharmacokinetic parameters for each dose are presented in Table 3. At the amphotericin B dose of 1.0 mg/kg, mean C max and AUC values amounted to 1.15 g/ml and 12.7 g ⅐ h/ml, respectively. For patients treated with AmB/L, significant correlations could be observed between the dose of amphotericin B on the one hand and C max (R ϭ 0.90; P ϭ 0.04) and AUC values (R ϭ 0.90; P ϭ 0.04) on the other hand.
Comparative analysis between the pharmacokinetics of AmB/G and AmB/L. A comparison between the standard application of AmB/G and treatment with AmB/L was performed with nine patients (patients 1 to 7 and 11 and 12) who were first treated with AmB/G and who subsequently received the identical dose of AmB/L ( Table 4).
The pharmacokinetic patterns were comparable in both groups. Amphotericin B showed biphasic elimination characteristics, with a short initial and a prolonged terminal half-life. A matched-pairs analysis (Wilcoxon test) revealed a significantly lower C max (1.30 versus 2.94 g/ml; P ϭ 0.008) for the AmB/L group compared to that for the AmB/G group. Accordingly, AUC values were also significantly lower in the AmB/L group (16.0 versus 24.9 g ⅐ h/ml; P ϭ 0.015). In fact, the mean C max decreased to 44.2% and the mean AUC was reduced to 64.3% when AmB/L was used. Figure 1 demonstrates the comparative pharmacokinetics obtained for patient 1. The volume of distribution was greater in the AmB/L group (1.88 versus 1.34 liters/kg); this difference, however, was statistically not significant. Interestingly, the initial half-life of distribution was significantly longer (P ϭ 0.03) in the AmB/L group than in the AmB/G group (0.52 versus 1.0 h). No significant differences between groups were observed with regard to the clearance (1.49 versus 0.88 ml/min) and the elimination half-life (t 1/2␤ ) of amphotericin B (27.1 versus 27.5 h).
Effect of hemodialysis on the pharmacokinetics of AmB/L. The effect of hemodialysis on the pharmacokinetics of AmB/L was investigated in patient 9. This patient was monitored on 2 subsequent days. On the first day, amphotericin B (1.0 mg/kg) was applied but hemodialysis was not performed. On day 2, a hemodialysis treatment was started, and the identical dose of amphotericin B was applied as a 1-h infusion during hemodialysis. Hemodialysis was carried out for 2.5 h without a negative fluid balance. When the data obtained before and during hemodialysis were compared, it became apparent that hemodialysis did not affect the C max of amphotericin B (0.8 versus 0.86 g/ml). The t 1/2␤ was, however, shorter during dialysis (39.3 versus 16.7 h), and accordingly, the clearance of amphotericin B was greater (3.75 versus 4.08 ml/min). As a result, the AUC was slightly lower during dialysis (4.45 versus 4.08 g ⅐ h/ ml). The reduced t 1/2␤ cannot be ascribed to hemodialysis since, during the terminal elimination phase, hemodialysis had already been terminated.
Effect of Intralipid on amphotericin B-mediated membrane toxicity. The membrane toxicity of amphotericin B was evaluated by determination of LDH release from human erythrocytes. Amphotericin B was dissolved either in 5% glucose or in a lipid emulsion like Intralipid 20% or Lipofundin 20%. Both lipid emulsions contained identical amounts of phospholipids (12 g/liter). However, the emulsions were different in that 20% Lipofundin contained 100 g of middle-chain triglycerides per liter and 100 g of soybean oil per liter, while 20% Intralipid contained 200 g of soybean oil per liter, but no middle-chain triglycerides.
It was demonstrated (Fig. 2) that increasing amphotericin B concentrations induced a sigmoidal increase in LDH release which reached 100%, i.e., complete lysis of erythrocytes, at an amphotericin B concentration of 40 g/ml. The release of LDH from erythrocytes was completely prevented by 20% Intralipid at an amphotericin B/Intralipid ratio of 2 mg/1 ml. The identical result was obtained for Lipofundin.
In a subsequent analysis, the impact of the ratio of amphotericin B/Intralipid on membrane toxicity was examined (Fig.  3). This assay was performed at final amphotericin B concentrations of 100 and 200 g/ml. A decreased ratio of amphotericin B/Intralipid decreased the amount of LDH released, and the combination was more effective at the lower amphotericin B concentration of 100 g/ml. For both amphotericin B con-  a Nine patients first received the standard preparation of AmB/G; after 24 h this was followed by administration of the identical dose of AmB/L. The pharmacokinetic parameters for both treatment modalities were compared by use of the Wilcoxon test. V, volume of distribution; NS, not significant. a Amphotericin B pharmacokinetics were evaluated at three different doses in patients receiving AmB/G, while patients treated with AmB/L were additionally evaluated after the administration of a dose of 2.0 mg/kg. Data are presented as means of the respective determinations indicated in Table 2. V, volume of distribution; the other abbreviations are defined in the text.
b Number of patients receiving each formulation.
centrations, the release of LDH was completely prevented at an amphotericin B/Intralipid ratio of 1 mg/0.3 ml.

DISCUSSION
Application of amphotericin B in lipid emulsions like Intralipid results in a significant reduction of treatment-associated toxicity, while antifungal activity remained unaffected in animal models (12,14). In an erythrocyte lysis assay, it was demonstrated that Intralipid and Lipofundin effectively prevented amphotericin B-mediated membrane toxicity (Fig. 2). An equally protective effect of Intralipid and Lipofundin was expected since both preparations contain the same amount of phospholipids. The optimal ratio of amphotericin B/Intralipid was established to be Ն1 mg/0.3 ml (Fig. 3) which compares favorably to the clinically recommended ratio of 1 mg/0.5 ml (6).
The following question arises, however: may the reduced clinical toxicity of AmB/L also be attributed to a modulation of the pharmacokinetics of amphotericin B or is it only caused by an alteration of the aggregation status of amphotericin B and consequently results from a modified drug interaction with cell membranes? The present analysis characterizes the pharmacokinetics of AmB/L and compares it to the standard preparation by which AmB/G is applied.
The pharmacology of AmB/G was evaluated at three different doses of amphotericin B, namely, 0.6, 1.0, and 1.5 mg/kg (Table 2), while AmB/L-treated patients were additionally evaluated with a dose of 2.0 mg/kg. The increased dose of 2.0 mg/kg was chosen since the ongoing analysis demonstrated a decrease in peak amphotericin B concentrations in the AmB/L group. For patients treated with AmB/L, a significant correlation between the indicated amphotericin B doses and the respective C max (P ϭ 0.04) and AUC values (P ϭ 0.04) was demonstrated, an observation not obtained after AmB/G treatment.
A comparative analysis of the pharmacokinetics of AmB/G and AmB/L was performed with nine patients who were treated with the respective regimens on subsequent days (Fig.  1). This evaluation demonstrated that mean C max and AUC values were significantly reduced in the AmB/L group, reaching 44.2% (P ϭ 0.008) and 64.3% (P ϭ 0.015) of the values observed in the AmB/G group, respectively (Table 4). Since the interval of amphotericin B application was only 24 h, the possibility of a period effect inducing a bias into the pharmacokinetic parameters cannot be excluded. However, there was no indication for a cumulative rise in amphotericin B concentrations in the sera of patients treated with AmB/G alone on subsequent days (data not shown). Due to the severity of the disease, a prolongation of drug application intervals was generally not regarded as being feasible.
From the data presented above, it was concluded that the reduced toxicity of AmB/L may, at least partly, be accounted for by a reduction of the peak amphotericin B concentration and AUC values.
The question of which mechanism is responsible for the significant decrease in C max and AUC values then arises. Given the assumption of a reduced interaction of AmB/L with mammalian cell membrane cholesterol (8,14,15), a reduced uptake by peripheral tissues and consequently a prolonged elimination from the circulating blood compartment would be expected. In fact, this pharmacological characteristic is observed with liposomal formulations of amphotericin B; however, it is not observed with AmB/L. Janoff et al. (10) reported that amphotericin B formed nonliposomal, ribbon-like structures when interacting with lipids in an aqueous environment. Comparable structures were also observed with a new amphotericin B-lipid complex (ABLC) (11). Interestingly, ABLC is characterized by significantly lower C max and AUC values compared to those obtained with amphotericin B deoxycholate. This relates to a rapid elimination of ABLC by organs of the reticuloendothelial system and a correspondingly greater accumulation of ABLC in liver, lungs, and spleen (15). By analogy, it may be suggested that the comparatively low concentrations of AmB/L in serum are, in fact, due to a rapid initial capture of amphotericin B-lipid aggregates by the reticuloendothelial system. In conclusion, AmB/L may be defined as a formulation characterized by low C max and AUC values in serum and a correspondingly faster deposition in tissue. Effect of Intralipid and Lipofundin on amphotericin B-mediated membrane toxicity. Amphotericin B-mediated membrane toxicity was evaluated by analysis of LDH release from human erythrocytes as described in Materials and Methods. Erythrocytes were exposed for 2 h to the indicated concentrations of amphotericin B. The drug was dissolved in 5% glucose (closed triangles), 20% Intralipid (open squares), or Lipofundin (closed squares). For the lipid emulsion, the ratio of 2 mg of amphotericin B/1 ml of 20% Intralipid was kept constant.
The differential importance of the concentrations of amphotericin B in serum compared to the concentrations of amphotericin B in tissue remains controversial. However, the Intralipid-mediated decrease in C max and AUC values combined with the lower toxicity of AmB/L may allow for the use of increased doses of the drug. The application of greater amphotericin B doses is guided by the concept that fast achievement of effective tissue drug concentrations is an important determinant of treatment outcome in invasive fungal infection.
With regard to the clinical application of AmB/L, it was of interest to analyze the effect of hemodialysis on the pharmacokinetics of this formulation. Previous reports demonstrated that the pharmacokinetics of amphotericin B deoxycholate in serum were essentially independent of renal function and were not affected by hemodialysis (7). The comparative pharmacokinetic analysis of a single patient before and during hemodialysis indicates that C max and AUC values remained essentially unaffected by hemodialysis. Adjustment of the AmB/L dose during hemodialysis therefore does not appear to be necessary.