Pharmacokinetics of Inhaled Liposome-encapsulated Fentanyl

Decades of research in liposome technology permit a reasonable prediction of the rate of drug release based on the composition of the liposome formulation. The rate of drug release is primarily dependent on the nature of the phospholipids, e.g., hydrogenated or unhydrogenated, or the phospholipid/cholesterol ratio (the higher this ratio, the faster the rate of release) and the hydrophilic/lipophilic properties of the active ingredients. Based on our previous work, [7]the following formulation was used for this study to provide a slow rate of fentanyl release: phospholipon 90-G-2%, cholesterol 0.2%, fentanyl 0.04%, saline solution sufficient quantity 100%.*

The encapsulation procedure of fentanyl was based on a well established and patented method* under Good Manufacturing Practice conditions. This method provides efficient drug entrapment (encapsulation efficiency of approximately 50%) and a sterile and stable liposomal product. All components of the products are of pharmacopoeia grade. These include phospholipon 90-G (used in intravenous hyperalimentation), cholesterol, United States Pharmacopeia, and sterile saline solution. During the studies, freshly prepared products were used, eliminating the need for preservatives, antioxidants, or other auxiliary agents in the formula.

The following procedure was used to characterize the FLEF products that contained 2,000 micro gram fentanyl in a volume of 5 ml: (1) Sephadex G-50 chromatographic column with ultracentrifugation were used to determine encapsulation efficiency. (2) The size of lipid vesicle was controlled to 100-1,000 nm by filtering the final product through a polycarbonate filter of 0.8-micro pore size. Optical microscopy and freeze-fracture electron microscopic techniques were used to determine the size distribution of the lipid vesicles. Taylor and McCalden demonstrated that an air-driven nebulizer did not alter the liposome-encapsulated preparations and found no increase in the amount of unencapsulated drug after aerosol generation. [8,9](3) Physical and chemical stability was monitored by the standard pharmaceutical techniques routinely used in our laboratory.*

After institutional approval, healthy volunteers with no history of cardiovascular, respiratory, hepatic, or renal dysfunction were recruited for a crossover study. Subjects with a history of analgesic abuse, opioid addiction, or fentanyl allergy were excluded. After obtaining informed consent, the volunteers fasted for 5 h before the study. The studies were conducted in the postanesthesia care unit with monitoring of blood pressure, heart rate, and hemoglobin oxygen saturation. A 16-G intravenous catheter was inserted under local anesthesia and was kept patent using diluted heparin solution to facilitate blood sampling. During phase I of the study, each volunteer received an injection of 200 micro gram fentanyl over 1 min through a 21-G butterfly needle in the contralateral forearm. Venous blood samples (3 ml each) were drawn at 2, 4, 6, 8, 10, 15, 20, 25, 30, 60, 90, and 120 min and at 4, 6, 8, 12, 18, and 24 h. The plasma was separated immediately after blood collection and stored at -20 degrees Celsius until analyzed.

Phase II of the study was conducted under similar conditions 4 weeks later. Each volunteer received 2,000 micro gram of free (50%) and liposome-encapsulated (50%) fentanyl in 5 ml via a nebulizer (Power Mist, Hospitak, Lindenburst, NY) with 6 l *symbol* min sup -1 flow of oxygen over 15 min. The volunteers were instructed to breathe normally with their mouths closed during the aerosol administration. Venous blood (3 ml) was drawn at 5, 10, 15, 20, 25, 30, 60, 90, and 120 min and at 4, 8, 12, 16, 20, 24, 32, 40, and 48 h. The plasma was separated immediately after the blood collection and stored at -20 degrees Celsius until analyzed. All plasma fentanyl concentrations (C^fens) were determined using a modified radioimmunoassay technique as described by Michiels et al. [10]The sensitivity of the fentanyl assay was 0.1 ng *symbol* ml sup -1 with an average coefficient of variation of 3.3% in a concentration range of 0.1-10.0 ng *symbol* ml sup -1.

The time to peak C^fen(T^max) and peak C^fen(C^max) were determined from the C^fenversus time profiles. For the intravenous study, C^fenversus time profiles were fitted to a three-compartment mamillary model using the extended least-squares nonlinear regression method (MKMODEL).** The pharmacokinetic parameters (V sub 1, V², V³, CL¹, CL², and CL³) were estimated, where V¹represents the initial volume of distribution and CL¹is the clearance (elimination) from this compartment. The remaining volumes of distribution (V²and V³) represent the rapid and slow redistribution, respectively. The CL²and CL³are the respective intercompartmental clearances.

The area under the C^fenversus time profile (AUC⁰-t) after the intravenous and inhalation administrations were calculated from the time of fentanyl administration to the last measurable plasma concentration by employing an approximate integration formula (linear trapezoidal method). [11]The area under the C^fenversus time curve from the last measurable C^fento infinity (AUC^t-infinity) was calculated by dividing the last measurable C^fenby the first-order rate constant of the terminal phase. The total area under the curve was estimated by the summation of these two components: {AUC⁰-infinity = AUC⁰-t + AUC^t-infinity}. The clearance^ivof fentanyl was calculated as DOSE^iv/AUC^iv. Assuming that the clearance of fentanyl by a subject is unchanged over a 4-week period, the amount of fentanyl absorbed from inhalation was calculated as the product of clearance^ivand AUC^FLEF. The bioavailability was calculated from the ratio of the dose-normalized areas under the curve: {AUC^FLEF/Dose^FLEF}/{AUC^iv/Dose^iv}. The absorption profile after inhalation of liposome-encapsulated fentanyl was determined using the constrained numeric deconvolution method of Verotta. [12]The constrained-numeric deconvolution method is a model-independent analytic deconvolution technique that uses the plasma concentrations from the intravenous portion of the study to extract absorption information from the plasma concentrations after the inhalation of FLEF. [13]The deconvolution was performed with the constraint that the absorption profile be a positive function. The potential advantage of the constrained numeric deconvolution method over the analytic deconvolution method of Loo-Riegelman [14]is that the characteristics of absorption may not be modeled easily by a simple absorption rate constant because of the heterogeneity of distribution of liposomes within the pulmonary tissue. For confirmation of the absorption profile obtained by deconvolution, bioavailability calculated from this deconvolution was compared to bioavailability from dose-normalized areas under the curve. The C^maxand C^fenat 8 and 24 h after aerosol and intravenous administration were compared using the analysis of variance with repeated measures and post hoc Bonferroni t test (P < 0.05).

Ten volunteers (five men and five women) were studied. Their mean (plus/minus SD) age, weight, and height were 31.5 (plus/minus 3.7) yr, 70.3 (plus/minus 10.4) kg, and 169.8 (plus/minus 7.0) cm, respectively. Apart from drowsiness, nausea (five subjects) and vomiting (one subject), none of the subjects had any complications during the study. Some of the volunteers commented on the "greasy smell" of the liposomal preparation. In general, the aerosol administration of the liposomal preparation was well tolerated, and there was no evidence of pulmonary irritation during the administration. The hemoglobin oxygen saturation of the subjects was maintained greater than 85% during the study while breathing room air. There were no significant hemodynamic changes during the study.

The C^fenversus time profiles after the intravenous and aerosol administrations are shown in Figure 1and Figure 2, respectively. After intravenous administration of fentanyl, the peak plasma concentration (C^max) was substantially greater than that after aerosol administration (Figure 3).

The mean (plus/minus SD) of C^max, T^max, and C^fenat 8 and 24 h after the aerosol and intravenous administration are summarized in Table 1. After the inhalation of FLEF, C^maxwas reached gradually in 22 min compared to 3.6 min for intravenous fentanyl. However, the mean C^maxfor intravenous was four times greater than that for aerosols (4.67 versus 1.15 ng *symbol* ml sup -1). The mean (plus/minus SD) C^fenat 8 h after FLEF (0.25 plus/minus 0.14 ng *symbol* ml sup -1) was sustained close to the therapeutic postoperative analgesic concentrations (0.2 to 1.2 ng *symbol* ml sup -1) [15]compared to intravenous fentanyl (0.16 plus/minus 0.10 ng *symbol* ml sup -1).

The estimates of model-dependent pharmacokinetic parameters of intravenous fentanyl are shown in Table 2. There was a large intersubject variability. The mean (plus/minus SD) volume of distribution at steady-state and clearance were 435.1 plus/minus 182.11 and 0.584 plus/minus 0.210 l *symbol* min sup -1.

The rate of absorption of fentanyl versus time after the inhalation of FLEF is shown in Figure 4. There was a substantial variability in the rate of absorption of fentanyl after FLEF aerosol. The peak absorption of fentanyl occurred within the first 30 min after the start of the aerosol administration, and the fentanyl absorption appeared to be complete by 60-90 min in most cases. The mean (plus/minus SD) peak absorption rate, and time to peak absorption after FLEF aerosol are summarized in Table 3. The bioavailabilities of FLEF determined by the dose-normalized areas under the curve and numeric deconvolution techniques also are included in Table 3. These methods produced comparable results with an overall bioavailability of 12%.

Although the analgesic efficacy of fentanyl is well established, the potential ventilatory depressant remains a concern in patients receiving fentanyl. This study has demonstrated several advantages of inhaled liposome-encapsulated fentanyl over other routes of administration of opioid in providing pain relief. Administration by inhalation is simple and noninvasive. It avoids the oral route for patients with gastrointestinal dysfunction, such as nausea and vomiting. It also avoids the first-pass metabolism by the liver after intestinal absorption. C^maxs can be achieved rapidly. In addition, plasma concentrations can be sustained higher for a longer duration compared to intravenous administration (Figure 3). These higher C^fens would have greater potential to provide postoperative analgesia. Although transmucosal fentanyl (lollipop) can provide a similar rapid increase of C^fen, it does not provide a sustained plasma concentration after its administration compared to inhaled liposome-encapsulated fentanyl. [13].

After the inhalation of FLEF, absorption, distribution, and elimination of fentanyl take place simultaneously. It would be difficult to distinguish these processes by examining the plasma C^fenversus time profile alone. Because the fentanyl distribution and elimination characteristics were determined for each subject in a separate intravenous fentanyl study, it was possible to use a deconvolution technique to extract the absorption profile for inhaled FLEF from the plasma C^fenversus time data. Numeric deconvolution is predicted on the assumption that the individuals' disposition pharmacokinetics are the same for both the intravenous and aerosol studies. Attempts were made to minimize the interval between the intravenous and FLEF studies to 4 weeks and to minimize diurnal variation by performing the experiments at the same time of the day for each of the subjects.

The overall bioavailability of fentanyl after FLEF inhalation was approximately 12%. This is in agreement with the bioavailability of drugs administered through the pulmonary system, which ranges 10-20%. [16]The bioavailability or the amount of drug delivered to the lungs can be improved with the use of a large initial volume of solution placed in the nebulizer, a higher compressed gas flow rate (12 l *symbol* min sup -1) to produce a higher percentage of small droplets (1-5 micro), deep inhalation with breath holding, and the use of positive expiratory pressure during the aerosol therapy (Resistex, DC Lung, Sebastopol, CA). [17]Although the 2,000 micro gram-fentanyl dose used during the study was large, only 10-20% of the dose would be absorbed.

Decades of research in liposome technology have indicated that liposome-encapsulation is an effective and safe drug delivery system. Liposomes have been used clinically as intravenous drug carrier systems in enzyme replacement therapy, [18]antifungal therapy, [19]and chemotherapy. [20]Liposomes also have been investigated as sustained-release carriers for inhalational medications. [7,21]Taylor et al. demonstrated that prolonged systemic absorption was obtained with inhalation of liposome-encapsulated sodium cromoglycate in human volunteers. [22]Successful clinical trials have been conducted on the administration of liposome-encapsulated orciprenaline and salbutamol into the respiratory system to provide a sustained bronchodilation effect in patients with chronic obstructive lung disease. [23]Liposomal drug delivery to the lungs appears to be well tolerated in both animals and humans. In animal models with rabbits and mice, acute and chronic inhalation of liposome aerosol did not adversely affect lung appearance, cell consistency, or pulmonary histopathology. [24,25]No adverse effects have been associated with inhalation of exogenous phospholipids to preterm infants with respiratory distress syndrome. [26]Thomas et al. [27]reported no hemoglobin oxygen desaturation, deterioration in pulmonary function, or side effects associated with the inhalation of liposomes in healthy human volunteers. In summary, an extensive series of studies in both animals and humans has failed to document any toxicity to the respiratory tract from administration of liposomes.

C^fens were determined from venous blood sampling in this study. Ideally, blood sampling would have been arterial rather than venous, because arterial fentanyl concentrations more accurately reflect fentanyl concentrations at the site of drug effect in the central nervous system, i.e., the biophase concentrations. During the initial distribution phase of fentanyl, peripheral venous fentanyl concentrations correlate poorly with the C^fenat the biophase concentrations. During the terminal elimination phase of fentanyl, however, the fentanyl concentrations from the peripheral venous blood will approximate the biophase concentrations. Because the focus of this study was to determine fentanyl concentrations occurring hours after the administration of FLEF, venous blood sampling was adequate. The use of venous sampling may explain the delay in the C^maxafter the bolus intravenous administration (Table 1). Slow venous mixing secondary to initial uptake by the peripheral tissues (lungs, brain, fats) and subsequent release of fentanyl from these tissues may cause a delay in the venous C^fento reach its peak. Because the sites of blood sampling for the intravenous and aerosol administrations were similar, the delay also occurred after the aerosol. Therefore, the effect of this delay on the deconvolution analysis is insignificant.

Only 12% of the administered fentanyl dose was absorbed into the circulation after the inhalation of FLEF. It is difficult to determine the amount of fentanyl absorbed from the pulmonary (inhaled) and the gastrointestinal systems (oral mucosa or swallowed) from this study. The remaining 88% of the dose presumably is metabolized by the lungs, lost into the gastrointestinal tract or the environment. Contamination of the environment with the opioid and the risk to health-care personnel administering the aerosol is a real concern. During three of the aerosol studies, blood samples were collected every 15 min for 2 h from three volunteers who sat at 1 m away from the nebulizer of the studied subjects. No fentanyl was detected in the plasma of any of these samples. These results suggest that the amount of opioid absorbed by health-care personnel is clinically insignificant.

As with any delivery system, there is a potential for abuse with this method of drug delivery. Recently, a patient was reported to have a respiratory arrest after an injection of the aspirate from a transdermal fentanyl patch through his central venous catheter. [28]Needless to say, addicted patients will be at risk for drug abuse with the inhaled liposome-encapsulated fentanyl delivery system. However, a high index of suspicion, proper education, and strict narcotic control can effectively minimize the risk of potential drug abuse and addiction.

Although this drug delivery system has potential clinical applications, many issues must be addressed. The sterility, encapsulation efficiency, and stability of the liposomal preparation must be resolved. The pharmacokinetics of multiple aerosol administrations and the fate of the fentanyl deposited elsewhere outside the alveoli, such as the oral cavity and the tracheobronchial tree, must be determined. Most importantly, clinical studies relating to its administration to achieve a therapeutic effect and its potential complications must be evaluated. Future studies are necessary to address these important questions.

In summary, the inhaled liposome-encapsulated opioid delivery system has many advantages. It is noninvasive, simple to administer, and inexpensive to produce. In addition, it does not produce an initial high C^fenand has the ability to maintain a prolonged therapeutic concentration of fentanyl. Although this was our initial attempt at the liposome formulation and ratio of free fentanyl to encapsulated fentanyl, the preliminary results of this new method of opioid delivery are promising. Present and future studies are being conducted to determine an optimal liposome formulation as well as the fentanyl dose that can provide predictable and sustained therapeutic C^fens over 24-48 h.

* Mezei M, Nugent FI: Method of encapsulating biologically active materials in multilamellar lipid vehicles (MLV). U.S. Patent no. 4,485,054, 1984; Canadian Patent no. 1205383, 1986.

** Holford N: MKMODEL. Cambridge, Elsevier-Biosoft, 1986.