Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Proleukin (aldesleukin) recombinant human interleukin-2 (IL-2) has antitumor activity in patients, with significant responses occurring most frequently in renal cell carcinoma and malignant melanoma (West et al., 1987; Rosenberg et al., 1989). The IL-2 receptor-positive T-lymphocytes are thought to be primarily, but not exclusively, associated with efficacy and reside largely in the lymphoid organs (Stites et al., 1994). On the other hand, after IL-2 exposure, natural killer cells and neutrophils in plasma produce cytokines, reactive oxygen intermediates, and proteases, all of which have been shown to be necessary, but not sufficient, to produce the full spectrum of IL-2 toxicities (Caligiuri, 1993). Therefore, adverse in vivo activity of IL-2 may be related to the plasma concentrations, but beneficial activity may be related to lymph concentrations.

Absorption of macromolecules like IL-2 can be targeted to the lymphatic system by s.c. rather than i.v. administration, because the plasma concentrations of macromolecules are dependent on capillary and lymphatic absorption processes after s.c. dosing (Bocci et al., 1986; Supersaxo et al., 1988, 1990). Macromolecules like IL-2 diffuse through the interstitium and enter both blood and lymph capillaries (Guyton, 1981a). Proteins circulate within the lymph and are gradually returned to the blood (Guyton, 1981b). Because the primary and secondary lymphoid organs may be the site of action of IL-2, the intensity of a pharmacological response may depend on both the systemic exposure (as measured by the plasma concentrations) and the route of administration. By simultaneously measuring the plasma and lymph concentrations, the rate of absorption directly from the injection site into the blood and lymph and the transfer rate from the lymph to the blood can be measured.

As a small protein (<50,000), IL-2 is rapidly cleared from the body by glomerular filtration, peritubular extraction (Gibbons et al., 1995), and, in humans, an inducible receptor-mediated mechanism (Piscitelli et al., 1996). Therefore, frequent dosing is required for efficacy. To decrease IL-2 clearance, its molecular weight was increased by the addition of monomethoxy polyethylene glycol (PEG) molecules to form PEG-IL-2 (Katre et al., 1987; Knauf et al., 1988). PEG-IL-2 has an apparent molecular weight of 95,000 to 250,000, whereas IL-2 has a molecular weight of 15,000.

There is a direct relationship between molecular weight up to 19,000 and the proportion of a dose transported lymphatically after the s.c. administration of a neutral, water-soluble compound (dextran; Supersaxo, 1990), and a diverse set of proteins between molecular weights 7,500 and 75,000 (Xie and Hale, 1996). Xie and Hale (1996) also found that positively charged proteins had decreased lymphatic absorption relative to negatively charged proteins with similar molecular weight. The addition of PEG to IL-2 increased the molecular weight and resulted in an overall decrease in positive charge of IL-2 after PEG attachment to lysines (Knauf et al., 1988). It was expected that PEG-IL-2 would be preferentially absorbed via the lymphatics compared with IL-2. However, the most profound consequence of changing IL-2 to PEG-IL-2 was the increased water solubility (Katre et al., 1987). Therefore, the extent of absorption of PEG-IL-2 compared with IL-2 into the lymphatic system is not predictable from previously conducted studies.

The purpose of this study was 2-fold: first, to characterize the pharmacokinetics of IL-2 and PEG-IL-2 in Yorkshire pigs after i.v. and s.c. administration using relevant clinical regimens and doses; second, to determine the distributional advantage of IL-2 and PEG-IL-2 into lymph after i.v. and s.c. administration. The results of this study will help determine whether extravascular routes of administration preferentially favor lymphatic exposure of IL-2 and PEG-IL-2. If there is a difference in biodistribution between the i.v. and s.c. administration of IL-2, the toxicity of IL-2 may be reduced by modifying the route and schedule of administration (Whittingdon and Faulds, 1993; Anderson and Sorenson, 1994). Furthermore, equivalent dosing regimens of IL-2 and PEG-IL-2 may be designed that minimize toxicity based on the plasma pharmacokinetics and result in similar in vivo activity based on the lymphatic pharmacokinetics.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials

Escherichia coli-derived IL-2 (Kato et al., 1985), a 132-amino acid, nonglycosylated protein, was supplied in vials as a sterile lyophilized powder. After reconstitution with sterile water, the solution contained 1.1 mg/ml (18 × 10⁶ I.U., MIU/ml) IL-2 protein. The specific activity of IL-2 has been assigned based on the in vitro activity of native IL-2 in a cell proliferation bioassay and was 16 MIU/mg.

A chemical modification of IL-2 (PEG-IL-2) was prepared by the addition of three or four polyethylene glycol molecules (molecular mass, approximately 7000 Da) as previously described (Katre et al., 1987; Knauf et al., 1988). PEG-IL-2 was also supplied in vials as a sterile lyophilized powder. After reconstitution with sterile water, the solution contained 0.45 mg/ml (1.8 MIU/ml) PEG-IL-2 protein. PEG-IL-2 recognizes and interacts with both the high- and intermediate-affinity IL-2 receptor complexes; however, the biological specific activity of PEG-IL-2 show 4- to 6-fold less activity compared with IL-2. The specific activity of PEG-IL-2 used in these studies was 4 MIU/mg. To compensate for the reduced activity, both materials were dosed based on activity.

Experimental Design

The study was conducted in two parts using Yorkshire pigs. Part I was a crossover study conducted at the University of California at Davis (Davis, CA) that evaluated the plasma pharmacokinetics of IL-2 after the i.v. and s.c. bolus administration of 0.16 and 1.6 MIU/kg (Table 1). Three animals received all four treatments sequentially, and an additional animal received two s.c. doses of 0.16 MIU/kg. There was a 1- to 3-day washout period between dosing. The IL-2 was administered undiluted, and the injections were made as a bolus into a femoral vein catheter or under the skin of an area of the thigh tented by the restraining thumb and forefinger for i.v. or s.c. administration, respectively.

Part II was a single-dose study conducted at Primedica Corporation (Worcester, MA), which evaluated both the plasma and lymph pharmacokinetics of IL-2 and PEG-IL-2. The animals (three or four per group) received a single injection of either IL-2 or PEG-IL-2 as a 15-min i.v. infusion or s.c. bolus injection; one group received IL-2 as an i.v. bolus injection (Table 1). The IL-2 and PEG-IL-2 were administered undiluted or diluted with 5% dextrose in water to keep injection volumes similar. Injections were made through a catheter in a marginal ear vein or as a bolus under the skin of an area of the thigh tented by the restraining thumb and forefinger for i.v. or s.c. administration, respectively. The left side was chosen because the right side was involved in the surgical procedures.

Animals

Yorkshire pigs were obtained from local farmers near Davis, CA (part I), or from Earle Parsons and Sons, Inc. (Hadley, MA; part II). During a quarantine period of at least 7 days, the pigs were handled daily to acclimate them to close human contact. In part II, the animals were also acclimated to wearing an aluminum protective jacket during the quarantine period.

Cannulations

Before the study and while the animals were under surgical anesthesia, catheters were placed in the jugular vein for blood sampling and in the femoral vein for i.v. dosing (part I). In part II, surgery to insert a cannula in both the external jugular vein for blood collection and thoracic duct for lymph collection was performed 1 to 3 days before dosing and was based on a previously published method (Jensen et al., 1990).

Briefly, the surgical method to insert the jugular vein and thoracic duct catheters was as follows: the external jugular vein was ligated cranially, and a catheter was inserted and advanced caudally approximately 10 cm so that the tip would be positioned in the superior vena cava, close to the right atrium. The catheter was tunneled s.c. to an area midway between the scapulae.

A right lateral thoracotomy was then performed through the fifth to seventh intercostal space, and for most of the animals, a rib was removed. The thoracic duct was identified subpleurally between the aorta and the vertebral bodies. The thoracic duct was ligated cranial to the insertion point. An appropriately sized Silastic catheter or Hydrocath was inserted into the duct and advanced approximately 1 to 2 cm caudally, so that the tip was positioned at approximately the level of the 10th thoracic vertebral body. The catheter was filled with heparin (5000 I.U./ml) to prevent coagulation. The catheter was passed through the right 7th intercostal space to a site 15 to 20 cm caudal to the site of exit of the jugular catheter.

The thoracic duct catheter was attached to the jugular vein catheter by a single three-way stopcock. The thoracotomy and exit sites were closed in a standard fashion. An aluminum jacket was placed on the animals, which were allowed to recover from anesthesia. The animals wore this jacket continuously throughout the study period; as the animals grew, the jacket was replaced or adjusted as necessary.

The jugular and thoracic duct catheters were flushed twice daily by a retrograde infusion of heparin into the lymph catheter and an antegrade infusion of heparin into the venous catheter to maintain patency.

Sample Collection Procedures

At each time of collection, blood (2 ml each) was obtained from the cannula in the jugular vein and placed in labeled tubes containing heparin (part I) or EDTA (part II). The collection times were pretreatment and 11 to 15 times for up to 48 h after dosing. After centrifugation, plasma samples were placed into sample storage tubes and frozen at 70°C until assayed.

In addition to blood samples, lymph samples (2 ml each) were collected from the externalized thoracic duct cannula and placed in labeled tubes containing EDTA (part II). The collection times were the same as the blood samples. The lymph samples were stored frozen at 70°C until assayed.

Assays

In part I, concentrations of IL-2 in plasma were measured in a bioassay adapted from Gillis et al. (1978). Cell proliferation was quantified by the incorporation of [³H]thymidine into an IL-2-dependent cell line, HT2A5E, which was a subclone of murine lymphocytes (Watson, 1979). The plasma and lymph samples collected from IL-2-treated pigs in part II were assayed using a commercially available double-antibody sandwich enzyme-linked immunosorbent assay (ELISA) specific for IL-2 (Amersham, Arlington Heights, IL). The quantification of PEG-IL-2 in plasma and lymph samples used an ELISA specific for IL-2 that was developed in-house. A murine monoclonal antibody was used for capture, and a horseradish peroxidase-conjugated affinity-purified rat anti-human IL-2 polyclonal antibody was used for detection. For the bioassay, standards were spiked into pig plasma, but the controls were made in buffer. For each ELISA, both the standards and controls, either IL-2 or PEG-IL-2 was spiked in pig plasma and lymph; the performance of each assay is summarized in Table 2.

Pharmacokinetic Analysis

Noncompartmental Analysis. The plasma and lymph concentration-time data were analyzed separately using Pharm-NCA (Noncompartmental Pharmacokinetic Data Analysis Software, Simed S.A., Creteil, France). The area under the plasma and lymph concentration-time curves (AUC_P and AUC_L, respectively) were calculated using linear-log trapezoidal rules, with extrapolation to time infinity using the estimated terminal elimination half-life (t_1/2).

Compartmental Analysis. A population-based analysis of the pharmacokinetics data was performed using the computer program NONMEM version IV level 1.0 with PREDPP version III level 1.0 and NMTRAN version II level 3.0 (Beal and Sheiner, 1992). All of the available data were analyzed simultaneously for each test article to provide estimates of the average population values of the pharmacokinetic parameters (fixed effects) as well as the degree of variation between and within individuals (random effects). In all, 506 observations in 22 animals constituted the database for IL-2, and 298 observations in 14 animals constituted the database for PEG-IL-2. The observation that the relative concentrations of IL-2 and PEG-IL-2 in plasma and lymph were different depending on the route of administration was justification for assigning them to separate compartments.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Models used in NONMEM analysis of plasma and lymph data. Model parameters were volume of the central compartment, Vc; volume of the lymph compartment, Vl; volume of the peripheral compartment, Vt; clearance from the central to the lymph compartment, CL12; systemic or elimination clearance, CL10; and intercompartmental clearances between sampling and peripheral compartments, CL13, CL25. For s.c. dosing, an s.c. dosing compartment, the absorption rate constant from the dosing site into lymph is indicated by kalymph, and the absorption rate constant from the dosing site into the central compartment is indicated by kacentral.

(1)

(2)

(3)

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Noncompartmental Analysis

Initially, the pharmacokinetic parameters for the IL-2 plasma data from the animals in parts I and II were evaluated separately (Table 3). In part I, there was no significant difference by dose in CL or V_ss, but both parameters differed when the route of administration was compared using ANOVA. This indicated that the bioavailability of the s.c. dose was less than complete. Surprisingly, there were no significant differences by dose or route of administration for CL or V_ss for IL-2 in part II. The pharmacokinetic parameters from the two parts were combined to determine whether the ANOVA results were indicative of the fundamental differences in methodology between the two studies (i.e., plasma assay method and surgical implantation of a lymph catheter).

When the data from the two studies were combined, the noncompartmental estimates of CL/F evaluated by ANOVA showed differences that were statistically significant when grouped according to treatment. CL/F was found to be similar (average, 250-300 ml/h/kg) when IL-2 was administered as an i.v. bolus at the two higher dose levels (0.16 and 1.6 MIU/kg) but was approximately 2-fold faster (680 ml/h/kg) at the lowest bolus dose level (0.10 MIU/kg). This parameter was an additional 2-fold faster (1600 ml/h/kg) when IL-2 was administered by infusion at a corresponding low dose (0.10 MIU/kg). There were no statistically significant differences among all pairs of mean CL/F values (550-1250 ml/h/kg) for the five s.c. dose levels (0.10-3.0 MIU/kg). These s.c. results suggested that there was no significant difference between the methodologies used in the two studies. Instead, the significant differences in CL/F were attributed to changes in bioavailability related to route of administration and dose, suggesting that a nonlinear model may be required to characterize the pharmacokinetics of IL-2. This assumption was the basis of the compartmental model adopted in the population pharmacokinetics analysis.

Noncompartmental estimates of the volume of distribution (V_ss/F) were not different when examined by treatment. Although there were large differences in the mean values of the treatment groups, the intragroup variance, especially in the s.c. dosed animals, was extremely large, with coefficient of variations ranging up to 125%. The lack of a significantly different V_ss/F in the face of significantly different CL/F values for the two highest bolus doses is not explained.

The corresponding results for PEG-IL-2 are given in Table 4. Estimates of CL/F and V_ss/F for PEG-IL-2 obtained by noncompartmental analysis were significantly different when grouped by route of administration and s.c. dose level. The post hoc test showed the CL/F value at the highest dose (0.10 MIU/kg) was different from that at the lowest dose (0.01 MIU/kg) after s.c. administration. However, it is noted that the estimates of CL for the i.v. dose of PEG-IL-2 are comparable to those seen when the lowest and middle doses are administered s.c. This suggests that bioavailability of the s.c. doses may be essentially complete.

The post hoc test showed the V_ss/F value of PEG-IL-2 at the highest s.c. dose was different from that after i.v. infusion but not different from that after the lowest s.c. dose. Because the estimates of V_ss/F for the i.v. dose of PEG-IL-2 are also comparable to those seen when the lowest and middle doses are administered s.c., this further suggests that bioavailability of the s.c. doses may be essentially complete at low doses. It is possible, of course, that the extent of systemic absorption may be reduced as the dose is increased; this would be consistent with the observed increase in the apparent CL and V_ss values of PEG-IL-2.

Significant increases in the AUC in lymph were seen when IL-2 was administered s.c. over that noted when the same dose was administered i.v. (Table 3). For example, the AUC in the lymph (mean ± S.D.) was 43.5 ± 3.1 ng·h/ml for s.c. dosing, whereas the lymph AUC was only 2.71 ± 1.46 and 6.56 ± 1.32 ng·h/ml when the same dose was administered by infusion and bolus, respectively. The AUC_L/AUC_P ratios are also presented in Table 3. It is clear that s.c. dosing provides significantly higher lymph levels than i.v. dosing. For example, the AUC_L/AUC_P ratio increases from 0.65 to 0.70 for i.v. infusion-bolus dosing to 3.5 when the same dose (0.1 MIU/kg) was administered s.c. Thus, the distribution advantage (DsA_L) is the ratio of these ratios, or approximately 5.

Table 4 contains the lymph PEG-IL-2 data. The AUC_L/AUC_P ratio was found to increase from 0.33 to 1.3 when comparing i.v. and s.c. doses. The corresponding estimate of DsA_L was 3.8.

Compartmental Analysis

There were significant differences in the structure of the final NONMEM models that described the plasma and lymph concentrations of IL-2 and PEG-IL-2. A noneliminating peripheral compartment reversibly connected to the lymph compartment was added to the model for IL-2, indicating that the distribution of PEG-IL-2 from the lymph was restricted. The NONMEM population estimates for the pharmacokinetic parameters for IL-2 and PEG-IL-2 are in Tables 5 and 6, respectively.

IL-2. In examining the i.v. 0.1 MIU/kg (lowest dose level) data shown in Fig. 2 (A and B, bolus; C and D, infusion), it appears that in general, the plasma and lymph concentration data fall off more rapidly than the model-predicted terminal decline. For the i.v. bolus data, some underprediction of the initial time course is also noted. This may be related to an underestimate of bioavailability (NONMEM estimate of F_bol, lo = 0.39) in this dosing group (Table 5). At the intermediate i.v. bolus dose level (Fig. 3A), the correspondence between data and model-predicted function in the plasma concentration in the terminal phase is excellent, whereas at the highest i.v. bolus dose level, there appears to be slight underprediction of the terminal phase plasma concentration (Fig. 3A). No lymph data were available in the crossover experiments portrayed in Fig. 3. Although some misspecification was noted, the model reasonably described the plasma and lymph concentration-time data.

View larger version (29K): [in this window] [in a new window]   Fig. 2.   Observed (symbols) and predicted (line) plasma and lymph concentrations after i.v. bolus (A and B) and 15-min i.v. infusion administration (C and D) of 0.1 MIU/kg IL-2 in part II. Observed data are from individual animals using the same symbols for the plasma and lymph concentrations.

View larger version (23K): [in this window] [in a new window]   Fig. 3.   Observed (symbols) and predicted (line) plasma concentrations after i.v. bolus (A) and s.c. administration (B) of 0.16 MIU/kg (open symbols) and 1.6 MIU/kg (filled symbols) IL-2 in part I. Observed data are from individual animals using the same symbols for each animal.

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Observed (symbols) and predicted (line) plasma and lymph concentrations after s.c. administration of 0.1 MIU/kg (A and B), 1.0 MIU/kg (C and D), and 3.0 MIU/kg (E and F) IL-2 in part II. Observed data are from individual animals using the same symbols for the plasma and lymph concentrations.

PEG-IL-2. Graphs of the plasma and lymph concentration-time data and of the corresponding NONMEM-predicted functions are given by sampling site for the i.v. infusion treatment group (Fig. 5, A and B), and s.c. dosing groups (Fig. 6, A-F) for PEG-IL-2. Some model misspecification can be observed in plots of predicted versus observed plasma concentrations in the case of PEG-IL-2. In contrast to IL-2, higher plasma concentrations tended to be overpredicted, and lower values were underpredicted, which is consistent with the nonlinearity in the data observed when subjected to noncompartmental analysis.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Observed (symbols) and predicted (line) plasma and lymph concentrations after 15-min i.v. infusion of 0.01 MIU/kg PEG-IL-2 (A and B). Observed data are from individual animals using the same symbols for the plasma and lymph concentrations.

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Observed (symbols) and predicted (line) plasma and lymph concentrations after s.c. administration of 0.01 MIU/kg (A and B), 0.03 MIU/kg (C and D), and 0.1 MIU/kg (E and F) PEG-IL-2. Observed data are from individual animals using the same symbols for the plasma and lymph concentrations.

Lymph Delivery Advantage Afforded by s.c. Dosing

The lymph delivery advantage (DvA_L) for IL-2 was estimated noncompartmentally and compartmentally (Table 7). At a dose of 0.1 MIU/kg, the calculated DvA_L is greater for s.c. dosing when defined with reference to infused IL-2 instead of a bolus dose because the bioavailability of infused IL-2 is less than that of a bolus ([F_s.c./F_inf] > [F_s.c./F_bol]; see eq. 2). For instance, the values for AUC_L for s.c. and i.v. bolus dosing at a dose level of 0.1 MIU/kg are calculated to be 33.39 and 5.243 ng·h/ml, respectively (Table 3). Their ratio (DvA_L) is 6.37 (Table 7). This is identical with the value of the lymph delivery advantage (DvA_L = 6.4; Table 7) calculated from the right-hand term where NONMEM clearance and bioavailability parameters are used.

Values of DvA_L for IL-2 could not be estimated noncompartmentally for doses of 0.16 MIU/kg or larger because no lymph data were available after i.v. dosing in this range. However, compartmental estimates are reported here, using the clearance values estimated in NONMEM and the values of F estimated or, in the case of high bolus doses, assumed. Clearly, studies that permit lymph as well as plasma sampling for the assay of IL-2 in the higher dosage range would provide more reliable estimates of the lymph delivery advantage.

For PEG-IL-2, noncompartmental estimates of AUC_L/AUC_P increase from 0.33 to approximately 1.2 when comparing i.v. and s.c. doses (Table 4). The corresponding noncompartmental estimate of DvA_L is 3.8 (Table 7). The NONMEM estimate of DvA_L is substantially lower than the value estimated noncompartmentally for PEG-IL-2 (Table 7). This may be due to the assumption of linearity, which seems not to hold for PEG-IL-2 at the highest dose as demonstrated by the results of the ANOVA. The difference may also stem from uncertainty in the estimates of ka_lymph and ka_central, which determine F_L (eq. 3).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Animal Model Selection. The main consideration in selecting the appropriate species to study lymphatic absorption was the ability to predict human data. Before using pigs, the pharmacokinetics of IL-2 and PEG-IL-2 had been characterized in mice, rats, rabbits, sheep, cynomolgus macaques, and cancer patients after i.v. administration. Male/female differences in the pharmacokinetics of IL-2 and PEG-IL-2 have never been reported. Allometric relationships between clearance and body weight could be established, and the exponents were 0.7, which are typical for compounds where interspecies scaling has been applied (Chappell and Mordenti, 1991; Mordenti et al., 1991).

Comparison between IL-2 and PEG-IL-2. The addition of PEG molecules decreased the clearance by increasing the hydrodynamic radius of IL-2, which led to a decrease in renal elimination (Knauf et al., 1988). The addition of PEG molecules to IL-2 not only slowed the elimination clearance but also all of the other pharmacokinetic parameters differed between IL-2 and PEG-IL-2. The clearance from central to lymph compartments was about 130-fold less for the higher molecular weight PEG-IL-2, and the clearance from lymph to plasma was approximately 65-fold less (Tables 5 and 6). Thus, AUC_L relative to AUC_P was about twice as great (0.67, Table 3) for IL-2 than it was for PEG-IL-2 (0.33, Table 4).