Abstract
The pharmacokinetics of the tumor necrosis factor receptorimmunoglobulin fusion protein, lenercept, were assessed in rats, rabbits, dogs and cynomolgus monkeys. Pharmacokinetic parameters were extrapolated to humans by allometric scaling. Lenercept was dosed i.v. at doses ranging from 0.1 to 5 mg/kg. Consistent with its all-human sequence, lenercept elicits an immune response in laboratory animals usually 6 to 10 days after dosing. The resulting period of more rapid clearance caused by the immune response was excluded from the pharmacokinetic evaluation. Lenercept showed a very low and similar clearance in all species tested (0.0071–0.0097 ml·min/kg). The volume of distribution was estimated at values between 61 and 90 ml/kg, whereas the terminal half-life ranged from 3.4 days in rabbits to 6.5 days in rats. Thus, lenercept was characterized by similar pharmacokinetic properties across species, irrespective of their particular body weight. Accordingly, both clearance (ml/min) and volume of distribution (ml) scaled with an allometric exponent close to 1, whereas half-lives (including literature data in mice) yielded an allometric exponent close to 0. The predicted parameters in humans agree well with the observed values. Overall, the results demonstrate an allometric scaling for lenercept different from that for other therapeutic proteins, in that lenercept displays a similar pharmacokinetic behavior across species. Despite an early and pronounced immune response against this all-human protein in laboratory animals, the pharmacokinetic data were found to be predictive for humans, given that the more rapid immune-modulated clearance component in animals could be identified and excluded from the pharmacokinetic evaluation.
During the last decades it has been shown that pharmacokinetic parameters of small molecular weight drugs scale across species as a function of body weight, using a power function of the general form: P = a Wb , where P is the pharmacokinetic parameter, W the body weight, a the allometric coefficient, and b the allometric exponent. After log-transformation the power function can be written as a linear equation: log P = log a + b logW, where log a is the intercept and bthe slope.
Allometric scaling of small molecular weight xenobiotics has been applied to compounds excreted by physical processes such as glomerular filtration of unchanged drug (Sawada et al., 1984; Mordenti, 1985), but also to metabolized compounds if suitable measures are taken to correct for interspecies differences in metabolic clearance (Ings, 1990; Lave et al., 1996). Pharmacokinetic parameters of nonmetabolized compounds usually scale with body weight according to power functions with a certain allometric exponent, depending on the pharmacokinetic parameter (Mordenti, 1986). Clearance tends to scale with an exponent of 0.6 to 0.8, whereas volumes of distribution and half-lives scale with exponents of approximately 0.8 to 1.0 and 0.2 to 0.4, respectively. It has recently been shown that the principles of allometric scaling established for nonmetabolized small molecular weight compounds can also be applied to therapeutic proteins covering a molecular weight range from 6 to 98 kDa (Mordenti et al., 1991; McCarthy et al., 1993).
Lenercept is a recombinant fusion protein consisting of the extracellular domain of two human p55 tumor necrosis factor (TNF)1 receptors and the hinge as well as the constant domain C2 and C3 sequences of the human immunoglobulin G1 (IgG1) heavy chain (Ashkenazi et al., 1991). Lenercept, previously referred to also as Ro 45–2081 or TNF receptor immunoadhesin, binds TNF with high affinity and a high kinetic stability of the resulting complex (Evans et al., 1994). Its therapeutic benefits derive from its capability of neutralizing TNF activity under disease conditions such as severe sepsis and septic shock, which are characterized and mediated by an overproduction of TNF. The protective effect of lenercept in severe sepsis and septic shock has been demonstrated in laboratory animals (Lesslauer et al., 1991; Van Zee et al., 1996) and, recently, in clinical trials (Abraham et al., 1997). Lenercept is purified from eukaryotic cell expression and is present as a disulfide-bonded homodimer with a molecular weight of about 120 kDa and eight potential asparagine-N-linked glycosylation sites. In the course of the development of lenercept, its pharmacokinetics were characterized in several animal species. This allowed for both better planning of toxicology trials and for predicting the pharmacokinetics in humans by allometric scaling, in order to help in the design of first clinical trials.
In this paper we describe the allometric scaling of lenercept. A first allometric scaling for lenercept was performed prospectively to help in the design of first clinical protocols. This prediction was based on preliminary pharmacokinetic data in several animal species. Here we communicate results from a refined, retrospective scaling, which was performed when additional data on animal pharmacokinetics became available. The results of the experiments showed that the pharmacokinetics of lenercept do not follow the described scales of allometry; rather, the drug shows similar pharmacokinetic behavior across species from rat to human, i.e., both clearance and volume of distribution scale with a slope factor close to 1.
Materials and Methods
Test Material.
Lenercept was made by Genentech, Inc. (South San Francisco, CA). It was produced in Chinese hamster ovary cells and purified by ion-exchange and affinity chromatography, as previously described (Ashkenazi et al., 1991). The test substance was formulated as an aqueous solution at a concentration of 5 mg/ml. This formulation was used as such for i.v. bolus administration to rats and rabbits, or was further diluted for administration by i.v. short infusion to cynomolgus monkeys and dogs.
Animal Experiments.
Lenercept was administered i.v. either by bolus administration or short infusion. Short infusion was chosen for those animal species (cynomolgus monkey, dog) that had not been exposed to lenercept before the described experiments.
Eight male RoRo rats (mean weight 0.266 kg) were dosed a single i.v. bolus of lenercept by injection into the tail vein at dose levels of 0.2 or 5 mg/kg (n = 4/dose group). Blood samples (0.5 ml) were collected by retro-orbital bleeding from two rats/dose group at each of the following times: 1, 8, 24, 48, 72, 144, 168, and 192 h after administration. At the last sampling time (312 h), terminal blood samples were obtained from all rats. Blood was collected over EDTA/NaF as anticoagulant, and plasma was prepared and stored frozen until analysis.
Three female Himalayan rabbits (mean weight 2.81 kg) were administered a single i.v. bolus dose of lenercept (5.0 mg/kg) through the ear vein. Blood samples (1 ml) were collected from the ear vein at: predose, 0.5, 1.5, 4, 8, 24, 26, 32, 48, 72, 96, 102, 168, 192, 216, 240, 264, 271, 288, and 312 h. Plasma was prepared as described above and stored frozen until analysis.
Four male cynomolgus monkeys (mean weight 3.2 kg) were dosed i.v. with a single dose of lenercept at dose levels of 4.0 or 5.0 mg/kg (n = 2 per dose level), administered by infusion (ca. 30 min) into the cephalic vein. Blood samples (1 ml) were collected from the brachial vein at predose, 0.5 (end of infusion), 1, 2, 3, 6, 8, 12, 24, 48, 72, 96, 120, 144, 168, 240, and 336 h after start of the infusion. Additional samples were taken from the 4 mg/kg group at 10, 192, and 288 h. Blood was collected over EDTA as anticoagulant. Plasma was stored frozen until analysis.
Two male beagle dogs (mean weight 13.9 kg) received lenercept as a 20 min i.v. infusion (0.11 and 0.14 mg/kg). Blood samples of 2 ml each were taken from the cephalic vein at predose, 0.5, 1.5, 3, 7, 23.5, 31, 48, 72, 95.5, 120, 168, 175, 192, 216, 240, 264, 271, 341, 365, 389, 413, and 437 h after administration. Blood samples were allowed to clot, then centrifuged. The serum was stored frozen until analysis. In contrast to that in the other species, the pharmacokinetics in the dog was studied only at a low dose level close to the expected therapeutic one, since dogs were not intended to be used in the safety evaluation of lenercept.
Drug Assay Method.
Lenercept was analyzed by means of an enzyme-linked immunological and biological binding assay. In brief, in a single step lenercept is bound to a microtiter plate coated with mouse monoclonal antibodies against the human sTNFR-55 portion of lenercept (clone htr-20), and at the same time the free TNF binding sites of lenercept are labeled with its ligand TNF-α, which is coupled to horseradish peroxidase. Nonbound material is removed by washing and the quantity of peroxidase bound to the microtiter plate is measured enzymatically. Measurement range was 0 to 20 ng/ml lenercept, with a detection limit of 0.2 ng/ml. Biological samples were analyzed after appropriate dilution with a suitable buffer solution.
Determination of Antibodies against Lenercept.
For the determination of neutralizing antibodies against lenercept a modified sandwich-type assay was used. Different dilutions of serum or plasma were coincubated with 10 ng/ml lenercept in microtiter plates coated with noninhibitory mouse monoclonal antibodies against the human sTNFR-55 portion of lenercept (clone htr-20), then a TNF-α-peroxidase conjugate was added to label still active lenercept. After washing the plates the amount of bound peroxidase was determined enzymatically. In the presence of antibodies the measurable amount of lenercept was less than the 10 ng/ml added. The difference between this 10 ng/ml and the measured concentration gave the amount of lenercept that was neutralized by antibodies. Antibody levels are reported as the concentration of lenercept that can be neutralized.
Pharmacokinetic Analysis.
Pharmacokinetic parameters were calculated by noncompartmental methods, using the computer program TOPFIT 2.0 (Heinzel et al., 1993). For rats, parameters were determined from composite plasma concentration data. The area under the plasma or serum concentration-time curve (AUC) was calculated using the linear trapezoidal rule and extrapolated to time infinity. However, both AUC and area under the first moment curve (AUMC) were estimated by extrapolation from the apparent linear portion of the semilogarithmic plot before the change in the kinetics due to the onset of the immune response against the test substance (for rationale, see Discussion). The apparent half-life (T1/2) was calculated using the equation T1/2 = ln 2/β. The apparent rate constant β was determined from the linear portion of the log plasma or serum concentration-time curve before the onset of the immune response. Total clearance (Cl) was calculated as Dose/AUC. The volume of distribution at steady state (Vss) was calculated as Dose × AUMC/(AUC)2. The initial volume of distribution (Vi) was calculated as Dose/concentration of the first sample taken after administration.
Allometric Scaling.
The mean pharmacokinetic parameters (Cl,Vss, and T1/2) for lenercept in laboratory animals were log-transformed and correlated with the log-transformed mean body weights (W) using the log-transformed allometric equations of the general form: logP = log a + b log W, where P is the pharmacokinetic parameter, W the body weight, a the allometric coefficient, and bthe allometric exponent. The values of allometric coefficientsa and exponents b were estimated by linear least-squares regression. The allometric equations obtained were used to calculate the respective pharmacokinetic parameter for a 70-kg human.
Results
Pharmacokinetic Experiments.
Characteristic individual plasma or serum concentration-time curves of lenercept and antibodies against lenercept following i.v. administration of lenercept to rabbits, cynomolgus monkeys and dogs are shown in Fig. 1. Composite plasma concentration-time curves in rats are depicted in Fig.2. All derived pharmacokinetic parameters are presented in Table 1.
Concentration-time curves were characterized by a triphasic profile with an initial rapid decrease in the concentrations followed by a slower second phase and then an accelerating rate of decline. The third phase, however, was not observed in the rats at a dose level of 5 mg/kg. In all other experiments the onset of this third phase occurred between 6 and 10 days after administration of the test substance, indicating an immune response against lenercept. An antibody assay to confirm formation of neutralizing antibodies could only be performed once lenercept was no longer detectable in plasma or present at very low concentrations only. After the third phase, neutralizing antibodies were measurable in all species, which clearly indicated that the third phase was due to the additional immunological clearance mechanism (Figs. 1 and 2). Levels of non-neutralizing antibodies were not assessed in these experiments. The beginning of the third phase differed from animal to animal. Thus, in individual rabbits the onset of the third phase was observed at days 4, 7, or 11 postdose, respectively.
The pharmacokinetic parameters were calculated by neglecting this third phase. They were estimated based on the second phase, which was considered the true terminal phase with regard to ‘normal’ pharmacokinetic processes, i.e., in the absence of an additional immunological clearance component. Thus, pharmacokinetic parameters in dogs and monkeys can be directly compared with the ones in rats at 5 mg/kg, in which this third phase was missing. Furthermore, pharmacokinetic parameters could be directly used for allometric scaling without any correction for the missing third phase in rats. In all species, lenercept was cleared very slowly, with mean values ranging from 0.0071 ml·min/kg in rats to 0.0097 ml·min/kg in rabbits. The initial distribution occurred within the plasma space, as indicated by Vi values close to the plasma volume. The values for Vss were slightly higher, ranging from 61 ml/kg in rabbits to 90 ml/kg in rats. Long apparent half-lives were observed with mean values (range) of 6.5, 3.4 (2.6–4.2), 5.1 (4.3–6.1), and 5.0 (3.9–6.1) days in rat, rabbit, cynomolgus monkey, and dog, respectively.
Dose linearity of the pharmacokinetics over a wide dose range (0.2–5 mg/kg) was demonstrated in the rat. Plasma concentration-time curves were superimposable before the onset of the immune response (composite data from four rats from each dose level). The 25-fold increase in dose led to a 21-fold increase in the AUC0–72h (177 and 3770 μg·hr/ml, respectively). In the 0.2-mg/kg group, however, no further pharmacokinetic evaluation was possible, because the second phase in the concentration-time curve was too short to allow any evaluation of pharmacokinetic parameters. The latter was also true for one out of three rabbits, in which the onset of the immune response occurred at 4 days postdose.
Allometric Scaling.
Table 2 shows the results of the linear least-squares fitting of log Cl, logVss, and logT1/2 versus log W data in animals, including the predicted and observed values in humans.
For both Cl and Vss, the slopes of the regression lines were close to 1 (1.06 and 0.969, respectively) (Fig. 3). For both parameters the correlation was excellent (coefficients of correlation: forCL, r2 = 0.997; forVss, r2 = 0.990). The predicted values of Cl andVss in humans (0.66 ml/min and 70 ml/kg, respectively) were in reasonable agreement with the observed values at dose levels from 1 to 100 mg (0.33 ml/min and 70 ml/kg, respectively) (Kneer et al., 1996). For T1/2 the slope of the regression line was −0.0789. When previously reported data on the pharmacokinetics of lenercept in mice (T1/2 = 5.4 days) were included (Haak-Frendscho et al., 1994), the slope changed to −0.0375 (Fig. 3). Consistent with the flat slope, poor coefficients of correlation were obtained (r2 = 0.229, including mice: 0.173), i.e.,T1/2 hardly appears to be correlated with body weight. Nevertheless, the predicted T1/2 in humans (4.2 days, including mice data) agreed well with the observed value (7.0 days).
Discussion
Preclinical studies with human proteins are often accompanied by an immune response against the protein, leading to the formation of antibodies (Working, 1992). Due to the human amino acid sequence and structure of lenercept, an immune response against the test substance was fully expected in laboratory animals. Indeed, in all species tested, except in the rat at 5 mg/kg, an immune response was observed usually within 6 to 10 days after single dosing, as evident from antibody analysis and a rapid drop in the plasma concentration-time curves. The apparent absence of an immune response in the rat at 5 mg/kg probably reflects the interindividual variability of the immune response rather than a species difference. Thus, in other pharmacokinetic experiments with lenercept in the rat at this dose level an immune response was observed (data not shown).
The rapid drop in the plasma concentration-time curves caused by the immune response was excluded from the pharmacokinetic analysis. This exclusion was justified because the immune response in nonhomologous species bears no relevance to humans, so that pharmacokinetic parameters from an analysis including the immune response cannot be extrapolated to humans. However, the onset of this immune response meant that the pharmacokinetics could only be followed for about one to two half-lives, leading to some imprecision in the values estimated for the parameters. This is especially true forVss because the contribution of the extrapolated portion to the AUMC was very high. Furthermore, an elimination from the central compartment had to be assumed for the estimation of Vss by noncompartmental methods.
Nevertheless, the pharmacokinetics of lenercept could be characterized in laboratory animals. In all species tested, lenercept was cleared very slowly, at rates far below physiological flow rates like liver or kidney plasma flow. The estimated values forVss were small, about twofold the plasma volume, indicating a very limited distribution of lenercept to tissues. The extremely low clearance was reflected in long apparent half-lives ranging from 3.4 days in the rabbit to 6.5 days in the rat.
Despite the difficulties in the pharmacokinetic assessment of lenercept caused by the immune response, allometric scaling of its pharmacokinetic parameters was reasonably predictive of the human pharmacokinetics (see Table 2). The greatest deviation of the predicted values from the observed ones was found for clearance, where the predicted value was twice as high as the observed one. The allometric equations turned out to be different from the ones reported for other proteins (Mordenti et al., 1991), the allometric scalings of which tend to follow the principles established for nonmetabolized small molecular weight compounds. Thus, for rCD4, CD4-IgG, rhGHm, rt-PA and relaxin the exponents in the allometric equation for Cl andVss vary from 0.65 to 0.84 and 0.84 to 1.02, respectively. By contrast, with lenercept the allometric exponents for both Cl and Vsswere close to 1.0. Consequently, the allometric exponent forT1/2 was close to 0, i.e., a similar half-life is seen in all species regardless of weight. The deviations of the predicted human parameters from the observed ones may be partially explained by the discussed shortcomings of the pharmacokinetic assessment in laboratory animals. Due to the immune response, the terminal phases of most concentration-time curves could not be well characterized. Thus, some Cl values might be overestimated, while some T1/2 values were underestimated. It is noteworthy that the longest period available for pharmacokinetic assessment (13 days in the rat) was associated with the longest estimated T1/2 (6.5 days), which was also closest to the observed T1/2 in humans. In all other species, only 6 to 10 days postdose were available for pharmacokinetic assessment before the onset of the immune response.
Possible reasons for the uncommon allometric scaling of lenercept’s clearance may include its extremely slow clearance, which represents only a marginal fraction of physiological flow rates to liver or kidney in all species studied. Accordingly, the allometric scaling of lenercept pharmacokinetics does not follow the allometric scaling of these flow rates; this is in contrast to results reported for smaller, more rapidly cleared proteins such as rt-PA (Mordenti et al., 1991). Other possible explanations for the uncommon allometric scaling may include potential specific clearance pathways of lenercept, suggested by its molecular structure. Due to its high molecular weight above the glomerular filtration threshold of about 70 kDa (Knauf et al., 1988) and its high number of glycosylation sites, lenercept is probably cleared, at least partially, via glycoprotein receptors. In addition, Fc-receptor mediated clearance pathways appear to be possible due the Fc-portion of lenercept. Interspecies differences of these clearance routes, however, are largely unknown for low intrinsic clearance proteins. Therefore, a full explanation for the unusual allometric scaling of lenercept cannot be provided yet. However, future examples of proteins with such allometric scaling properties may enhance the understanding, for which therapeutic proteins such extrapolation behavior can be expected.
Overall, the results from this study demonstrate an uncommon allometric scaling of the pharmacokinetics of lenercept, in that it shows similar half-lives from mice to humans. Despite an immune response against this all-human protein in laboratory animals the pharmacokinetic data in animals were found to be predictive for humans, given that the more rapid immune-modulated clearance in animals could be identified and excluded from the pharmacokinetic evaluation.
Acknowledgments
We thank Marie-Stella Gruyer and Arthur Wälchli for their skillful technical assistance. The support of Dr. Zühlke at Covance Laboratories, Münster (Germany) for the conduct of in-life experiments with cynomolgus monkeys is gratefully acknowledged.
Footnotes
-
Send reprint requests to: Wolfgang F. Richter, Pharma Division, Preclinical Research, F. Hoffmann-La Roche Ltd., Grenzacherstrasse 124, CH-4070 Basel, Switzerland. E-mail:wolfgang.richter{at}roche.com.
- Abbreviations used are::
- TNF
- tumor necrosis factor
- AUC
- area under the plasma or serum concentration-time curve
- AUMC
- area under the first moment curve
- Received April 6, 1998.
- Accepted July 28, 1998.
- The American Society for Pharmacology and Experimental Therapeutics