A novel pharmacokinetic model based on the complex elimination of monoclonal antibodies for bevacizumab pharmacokinetic study in rabbits

International Immunopharmacology 31 (2016) 39–44

Contents lists available at ScienceDirect

International Immunopharmacology journal homepage: www.elsevier.com/locate/intimp

A novel pharmacokinetic model based on the complex elimination of monoclonal antibodies for bevacizumab pharmacokinetic study in rabbits Lei Wang, Shan Ji, Meizhen Li, Zeneng Cheng ⁎ Research Institute of Drug Metabolism and Pharmacokinetics, School of Pharmaceutical Sciences, Central South University, Changsha, Hunan 410013, China

a r t i c l e

i n f o

Article history: Received 8 September 2015 Received in revised form 28 November 2015 Accepted 10 December 2015 Available online 15 December 2015 Keywords: Bevacizumab Estimation Monoclonal antibody Nonlinear elimination Pharmacokinetics

a b s t r a c t Monoclonal antibodies (mAbs) complex pharmacokinetic (PK) properties including a nonlinear pharmacokinetics and a significant variation in individual PK process cannot be appropriately described by classic PK models, probably derived form a poor understanding of the complex elimination of mAbs. In this study, a novel PK model based on mAbs' complex drug elimination was established. Subsequently, this new model was used to fit bevacizumab plasma concentration data from PK rabbits, and the outcomes of model fitting were compared with those came from a fit with classic models. In addition, the variations existing in the parameters set in the new model were analyzed. As a result, this novel model reasonably described the single-dose PK profiles of bevacizumab in rabbits, and its fitting efficiency was greatly improved compared with those fitted with classic PK models in terms of the weighted residual sum of squares. Moreover, the variations existing in the new model's parameters CA(antibody) and K0 could reasonably explain the individual variations of bevacizumab's PK profiles. In conclusion, the novel model reasonably explained the elimination of bevacizumab, and exhibited a potential as a useful tool for the PK studies of bevacizumab and other mAbs in practice. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Antibodies are large proteins utilized by the immune system for an identification and neutralization of foreign objects such as bacteria and viruses. Monoclonal antibody (mAb) preparations, which derive from a single progenitor through cloning the hybridoma cells, are homogeneous with respect to antibody isotype, primary amino acid sequence, affinity, and specificity [1–2]. Due to a specific binding to targets relevant to disease progress, mAbs can offer considerable advantages over small-molecule drugs, by having fewer adverse effects and/or possibly by increasing the efficacy of treatment than conventional therapy [3–4]. In recent years, more and more mAbs have been approved as therapeutic drugs and entered into the clinic for various diseases, such as cancer, inflammatory diseases and hematological disorders [5–7]. In particular, a considerable clinical success has been achieved with mAbs in cancer therapy for its remarkable pharmacological characteristics, such as high potency, limited off-target toxicity, and long serum half-lives [8–9]. Bevacizumab, a recombinant humanized IgG1 antibody, has been widely employed for the treatment of metastatic colorectal cancer and non-small cell lung cancer as a kind of anti-angiogenic Abbreviations: ADCC, antibody dependent cellular cytotoxicity; AUC, area under the concentration–time curve; CL, clearance; mAbs, monoclonal antibodies; M–M, Michaelis–Menten; PBS, phosphate-buffered saline; PK, pharmacokinetic; TMB, tetramethylbenzidine; TMDD, target-mediated drug disposition; T1/2, half life; V, apparent volume of distribution; VEGF, vascular endothelial growth factor. ⁎ Corresponding author. E-mail address: [email protected] (Z. Cheng).

http://dx.doi.org/10.1016/j.intimp.2015.12.016 1567-5769/© 2015 Elsevier B.V. All rights reserved.

agent for inhibiting effects induced by vascular endothelial growth factor (VEGF). Pharmacokinetic (PK) analyses are essential components of the drug discovery and development process [10–11]. Particularly in early clinical trials, it is of great importance to study PK to serve itself as an “auxiliary biomarker”, helpful for the selection of doses in further trials and promotion of clinical rational drug use [12–14]. Traditional PK models including compartment models and Michaelis–Menten (M–M) model were widely used in the PK analyses of various drugs, especially in small-molecule drugs [15–16]. They provided a convenient way to character drugs' properties with model parameters, such as half-life (T1/2), maximum concentration (Cmax), area under the concentration–time curve (AUC), clearance (CL), and so on. However, much more complex PK properties were showed by mAbs compared to those typically associated with small-molecule drugs, and traditional PK models failed to well describe the complex PK process of mAbs. For instance, complex nonlinear PK was encountered frequently by mAbs [17]. The drug exposure or responses did not proportionally vary with dose increasing, and PK parameters such as CL and apparent volume of distribution (V) were not constant in different dosage. Moreover, PK parameters derived from a compartment model fitting and/or an M–M model fitting were significantly varied among patients, in the case they were hard to provide useful information for personalized medicine [18]. For this reason, some new models have been proposed to fit the nonlinear PK of mAbs in recent years. Mager's group has made a successful attempt by building a target-mediated drug disposition (TMDD) model from the perspective of an interaction between antibodies and their targets. The

40

L. Wang et al. / International Immunopharmacology 31 (2016) 39–44

TMDD model deduced the final mathematical model of antibodies through making a comprehensive analysis about the kinetics of targets, antibodies and target-antibody complexes [17,18]. However, the TMDD model consisted of numerous parameters and the value of some parameters were only available through other complicated experiments such as fluorescence-activated cell sorting analysis and so on, thus probably limiting its extensive applications in clinical practice [19]. Monoclonal antibody drugs exhibits much more complex PK properties probably due to significant differences from small-molecule drugs not only in their pharmacological mechanism of action but also in their elimination mechanism [18–20]. Compared to small-molecule drugs, renal elimination is relatively unimportant for mAbs, as its large size prevents efficient filtration through the glomerulus. Specific to its targets such as antigen, receptor, and some protein or polypeptide, mAbs will encounter a target-mediated endocytosis as most kinds of endogenous IgG [20]. Moreover, a phagocytosis of lymphocyte and even some incompletely explained pathways further complicate the elimination of mAbs [19]. In this work, a novel PK model based on the complex elimination of mAbs was established, and its mathematical expressions were also put forward. Subsequently, bevacizumab was selected as a model drug and the new PK model was used to fit the bevacizumab plasma concentration data from pharmacokinetic rabbits. 2. Materials and methods 2.1. PK model based on complex drug elimination A novel PK model was constructed based on the complex elimination of mAbs. The model diagram is presented in Fig. 1. When mAbs were firstly administrated, a sharp decrease in concentration occurred due to a specific binding to pre-existing targets. The rest of mAbs were then eliminated through two main pathways. On the one hand, a degradation and metabolism of protein and/or a phagocytosis of lymphocyte could induce mAbs' concentration decrease, due to mAbs were kinds of protein or peptides and they would encounter a general metabolism in vivo like most kinds of proteins or peptides [19,21–22]. This action could happen all the time and it was a concentration-depended process to some extent, so that the apparent elimination rate of this pathway was assumed to be first-order. On the other hand, most targets were endogenous and generated all the time, and a continuous bind with newly generated targets would trigger antibody dependent cellular cytotoxicity (ADCC) and/or complement activity to induce mAbs' another pathway of elimination [23–26]. The rate of this pathway was assumed to be zero-order because targets' zero-order generation was the ratelimiting step in this pathway. A superposition of a first-order process and another zero-order process complicated in vivo elimination of mAbs.

Fig. 1. Model diagram of the novel model used to describe the in vivo PK of mAbs in rabbits receiving an intravenous administration. As administrated intravenously, mAbs (X0(antibody)) firstly go through an initial elimination by binding with the pre-existing targets, thus causing a decrease in its concentration (F). F can be estimated with the difference value between X0(antibody) and remaining mAbs (XA(antibody)). Subsequently, XA(antibody) underwent a combined elimination of one first-order process (K) and another zeroorder process (K0).

The PK models were described by the following equations: C ¼ CAðantibodyÞ e−Kt −


K0 1−e−Kt KV

ð1Þ

Eq. (1) is applied to describe the PK process with an intravenous administration. CA(antibody) represents the maximum drug concentration in the plasma after drug is firstly administrated and occurs a quick binding with pre-existed targets; K represents the first-order elimination rate of mAbs, and it describes the kinetics of the pathway where mAbs eliminated through a phagocytosis of lymphocyte and/or a degradation and metabolism in liver; K0 represents the zero-order elimination rate, describing the kinetics of the process where mAbs eliminate through binding with continuously generated targets, and in fact, it describes the zero-order generation rate of targets for the assumption that mAbs are dominant and newly generated targets are immediately captured; V is the volume of apparent distribution. Actually, parameters CA(antibody) and K0 describe the kinetics of endogenous targets, and parameter K reflects the stability of mAbs in plasma. 2.2. Materials Commercial immunoassay kits (Boster Biotechnology, China) were utilized to detect the plasma vascular endothelial growth factor (VEGF) concentration in rabbits. Ninety-six-well plates (Greiner, Germany) were used for an immunoassay of bevacizumab. Recombinant human VEGF165 (Peprotech, USA) was immobilized on solid phase surface to capture bevacizumab. Bevacizumab (Avastin, 100 mg/4 ml) was obtained from the manufacturer (Genentech, CA, USA). Nonfat dried milk (Dingguo Changsheng Biotechnology, China) was selected as a sealed liquid, and phosphate-buffered saline (PBS) (Dingguo Changsheng Biotechnology, China) with an addition of 0.5% Tween-20 (Damao Chemical Reagent Factory, China) worked as a wash solution. Horseradish peroxidase-goat anti-human IgG (H + L) conjugate (ZSGB-BIO, China) was obtained to detect bevacizumab. Tetramethylbenzidine (TMB) (Solarbio, China) was purchased as a substrate solution, and 1 mol/L hydrogen chloride (Sinopharm Chemical Reagent, China) was prepared in the laboratory to work as TMB stop solution. 2.3. Pharmacokinetics in rabbits The animal studies were approved by the Animal Ethics Committee of the Third Xiangya Hospital of Central South University. All experiments were conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. Twenty New Zealand rabbits with half males and half females, weighing 1.7 to 2.5 kg, were obtained from Slac Jingda Laboratory Animal Co., Ltd. (Changsha, China). All animals were provided with a 12 h light–dark cycle at an ambient temperature of 21–22 °C, and offered standard laboratory diet and water. The rabbits were randomly divided into three groups among which no significant differences existed with respect to body weight and sex, and then received an intravenous infusion treatment of 15 mg/kg (group A), 5 mg/kg (group B), and 1 mg/kg (group C) of bevacizumab respectively. 1 ml blood was collected via ear vein before and 0.25, 1, 2, 4, 8, 12, 24, 48, 72, 120, 168, 216, 264, 336, 408, 480, 600, and 720 h after a single-dose administration. All plasma samples were separated following centrifugation at 3500 rpm for 10 min after collection, and then stored at −20 °C until analyzed. 2.4. Analytical method An enzyme-linked immunosorbent assay was utilized to measure bevacizumab concentration as previously described with slight modification [27]. Firstly, Ninety-six-well plates were coated with

L. Wang et al. / International Immunopharmacology 31 (2016) 39–44

recombinant human VEGF165 at a concentration of 1 μg/mL in 50 mmol/ L carbonate buffer, pH 9.6, incubating at 37 °C for 1 h and then overnight at 4 °C (100 μL/well). Secondly, after washing three times with phosphate-buffered saline containing 0.05% Tween-20 (PBST), the wells were blocked with 5% nonfat dried milk/PBST with an incubation at 37 °C for 2 h (200 μL/well). Thirdly, block solution was clean out and aqueous plasma diluted in 1% nonfat dried milk/PBST was added to the plates with a 1-hour incubation at 37 °C (50 μL/well). Fourthly, the wells were washed five times with PBST and then bevacizumab was detected by horseradish peroxidase goat anti-human IgG (H + L) conjugate with a concentration of 1 μg/mL after incubating at 37 °C for 1 h. Lastly, after washing wells with PBST five times, 100 μL tetramethylbenzidine substrates (3,3′,5,5′-tetramethylbenzidine substrate) was add to each well to perform color development, and the reaction was finally stopped by the addition of 1 mol/L sulfonic acid (100 μL). Optical density was measured at 450 nm with correction at 570 nm. This assay measures the free bevacizumab, and all measurements were performed twice. A standard curve was prepared with bevacizumab ranging from 25 to 800 pg/mL. The plasma concentration of VEGF before administration was measured with commercial immunoassay kits according to the manufacturer's protocol. The optical density was determined at 450 nm with the absorption spectrophotometer with the correction wavelength set at 570 nm. VEGF concentration in each sample was measured twice. A standard curve was prepared, with VEGF ranging from 20 to 640 pg/mL.

41

increasing bevacizumab dose when dose rate was low, further suggesting a nonlinear elimination of bevacizumab from the plasma. However, as the dose rate increased, the Cmax and AUC seemed to proportionally increase, in contrast with the data of medium and high dose groups. Overall, the results of non-compartmental analysis indicated a nonlinear elimination of bevacizumab in rabbits.

3.2. M–M model fitting The M–M model has been frequently used to describe the nonlinear PK of such drugs whose distribution and/or clearance are affected by its target due to high binding affinity and limited capacity. Bevacizumab had exhibited high binding affinity with its specific target (VEGF) in plasma, thus its PK data was tried to fit with an M–M model to explore whether the nonlinear PK just derived from its target's limited capacity. The M–M model was described by the following equation:

−

dC Vm  C ¼ : dt Km þ C

ð2Þ

The Lineweaver–Burk equation is obtained through a mathematical transformation, which replaces instantaneous velocity dC/dt with average velocity ΔC/Δt, and instantaneous concentration C with the average concentration Cmean at an interval:

2.5. PK modeling and parameters analyses The AUC of different groups were calculated according to a noncompartment model using DAS 2.0 (provided by Clinical drug evaluation center of Anhui province, China) software. The M–M model was also utilized to fit the rabbit PK data for a comparison. Compartment models and our novel model were used to fit the rabbit PK data using DAS 2.0 and Matlab 7.0 (MathWorks, USA) software respectively, and according weighted residual sum of squares were obtained from the mold fitting outputs. The reciprocal of model-predicted concentration was selected as the weighted factor and weighted residual sum of squares were set as the evaluation index in the comparisons of fitting efficiency of both models. The variable coefficient of CA(antibody), K, and K0 were calculated to describe the variations in individuals, and the relationship between K value and drug dose was analyzed by a variance test of the means of K value among different dose groups using SPSS 17.0 (IBM, USA) software. A correlation analysis was also performed between CA(antibody) and pre-existing VEGF concentration to verify a relationship between them. 3. Results

−

1 Km 1 1  : ¼ þ ΔC Vm Cmean Vm Δt

ð3Þ

As Eq. (3) exhibited, there ought to be a linear relationship between 1/(ΔC/Δt) and 1/Cmean. Accordingly, the bevacizumab PK data was transformed and a regression analysis between 1/(ΔC/Δt) and 1/Cmean was made. The results in Fig. 2 demonstrated that 1/(ΔC/Δt) was not linearly related to 1/Cmean, and the relationship between 1/(ΔC/Δt) and 1/Cmean is not as same as the Lineweaver–Burk equation predicted. That is, the M– M model cannot appropriately fit the bevacizumab PK data. The M–M model was to mechanistically explain about the affinity of the ligand towards the target receptors and the receptor abundance, and predict the PK in healthy or disease state population. The failure of the M–M model fitting suggested that target's limited capacity was not the only reason for the nonlinear PK of bevacizumab, and a consideration of a more complex elimination consisted of target-mediated elimination was essential to the PK modeling of bevacizumab and possible other mAbs.

3.1. Noncompartment analysis A noncompartmental analysis was initially performed on the plasma concentration data of bevacizumab at three dose levels. Pharmacokinetic parameters (AUC, Cmax, and MRT) of each rabbit were listed in Table 1. It exhibited a greater than proportional increase in Cmax and AUC with

Table 1 Pharmacokinetic results from the noncompartmental analysis of the bevacizumab plasma concentration data after an intravenous infusion. Parameters

High-dose (15 mg/kg)

Media-dose (5 mg/kg)

Low-dose (1 mg/kg)

AUC0 −

63,362.53 ± 53,672.85 23,420.36 ± 9302.09 327.29 ± 89.88

∞(mg/L∗h)

Cmax(mg/L) MRT (h)

511.21 ± 294.75 188.51 ± 25.52

320.59 ± 128.54 124.97 ± 21.80

4.71 ± 2.05 105.65 ± 21.97

Fig. 2. Evolution of 1/(ΔC/Δt) with 1/Cmean. ΔC/Δt represented the average velocity at every interval, and Cmean stood for the average concentration at the same interval. The concentration data of group A were used for this evolution, and similar results were obtained in other two groups.

42

L. Wang et al. / International Immunopharmacology 31 (2016) 39–44

3.3. Novel model fitting and comparisons with the results of compartmental model fitting A novel PK model based on complex drug elimination was constructed to fit the PK data of bevacizumab in rabbits. Fig. 3 showed the model predicted plotted against observed PK data at different dose levels. As can be seen from the figure, the specified model reasonably described the single-dose PK profiles of bevacizumab in rabbits. In addition, all the results demonstrated that the fitting efficiency of the novel model was greatly improved compared to compartmental models, because its weighted residual sum of squares of bevacizumab PK modeling decreased significantly. Additionally, the weighted residual sum of squares of PK modeling with individual PK data were also obtained and listed in Table 2. As shown in the results, the weighted residual sum of squares of new model were generally decreased, proving that new model predicted profile matched well against observed individual concentration–time profiles of bevacizumab in rabbits. 3.4. Pharmacokinetic parameter analysis The new model contains three important PK parameters including CA(antibody), K, and K0/V. First of all, a correlation analysis between CA(antibody) and pre-existing VEGF concentration was performed to characterize their internal relations. As was illustrated in Fig. 4, CA(antibody) correlated negatively with VEGF concentration. It demonstrated that a high pre-existing VEGF concentration would induce a low concentration of bevacizumab in the first administration, because large amount of bevacizumab had bound with VEGF thus causing a low value of CA(antibody). In other words, CA(antibody) was a valuable kinetic parameter reflecting the pre-existing targets' concentration, which is also called the baseline of targets in clinical practice. Subsequently, the CV of CA(antibody), K, and K0/V were calculated and listed in Table 3. As can be seen from the table, the CV of K was smaller than that derived from CA(antibody) and K0/V. Moreover, a one-way analysis of variance was performed on the mean K values of each dose group, and there was no significant difference among them.

Table 2 Weighted residual sum of squares of model fitting with individual PK data by compartmental model and new model. ID Dose = 15 mg/kg A-3 A-9 A-11 A-13 A-15 A-16 A-18 Dose = 5 mg/kg B-1 B-4 B-12 B-21 B-24 B-26 Dose = 1 mg/kg C-2 C-5 C-8 C-10 C-19 C-22 C-25

Compartmental model

New model

443,595.8 872.5 15,382.0 11,810.2 7435.7 5800.6 101,424.1

127,868.1 743.4 5778.9 5367.3 23,648.9 1913.7 35,358.8

52,657.7 655.1 3073.4 3042.4 4536.0 5464.3

43,253.3 242.2 1372.4 834.0 1502.3 2307.7

0.170 0.298 0.303 0.094 0.059 0.256 1.036

0.315 0.267 0.297 0.655 1.083 0.252 0.856

4. Discussion As we all know, classic compartment models are built based on the hypothesis that the elimination of drug is a first-order process in vivo, which representing a linear profile in the logarithmic concentration– time curve. However, as was shown in Fig. 3, there was an inflection of the logarithmic concentration–time curve in the end elimination of bevacizumab, namely the elimination of bevacizumab in rabbits was not a simple first-order process. It hinted that bevacizumab might not abide by first-order elimination rule as small-molecule drugs. The new model is built in the perspective of that mAbs underwent complex other than simple first-order elimination in vivo. Firstly, an

Fig. 3. Mean model predicted versus observed concentration–time profiles of bevacizumab in rabbits following a single-dose intravenous administration (15 mg/kg (A and a), 5 mg/kg (B and b), and 1 mg/kg (C and c)). Predicted models were one-compartment (A–C) model and the new model (a–c) respectively.

L. Wang et al. / International Immunopharmacology 31 (2016) 39–44

43

Fig. 4. Evolution of the value of CA(antibody) with the pre-existing VEGF concentration. CA(antibody) correlated negatively with VEGF concentration and according correlation coefficients in different dose groups were obtained.

initial elimination affected the bevacizumab PK. As bevacizumab firstly administrated, a moderate decrease in concentration occurred due to a specific bind to pre-existing targets VEGF, thus causing a decrease in drug exposure. Accordingly, there was non-proportional increase in Cmax and AUC with increasing bevacizumab dose, because bevacizumab consumption in the initial elimination distorted the proportional increase of drug exposure in vivo. However, the initial consumption weighted a little to the exposure when bevacizumab was in a high dose level, so that the Cmax and AUC might proportionally increase with dose rising as the results showed in Table 1. In addition to an initial elimination, a complex elimination which consists of one first-order elimination and the other zero-order elimination followed. As exhibited in Fig. 3, there was a turning point at the elimination phase in observed concentration–time profiles of bevacizumab. The inflection of the concentration–time curve in the end elimination of bevacizumab can be well explained by a superposition of a first-order process and a zero-order process. When the concentration of bevacizumab was high, its first-order elimination that bevacizumab underwent a phagocytosis of lymphocyte and/or a degradation and metabolism in liver as a kind of protein, was so fast that it covered the zero-order elimination, due to the first-order elimination rate was related to the concentration [19]. The apparent elimination seemed to be a first-order process, and it exhibited a linear concentration–time profile. As the concentration decreasing, the rate of firstorder elimination slowed while zero-order elimination rate was constant, because bevacizumab's targets (that is VEGF) were generated all the time in a relatively constant rate, thus causing a continuous change in their ratio [25]. The weight of zero-order elimination was increasing and it cannot be ignored when the concentration fell to a relatively low rate. Accordingly, the apparent elimination was deviating from the original first-order process and an inflection in the concentration–time profile appeared. Therefore, no matter the first-order process based compartment models or the saturation process based M–M model failed to reasonably

Table 3 Pharmacokinetic results from the novel model fitting of bevacizumab plasma concentration data in different dose groups. Parameters 15 mg/kg K (1/h) K0/V (kg/h∗L) CA(antibody) (mg/L) 5 mg/kg K (1/h) K0/V (kg/h∗L) CA(antibody) (mg/L) 1 mg/kg K (1/h) K0/V (kg/h∗L) CA(antibody) (mg/L)

Mean

Coefficient of variation

0.00476a 0.0553 284.035

0.786 0.935 0.841

0.00582 0.0432 162.306

0.321 0.303 0.524

0.00634 0.00112 2.560

0.292 0.574 0.460

a A variance test was performed on the mean values of each dose group, and according P value was bigger than 0.01.

describe the non-linear PK properties of bevacizumab because of an incomplete explanation of its elimination. The novel model made comprehensive consideration of complicated elimination pathways so that it achieved a better PK modeling of bevacizumab PK data. The new model contains three important PK parameters including CA(antibody), K, and K0/V. As described above, CA(antibody) represents the maximum drug concentration in the plasma after drug is initially administrated. K represents the first-order elimination rate of mAbs, and it describes the kinetics of the pathway where mAbs eliminate through a general approach regardless of their binding domain. K0 represents the zero-order elimination rate, and in fact, it demonstrates the zeroorder generation rate of targets because mAb is dominant and newly generated targets will be immediately captured. K0/V works as an index of zero-order generation rate of targets in the model fitting. As discussed above, CA(antibody) and K0/V were the parameters actually describing the kinetics of targets. The expression of targets was of individual difference and the disease development will embody in the variation of the kinetics of disease-related targets when individuals were in disease state. Therefore, both CA(antibody) and K0/V would be affected by individual conditions and disease development, thus a significant variation existed among individuals as the results displayed. On the contrary, K described an intrinsic property of drugs to be independent of dose rates and individual conditions, that is, K represents the stability of mAbs in plasma, so that it exhibited little variation among different individuals and dose rates. All of the results demonstrated that the significant variation of mAbs PK in a specified species mainly derived from a disparity of individual target's kinetics including its pre-existing concentration and generating rate, which could be reflected by the value of CA(antibody) and K0/V. The specificity of mAbs mainly derives from their antigen-binding fragments (Fab) which only bind with the complementary structures of specific targets just like one key applied to one lock [28]. Every mAb is specific to one kind of disease-related target, which may be varied in its state such as soluble or cell-bound [29–30]. However, no matter whether the target is soluble in body fluid or bound in cell surface, its binding with mAbs will trigger ADCC and/or complement activity to induce the zero-order elimination of mAbs [23,26]. Additionally, the firstorder elimination of mAbs derives from a general metabolism for its nature of proteins or peptides, and it will be similar among different mAbs in vivo. Moreover, some new pharmacokinetic methods like limited sampling strategy have been proposed to provide reliable and accurate estimation of PK parameters with simplified equations and reduced amount of samples [31]. Therefore, although the new model is only employed to fit with the PK data of bevacizumab, which directed against its soluble target of VEGF, it may be also available for the description of the PK profiles of other mAbs combined with other useful pharmacokinetic methods. 5. Conclusion In summary, the new model has exhibited strong fitting capacity for the nonlinear PK profiles of bevacizumab, on the basis of a successful

44

L. Wang et al. / International Immunopharmacology 31 (2016) 39–44

explanation about its complex elimination consisting of one first-order elimination and the other zero-order elimination. Although some details of the complex mechanism of bevacizumab elimination required a further study, this novel model has manifested its potential as a powerful tool for the PK studies of bevacizumab and other mAbs in practice, and it provides us with a new PK perspective to improve the rational use of mAbs. Conflict of interest

[13]

[14] [15]

[16]

The authors declare no conflict of interest. Acknowledgment

[17]

The work was supported by National Natural Science Foundation of China (81573498).

[18]

References

[20]

[1] G. Kohler, C. Milstein, Continuous cultures of fused cells secreting antibody of predefined specificity, Nature 256 (1975) 495–497. [2] J. Reichert, A. Pavolu, Monoclonal antibodies market, Nat. Rev. Drug Discov. 3 (2004) 383–384. [3] J.M. Reichert, Monoclonal antibodies as innovative therapeutics, Curr. Pharm. Biotechnol. 9 (2008) 423–430. [4] J.M. Reichert, C.J. Rosensweig, L.B. Faden, M.C. Dewitz, Monoclonal antibody successes in the clinic, Nat. Biotechnol. 23 (2005) 1073–1078. [5] I. Brana, R. Berger, T. Golan, P. Haluska, J. Edenfield, J. Fiorica, et al., A parallel-arm phase I trial of the humanised anti-IGF-1R antibody dalotuzumab in combination with the AKT inhibitor MK-2206, the mTOR inhibitor ridaforolimus, or the NOTCH inhibitor MK-0752, in patients with advanced solid tumours, Br. J. Cancer (2014). [6] P. Nair, Anti-interleukin-5 monoclonal antibody to treat severe eosinophilic asthma, N. Engl. J. Med. 371 (2014) 1249–1251. [7] A. Naing, P. Lorusso, S. Fu, D. Hong, H.X. Chen, L.A. Doyle, et al., Insulin growth factor receptor (IGF-1R) antibody cixutumumab combined with the mTOR inhibitor temsirolimus in patients with metastatic adrenocortical carcinoma, Br. J. Cancer 108 (2013) 826–830. [8] M. Javle, E.C. Smyth, I. Chau, Ramucirumab: Successfully Targeting Angiogenesis in Gastric Cancer, Clin Cancer Res, 2014. [9] Z.X. Xuan, L.N. Li, Q. Zhang, C.W. Xu, D.X. Yang, Y. Yuan, et al., Fully human VEGFR2 monoclonal antibody BC001 attenuates tumor angiogenesis and inhibits tumor growth, Int. J. Oncol. (2014). [10] H. Jones, K. Rowland-Yeo, Basic concepts in physiologically based pharmacokinetic modeling in drug discovery and development, CPT Pharmacometrics Syst. Pharmacol. 2 (2013), e63. [11] P.J. Webborn, The role of pharmacokinetic studies in drug discovery: where are we now, how did we get here and where are we going? Future Med. Chem. 6 (2014) 1233–1235. [12] J. Baselga, A.C. Mita, P. Schoffski, H. Dumez, F. Rojo, J. Tabernero, et al., Using pharmacokinetic and pharmacodynamic data in early decision making regarding drug

[21]

[19]

[22]

[23] [24]

[25] [26]

[27]

[28] [29] [30]

[31]

development: a phase I clinical trial evaluating tyrosine kinase inhibitor, AEE788, Clin. Cancer Res. 18 (2012) 6364–6372. J. Maringwa, M. Kagedal, U.W. Hamren, P. Martin, E. Cox, B. Hamren, Pharmacokinetic-pharmacodynamic modelling of fostamatinib efficacy on ACR20 to support dose selection in patients with rheumatoid arthritis (RA), J. Clin. Pharmacol. (2014). N. Cook, A.R. Hansen, L.L. Siu, R.A. Abdul, Early phase clinical trials to identify optimal dosing and safety, Mol. Oncol. (2014). H. Kusuhara, H. Furuie, A. Inano, A. Sunagawa, S. Yamada, C. Wu, et al., Pharmacokinetic interaction study of sulphasalazine in healthy subjects and the impact of curcumin as an in vivo inhibitor of BCRP, Br. J. Pharmacol. 166 (2012) 1793–1803. T. Takahashi, N. Boku, H. Murakami, T. Naito, A. Tsuya, Y. Nakamura, et al., Phase I and pharmacokinetic study of dacomitinib (PF-00299804), an oral irreversible, small molecule inhibitor of human epidermal growth factor receptor-1, -2, and -4 tyrosine kinases, in Japanese patients with advanced solid tumors, Investig. New Drugs 30 (2012) 2352–2363. K.T. Luu, S. Bergqvist, E. Chen, D. Hu-Lowe, E. Kraynov, A model-based approach to predicting the human pharmacokinetics of a monoclonal antibody exhibiting target-mediated drug disposition, J. Pharmacol. Exp. Ther. 341 (2012) 702–708. G. Paintaud, Pharmacokinetics (PK) of mAbs, Med. Sci. (Paris) 25 (2009) 1057–1062. W. Wang, E.Q. Wang, J.P. Balthasar, Monoclonal antibody pharmacokinetics and pharmacodynamics, Clin. Pharmacol. Ther. 84 (2008) 548–558. M.A. Tabrizi, C.M. Tseng, L.K. Roskos, Elimination mechanisms of therapeutic monoclonal antibodies, Drug Discov. Today 11 (2006) 81–88. Y. Vugmeyster, X. Xu, F. Theil, L.A. Khawli, M.W. Leach, Pharmacokinetics and toxicology of therapeutic proteins: advances and challenges, World J. Biol. Chem. 3 (2012) 73. I.H. Patel, X. Zhang, K. Nieforth, M. Salgo, N. Buss, Pharmacokinetics, pharmacodynamics and drug interaction potential of enfuvirtide, Clin. Pharmacokinet. 44 (2005) 175–186. L.M. Rogers, S. Veeramani, G.J. Weiner, Complement in monoclonal antibody therapy of cancer, Immunol. Res. (2014) 1–8. R. Dienstmann, J. Tabernero, Necitumumab, a fully human IgG1 mAb directed against the EGFR for the potential treatment of cancer, Curr. Opin. Investig. Drugs 11 (2010) 1434–1441. N. Ferrara, H. Gerber, J. LeCouter, The biology of VEGF and its receptors, Nat. Med. 9 (2003) 669–676. S.C. Lee, R.M. Srivastava, A. López-Albaitero, S. Ferrone, R.L. Ferris, Natural killer (NK): dendritic cell (DC) cross talk induced by therapeutic monoclonal antibody triggers tumor antigen-specific T cell immunity, Immunol. Res. 50 (2011) 248–254. T. Miyake, O. Sawada, M. Kakinoki, T. Sawada, H. Kawamura, K. Ogasawara, et al., Pharmacokinetics of bevacizumab and its effect on vascular endothelial growth factor after intravitreal injection of bevacizumab in macaque eyes, Invest. Ophthalmol. Vis. Sci. 51 (2010) 1606–1608. W. Wang, S. Singh, D.L. Zeng, K. King, S. Nema, Antibody structure, instability, and formulation, J. Pharm. Sci. 96 (2007) 1–26. B.A. Teicher, Targets in small cell lung cancer, Biochem. Pharmacol. 87 (2014) 211–219. N. Papadopoulos, J. Martin, Q. Ruan, A. Rafique, M.P. Rosconi, E. Shi, et al., Binding and neutralization of vascular endothelial growth factor (VEGF) and related ligands by VEGF Trap, ranibizumab and bevacizumab, Angiogenesis 15 (2012) 171–185. D. Capone, G. Tarantino, I. Kadilli, G. Polichetti, V. Basile, S. Federico, et al., Evalutation of mycophenolic acid systemic exposure by limited sampling strategy in kidney transplant recipients receiving enteric-coated mycophenolate sodium (EC-MPS) and cyclosporine, Nephrol. Dial. Transplant. 26 (2011) 3019–3025.