QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: dmd.107.019182v1 36/6/1153    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Fang, L. Articles by Sun, D. Search for Related Content PubMed PubMed Citation Articles by Fang, L. Articles by Sun, D.

Predictive Physiologically Based Pharmacokinetic Model for Antibody-Directed Enzyme Prodrug Therapy

Division of Pharmaceutics, College of Pharmacy, the Ohio State University, Columbus, Ohio

(Received October 10, 2007; Accepted March 19, 2008)

	Abstract

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 
Antibody-directed enzyme prodrug therapy (ADEPT) using anti-TAG-72 antibody and geldanamycin (GA) prodrug were validated in vitro. To understand the complexity and to explore optimal therapeutic regimens for ADEPT in vivo, a physiologically based pharmacokinetic model (PBPK) is applied to analyze each anatomical component/organ. The baseline model predicts that active drug tumor/plasma exposure (AUC) ratio is 2-fold, although antibody-enzyme conjugates (AbE) are distributed into tumors up to 9-fold higher than in plasma. However, the active drug tumor/plasma AUC ratio can be increased up to 100-fold when AbE are depleted from plasma. Similarly, the active drug tumor/plasma AUC ratio can be increased from 2- to 6-fold when the intrinsic clearance of AbE is accelerated by 10-fold. Several sensitive parameters are identified: 1) increasing flow inside tumor (J_iso,tumor) significantly increases active drug tumor/plasma AUC ratio; 2) increasing permeability of prodrug (from range 1.4 x 10^–6 to 1.4 x 10^–4 cm/s) increases active drug tumor/plasma AUC ratio significantly, whereas active drug permeability enhancement (from range 5 x 10^–4 to 5 x 10^–2 cm/s) has minimal effect; 3) decreasing E_max and increasing EC₅₀ for converting prodrug to active drug increase tumor/plasma AUC ratio for active drug. The PBPK model predicts that the optimal dosing interval between AbE and prodrug administration is 5 days, the optimal AbE dose is 0.1 B_max, and the optimal dose for GA prodrug is 60 mg/kg. The current PBPK model successfully identifies sensitive parameters and predicts an optimal dosing regimen for ADEPT.

This is a rather complex therapeutic regimen considering the following facts (Xu and McLeod, 2001). First, at least two components or steps are indicated in ADEPT, (i.e., AbE should be specifically delivered to the desired tumor site and the nontoxic prodrug should be activated in the target site by the predelivered enzyme). Increasing therapeutic steps always makes it harder to control and predict the outcome, so better understanding of the therapy itself and optimal design are necessary. Second, high expression level of antigen in tumors but low level in normal tissues is required for AbE tumor targeting. Third, AbE might not penetrate into the target tissues. Fourth, elimination of the biological molecules (AbE) might be nonlinear. Fifth, the prodrug needs to be enzyme-specifically activated with minimal nonspecific activation. Large difference of cytotoxicity and other properties between the prodrug and activated drug is desired. For example, low permeability to the cell membrane is preferred for the prodrug, whereas high permeability is preferred for the activated drug (Syrigos and Epenetos, 1999).

Because of the complexity of ADEPT, it would be ideal to predict the therapeutic outcome before the large-scale lengthy and costly preclinical and clinical studies are undertaken. Application of pharmacokinetics (PK) and pharmacodynamics (PD) studies has been shown to guide and expedite drug development (Galluppi et al., 2001). The urgency of PK/PD modeling is clearly demonstrated in U.S. Food and Drug Administration guidance (http://www.fda.gov/cber/gdlns/popharm.htm). It is without question that using mathematical modeling and deconstructing the complex system into elementary components will help us understand the complex ADEPT and optimize the therapeutic regimen. Few articles have been published that used the mathematical modeling method to provide insight and guidance for ADEPT therapy (Yuan et al., 1991; Baxter and Jain, 1996; Varner, 2005). However, these studies assumed simple compartment model for AbE, prodrug, and active drug. Although informative, the compartment model is very limited in its ability to describe the pharmacokinetic properties of biological molecules. PBPK has been proposed to predict the pharmacokinetics of antibodies itself (Baxter et al., 1994; Baxter et al., 1995; Friedrich et al., 2002; Ferl et al., 2005). PBPK approach is more physiologically relevant and allows for integration of all anatomical compartments together as they are in the biological system. This will enable the accurate predication of the system change with each compartment variation. The current study intends to apply PBPK modeling to simulate and predict the outcome of ADEPT. PBPK modeling will be able to incorporate special features of ADEPT components, such as convection, which is more likely to contribute to the AbE distribution process, FcRn, which is important for antibody recycling and elimination, and tumor size, which might change significantly during the ADEPT course.

In our previous study, we used anti-TAG-72 (tumor-associated glycoprotein 72) antibody (HuCC49CH2) to deliver a drug-activating enzyme for GA-prodrug activation in ADEPT (Cheng et al., 2005; Fang et al., 2006). It was shown that HuCC49CH2-β-galactosidase conjugates were highly specific and enzymatically active. The conjugates were demonstrated to activate the GA prodrug and reduce the IC₅₀ of prodrug to 1 µM from 25 µM against LS174T, colorectal cancer cells with high levels of TAG-72 expression. Several important factors contribute to the advantages of using the anti-TAG-72 antibody and GA prodrug. First, the TAG-72 antigen is primarily expressed in tumors with almost no detectable expression in normal tissues. Second, the humanized anti-TAG-72 antibody (HuCC49CH2) was used for tumor detection in human clinical trials and showed no immunological response. Third, HuCC49CH2 was shown to localize in the extracellular matrix. The large amount of antigen in the extracellular space may trap a large amount of AbE (Knox RJ, 1999). Unlike other antibodies against tumor cell membrane antigen, HuCC49CH2 may avoid the problem of internalization of AbE for lysosomal degradation. Fourth, the GA prodrug with a sugar moiety is more polar and less membrane permeable than the activated drug with high lipophilicity (Tietze et al., 2002).

The current study aims to use the PBPK approach to understand the complexity of ADEPT therapy and to guide AbE selection, AbE dose regimen (dose and dosing interval before prodrug administration), prodrug design, and prodrug dose. The objectives are the following: 1) to build three levels of PBPK model for AbE, prodrug, and active drug and to provide a quantitative basis to carry out ADEPT therapy in vivo; 2) to identify the sensitive factors or parameters controlling the ADEPT process including antigen expression in tumor tissues, antibody affinity to antigen, clearance and depletion of AbE, and prodrug and active drug permeability; and 3) dosing regimen optimization for AbE and prodrug administration to achieve maximal active drug concentration in tumors and minimal active drug in blood and normal tissues.

	Materials and Methods

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 
Model Development. This model was built based on the LS174T colorectal cancer cell xenograft mouse model. The model was developed to study the distribution and targeting of AbE and the activation of prodrug to active drug based on previously published models (Baxter et al., 1994; Baxter et al., 1995; Ferl et al., 2005). A whole-body physiologically based model is used in which the major organs of the mice are represented as compartments. Each compartment is connected in an anatomical manner by the systemic and lymphatic circulation (Fig. 1). The organ compartments are then further subdivided into vascular and extravascular subcompartments.

View larger version (20K): [in this window] [in a new window]   FIG. 1. Basic PBPK structure model for ADEPT therapy in LS174T colorectal caner xenograft model.

The main transport processes for prodrug and active drug model include the following: 1) transport of prodrug and active drug is diffusion-limited across the endothelial wall of vascular and extravascular spaces; 2) the metabolism of prodrug and active drug is predominantly through liver; 3) the connection among AbE, prodrug, and active drug is that the predisposed AbE convert prodrug to active drug in each specified organ subcompartment.

The tumor mass was found to be highly variable based on the previous studies. Thus, a variable tumor mass submodel was adopted to reflect the real tumor physiology (Ferl et al., 2005). This phenomenon was also confirmed in rapidly growing LS174T xenograft model based on our own experience. The model nomenclature is given in Appendix A and mathematical equations accounting for the mass balance are presented in Appendix B, respectively.

Model Baseline Parameters. All the widely accepted physiological parameters for mice are from previous publications (Baxter et al., 1995; Baxter and Jain, 1996; Ferl et al., 2005). These physiological parameters are summarized in Appendix A. For AbE-related parameters, the baseline values of all parameters are obtained or estimated from the results of Ferl et al. (2005). Considering that the values of these constants may vary with different antibodies and selected enzyme and tumor cell lines, we arbitrarily chose baseline values based on literature values and our own experimental results for simulation studies. For prodrug and active drug, most parameters are from the literature and our own experimental data for GA prodrug (Fang et al., 2006). The effective permeability coefficients (P_A) of prodrug and active drug are assumed to be very different based on the high hydrophilic prodrug moiety and hydrophobic properties for active drug. This assumption is based on the fact that active drug needs to be more permeable than prodrug in order to penetrate into the tumor tissues in ADEPT system. All baseline values used are summarized in Table 1.

Time and Dosing Schedules of Simulation Studies. In the sensitivity analysis, the time interval between antibody-enzyme conjugate (AbE) and prodrug is fixed at 3 days based on the following rationale. First, antibody alone distributed into tumor tissues at the highest level at 3 days after antibody injection in the previous publication (Ferl et al., 2005). Second, this fixed interval allows us to study the effect of other sensitive parameters separately (such as antigen expression levels, AbE affinity to antigen, clearance and depletion of AbE, AbE enzymatic activity, and prodrug permeability). Both AbE and prodrug are administered as a single bolus dose injection in this simulation.

The baseline dose of a single bolus AbE injection was chosen to be 1/10 of the maximum binding capacity of the tumor tissues (B_max). This selection was based on the result from a previous study (Yuan et al., 1991). The baseline dose of prodrug injection was chosen to allow the active drug concentration above IC₅₀ in the tumor tissues.

Model Simulation and Simulation Criteria. Three factors are critical for ADEPT to succeed: 1) active drug tumor/plasma exposure (AUC) ratio, 2) minimum effective active drug tumor AUC, and 3) the absolute tolerable plasma AUC for active drug. As indicated in Varner's study (Varner, 2005), a tumor-to-plasma ratio of 10 to 1 for active drug concentration or AUC is sufficient for ADEPT efficacy, whereas the absolute tumor and tolerable plasma exposure will depend on the specific active drug's toxicity and efficacy. In the current study, we set the absolute active drug plasma concentration below IC₅₀ (which is 1 µM for activated GA prodrug against colon cancer cells LS174T) in plasma and normal tissues and tumor active drug concentration above 1 µM. The reason for setting IC₅₀ rather than exposure is because IC₅₀ is an easier parameter to get in the early stage of preclinical study when PBPK modeling is actually needed to guide further animal or human study design.

Simulation Software. Berkeley Madonna version 8.3.9 (http://www.berkeleymadonna.com) was used in the current simulation study. Rosenbrock (stiff) was the selected integration method.

	Results

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 
AbE, Prodrug, and Active Drug Distribution with the Baseline Parameter Assumption. We first analyzed the distribution of AbE, prodrug, and active drug with the baseline physiological, biological, and physical parameters (Fig. 2). As indicated in Fig. 2, A and B, the tumor/plasma AbE ratio rises rapidly before day 4, followed by a slowly rising phase up to day 8. Therefore, the selected baseline interval between AbE and prodrug injection is 3 days.

View larger version (22K): [in this window] [in a new window]   FIG. 2. Antibody-enzyme conjugate, prodrug, and active drug distribution into tumor and plasma upon baseline parameters. A, AbE concentration (percentage of injected dose per gram tissue) in the tumor (solid line) and plasma (dotted line). B, tumor/plasma AbE concentration ratio. C, prodrug concentration in the tumor (solid line) and plasma (dotted line). D, tumor/plasma concentration (dotted line) and AUC (solid line) ratio of active drug.

Depletion of AbE from Plasma on Day 3, when the Prodrug Is Injected, Dramatically Increased Tumor/Plasma Concentration Ratio of the Active Drug. It was reported previously that tumor/plasma ratio of AbE exceeded 10,000:1 in a clinical trial study if a second antibody against AbE conjugate was applied to increase AbE clearance from plasma (Napier et al., 2000). This high ratio of tumor/plasma or tumor/normal tissues for AbE level provided a favorable environment for prodrug activation in tumors in theory.

However, it is not clear how this high tumor/plasma ratio of AbE will affect the active drug concentration in tumors and tumor/plasma ratio. Therefore, we used the PBPK model to validate the effect on the active drug concentration in the tumor and plasma by removing AbE from the plasma by a secondary clearing antibody. On day 3, when prodrug is injected into the system, the AbE concentration in the plasma is set to 0 to reflect the removal. As shown in Fig. 3A, active drug concentration in the plasma is dramatically reduced but not to zero when AbE in the plasma is removed. This is different from the previous study in which no conversion of prodrug to active drug was assumed in the plasma and, thus, the plasma concentration of the active drug is set close to zero (Yuan et al., 1991). Our simulated result is considered more reasonable since AbE, prodrug, and active drug can all be collected back into the circulation from organs such as liver, kidney, lung, and tumor physiologically. Our PBPK model in this study can reflect this phenomenon: the plasma active drug concentration peak time is delayed when AbE is removed, indicating that the active drug in other organs can "leak" back into the plasma.

View larger version (19K): [in this window] [in a new window]   FIG. 3. The effect of AbE depletion on the active drug tumor and plasma concentration. A, active drug concentration in the plasma when the plasma AbE is removed at day 3 (dotted line) compared with AbE that remained in the plasma (solid line). B, active drug concentration in the tumor when the plasma AbE is removed at day 3 (dotted line) compared with AbE that remained in the plasma (solid line). C, active drug concentration in the tumor (solid line) and plasma (dotted line) when the plasma AbE is removed at day 3.

Acceleration of AbE Intrinsic Clearance Significantly Increased Tumor/Plasma AUC Ratio of the Active Drug. Although a second antibody can deplete AbE from the plasma and increase the tumor/plasma ratio of active drug concentration, the second antibody may cause immune response and make ADEPT harder to control. Therefore, another approach is to increase AbE intrinsic clearance to accelerate removal of AbE from blood circulation (and normal tissues/organs) and reduce residual conjugate activity. We have studied an engineered humanized antibody HuCC49CH2 with clearance of 1.5 x 10^–3 ml/min (Fang et al., 2007). The high clearance of this humanized antibody was due to the deletion of CH2 domain and glycosylation site.

However, increasing the clearance of AbE might also reduce the uptake and retention of AbE in tumor tissues since the remained AbE inside the body is reduced. Therefore, a delicate balance must be established in order to maximize the absolute tumor retention and tumor/plasma ratio of AbE. PKPB model was used to simulate the effect of increasing AbE clearance on active drug concentration in tumors and plasma and their ratios.

As predicted, increased AbE clearance from circulation leads to reduced AbE concentration (or distribution) in tumor and plasma (Fig. 4, A and B). However, the AUC ratio of active drug in tumor/plasma is significantly increased when intrinsic clearance of AbE is increased from 1.5 x 10^–5 to 1.5 x 10^–3 ml/min (Fig. 4C). It is worth noting the nonlinear distribution of AbE in tumor in the PBPK model. AbE distribute into tumor or other tissues in a nonlinear manner because of the capacity-limited binding process. Because of the distribution equilibrium between tumor (or other tissues) and circulation, the nonlinear distribution of AbE in tumor (or other organs) causes nonlinear removal from circulation. Consequently, when intrinsic clearance of AbE is increased from 1.5 x 10^–5 to 1.5 x 10^–4 ml/min, AbE concentration in the plasma is reduced to a small extent (Fig. 4B). However, when clearance is increased from 1.5 x 10^–4 to 1.5 x 10^–3 ml/min, the AbE concentration in the plasma decreased significantly (Fig. 4B). Therefore, the nonlinear distribution may have a "buffering" effect when AbE clearance is increased from 1.5 x 10^–5 to 1.5 x 10^–4 ml/min, whereas this "buffering" effect is saturated when AbE clearance is further increased to 1.5 x 10^–3 ml/min. Overall, the active drug tumor/plasma AUC ratio increases up to over 6-fold when AbE clearance is increased to 1.5 x 10^–3 ml/min.

View larger version (18K): [in this window] [in a new window]   FIG. 4. The effect of AbE clearance acceleration on the AbE and active drug distribution in the tumor and plasma. A, tumor AbE concentration at different clearance. B, plasma AbE concentration at different clearance. C, active drug tumor/plasma AUC ratio with changing AbE clearance.

Parameter Sensitivity Analysis in PBPK Modeling of ADEPT. For complex ADEPT therapy, it is beneficial to know the effect of each unknown or estimated parameter on the model for the dose regimen optimization. These parameters include AbE enzymatic activity, AbE binding affinity to antigen, permeability of prodrug and active drug, and tumor physiological conditions. Since AUC is a more relevant parameter to measure the therapeutic efficacy of active drug in the tumor, tumor/plasma AUC ratio for the active drug was used in the following study.

As shown in Fig. 5, a decrease of the enzymatic activity (E_max) of AbE for prodrug conversion rate from baseline 6000 min^–1 to 600 min^–1, the tumor/plasma AUC ratio of active drug increases from 2- to 3-fold. Further decrease of E_max has minimal effect (Fig. 5A). This is consistent with the previous finding that the prodrug concentrations in both tumor and plasma are constant when E_max is very low, thus the prodrug conversion is limited by the AbE concentration in the individual organ (Yuan et al., 1991). Consequently, initial reduction of E_max leads to increased tumor/plasma AUC ratio, whereas further reduction of E_max has no effect. However, the effect of EC₅₀ on the tumor/plasma AUC ratio of active drug is opposite that of E_max. Initial increase of EC₅₀ causes increase of active drug tumor/plasma AUC ratio, whereas further increase has minimal effect (Fig. 5B).

View larger version (33K): [in this window] [in a new window]   FIG. 5. Parameter sensitivity analysis on Emax (A) and EC50 (B) of AbE for prodrug conversion; permeability coefficient of prodrug (C) and active drug (D); and inside tumor flow (E).

To best optimize ADEPT, the permeability of active drug should be larger than that of prodrug since it is desired that active drug can penetrate into the tumors in depth. To best define the permeability range of the prodrug and active drug, we tested the sensitivity of permeability in this PBPK model.

For the defined prodrug permeability range (1.4 x 10^–6 cm/s to 1.4 x 10^–4 cm/s), increasing the prodrug permeability has a significant effect on the active drug tumor/plasma AUC ratio (Fig. 5, C and D). As predicted, increasing the prodrug's permeability coefficient from 1.4 x 10^–6 cm/s to 1.4 x 10^–4 cm/s consistently increases the active drug tumor/plasma AUC ratio, especially in the first 10 min. This can be explained by the fact that enhancement of the prodrug permeability can increase the penetration of prodrug into the organs and tumor and increase the prodrug concentration in the organs and tumor, thus reducing the prodrug concentration in the plasma. As a consequence, the prodrug conversion is increased in the tumor while it is decreased in the plasma. Therefore, a high prodrug permeability should be preferred in the allowed permeability range (1.4 x 10^–6 cm/s to 1.4 x 10–⁴ cm/s). However, for the defined active drug permeability range (5 x 10^–4 cm/s to 5 x 10^–2 cm/s), enhancement of the permeability coefficient beyond 5 x 10^–3 cm/s has little effect. This may be due to the fact that significantly high permeability of the active drug can increase the leak of active drug from tumor to plasma, and thus active drug tumor/plasma AUC ratio cannot be further enhanced. Therefore, there is no need to increase active drug permeability to beyond 5 x 10^–3 cm/s.

Tumor physiology may also significantly affect the outcome of ADEPT therapy. Fluid recirculation flow rate between vascular and interstitial compartment of tumor (J_iso,tumor) is a parameter to evaluate pressure inside the tumor. When the pressure inside the tumor is very high, J_iso,tumor is very small. On the other hand, when the tumor is reasonably small and the inside pressure is still low, the blood supply to the tumor is fairly large and J_iso,tumor is much bigger. As seen in Fig. 5E, the enhancement of J_iso,tumor can significantly increase active drug tumor/plasma AUC ratio. This suggests that the PBPK model is valid in predicting that smaller tumors have much better therapeutic outcome after ADEPT treatment.

The Determination of Time Interval between AbE and Prodrug Administration. The PBPK model showed that active drug tumor/plasma AUC ratio increases as a result of increasing the AbE and prodrug dosing interval up to 5 days. When the interval is longer than 5 days, the AUC ratio decreases with increasing interval time (Fig. 6). Thus, the most optimal dosing interval of AbE and prodrug is 5 days in the current study. This is to the contrary of the previous finding that increasing the dosing interval between AbE and prodrug administration may increase active drug tumor/plasma concentration ratio consistently (Yuan et al., 1991). It was shown in the previous study that the concentration ratio increased from under 1 (interval = 0 day) to above 10 (interval = 14 days). The current PBPK model is believed to provide more accurate prediction. In the baseline parameters, the trapping of the active drug by the tumor cells was not considered (Kp = 1). Thus, as discussed previously, the only way to increase the active drug tumor/plasma AUC or concentration ratio is to increase the difference of the prodrug conversion (through AbE) between the tumor and plasma (Yuan et al., 1991). Meanwhile, the absolute tumor AbE concentration is also very critical. The actual tumor AbE concentration cannot be too low; otherwise, the effective tumor prodrug conversion is so slow that the advantage is diminished even though the tumor/plasma AbE concentration ratio might be high. The best dosing interval should be balanced between the tumor/plasma AbE concentration ratio and the absolute tumor AbE concentration.

View larger version (20K): [in this window] [in a new window]   FIG. 6. Active drug tumor/plasma AUC ratio when prodrug is injected at days 1, 3, 5, and 8 after AbE injection. Time interval of 5 gives the highest tumor/plasma AUC ratio of active drug.

As shown, the tumor localized AbE is highest around day 4 (Fig. 2A), and the tumor/plasma AbE concentration ratio is continuously increasing beyond day 4 after AbE injection (Fig. 2B). Since the absolute concentration of AbE in tumor becomes lower after day 5, it is reasonable to expect that an optimal dosing interval of 5 days between AbE and prodrug is predicted by the current PBPK model.

ADEPT Based on the Most Optimal Parameters. The most optimal parameters from the sensitivity analysis replace the baseline values in the developed PBPK model for ADEPT. The combination of increased AbE clearance (1.5 x 10^–3 ml/min) and prolonged interval (Tau = 5 days) gives the highest tumor/plasma AUC ratio of the active drug. Thus, the model with these two newly identified parameters is the selected optimal PBPK model of ADEPT system. Although there are other sensible parameters, replacement of them into the optimal system will have only minimal effect. As shown in Fig. 7A, tumor/plasma AUC ratio for the active drug is increased up to 10-fold in the optimal model compared with 2-fold using the baseline parameters. In addition, the absolute tumor concentration for the active drug is above the preset IC₅₀ (1 µM), and the plasma active drug concentration is below 1 µM (Fig. 7B).

View larger version (18K): [in this window] [in a new window]   FIG. 7. ADEPT based on the most optimal parameters. Active drug tumor/plasma AUC ratio is depicted in (A), and absolute active drug concentration in the tumor (solid line) and plasma (dotted line) are represented in (B).

The model shows that increasing AbE dose from 1 x 10^–11 mols (0.1 B_max) to 1 x 10^–10 mols (1 B_max) reduces the tumor/plasma AUC ratio of active drug and increases both tumor and plasma active drug concentrations above 1 µM. On the other hand, reducing AbE dose from 1 x 10^–11 mols (0.1 B_max) to 1 x 10^–12 mols (0.01 B_max) increases the tumor/plasma AUC ratio, but tumor active drug concentration is below 1 µM. Since 1 µM is the required concentration for active drug to achieve efficacy against cancer cells, the optimal dose for AbE is 1 x 10^–11 mols (0.1 B_max) (Fig. 8, A–C).

View larger version (42K): [in this window] [in a new window]   FIG. 8. Dose selection for AbE (A–C) and prodrug (D–F).

	Discussion

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 
Irrespective of the expected complexity, ADEPT therapy has been widely tested by many research groups. Previous studies established systemic approaches to screen effective ADEPT therapeutic regimen from in vitro to in vivo and further to human clinical trials. In the early clinical trials from 1997 to 2000, limited colorectal cancer patients were enrolled in the study (Martin et al., 1997; Napier et al., 2000). A5B7 F(ab')₂ antibody (against carcinoembryonic antigen) conjugated to carboxypeptidase G2 was used in the clinical study. To accelerate the clearance of AbE from blood circulation, a clearing antibody SB43-gal (against the active site of carboxypeptidase G2) was applied. The high tumor/plasma ratio of AbE (exceeding 10,000:1) provided an advantage for prodrug administration around 50 h. Benzoic acid mustard-glutamate, which can be converted to the alkylating agent, was the selected prodrug. However, although there was evidence of tumor response by the active drug, the dosing-limiting myelosuppression was observed for the majority of the patients. This myelosuppression was attributed to the presence of active drug in the plasma since prodrug itself was not shown to cause toxicity in their previous study. It was suggested that active drug in the circulation was the result of "leakback" from tumor given that no active enzyme was found in plasma.

In 2002, a new phase I clinical trial using a new prodrug and galactosylated AbE was undertaken (Francis et al., 2002). The galactosylated AbE was expected to be cleared faster from circulation without the help of a second clearing antibody. The elegantly designed prodrug was converted to highly potent active drug with a short half-life, which was expected to prevent the leakback of active drug to the circulation observed in the previous clinical trial. The result of this clinical trial was rather unexpected. Inadequate tumor localization of AbE (median tumor/normal tissue ratios are less than 1) was observed. Therefore, selective prodrug conversion in the tumor tissue was not observed for almost all the patients except one. The study concluded that each component (or step) of ADEPT must be strictly monitored in order to achieve the optimal therapeutic outcome. AbE with more efficient tumor localization and less immunogenicity was suggested if second clearing antibody was not used.

One critical study was published in 2005 (Sharma et al., 2005). In this study, a recombinant genetic fusion protein, composed of a single-chain Fv antibody and an enzyme, was expressed in Pichia pastoris, and a glycosylated protein with branched mannose resulted. The study successfully confirmed the hypothesis in two morphologically different human colon carcinoma xenografts (LS174T and SW1222) that mannosylated fusion protein's clearance from the circulation was accelerated (rapid clearance from plasma within 6 h), via mannose receptors, without inducing toxicity. Also, the high AbE tumor-to-plasma ratios of 1400:1 and 339:1 were observed for LS174T and SW1222 models, respectively. In 2006, the clinical trial study result was published (Mayer et al., 2006). Eighty-fold dose escalation from the starting dose of prodrug was carried out, and the therapeutic benefit was evaluated.

Although ADEPT was demonstrated to have great potential, improvement on the optimal conditions for better outcome is necessary. From the long development journey, it was suggested that efficient approach to expedite the development process was desired. The use of large molecular proteins (such as AbE) introduces much more complexity to predict the outcome of ADEPT. The PK/PD properties of therapeutic proteins, such as monoclonal antibodies, are far more complicated than small molecular weight therapeutic agents. Specifically, the absorption is very variable, thus most antibodies on the market are administered intravenously and subcutaneously; the distribution is often mediated by convection and receptor-mediated transport; there is capacity-limited binding between antigen and antibody; and the elimination is usually expected to be nonlinear. Due to the expected complexity, the best approach to predict PK properties of biological therapeutic agents is PBPK. Many advantages are provided by PBPK modeling: 1) it allows integration of many physiological, physical, and biological parameters and translation of them into the in vivo therapeutic efficacy prediction; 2) it estimates the tissue-specific concentration changes as the function of time; 3) it predicts the kinetic changes by variations of physiological parameters, and thus 4) it enables the dose regimen scale-up from animal models to future clinical trials (Zhu et al., 1997).

In our current study, we successfully constructed PBPK simulation models for AbE, prodrug, and active drug. All three molecules' pharmacokinetics were interconnected through prodrug activation to active drug by the pretargeted AbE. We believe this PBPK model is advantageous compared with the previously published compartmental model in the following aspects. First, the current PBPK model is more physiologically relevant. Our PBPK model integrates all the tissues/organs into the system as they are in the biological body. Thus they reflect the real biological system much better than the compartmental model. For example, the previous compartmental model does not consider the flow collection back from normal organs into the circulation, whereas this is captured in the current PBPK model (Yuan et al., 1991). Indeed, this phenomenon is confirmed by the previously published clinical trial study, the leakback of active drug from regular organs or tumor to the circulation was the main reason for the observed toxicity (Napier et al., 2000). If PBPK model was introduced before this clinical study, the leakback issue would be captured and understood better. Second, the model captures and predicts individual organ/tissue concentration-time response based on each organ's characteristics. For example, the skin and muscle are unique compared with other organs in the current study since FcRn binding sites were found to be in these organs. It will be difficult for previous compartmental models to incorporate those unique features. Thus the current PBPK model is much more flexible in this aspect.

It is also important to note that further refinement of the current PBPK model is necessary. For example, active drug trapping in the tissues is not considered; however, it is more likely that the active drug can be trapped in the tumor cells. It was previously reported that the active metabolite of 17-AG was trapped in the tumor more than the parent compound 17-AAG itself (Xu et al., 2003). It is reasonable to expect that active drug would behave similarly since active drug may bind to its target inside the tumor cells. Consequently, the active drug tumor/plasma AUC ratio might be even higher in the real situation. In addition, FcRn binding sites can be extended to other organs in addition to skin and muscle. In the current PBPK model, only skin and muscle are considered to contain FcRn binding sites according to a previous study (Ferl et al., 2005).

In our previous study, we successfully synthesized GA prodrugs suitable for ADEPT therapy (Cheng et al., 2005; Fang et al., 2006). In addition, humanized anti-TAG-72 antibody HuCC49CH2 and β-galactosidase were chemically conjugated and tested for specific activation of GA prodrugs. We expect the current PBPK modeling study will shed light on our future ADEPT experiment design and further therapeutic outcome.

	Appendix A: Nomenclature

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 

Q_organ

: Plasma flow rate of each organ (ml/min).

L_organ

: Lymph flow rate of each organ (ml/min/g).

J_iso,organ

: Fluid recirculation flow rate for each organ (flow rate through large pore to the interstitial space for L = 0; ml/min/g).

Mass_organ

: Organ weight (gram).

V_v,organ

: Vascular space of each organ (ml).

V_i,organ

: Interstitial space of each organ (ml).

V_organ

: Total volume of each organ (ml).

L, S

: Osmotic reflection coefficient for large and small pores, respectively.

L, S

: Fraction of extravasation via large and small pores, respectively.

PS_L,organ, PS_S,organ

: Permeability-surface area product for each organ (ml/min) for large and small pores, respectively.

Pe_L,organ,Pe_S,organ

: Péclet number, ratio of convection to diffusion across large and small pores [J(1–)/PS].

J_L,organ, J_S,organ

: Transcapillary fluid flow rate (vascularinterstitial) for each organ (ml/min) via large and small pores, respectively.

B_max

: Maximum tumor-associated antigen concentration inside tumor tissue (mol/ml).

K_fCEA

: Association rate constants for the binding of antibodies (ml/mol/min).

K_rCEA

: Dissociation rate constants for the binding of antibodies (1/min).

FcRn_tot

: Total FcRn binding capacity in the skin and muscle (mol/ml).

K_int,FcRn

: Nonspecific internalization rate constant of free antibody into endosome for binding to FcRn sites (1/min).

K_rec,FcRn

: Recycling rate of FcRn bound antibody back to vascular space (1/min).

K_on,FcRn

: Binding constant for IgG-FcRn interaction (ml/mol/min).

K_off,FcRn

: Dissociation constant for IgG-FcRn interaction (1/min).

K_deg,FcRn

: Degradation rate constant of unbound endosomal IgG (1/min).

CL_{deg Liver}

: Clearance of antibody-enzyme conjugate (AbE) from liver (ml/min).

C_pl

: AbE concentration in plasma (mol/ml).

C_v,organ

: AbE concentration in the vascular space of each organ (mol/ml).

C_if,organ

: Free AbE concentration in the interstitial space of each organ (mol/ml).

C_ib,organ

: Bound AbE concentration in the interstitial space of each specified organ (mol/ml).

C_ef,organ

: Free AbE concentration in the endosome (mol/ml).

C_eb,organ

: Bound AbE concentration in the endosome with FcRn (mol/ml).

C_tot,organ

: Total AbE concentration in each organ (mol/ml).

f_up,prodrug, f_up,drug

: Unbound free fractions in plasma for prodrug and active prodrug, respectively.

E_max,prodrug

: Maximum rate constant for converting prodrug to active drug by antibody-enzyme conjugate (AbE) (1/min).

EC_{50 prodrug}

: EC₅₀ for converting prodrug to active drug by antibody-enzyme conjugate (AbE) (mol/ml).

Cl_{int_prodrug},Cl_{int_drug}

: Liver intrinsic clearance for prodrug and active drug, respectively (ml/min).

P_prodrug, P_drug

: Effective permeability coefficient for prodrug and active drug, respectively (1/min).

PS_prodrug,PS_drug

: Permeability-area product for prodrug and active drug, respectively (ml/min).

R_{prodrug,organ}, R_drug,organ

: Tissue to plasma concentration ratio of each organ for prodrug and active drug, respectively.

C_{pl_prodrug}, C_{pl_drug}

: Plasma concentration for prodrug and active drug, respectively (mol/ml).

C_{v_prodrug,organ}, C_{v_drug,organ}

: Prodrug and active drug concentration in the vascular space of each organ (mol/ml).

C_{i_prodrug,organ}, C_{i_drug,organ}

: Prodrug and active drug concentration in the interstitial space of each organ (mol/ml).

C_{tot_prodrug,organ}, C_{tot_drug,organ}

: Total prodrug and active drug concentration in each organ (mol/ml).

AUC_{prodrug,organ}, AUC_drug,organ

: AUC for prodrug and active drug in each organ (mol · min/ml).

	Appendix B: Mathematical Model and Governing Equations

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 
The mass balance equations for the pharmacokinetic model describe the circulation of antibody-enzyme conjugate (AbE), prodrug, and active drug throughout the body of a 20-g nude mouse (Fig. 1). For large molecular weight molecules, AbE, each organ (except tumor, skin, and muscle) is further divided into two subcompartments, the vascular space and extravascular space. Nonspecific binding is not considered in those organs. A specific, saturable, and reversible binding compartment is added in the extravascular space in the tumor tissue. Similarly, two endosome bound and free compartments are added as a continuation of vascular space in skin and muscle tissues. Inside all these organs or tissues, the net flux of AbE across the capillary between vascular and extravascular fluid is determined by the two-pore mechanism proposed by Rippe and Haraldsson (1987).

For small molecules, prodrug, and active drug, only vascular and extravascular spaces were considered. In each subcompartment, the predisposed conjugate will convert the exiting prodrug to the active drug. In tumor tissue, both free and bound AbE will convert existing prodrug to the active drug. In skin and muscle, the predelivered endosome free and bound AbE will convert the prodrug in the vascular space to the active drug.

B.1: Mass Balance Equation for Antibody-Enzyme Conjugate. B.1.1: Mass balance equation for plasma. According to the circulation scheme illustrated, the mass balance equation for the plasma compartment is

There is an additional constraint on the volumetric flow rates as

B.1.2: Mass balance equation for noneliminating organs. Vascular space:

Interstitial space:

B.1.3: Mass balance equation for tumor. Vascular space:

Interstitial free concentration:

Interstitial bound concentration:

B.1.4: Mass balance equation for skin (muscle is similar). Vascular space:

The endosome is assumed to be the continuous phase of vascular space, so the whole endosome concentration is equal to the vascular concentration for the conjugate. Thus:

Interstitial free concentration:

Endosome free concentration:

Endosome bound concentration:

B.1.5: Mass balance equation for liver. Vascular space:

Interstitial space:

In each organ, the total (or average) concentration is the weighted average of the concentrations within each subcompartment

or for skin and muscle

B.2: Mass Balance Equation for Prodrug and Active Drug Diffusion-limited model is assumed for both prodrug and active drug. Liver is considered as the only eliminating organ.

Prodrug in vascular space:

Active drug in vascular space:

Prodrug in the interstitial space:

Active drug in the interstitial space:

	Acknowledgments

 
The authors thank Dr. Nick Holford for suggestions and kind help in the current study.

	Footnotes

 
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.

doi:10.1124/dmd.107.019182.

ABBREVIATIONS: ADEPT, antibody-directed enzyme prodrug therapy; PBPK, physiologically based pharmacokinetic model; GA, geldanamycin; FcRn, neonatal Fc-receptor; AbE, antibody-enzyme conjugates; TAG-72, tumor-associated glycoprotein 72; P_A, effective permeability coefficients; AUC, area under the curve; PK, pharmacokinetics; PD, pharmacodynamics; Tau, time interval between AbE and prodrug.

¹ Current affiliation: Clinical DMPK, Merck & Co., Inc., West Point, PA.

Address correspondence to: Dr. Duxin Sun, Division of Pharmaceutics, College of Pharmacy, The Ohio State University, 500 W 12th Ave, Columbus, OH 43210. E-mail: sun.176{at}osu.edu

	References

Top Abstract Materials and Methods Results Discussion Appendix A: Nomenclature Appendix B: Mathematical Model... References

 


Bagshawe KD (1987) Antibody directed enzymes revive anti-cancer prodrugs concept. Br J Cancer 56: 531–532.[Medline]

Bagshawe KD, Sharma SK, Burke PJ, Melton RG, and Knox RJ (1999) Developments with targeted enzymes in cancer therapy. Curr Opin Immunol 11: 579–583.[Medline]

Bagshawe KD, Springer CJ, Searle F, Antoniw P, Sharma SK, Melton RG, and Sherwood RF (1988) A cytotoxic agent can be generated selectively at cancer sites. Br J Cancer 58: 700–703.[Medline]

Baxter LT and Jain RK (1996) Pharmacokinetic analysis of the microscopic distribution of enzyme-conjugated antibodies and prodrugs: comparison with experimental data. Br J Cancer 73: 447–456.[Medline]

Baxter LT, Zhu H, Mackensen DG, Butler WF, and Jain RK (1995) Biodistribution of monoclonal antibodies: scale-up from mouse to human using a physiologically based pharmacokinetic model. Cancer Res 55: 4611–4622.[Abstract/Free Full Text]

Baxter LT, Zhu H, Mackensen DG, and Jain RK (1994) Physiologically based pharmacokinetic model for specific and nonspecific monoclonal antibodies and fragments in normal tissues and human tumor xenografts in nude mice. Cancer Res 54: 1517–1528.[Abstract/Free Full Text]

Cheng H, Cao X, Xian M, Fang L, Cai TB, Ji JJ, Tunac JB, Sun D, and Wang PG (2005) Synthesis and enzyme-specific activation of carbohydrate-geldanamycin conjugates with potent anticancer activity. J Med Chem 48: 645–652.[CrossRef][Medline]

Fang L, Battisti RF, Cheng H, Reigan P, Xin Y, Shen J, Ross D, Chan KK, Martin EW, Jr., Wang PG, et al. (2006) Enzyme specific activation of benzoquinone ansamycin prodrugs using HuCC49DeltaCH2-beta-galactosidase conjugates. J Med Chem 49: 6290–6297.[CrossRef][Medline]

Fang L, Holford NH, Hinkle G, Cao X, Xiao JJ, Bloomston M, Gibbs S, Saif OH, Dalton JT, Chan KK, et al. (2007) Population pharmacokinetics of humanized monoclonal antibody HuCC49deltaCH2 and murine antibody CC49 in colorectal cancer patients. J Clin Pharmacol 47: 227–237.[Abstract/Free Full Text]

Ferl GZ, Wu AM, and DiStefano JJ, 3rd (2005) A predictive model of therapeutic monoclonal antibody dynamics and regulation by the neonatal Fc receptor (FcRn). Ann Biomed Eng 33: 1640–1652.[CrossRef][Medline]

Francis RJ, Sharma SK, Springer C, Green AJ, Hope-Stone LD, Sena L, Martin J, Adamson KL, Robbins A, Gumbrell L, et al. (2002) A phase I trial of antibody directed enzyme prodrug therapy (ADEPT) in patients with advanced colorectal carcinoma or other CEA producing tumours. Br J Cancer 87: 600–607.[CrossRef][Medline]

Friedrich SW, Lin SC, Stoll BR, Baxter LT, Munn LL, and Jain RK (2002) Antibody-directed effector cell therapy of tumors: analysis and optimization using a physiologically based pharmacokinetic model. Neoplasia 4: 449–463.[CrossRef][Medline]

Galluppi GR, Rogge MC, Roskos LK, Lesko LJ, Green MD, Feigal DW, Jr., and Peck CC (2001) Integration of pharmacokinetic and pharmacodynamic studies in the discovery, development, and review of protein therapeutic agents: a conference report. Clin Pharmacol Ther 69: 387–399.[CrossRef][Medline]

Green AJ, Johnson CJ, Adamson KL, and Begent RH (2001) Mathematical model of antibody targeting: important parameters defined using clinical data. Phys Med Biol 46: 1679–1693.[CrossRef][Medline]

Martin J, Stribbling SM, Poon GK, Begent RH, Napier M, Sharma SK, and Springer CJ (1997) Antibody-directed enzyme prodrug therapy: pharmacokinetics and plasma levels of prodrug and drug in a phase I clinical trial. Cancer Chemother Pharmacol 40: 189–201.[CrossRef][Medline]

Mayer A, Francis RJ, Sharma SK, Tolner B, Springer CJ, Martin J, Boxer GM, Bell J, Green AJ, Hartley JA, et al. (2006) A phase I study of single administration of antibody-directed enzyme prodrug therapy with the recombinant anti-carcinoembryonic antigen antibody-enzyme fusion protein MFECP1 and a bis-iodo phenol mustard prodrug. Clin Cancer Res 12: 6509–6516.[Abstract/Free Full Text]

Melton RG and Knox RJ (1999) Enzyme-prodrug strategies for cancer therapy. Kluwer Academic/Plenum Publishers, New York.

Napier MP, Sharma SK, Springer CJ, Bagshawe KD, Green AJ, Martin J, Stribbling SM, Cushen N, O'Malley D, and Begent RH (2000) Antibody-directed enzyme prodrug therapy: efficacy and mechanism of action in colorectal carcinoma. Clin Cancer Res 6: 765–772.[Abstract/Free Full Text]

Rippe B and Haraldsson B (1987) Fluid and protein fluxes across small and large pores in the microvasculature. Application of two-pore equations. Acta Physiol Scand 131: 411–428.[Medline]

Sharma SK, Pedley RB, Bhatia J, Boxer GM, El-Emir E, Qureshi U, Tolner B, Lowe H, Michael NP, Minton N, et al. (2005) Sustained tumor regression of human colorectal cancer xenografts using a multifunctional mannosylated fusion protein in antibody-directed enzyme prodrug therapy. Clin Cancer Res 11: 814–825.[Abstract/Free Full Text]

Syrigos KN and Epenetos AA (1999) Antibody directed enzyme prodrug therapy (ADEPT): a review of the experimental and clinical considerations. Anticancer Res 19: 605–613.[Medline]

Tietze LF, Feuerstein T, Fecher A, Haunert F, Panknin O, Borchers U, Schuberth I, and Alves F (2002) Proof of principle in the selective treatment of cancer by antibody-directed enzyme prodrug therapy: the development of a highly potent prodrug. Angew Chem Int Ed Engl 41: 759–761.[CrossRef]

Varner JD (2005) Systems biology and the mathematical modelling of antibody-directed enzyme prodrug therapy (ADEPT). Syst Biol (Stevenage) 152: 291–302.[Medline]

Xu G and McLeod HL (2001) Strategies for enzyme/prodrug cancer therapy. Clin Cancer Res 7: 3314–3324.[Abstract/Free Full Text]

Xu L, Eiseman JL, Egorin MJ, and D'Argenio DZ (2003) Physiologically-based pharmacokinetics and molecular pharmacodynamics of 17-(allylamino)-17-demethoxygeldanamycin and its active metabolite in tumor-bearing mice. J Pharmacokinet Pharmacodyn 30: 185–219.[CrossRef][Medline]

Yuan F, Baxter LT, and Jain RK (1991) Pharmacokinetic analysis of two-step approaches using bifunctional and enzyme-conjugated antibodies. Cancer Res 51: 3119–3130.[Abstract/Free Full Text]

Zhu H, Baxter LT, and Jain RK (1997) Potential and limitations of radioimmunodetection and radioimmunotherapy with monoclonal antibodies. J Nucl Med 38: 731–741.[Abstract/Free Full Text]

This Article Abstract Full Text (PDF) All Versions of this Article: dmd.107.019182v1 36/6/1153    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Fang, L. Articles by Sun, D. Search for Related Content PubMed PubMed Citation Articles by Fang, L. Articles by Sun, D.