Optimizing preclinical study design in oncology research

Chemico-Biological Interactions 190 (2011) 73–78

Contents lists available at ScienceDirect

Chemico-Biological Interactions journal homepage: www.elsevier.com/locate/chembioint

Mini Review

Optimizing preclinical study design in oncology research Luke A. Wittenburg a,∗ , Daniel L. Gustafson a,b a b

Flint Animal Cancer Center, Department of Clinical Sciences, Colorado State University, 300 West Drake Road, Fort Collins, CO 80523-1620, United States University of Colorado Comprehensive Cancer Center, Anschutz Medical Campus, 13001 E. 17th Pl., P.O. Box 6511, Mail Stop F434, Aurora, CO 80045, United States

a r t i c l e

i n f o

Article history: Received 18 November 2010 Received in revised form 7 January 2011 Accepted 26 January 2011 Available online 4 February 2011 Keywords: Preclinical study Oncology Pharmacokinetics Pharmacodynamics

a b s t r a c t The current drug development pathway in oncology research has led to a large attrition rate for new drugs, in part due to a general lack of appropriate preclinical studies that are capable of accurately predicting efficacy and/or toxicity in the target population. Because of an obvious need for novel therapeutics in many types of cancer, new compounds are being investigated in human Phase I and Phase II clinical trials before a complete understanding of their toxicity and efficacy profiles is obtained. In fact, for newer targeted molecular agents that are often cytostatic in nature, the conventional preclinical evaluation used for traditional cytotoxic chemotherapies utilizing primary tumor shrinkage as an endpoint may not be appropriate. By utilizing an integrated pharmacokinetic/pharmacodynamic approach, along with proper selection of a model system, the drug development process in oncology research may be improved leading to a better understanding of the determinants of efficacy and toxicity, and ultimately fewer drugs that fail once they reach human clinical trials. © 2011 Elsevier Ireland Ltd. All rights reserved.

Contents 1. 2. 3. 4. 5. 6. 7.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Use of appropriate model system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PK evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Protein binding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Metabolism and detoxification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pharmacodynamics and target evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction One of the most serious challenges currently facing pharmaceutical research of novel anti-cancer therapeutics is the lack of translation of efficacy and safety from preclinical models to human clinical trials, leading to a large attrition rate of investigational compounds. For new oncology drugs, only about 5% of investigational new drug applications submitted progress beyond the investigational phase due to a general lack of preclinical systems that can accurately predict efficacy and toxicity of new agents [1]. This

∗ Corresponding author at: Flint Animal Cancer Center, Department of Clinical Sciences, Colorado State University, 300 West Drake Road, 168 Campus Delivery, Fort Collins, CO 80523-1620, United States. Tel.: +1 970 297 4093; fax: +1 970 297 1254. E-mail addresses: [email protected], [email protected] (L.A. Wittenburg). 0009-2797/$ – see front matter © 2011 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2011.01.029

73 74 74 75 76 76 77 77 77

problem can be addressed by correctly designing and optimizing preclinical studies such that the over—or underestimation of safety and efficacy is minimized. There are significant costs, as well as ethical considerations, associated with proceeding to human clinical trials with compounds that are either too toxic or ineffective in the target population, particularly when appropriately designed preclinical studies may have the potential to accurately predict toxicity and/or efficacy. A goal of these types of studies should therefore be to improve the process that is currently used to clarify the relationships between a compounds pharmacokinetic (PK) characteristics, which can be defined as the mechanisms and kinetics of the processes which define the drug level–time relationships in the body (absorption, distribution, metabolism, elimination) [2], its pharmacodynamic (PD) profile, defined as the effect produced by the compound at its site(s) of action [2], and its safety profile (Fig. 1). Clarification of these relationships may be done through refined integration of PK

74

L.A. Wittenburg, D.L. Gustafson / Chemico-Biological Interactions 190 (2011) 73–78

Fig. 1. Process of pharmacokinetic and pharmacodynamic integration. Pharmacokinetics is comprised of the dose (and dose route) and the processes in the body that dictate the levels of drug within various tissues. Pharmacodynamics is the effect that is obtained from exposure to the drug within these tissues. The portion that is overlapping represents the dose–response relationship. Shaded boxes and solid lines represent variables that can be directly measured in pre-clinical pharmacology studies while the dashed lines represent extrapolations that are often made between the pre-clinical model and humans.

and PD, also known as “quantitative pharmacology”, which enables identification of the interdependence of the compound’s pharmacological properties on its target physiological system and systemic exposure characteristics [3,4]. In fact, it is suggested that the PK/PD approach be viewed as the core activity of translational models where predictions on dose, exposure, and response are to be made in humans [5]. When planning preclinical studies, a few mandatory aspects of study design should be considered: use of an appropriate model system, assessment of drug exposure in the animal (preferably through plasma concentrations at different time points and different doses), measurement of plasma protein binding (vital if comparisons of drug action are to be made across species), evaluate the metabolism of the drug and consider whether active metabolites, or metabolites that are inactive in terms of intended effect but potentially responsible for toxicity, could be influencing the results being obtained, ensure that dose schedules and routes of administration are relevant to the way the drug will be used clinically to ensure translational validity, and consider the determinants of target engagement by the compound being studied.

2. Use of appropriate model system The most integral component of designing a study with translational validity is model selection, which is generally dictated by the objectives and goals of the study. In the case of oncology studies, orthotopic models that more closely recapitulate the setting found in human cancer with regard to stromal interactions and drug accessibility will often be superior to simple xenograft studies in predicting intra-tumoral drug levels, target engagement, and efficacy. One example would be the use of an orthotopic, spontaneously metastasizing, surgical adjuvant model of osteosarcoma (OS) in immunocompetent mice for evaluation of investigational therapeutics in the treatment of osteosarcoma [6]. In this case, a primary tumor is established in an anatomic site that is commonly affected in humans (proximal tibia), and therapeutic intervention is initiated following primary tumor removal (amputation) but prior to the onset of metastatic disease. The time to metastasis is then

evaluated, which is a more clinically meaningful endpoint than reduction of primary tumor size as the vast majority of OS patients succumb to metastatic disease and not recurrence of the primary tumor [7]. Although murine models such as the one described above have been useful for analyzing the biology of pathways involved in cancer initiation, promotion, and progression the majority of these models do not adequately recapitulate the features that define cancer in humans such as a long latency period, heterogeneity in both tumor cells and surrounding microenvironment, genomic instability, and recurrence and metastasis [8]. The use of spontaneously occurring cancers in pet dogs provides an opportunity to answer questions about the best use of novel drugs that likely cannot be answered with murine models. In these pet dogs, cancers develop spontaneously in the context of an intact immune system and these cancers share many characteristics with human cancers such as histologic appearance, tumor genetics, biological behavior, molecular targets, therapeutic response, and acquired resistance, recurrence, and metastasis [8,9]. In addition, the relatively larger size of pet dogs and their tumors makes this model more amenable to repeated biopsies and advanced imaging techniques in evaluation of PD endpoints. An example of the utility of spontaneously occurring cancers in pet dogs is seen in the advent of liposomal muramyl tripeptide phosphatidylethanolamine (LMTP-PE) in the treatment of osteosarcoma. Studies performed in dogs showing a significantly prolonged survival time guided the translation of this compound into humans, where similar responses were seen with improvements in both overall survival and event-free survival [10,11]. Increased utilization of cancer in pet dogs as an intermediary model between conventional preclinical models and human clinical trials carries the potential for better informed decisions in human clinical trials and decreased drug attrition rates.

3. PK evaluation In addition to proper model selection, the sampling schedule for PK studies is also dependent upon the goals of the study; however,

L.A. Wittenburg, D.L. Gustafson / Chemico-Biological Interactions 190 (2011) 73–78

Fig. 2. Theoretical plasma concentration–time curve illustrating importance of sampling at later time-points. Dashed line represents the terminal slope calculated without the addition of the later time-point, while the dotted line represents the terminal slope with inclusion of the later time-point. Calculation of PK parameters dependent upon the terminal slope will be more accurate with inclusion of the later time-point. Cmax = maximal plasma concentration; Tmax = time to maximal plasma concentration.

some basic principles apply for using specific modeling methodologies and generating valid PK parameters. The ability to define the PK of a given drug is often dependent upon the intended rate, route, and frequency at which the drug is to be administered. The timing of sample collection is then based on components of the plasma concentration versus time curve. For example, early and frequent sampling should be performed for drugs that will be administered as an IV bolus so that distribution from the central compartment to tissues can be defined. Then, samples collected at later time points and longer time intervals are utilized to define the elimination and elimination half-life (t1/2 ) components. A common problem in pharmacokinetic studies is the inadequate or incomplete measurement of drug elimination from the system, often due to premature termination of sample collection [12]. For an accurate interpretation of PK data, a true terminal disposition phase must be examined. For example, clearance and volume of distribution at steady-state are calculated using drug concentration–time curve areas (AUC and AUMC, respectively) which are prone to exaggerated error from an inaccurate terminal slope [12] (Fig. 2). Drugs given by an extravascular route have additional considerations such as estimation of the time to maximal plasma concentration (Tmax ) and thus Cmax which leads to more intense sampling around the range of Tmax . This does require some knowledge of the drug’s PK. For example, when compared to many small molecules, the absorption of large biologics such as monoclonal antibodies may be entirely through the lymphatic system and thus the Tmax can be expected to be delayed, potentially up to 24 h [5]. In a situation where there is no a priori knowledge the sampling schedule should be altered in an iterative manner based upon initial measurements. In single drug PK studies, the determination of PK parameters with multiple doses is important if extrapolation of exposure is to be made in humans. The use of multiple doses allows for the determination of linearity of drug elimination and distribution, which is critical as it will allow or disallow for dose extrapolation to predict exposure. Detection of dose dependence of the AUC

75

or Cmax can be utilized to estimate the linearity of drug elimination and distribution, respectively. It is important to remember that a subset of newer “targeted” anti-cancer agents will demonstrate non-linear PK because of more complex elimination that is often dose-dependent. For example, compared to small molecules that are primarily eliminated through renal filtration or oxidizing enzymes in the liver, elimination of many macromolecules such as monoclonal antibodies can occur through the target receptor, blood proteolysis, or the reticuloendothelial system [5]. This results in variations in PK parameters such as half-life and clearance with altered dose and routes of administration, which underscores the need to evaluate multiple doses and routes during preclinical studies. Preclinical oncology studies are often inherently more complicated than other types of studies because they often involve the evaluation of a novel agent when combined with a standard treatment protocol, often containing more than one drug. The goal of many of these studies is to determine the effect of one drug on the PK of others in the treatment protocol. To this end, determination must be made during study design as to whether single agent PK in individual animals will be performed followed by combined agent PK to interrogate potential interactions, or if PK of just the combination will be determined and compared to historic or literature values of the single agents. If historical data are to be used for comparison, then the type and quality of the data must be considered; ideally using similar dose ranges (unless linearity to dose has already been determined), analytical strategies, and modeling methodology. Sensitivity of the equipment and methodology being used to determine drug exposure are also linked to sample collection; the use of relatively insensitive detection methods would preclude sampling at much later time points when drug levels are likely to fall below the limit of detection or quantification. This may have profound implications in determination of PK parameters such as terminal half-life, as evidenced by a study that evaluated PK of docetaxel in human patients and found that the use of a more sensitive detection method (HPLC coupled to tandem mass spectrometry compared to UV detection) afforded sampling at later time points (48 h) which subsequently resulted in a significant increase in terminal half-life and allowed for more accurate predictions of patient plasma concentrations at later time points [13].

4. Protein binding An additional parameter that has the potential to significantly impact the distribution, metabolism, and elimination of drugs in preclinical models is binding of drugs to proteins in blood as well as tissues. The free, unbound concentration of a drug is a key determinant of both its pharmacokinetic and pharmacodynamic profiles [14]. Drug that is bound to proteins generally cannot distribute to tissues and therefore will not be metabolized or eliminated. This becomes important when multi-drug protocols are being evaluated as the extent of drug–protein interaction can provide a major potential for drug–drug interactions, particularly in the case of compound that is highly protein bound with a narrow therapeutic window. It is suggested, therefore, that effective concentration ranges, characterization of the concentration–response relationship, and safety margins be derived from unbound plasma concentrations which may necessitate sampling at low and high total plasma concentrations for ex vivo measurements of unbound and total concentrations [3]. Adequate information on tissue and plasma protein binding is certainly critical if the goal of the study is developing a predictive physiologically based model [14].

76

L.A. Wittenburg, D.L. Gustafson / Chemico-Biological Interactions 190 (2011) 73–78

5. Metabolism and detoxification Drug metabolizing enzymes play a key role in the pharmacokinetics of a compound and contribute to the toxicity within tumor and normal cells, as well as the balance between drug sensitivity and resistance within target tumor cells [15]. As mentioned previously, preclinical oncology studies often involve the evaluation of multiple drugs given in combination, and therefore, the role of metabolizing enzymes and transporters in the disposition of drugs as well as the potential modulation of expression of metabolic enzymes by these drugs must be considered. The metabolism of a drug may result in the formation of inactive, active, or toxic metabolites which can complicate the pharmacology of a given compound, particularly if the compound affects the expression or activity of enzyme systems responsible for its own metabolism. The hepatic microsomal P450 enzyme system forms the basis for the majority of oncology drug metabolism [16,17], and alterations in expression or activity of these enzymes can have a variety of compound-dependent effects on PK or pharmacodynamics. For example, treatment of rodents with the anthracycline chemotherapeutic doxorubicin has been shown to result in lowered hepatic P450 content and decreases in a variety of P450-catalyzed activities [18]. Many drugs are metabolized by multiple metabolic pathways, and thus total drug metabolism is the sum of the individual pathways. Alterations in activity in one of the pathways can tip the balance of metabolism for a compound and result in either increased toxicity or decreased efficacy. For example, the camptothecin derivative irinotecan is a pro-drug that requires metabolic activation by carboxylesterases [19] and then undergoes metabolic detoxification via two major metabolic pathways, one involving the P450 enzyme Cyp3A4 [20]. Co-administration of vinorelbine and irinotecan at clinically achievable concentrations in vitro has been demonstrated to result in reduced activity of the Cyp3A4-mediated catabolism of irinotecan, therefore this combination might be expected to result in increased toxicity in a clinical setting [20]. Thus, the clinical implication of induction or inhibition of metabolizing enzymes will depend upon the relative pharmacologic activities of the parent compound and the metabolite(s). If the parent compound exhibits greater activity than the metabolite, inhibition of the metabolism will increase plasma concentrations of the parent drug along with therapeutic and/or toxic effects; if the metabolite exhibits greater activity, any inhibition of metabolism may result in reduced efficacy [21]. In vitro studies aimed at determining which P450 isoforms have a role in the transformation of an anticancer agent can facilitate the prediction of interaction with other drugs. In addition to the P450 enzyme system, other components of the drug detoxification system can be involved in drug–drug interactions that may result in either increased toxicity or reduced anti-tumor efficacy. Efflux transporter proteins such as those of the ATP-binding cassette (ABC) family, including P-glycoprotein (Pgp or ABCG1), and the phase-II detoxification enzymes glutathione S-transferase (GST) and UDP-glucuronosyltransferases (UGT) also have roles in controlling the bioavailability, toxicity, and efficacy of a majority of cytotoxic chemotherapy drugs [22]. Expression of P-gp and CYP3A4 are coordinately regulated in cells lining the intestine and bile ducts as well as tumor cells, and an increased expression or activity of these proteins by one drug has the potential to reduce oral uptake and intracellular levels of other drugs in a combination protocol while increasing elimination, thereby reducing efficacy [22]. In contrast, it could then be anticipated that a drug that reduces the expression or activity of CYP3A4 and/or P-gp has the potential to cause an increase in exposure and toxicity of a co-administered drug by the same mechanisms. Drug interactions may also occur through alterations in the phase-II drug detoxifying enzymes GST and UGT or their regulators. Because these enzymes

are involved in detoxification of numerous cytotoxic chemotherapeutics within normal and malignant cells, attempts have been made to take advantage of this by reducing cellular glutathione levels to increase dose intensity and anti-tumor efficacy. For example, administration of the glutathione-reducing drug butathione sulfoximime (BSO) has been shown to increase the cytotoxicity of multiple chemotherapy agents, but also results in increased bone marrow toxicity because of systemic reductions in drug detoxification [23]. Conversely, administration of drugs that mimic or enhance the activity of GST or UGT could be expected to reduce systemic toxicity through increased detoxification but may also reduce anti-tumor efficacy. These examples illustrate the importance and the complexity of the interactions between phase-I (CYP) enzymes, phase-II (GST/UGT) enzymes, and drug efflux pumps such as P-gp in determining the overall distribution and subsequent efficacy and/or toxicity of chemotherapeutic agents, particularly when used in combination, and underscores the importance of consideration of these interactions in preclinical study designs.

6. Pharmacodynamics and target evaluation Another aspect of preclinical study design that is essential in the development of oncology drugs is that of target validation and modulation. With the shift in emphasis from traditional cytotoxic agents to molecularly targeted agents, there is a need to develop and make better use of biomarkers early in drug development as the strategies that are applied for cytotoxics may not apply to molecular targeted agents; many of these agents demonstrate an ability to inhibit tumor growth without significant cytotoxicity [1,24]. Preclinical identification and implementation of PK/PD endpoints has numerous advantages including facilitating proofof-concept studies for target modulation, enhancing the rational decision of dose and schedule, and reduced uncertainty that is associated with the prediction of efficacy and toxicity [25]. The interaction between PK and PD of a compound becomes even more important in studies evaluating molecular targeted agents. Studies of traditional cytotoxic drugs are often based upon the assumption that the mechanism of therapeutic effect and toxicity are the same; higher doses result in improved efficacy as well as toxicity. This is not always the case for molecular targeted agents which may be cytostatic and can have mechanisms of therapeutic effect that differ from those causing toxic effects. In these cases, biologic PD endpoints may become more heavily relied upon as surrogates for antitumor potential than toxicity or objective tumor response [24]. Ideally, validation of a target and target inhibition proof-of-concept is done in a disease-relevant model that closely mimics the clinical condition and uses a method of modulating the target that is comparable to the method that will ultimately be used [26]. This is in contrast to target validation models utilizing knock-out animals, as the sequelae of pharmacologic inhibition can vary drastically from that of knocking out the target. For example, the histone deacetylase enzymes (HDAC), which have become a promising therapeutic target in oncology, function to alter chromatin structure through removal of acetyl groups from histone proteins as well as functioning as transcriptional co-repressors by forming repressive complexes through protein–protein interactions [27]. While pharmacologic inhibition of HDAC blocks the catalytic activity responsible for removal of acetyl groups from histone proteins, it does not appear to affect the ability of HDAC to form multi-protein complexes, which may account for the differences seen between HDAC knock-outs and the clinical effects of HDAC inhibitors [28,29]. For optimal preclinical evaluation of oncology compounds it is therefore essential to have an idea of the exposure being achieved

L.A. Wittenburg, D.L. Gustafson / Chemico-Biological Interactions 190 (2011) 73–78

in the tissue of interest in the relevant disease model to know if the target is being appropriately modulated, and to determine if target modulation is leading to the desired biologic effect [25]. With regard to molecular targeted agents, the endpoints of interest may not necessarily be normal tissue tolerance (or maximum tolerated dose), but identification of a biologically effective dose. The assumption in determining a biologically effective dose is that there is a reliable method to evaluate target engagement and/or inhibition. In this regard, preclinical studies should aim to identify the expected effects of target inhibition, the function of the target in normal tissues, and whether the desired effect requires continuous or intermittent target inhibition [24]. As might be expected, the invasiveness of the methodology must be considered in the development of PK/PD biomarkers that are being translated from preclinical to clinical settings. For example, development of a method to determine target modulation in a surrogate tissue, such as peripheral blood mononuclear cells (PBMC), would be superior to repeated tumor biopsies from a logistic and patient compliance standpoint. However, before surrogate tissues can be used for target effect in tumor tissues, preclinical animal models must be developed and validated; a dose that inhibits the target in a surrogate tissue may not provide sufficient exposure in the tumor environment. Incorporation of advanced imaging modalities such as digital contrast-enhanced magnetic resonance imaging (DCE-MRI) and positron emission tomography (PET) is another strategy that could be utilized in defining optimal biologically effective doses [25]. In order to accurately make an assessment of target modulation, consideration must be given to optimal sampling schedules. Single time-point collection (i.e. at Cmax ) can lead to misleading results, particularly if the compound under investigation displays hormetic-like dose response curves, which have been reported for a number of chemotherapeutic agents [30]. In such cases, single time-point collection even at multiple dose levels can result in misleading information, and collection of both rising and declining response–time and concentration–time data is considered a more accurate approach [31]. Temporal differences between maximum pharmacodynamic effect and plasma concentrations are also commonly encountered, often with a lag between maximal plasma concentrations and maximal target modulation, or hysteresis. For drugs that exhibit hysteresis or act indirectly, the collection of PD data points should be done at steady-state, with at least two concentration-effect data pairs to assess PD variability [32]. Evaluating PD responses at steady-state requires consideration of the route of drug administration; once or twice per day oral gavage or subcutaneous injections carry the potential for increased stress to the animal, and often the use of an osmotic minipump can reduce this variable, provided the drug is soluble and stable at 37 ◦ C. The use of a minipump, with changes in the rate of infusion, also allows for simulating non-linear absorption or disposition patterns of a drug, both of which may can have an impact on the onset, intensity, and duration of a PD response [3]. If drug administration is to be done via a method other than continuous infusion (oral gavage, subcutaneous, intraveneous, or intraperitoneal injection), the timing of drug delivery should be such that collection of samples can be done in each animal at precise post-administration time-points, often accomplished by staggering or spacing treatment groups to allow sufficient time for sample collection. Proper timing of drug delivery and sample collection will help in reducing the apparent variability in PK/PD data collected, particularly in studies utilizing large numbers of subjects.

7. Conclusions The current “unidirectional” pathway in oncology drug development that involves assessment of novel agents in conventional

77

preclinical studies of efficacy and toxicity followed by Phase I and Phase II human trials leads to a large number of drugs that fail due to either lack of efficacy or unacceptable toxicity. Improvements must be made to the current method in which preclinical studies are designed, starting with proper model selection such that a maximum amount of information can be obtained prior to moving forward with development of novel drugs. Depending on the goals of the study, models may range from simple xenograft studies of target validation to complex models involving spontaneously occurring tumors in pets in a clinical setting. In addition to model selection, other factors affecting the disposition of a drug or drug combination must be considered prior to initiating preclinical studies. Information on protein binding and metabolism should be obtained so that informed decisions on potential drug interactions can be made. Accurate PK data with relevant exposures is essential in preclinical studies of drugs that are intended to be translated to human medicine and thus the dose, route of administration, and the rate and extent of drug absorption must be considered. Consideration must also be given to the timing of sample collection; sample collection should be timed such that PK/PD data are not over or under estimated by accounting for potential hysteresis or hormetic-like dose responses, and are appropriate for the sensitivity of the particular detection method being used. Particularly for the newer targeted agents, integrated PK/PD studies that aim to not only determine the PK parameters that dictate a particular PD response, but also validate the utility of target inhibition as well as develop biomarkers in surrogate tissues that may be used to evaluate efficacy in a minimally invasive manner, are likely to provide the most useful information for better informed decisions in human clinical trials. Conflict of interest statement None declared. References [1] S. Kummar, R. Kinders, L. Rubinstein, R.E. Parchment, A.J. Murgo, J. Collins, O. Pickeral, J. Low, S.M. Steinberg, M. Gutierrez, S. Yang, L. Helman, R. Wiltrout, J.E. Tomaszewski, J.H. Doroshow, Compressing drug development timelines in oncology using phase ‘0’ trials, Nature reviews 7 (2) (2007) 131–139. [2] M. Rowland, T.N. Tozer, Why clinical pharmacokinetics? in: D. Balado (Ed.), Clinical Pharmacokinetics: Concepts and Applications, Williams & Wilkins, Media, PA, 1995, pp. 1–7. [3] J. Gabrielsson, A.R. Green, P.H. Van der Graaf, Optimising in vivo pharmacology studies—practical PKPD considerations Journal of Pharmacological and Toxicological Methods 61 (2) (2010) 146–156. [4] J. Gabrielsson, A.R. Green, Quantitative pharmacology or pharmacokinetic pharmacodynamic integration should be a vital component in integrative pharmacology, The Journal of Pharmacology and Experimental Therapeutics 331 (3) (2009) 767–774. [5] B.M. Agoram, S.W. Martin, P.H. van der Graaf, The role of mechanismbased pharmacokinetic–pharmacodynamic (PK–PD) modelling in translational research of biologics, Drug Discovery Today 12 (23–24) (2007) 1018–1024. [6] J.L. Sottnik, D.L. Duval, E.J. Ehrhart, D.H. Thamm, An orthotopic, postsurgical model of luciferase transfected murine osteosarcoma with spontaneous metastasis, Clinical and Experimental Metastasis 27 (3) (2010) 151–160. [7] N Federman, N. Bernthal, F.C. Eilber, W.D. Tap, The multidisciplinary management of osteosarcoma, Current Treatment Options in Oncology 10 (1–2) (2009) 82–93. [8] I. Gordon, M. Paoloni, C. Mazcko, C. Khanna, The comparative oncology trials consortium: using spontaneously occurring cancers in dogs to inform the cancer drug development pathway, PLoS Medicine 6 (10) (2009) e1000161. [9] C. Khanna, K. Lindblad-Toh, D. Vail, C. London, P. Bergman, L. Barber, M. Breen, B. Kitchell, E. McNeil, J.F. Modiano, S. Niemi, K.E. Comstock, E. Ostrander, S. Westmoreland, S. Withrow, The dog as a cancer model, Nature Biotechnology 24 (9) (2006) 1065–1066. [10] I.D. Kurzman, E.G. MacEwen, R.C. Rosenthal, L.E. Fox, E.T. Keller, S.C. Helfand, D.M. Vail, R.R. Dubielzig, B.R. Madewell, C.O. Rodriguez Jr., et al., Adjuvant therapy for osteosarcoma in dogs: results of randomized clinical trials using combined liposome-encapsulated muramyl tripeptide and cisplatin, Clinical Cancer Research 1 (12) (1995) 1595–1601. [11] P.A. Meyers, C.L. Schwartz, M.D. Krailo, J.H. Healey, M.L. Bernstein, D. Betcher, W.S. Ferguson, M.C. Gebhardt, A.M. Goorin, M. Harris, E. Kleinerman, M.P. Link, H. Nadel, M. Nieder, G.P. Siegal, M.A. Weiner, R.J. Wells, R.B. Womer, H.E. Grier,

78

[12]

[13]

[14]

[15]

[16]

[17] [18]

[19]

[20]

L.A. Wittenburg, D.L. Gustafson / Chemico-Biological Interactions 190 (2011) 73–78 Osteosarcoma: the addition of muramyl tripeptide to chemotherapy improves overall survival—a report from the Children’s Oncology Group, Journal of Clinical Oncology 26 (4) (2008) 633–638. W.J. Jusko, Guidelines for collection and analysis of pharmacokinetic data, in: W.E. Evans, J.J. Schentag, W.J. Jusko (Eds.), Applied Pharmacokinetics: Principles of Therapeutic Drug Monitoring, Applied Therapeutics, Inc., Vancouver, WA, 1992, pp. 2-1–2-43. D.L. Gustafson, M.E. Long, J.A. Zirrolli, M.W. Duncan, S.N. Holden, A.S. Pierson, S.G. Eckhardt, Analysis of docetaxel pharmacokinetics in humans with the inclusion of later sampling time-points afforded by the use of a sensitive tandem LCMS assay, Cancer Chemotherapy and Pharmacology 52 (2) (2003) 159–166. J. Kapitulnik, O. Pelkonen, U. Gundert-Remy, S.G. Dahl, A.R. Boobis, Effects of pharmaceuticals and other active chemicals at biological targets: mechanisms, interactions, and integration into PB-PK/PD models, Expert Opinion on Therapeutic Targets 13 (7) (2009) 867–887. D.S Riddick, C. Lee, S. Ramji, E.C. Chinje, R.L. Cowen, K.J. Williams, A.V. Patterson, I.J. Stratford, C.S. Morrow, A.J. Townsend, Y. Jounaidi, C.S. Chen, T. Su, H. Lu, P.S. Schwartz, D.J. Waxman, Cancer chemotherapy and drug metabolism, Drug Metabolism and Disposition: The Biological Fate of Chemicals 33 (8) (2005) 1083–1096. C.D. Scripture, A. Sparreboom, W.D. Figg, Modulation of cytochrome P450 activity: implications for cancer therapy, The Lancet Oncology 6 (10) (2005) 780–789. J.H. Beijnen, J.H. Schellens, Drug interactions in oncology, The Lancet Oncology 5 (8) (2004) 489–496. D.J. Marchand, K.W. Renton, Depression of cytochrome P-450-dependent drug biotransformation by adriamycin, Toxicology and Applied Pharmacology 58 (1) (1981) 83–88. J.G. Slatter, P. Su, J.P. Sams, L.J. Schaaf, L.C. Wienkers, Bioactivation of the anticancer agent CPT-11 to SN-38 by human hepatic microsomal carboxylesterases and the in vitro assessment of potential drug interactions, Drug Metabolism and Disposition: The Biological Fate of Chemicals 25 (10) (1997) 1157–1164. V. Charasson, M.C. Haaz, J. Robert, Determination of drug interactions occurring with the metabolic pathways of irinotecan, Drug Metabolism and Disposition: The Biological Fate of Chemicals 30 (6) (2002) 731–733.

[21] P. Blower, R. de Wit, S. Goodin, M. Aapro, Drug–drug interactions in oncology: why are they important and can they be minimized? Critical Reviews in Oncology/Hematology 55 (2) (2005) 117–142. [22] I. Meijerman, J.H. Beijnen, J.H. Schellens, Combined action and regulation of phase II enzymes and multidrug resistance proteins in multidrug resistance in cancer, Cancer Treatment Reviews 34 (6) (2008) 505–520. [23] E.A. Neuwelt, M.A. Pagel, B.P. Hasler, T.G. Deloughery, L.L. Muldoon, Therapeutic efficacy of aortic administration of N-acetylcysteine as a chemoprotectant against bone marrow toxicity after intracarotid administration of alkylators, with or without glutathione depletion in a rat model, Cancer Research 61 (21) (2001) 7868–7874. [24] S Kummar, M. Gutierrez, J.H. Doroshow, A.J. Murgo, Drug development in oncology: classical cytotoxics and molecularly targeted agents, British Journal of Clinical Pharmacology 62 (1) (2006) 15–26. [25] P. Workman, E.O. Aboagye, Y.L. Chung, J.R. Griffiths, R. Hart, M.O. Leach, R.J. Maxwell, P.M. McSheehy, P.M. Price, J. Zweit, Minimally invasive pharmacokinetic and pharmacodynamic technologies in hypothesis-testing clinical trials of innovative therapies, Journal of the National Cancer Institute 98 (9) (2006) 580–598. [26] F. Sams-Dodd, Target-based drug discovery: is something wrong? Drug Discovery Today 10 (2) (2005) 139–147. [27] A.J. de Ruijter, A.H. van Gennip, H.N. Caron, S. Kemp, A.B. van Kuilenburg, Histone deacetylases (HDACs): characterization of the classical HDAC family, The Biochemical Journal 370 (Pt 3) (2003) 737–749. [28] O. Witt, H.E. Deubzer, T. Milde, I. Oehme, HDAC family: what are the cancer relevant targets? Cancer Letters 277 (1) (2009) 8–21. [29] H.M. Prince, M.J. Bishton, S.J. Harrison, Clinical studies of histone deacetylase inhibitors, Clinical Cancer Research 15 (12) (2009) 3958–3969. [30] E.J. Calabrese, L.A. Baldwin, Chemotherapeutics and hormesis, Critical Reviews in Toxicology 33 (3–4) (2003) 305–353. [31] J. Gabrielsson, H. Dolgos, P.G. Gillberg, U. Bredberg, B. Benthem, G. Duker, Early integration of pharmacokinetic and dynamic reasoning is essential for optimal development of lead compounds: strategic considerations, Drug Discovery Today 14 (7–8) (2009) 358–372. [32] G. Levy, Impact of pharmacodynamic variability on drug delivery, Advanced Drug Delivery Reviews 33 (3) (1998) 201–206.