Pharmacokinetics in Drug Discovery: An Exposure-Centred Approach to Optimising and Predicting Drug Efficacy and Safety

Part of the Handbook of Experimental Pharmacology book series (HEP, volume 232)

Abstract

The role of pharmacokinetics (PK) in drug discovery is to support the optimisation of the absorption, distribution, metabolism and excretion (ADME) properties of lead compounds with the ultimate goal to attain a clinical candidate which achieves a concentration–time profile in the body that is adequate for the desired efficacy and safety profile. A thorough characterisation of the lead compounds aiming at the identification of the inherent PK liabilities also includes an early generation of PK/PD relationships linking in vitro potency and target exposure/engagement with expression of pharmacological activity (mode-of-action) and efficacy in animal studies. The chapter describes an exposure-centred approach to lead generation, lead optimisation and candidate selection and profiling that focuses on a stepwise generation of an understanding between PK/exposure and PD/efficacy relationships by capturing target exposure or surrogates thereof and cellular mode-of-action readouts in vivo. Once robust PK/PD relationship in animal PD models has been constructed, it is translated to anticipate the pharmacologically active plasma concentrations in patients and the human therapeutic dose and dosing schedule which is also based on the prediction of the PK behaviour in human as described herein. The chapter outlines how the level of confidence in the predictions increases with the level of understanding of both the PK and the PK/PD of the new chemical entities (NCE) in relation to the disease hypothesis and the ability to propose safe and efficacious doses and dosing schedules in responsive patient populations. A sound identification of potential drug metabolism and pharmacokinetics (DMPK)-related development risks allows proposing of an effective de-risking strategy for the progression of the project that is able to reduce uncertainties and to increase the probability of success during preclinical and clinical development.

Keywords

ADME Candidate profiling Drug discovery Exposure Lead generation Lead optimisation Pharmacokinetics PK/PD Prediction 

1 Introduction

Drugs can only exert their desired effects if they are able to bind to the intended target proteins in the body. Although this drug–target engagement is not a guarantee for efficacy, it is a prerequisite for pharmacological effects in the target cells. Efficacy thus is not only dependent on the potency of a drug but also on the exposure of the drug to the pharmacologically active site. Exposure means that the drug must reach the target site at sufficiently high concentrations and for a sufficiently long period of time after it has been administered to the patient. Pharmacokinetics (PK) is the discipline that explores the absorption, distribution, metabolism and excretion (ADME) behaviour of drugs which are the processes that control the kinetics of the concentration–time profile in the blood circulation and the body tissues and organs.

Traditional dose–response concepts are insufficient for the understanding of drug effects if the body and target tissue exposure of the drug is not being considered (Figs. 1 and 2). In a retrospective analysis of Phase II clinical trials, Morgan et al. (2012) extracted three “pillars of survival” for clinical proof-of-concept studies in patients: demonstration of (1) drug–target exposure, (2) drug–target binding and (3) expression of pharmacological activity. The authors concluded that an integrated understanding of the fundamentals of the pharmacokinetic (PK) and pharmacodynamic (PD) principles of a drug is a key success factor with preclinical experimental evidence for at least two of the three pillars significantly enhancing the success rate of drug discovery programmes during clinical development. The conclusions of this analysis have recently been further supported and expanded by a retrospective analysis of AstraZeneca’s project portfolio (Cook et al. 2014).
Fig. 1

Schematic illustration contrasting the past dose–response paradigm with the concept of dose–exposure–response relationships incorporating pharmacokinetic (PK) processes controlling the concentrations in the body (blue) and the processes required to elicit a pharmacological response (green), with the unbound plasma and (target) tissue concentrations representing the exposure link between dose and response

Fig. 2

Illustration of the impact of target exposure on efficacy. Drug A and B which are almost equipotent (unbound cellular IC50s of 150 and 210 nM, respectively) have been applied daily at the same dose in a tumour xenograft study in mice. The inefficacy of Drug B can be explained by the unfavourable relation between the unbound plasma concentration vs. IC50, while the strong tumour growth inhibition (TGI) effect of Drug A is in accordance with the target coverage of the unbound concentrations

Addressing exposure aspects right from the start of a new project and throughout all phases of the drug discovery process has a significant impact on the selection and optimisation of compounds and the prospect to turn them into viable drug candidates for preclinical and clinical development. The “free drug hypothesis” (Smith et al. 2010) is fundamental to this concept.

The wide acceptance that efficacy (and safety) is not only a result of the potency of a drug at the target (and off-target) protein, but also depends on the exposure of and the engagement with these target proteins has secured pharmacokinetics an immanent role in the drug discovery process. More than a decade since PK representatives have become integral part of drug discovery projects, the attrition rate of projects during clinical development due to PK liabilities went down significantly from originally 40% to less than 10% today (Kennedy 1997; Frank and Hargreaves 2003; Empfield and Leeson 2010; DiMasi et al. 2010, 2013). The attrition at the point of transition to preclinical development is even less allowing viable projects to progress from the discovery to the development phase with a higher probability of success.

This chapter describes how pharmacokinetics supports drug discovery based on an exposure-centred approach by identifying and optimising those PK liabilities which enable both the efficacy and safety of drug candidates and by designing in those ADME properties that result in an adequate PK profile. The role of pharmacokinetics in the different phases of drug discovery is outlined in Fig. 3 which also serves to structure the chapter.
Fig. 3

Overview of the main tasks of DMPK support during the different phases of drug discovery. TIV target identification and validation, PK pharmacokinetics, DM drug metabolism, MC medicinal chemistry

2 Pharmacokinetics in Drug Discovery

In the following, the main tasks and activities of pharmacokinetics during the different phases of drug discovery are described, emphasising the exposure-driven approach of project support with integral PK/PD thinking and considerations. Other DMPK aspects are only touched upon and may be followed up in more detail elsewhere (Kerns and Di 2008; Tsaioun and Kates 2011; Zhang and Surapaneni 2012; Smith et al. 2012; Wang and Urban 2014). Although the chapter has been written with oncology projects in mind, the principles outlined below are also applicable to other indications and small molecule drugs.

2.1 Target Validation

For projects that are based on a novel, unprecedented disease hypothesis efforts to validate the new drug target start at the initiation of the project. For these so-called first-in-class projects, there are by definition no tool compounds available, and early evidence for the validity of the target also depends on the availability of suitable animal models which may either involve animals in which the target has been knocked out or significantly attenuated (si-RNA). In these projects, the confidence in the target accumulates throughout the entire drug discovery and development process with the ultimate evidence coming not until the clinical proof-of-concept (PoC) studies.

If the programme is going for a best-in-class approach, there is a high level of confidence in the target, and both tool compounds and relevant animal models are available. A rigorous interpretation of the role of the proposed target in the disease strongly benefits from the availability of exposure data of tool compounds in the animal studies. To assess the compound exposure, plasma samples are being collected from the animals, and, after sample preparation, the compound is quantitated by LCMS/MS analysis. Typically, a crude time course with just a few sampling time points covering the dosing interval suffices. Together with the in vitro data of the fraction unbound in plasma in the animal species, the unbound concentration–time profile in plasma is plotted against a relevant potency parameter, e.g. the unbound IC50 from in vitro tests to see whether the systemic levels reached in the animal study were in the range to cover the target and to elicit the desired effect (Fig. 2). Visualisations of this kind particularly help understanding (1) whether a negative experimental outcome was due to insufficient target exposure or an incomplete understanding of the target and (2) whether a positive outcome of the animal study is in line with the compound exposure at the proposed pharmacological target.

2.2 Lead Generation

Once there is a satisfactory level of evidence for a new drug target, a lead finding strategy is being pursued. This typically involves a high-throughput screen (HTS) of large compound libraries in order to identify hits/hit clusters that can serve as lead structures on which subsequently a full lead optimisation (LO) programme may be based on.

From a DMPK point of view, there are two main objectives in this phase: (1) support of medicinal chemistry in assisting hit cluster evaluation and ranking through identification of the DMPK liabilities and assessment of the optimisation potential of the compounds in each cluster and (2) support of in vivo pharmacology studies with exposure-based advice on dose and schedule design of PD and efficacy studies.

Once the hit-to-lead process has narrowed down the number of hit clusters to a few, about 3–5 compounds per cluster are being subjected to in vitro ADME assays, in particular metabolic stability in animal and human liver microsomes and/or hepatocytes, Caco-2 permeability and efflux as well as CYP inhibition using human liver microsomes and CYP induction potential based on the PXR assay (Roberts 2001, 2003; Tsaioun and Kates 2011; Zhang and Surapaneni 2012; Wang and Urban 2014). Suitable compounds which are also pharmacologically active are subsequently submitted to rodent PK studies in vivo to examine how the in vitro ADME liabilities translate to in vivo, to determine the PK parameters and to assess their behaviour in the whole organism (Li et al. 2013). This allows to elucidate what type of DMPK liabilities a hit cluster may carry, how many liabilities there are in a given cluster, whether all compounds tested show the same type of liabilities and whether there are structural variations which could serve as promising starting points for chemical optimisation. The power of this analysis increases with a clear differentiation between properties which are relevant for the whole cluster and properties which are seen in single compounds only. In addition, we have made the experience that HTS runs using cellular potency screens result in higher ADME quality hits compared to HTS based on biochemical assays. Although hits often seem more potent in the latter, hit clusters identified by cellular assays tend to be more drug-like right from the start, ultimately making it more likely to turn them into viable drug candidates.

Our experience has also shown that hit clusters carrying more than three independent compound liabilities are very difficult to optimise during the LO phase. Independent liabilities are, for instance, poor potency and selectivity, low aqueous solubility, strong efflux, high clearance, CYP inhibition as well as CYP induction. In contrast, low permeability, low metabolic stability and low oral bioavailability are not considered independent if the latter liability is the consequence of the former two, e.g. low bioavailability due to low intestinal absorption or high hepatic first-pass. Similarly, a short half-life may be the consequence of high clearance.

Programmes with a difficult chemical starting point generally tend to consume a substantial amount of effort in many functions involved without being able to deliver in a viable drug candidate even if the team is given extra capacity and an extended period of time. In such instances, it is advisable to explore the chemical space by synthesising new compounds around the hit structure to assess the tractability of the chemical space in terms of how feasible it may become to improve the various liabilities in one chemical structure/molecule. Achieving only independent improvements of each liability in separate molecules may not be sufficient to forecast the effort needed during the LO phase and to increase the probability of success to find a viable development candidate.

The sound identification of the key liabilities of the lead cluster allows the design of a powerful and efficient screening tree for the LO phase. A typical example is shown in Fig. 4. The power of the screening tree often increases when focussing on the most relevant assays from the different functions and disciplines placed into a logic which follows the criticality of the assay with regard to the optimisation goal, the throughput of the assay and the filtering effect of the results on compound progression to the next tier of assays or studies.
Fig. 4

Example of a screening tree for early lead optimisation. Typical DMPK liabilities which are most frequently seen in lead compounds are shown on the right. Tailored to the specific liabilities of the lead structures in each project, the most appropriate assays are selected and placed into the screening tree depending on the type and significance of the liabilities, and the filtering effect of the assay results to discard or progress compounds to the next tier of assays which also takes into account the complexity and throughput of the assays. The green and blue colours of the boxes reflect PD and DMPK assays, respectively. Highlighted in red are those assays and study types which provide input for the generation of PK/PD relationships between in vitro potency, unbound exposure, PD and/or efficacy readouts in animal studies

Besides the identification of the liabilities of the hit cluster, the project team also defines the desired profile of the drug candidate as optimisation goal for the LO phase. The stepwise generation of a basic understanding of the relationship between the PK/exposure and the pharmacodynamics/efficacy is extremely beneficial already during this early phase of a programme. If the new programme is going to be a follow-up of an advanced preclinical or clinical compound, this is a must and forms the basis for the definition of the optimisation goals, together with the information on those liabilities or properties of the front-runner where the follow-up candidate has to improve on or differentiate against.

For a new programme with little or no precedence, early conductance of dedicated in vivo PK/PD studies have proven particularly informative to learn about the role of the target and the dynamics of target engagement for in vivo efficacy. This relationship can be studied already with compounds which are still suboptimal in terms of their PK properties, as long as sufficient exposure can be achieved in animals with unbound plasma concentrations approaching unbound in vitro IC50 values of target inhibition. For instance, highly potent compounds which still suffer from high metabolic instability/clearance and/or low absorption due to high efflux may be given at high doses in pharmacological in vivo studies, thereby overrunning/saturating these mechanisms. Sufficient plasma exposure can thus be achieved in animal PD models, allowing exploration of the relationship between unbound plasma concentrations, in vitro potency and in vivo efficacy readouts from quite early on. In case of very insoluble compounds, unusual formulations may be used as long as the excipients applied do not interfere with PD/efficacy readouts. The purpose of such studies is not to rescue pharmacokinetically insufficient compounds, but to extract as early as possible what type of in vivo concentration–time profile best enables the efficacy of a given target mode-of-action (MoA). Indeed compounds with a very short half-life are especially well suited to explore time dependencies in PK/PD relationships, e.g. to examine on/off requirements for target engagement.

The earlier and the better the understanding of the PK/PD relationship of the target is developing, the better the guidance that can be given to the project team. The impact may be severalfold: (1) Often there are diverse types of in vitro potency assays encompassing different biochemical and cellular systems with sometimes different assay conditions. Understanding the PK/PD relation helps to select the in vitro potency assay which is most relevant for in vivo efficacy. This should subsequently become the assay on which to base on structure–activity relationships during the LO phase. (2) Upon varying doses and administration schedules of PD/efficacy studies, different unbound plasma concentration–time profiles can be linked to differences in the in vivo coverage of in vitro potency/IC50 values. This may serve as first indirect evidence for target engagement and thereby will further strengthen the confidence in the target. (3) The PK/PD relationship allows delineating the type and shape of the plasma concentration–time profile needed to elicit intended pharmacodynamic effects. This information will in turn help to define the desired PK profile of the development candidate and to define what PK properties of the lead compound(s) need(s) to be optimised in order to get there (Fig. 5).
Fig. 5

Illustration of the iterative approach of applying different doses/schedules of a compound to explore the pharmacological activity in vivo based on defined MoA or efficacy readouts in relation to the coverage of the potency (e.g. unbound in vitro IC50) by the corresponding unbound plasma concentration–time profile. This information triggers a learning cycle to establish (1) what type of in vitro potency assay is most relevant for the in vivo activity and (2) what shape a concentration–time profile should have to enable the desired level of target engagement

2.3 Lead Optimisation

The decision to start the optimisation of a lead structure class endorses a significant investment into a project assigning the allocation of large amounts of resources in medicinal chemistry, pharmacology, drug metabolism and pharmacokinetics and many other disciplines to embark on a multidimensional optimisation of the chemical starting matter to improve the liabilities of the lead structure. The ultimate goal of the optimisation is to generate a drug candidate molecule which carries substantial evidence not only to be efficacious in a well-defined indication and patient population but also to be able to be administered safely and conveniently to humans.

DMPK efforts during the LO phase concentrate on those aspects which are critical (1) to change the PK profile of the compounds such as to enable efficacy, (2) to avoid/reduce the potential of the compounds to elicit safety risks and drug–drug interactions (DDI) both as victim and perpetrator when given to patients with indication-specific co-medications and (3) to ultimately allow for a human efficacious dose and dosing schedule which can be formulated and administered in a way that is convenient for clinical use.

Based on the growing understanding of which elements in the concentration–time profile are driving efficacy (the so-called PK/PD driver) and the desired route and scheme of administration, it can be simulated which and how the PK parameters of the lead compound and its analogues need to be optimised in order to turn them into potential drug candidates (Fig. 5). In this phase, it is important to be able to answer key questions such as the following: Which pharmacological IC50 assay variant is most relevant and hence predictive for in vivo efficacy? How much and how long should the unbound plasma concentration stay above this in vitro IC50 to consistently see in vivo efficacy with minimal side-effects? The earlier and the better the team learns to answer these questions, the more they can concentrate their efforts on those parameters which really make a difference rather than getting lost in aspects which are not on the critical path and simply seem easy to be addressed at first sight. The more stringent a team is able to focus, the more rapid it will make progress.

Identification of the elements in the concentration–time profile that are driving efficacy (e.g. AUC, time over IC50, Cmax, Ctrough, concentration at a time) requires application of varying the doses and dosing schedules in animal efficacy studies that cover a dose range from full to medium and low/no efficacy. Split-dose studies (e.g. 50 mg/kg vs. 2 × 25mg/kg) are very useful as the AUC remains the same, but Cmax and time over IC50 differ. If both schedules are equally efficacious, efficacy is more likely to be driven by AUC, while a higher efficacy of the latter may be suggestive of time over IC50 if this was longer in this schedule. This information can be used to guide PK optimisation depending on what aspect in the plasma concentration–time profile in the current compounds is still suboptimal. If the optimisation goal is to elevate unbound drug concentrations, ADME mechanisms to be optimised are aqueous solubility, intestinal permeability/efflux and/or metabolic clearance, whichever is/are the limiting factor(s) for a given class of compounds (Wang and Urban 2014; Reichel 2014, 2015). Noteworthy, the fraction unbound in plasma or tissues is not an optimisation parameter (Reichel 2009; Smith et al. 2010). If the optimisation goal is to increase the half-life of the LO compounds in order to extent the time over IC50 achievable by a single dose, the parameters to look out for improvements are clearance and volume of distribution. Clearance can be reduced by increasing the metabolic stability of the compounds or reducing the renal or biliary elimination (whichever is the main clearance pathway). The volume of distribution may be increased by making the molecules more lipophilic or basic (Smith et al. 2012).

Focussing on the critical path from a PK point of view means to concentrate on those aspects which really impact the concentration–time course of the LO compounds and their technical DMPK profile in the desired direction. This involves the generation of structure–activity relationships (SAR) and structure–potency relationships (SPR) to explore the chemical space and to elucidate those structural moieties which carry the highest potential for potency and selectivity (via SAR) and optimal physicochemical, DMPK and safety properties (via SPR). It is beyond the scope of this chapter to discuss the standard ADME and safety in vitro assays which have been described extensively in the literature (Kerns and Di 2008; Meanwell 2011; Tsaioun and Kates 2011; Zhang and Surapaneni 2012; Smith et al. 2012; Wang and Urban 2014). It is important that these assays run at high throughput and with short turnaround times which are fully in line with design–make–test–analyse learning cycles (Plowright et al. 2012) that are jointly driven by medicinal chemistry, pharmacology and DMPK (Fig. 6).
Fig. 6

Illustration of the design–make–test–analyse cycle in the LO phase. The in vitro assay in pharmacology and DMPK have to run at the same throughput and turnaround times to be able to synchronise the generation of structure–activity (SAR) and structure–property relationships (SPR) to avoid bias in the chemical synthesis. Optimised compounds are then submitted to in vivo studies (1) to examine the overall PK behaviour of the compound and (2) to evaluate the pharmacological activity in vivo in relation to unbound plasma concentrations and the in vitro potency (see Fig. 5). If possible safety readouts can be included in the in vivo studies. Depending on the level of understanding of the pharmacology of the target and the animal disease model, mode-of-action studies which can be performed after single doses or more chronic efficacy studies with repeated dosing schedules are carried out

A small core team with one representative from each of the three functions has been recognised to be the most efficient set-up in this process, in particular if they work along a clear hypothesis-driven way: i.e. defining a decision-related question for each compound prior to synthesising and subsequently submitting it to the relevant screening assays so to be able to rationalise their thoughts and to gradually transform the rapidly expanding amount of information into working knowledge rather than simply filling data bases very efficiently.

Once promising compounds have emerged from the early optimisation cycles, animal PK studies will be performed to verify the improvements made in vitro in the whole animal in vivo. Establishing and verifying the link between in vitro and in vivo (IVIVC) is important to ensure that the screening assays have the desired impact and that the screening tree is still relevant. Otherwise, the screening tree needs readjustment to respond to the chemical space and issues of the current compounds and to return on the critical path. While there are simple and convenient tools to examine IVIVCs such as correlation analysis between two or a few more properties, more sophisticated tools such physiologically based PK (PBPK) models offer the advantage to propagate in vitro ADME data through an in silico representation of the whole organism which allows simulation of the impact a given property change may have on the concentration–time profile (Parrott et al. 2005; Lüpfert and Reichel 2005; Peters et al. 2009; Chen et al. 2012; Rostami-Hodjegan 2012; Jones et al. 2015). The purpose of the application of PBPK modelling in this phase is not to be predictive of human but to diagnose ADME liabilities that are key to be improved and subsequently to aid the identification of compounds with an optimal balance of properties (and an acceptable compromise of their insufficiencies) to enable the desired concentration–time and PK profile.

Compounds which are expected to provide sufficient target exposure in animal PD models qualify for efficacy studies to explore their pharmacological profile in vivo. Prior information from the in vivo PK and/or pilot exposure studies supports the design of these studies in terms of suggesting doses and schedules which are most likely to be efficacious and for suggesting suitable time points for sampling of plasma and/or tissue. Exposure measurements which allow to capture a time course throughout the dosing interval are important and allow to dynamically link the concentration–time profile with the effects seen (or not seen). Dedicated dose–response studies with different dosing schedules are particularly powerful for PK/PD modelling and simulation (Gabrielsson et al. 2009, 2010, 2011; Bueters et al. 2013; Tuntland et al. 2014). If the design of these studies has been carefully thought through, important pieces of information can be extracted giving answers to questions such as the following: Can efficacy be linked to the unbound plasma concentrations? What element(s) in the plasma concentration–time profile drives efficacy (i.e. the PK/PD driver)? For how long should the target be exposed/engaged to elicit a certain level of efficacy? What would be the unbound concentration–time profile we like to see in humans to be efficacious in the clinic?

Single-dose studies with MoA readouts in target tissue can be very informative of target exposure, target engagement and subsequent expression of pharmacology when pharmacological readouts are taken dynamically, i.e. followed along a time course after dosing together with plasma and tissue samples to be analysed by means of LCMS/MS for the corresponding compound concentrations. The relation between the compound concentrations in plasma and target tissue and the extent and duration of target inhibition needed for efficacy forms the basis for quantitative PK/PD models which ultimately will be used to estimate the human efficacious dose.

While it is important to keep a clear focus on the critical path and to develop a growing understanding of the relationship between pharmacokinetics/exposure and pharmacodynamics/efficacy, there are a number of pitfalls on the way to identify a viable drug candidate:
  1. 1.

    The potency trap. Care should be taken that the screening tree is not steering chemical synthesis into the potency trap. Highly potent compounds often carry too much lipophilicity resulting in a prohibitive ADME profile which in most cases cannot be improved without losing potency altogether. Using potency measures such as the ligand binding efficiency index, where potency gains merely by lipophilicity increase are punished, can be very useful indicators to avoid this pitfall (Reynolds et al. 2008; Hopkins et al. 2014). Also, putting due emphasis on cellular potency readouts with demonstrated relevance for in vivo efficacy rather than on potency values from biochemical assays with recombinant target proteins helps avoiding this trap. Instead, biochemical assays should be used primarily for improving the selectivity of the compounds.

     
  2. 2.

    The fast and easy way. Beyond the many easy to synthesise chemical modifications, try also to get hold of more challenging to synthesise molecules in the chemical series if they are indicated by SAR/SPR analyses. If small structural changes only bring about small improvements, larger chemical modifications might be more fruitful. They also allow exploration of the distant corners in the chemical space and may direct teams on the (unexpected) path to the ultimate development candidate (Lücking et al. 2013; Hartung et al. 2013).

     
  3. 3.

    Getting lost in too much. Attempting optimisation of all liabilities at once often proves to be overtly challenging, not so much in terms of capacity but in terms of rationalising too many SAR and SPR relationships simultaneously. It may be more satisfactory to focus on the resolution of the most crucial issues initially and address other issues once significant improvements have been achieved. The then new chemical matter might no longer suffer from other liabilities or may have other issues to take care of. The paradigm of “all compounds in all assays as early as possible” produces a lot of data and noise, and the overwhelming amount makes it difficult to extract useful knowledge for decision-making. Well-defined rational design–make–test–analyse cycles with a relevant question behind each compound and every assay the compound is submitted to turn out to be more instructive besides saving capacity (Ballard et al. 2012; Plowright et al. 2012).

     
  4. 4.

    Losing sight of the relevant issues. The validity of the screening assays in terms of delivering real improvements on the key liabilities should be repeatedly checked by in vivo studies during lead optimisation. For instance, IVIVC between solubility and/or Caco-2 permeability/efflux and oral bioavailability, and metabolic stability and total clearance should be confirmed throughout the LO phase to make sure that a significant improvement seen in vitro indeed brings about the anticipated improvement in vivo. This ensures that the screening tree is still on the right track and indicates if new issues have turned up in the current compounds which need to be taken care of in a different set of screening assays.

     
  5. 5.

    Clinging on for too long. Sometimes even excessive optimisation efforts in medicinal chemistry, pharmacology and DMPK do not result in sufficient improvements of the current compounds compared to the lead structures. Better results on one liability lead to a deterioration of one or more of the other liabilities and vice versa. One issue may be solved, but other issues surface up and no way out seems on the horizon where at least the most important liabilities can be resolved in one molecule. When the team has arrived at such a situation, it may be hard but unavoidable to ask the “killer question” (Cook et al. 2014) rather than continuing and hoping for the “lucky shot”. A stringent exploration of the chemical space at the virtual interface between ADME space and potency/selectivity space helps supporting at a clear Go/Nogo decision. Abandoning an un-optimisable lead structure in due course allows redirecting of valuable LO resources to other programmes where they will have more impact. Projects should not be terminated, however, without extracting lessons learned about the reasons for failure, being related either to the target or to the chemical matter, each having important but different implications for future programmes. Convincing evidence for the former reason is the demonstration of target exposure and expression of pharmacological activity (MoA) with consistently no impact on the efficacy in animal models.

     
  6. 6.

    Stay open minded. While the virtue of a rational, hypothesis-driven way of working with continuous integration of data and extraction of knowledge and understanding is very obvious, there are also situations where the data do not seem to make sense or do not seem to fit in the current thinking. Trying to understand such discrepancies instead of ignoring them is important and can be very rewarding. They might either be due to experimental issues which need resolving to make the assay more relevant or they may point to a missing piece in the understanding of the project team which can turn out to become crucial for the progression of the project. Face-to-face brainstorming discussions in small groups tend to be more productive than trying to resolve such issues by electronic communication means. Not only may interactive discussions be more productive, they are also more likely to bring out creative ideas and solutions and even may be starting points for novel compounds, novel targets and indeed new projects.

     

After several LO cycles, the current compounds will carry significant improvements on the liabilities seen in the original lead structures. Arrival at this stage starts the late LO phase which besides continued screening efforts embarks on a broader examination of the most promising compounds in vivo, e.g. more sophisticated animal PD and efficacy models, non-rodent PK studies as well as pilot toxicology and safety pharmacology studies. These activities gradually lead into the phase of candidate selection and profiling.

2.4 Candidate Selection and Profiling

While the LO phase embarks on changing particular properties in a molecule, the candidate selection phase concentrates on a broad characterisation of the compounds that are finally emerging from the optimisation efforts of the programme. The purpose of the characterisation is to examine whether one of the most promising molecules qualifies as potential drug candidate. Apart from an in-depth characterisation by pharmacology, pharmacokinetics and drug metabolism, the compounds are also scrutinised by toxicology, safety pharmacology and formulation development (see, for instance, Pelkonen et al. 2002, Zhang and Surapaneni 2012, and Wang and Urban 2014). In the following, attention is paid to those PK aspects which relate to efficacy and safety: (1) predicting the PK in human, (2) estimating the human therapeutic exposure, (3) predicting the therapeutic dose in patients and (4) estimating the therapeutic window in human. This phase will also identify the potential DMPK-related risks which have to be addressed specifically during preclinical and clinical development to de-risk the programme and to allow assessing of the future investments needed and the likely probability of success.

2.4.1 Prediction of Human PK

The anticipation of the PK behaviour in humans has the intention to predict the exposure and plasma concentration–time profile in humans for a given dose applied to patients via the intended route of application. This prediction is fundamental to estimate the human therapeutic dose as well as the expected therapeutic window.

The prediction of the PK in humans is a complex exercise and cannot be dealt with in detail in this chapter. It involves (1) the generation of in vitro ADME data both in animal and in human systems, in particular with regard to clearance mechanisms; (2) in vivo PK studies in at least 2–3 species, including rodent and non-rodents; and (3) the integration of this data into a framework which allows to predict the PK profile in humans (Beaumont and Smith 2009; van den Bergh et al. 2011; Grime et al. 2013). The methodologies and approaches have been reviewed and evaluated extensively across industry in a recent PHRMA initiative (Poulin et al. 2011). We have made good experiences with a combination of allometric interspecies scaling and human in vitro data including simultaneous fitting of the plasma concentration–time profiles of the animal species and physiologically based approaches. Human PK predictions are carried out for the following PK parameters: clearance (CL), volume of distribution (Vss), area under the plasma concentration–time curve (AUC), elimination half-life (t1/2), oral bioavailability (F%) and the highest and the lowest plasma concentration (Cmax and Cmin or Ctrough, respectively). In addition, plasma concentration–time profiles will be simulated for the intended route of application.

While allometry generally performs well for volume of distribution when corrected for plasma protein binding, this approach performs clearance less well, particularly if there are species differences in the rate and/or mechanism(s) of clearance which cannot be corrected for by plasma protein binding, intrinsic metabolic clearance or other correction factors (Mahmood 2005; Beaumont and Smith 2009). In such cases, single species scaling can be applied to derive different scenarios of the predicted human PK parameters and plasma concentration–time profile. Confidence in the predictions depends as much on experience, context information and tacit knowledge from this and other programmes as it does on actionable study suggestions which allow to add further confidence when going ahead towards or during preclinical development.

Another very important aspect, although only indirectly related to human PK prediction as such, should be mentioned here. Up to this point in time, the oral bioavailability generally has been examined from oral solutions based on rather undefined material. In order to estimate the oral bioavailability from well-defined compound material (e.g. microcrystalline material of defined particle size), the relative oral bioavailability between the two states is being determined in rats at an equivalent of the human therapeutic dose. A relative oral bioavailability of below 50% indicates the need for special formulation efforts to overcome potential dissolution limitations of the drug material, while high values indicate that a standard immediate release tablet formulation is feasible to achieve the anticipated exposure in humans (Muenster et al. 2011).

2.4.2 Estimation of Therapeutic Exposure in the Patient

Although tempting, the human therapeutic exposure cannot simply be taken from the efficacious exposure in an efficacy study, even if corrections are made for plasma protein binding to obtain the unbound AUC and plasma concentration–time profile and for possible species differences in potency.

In contrast, the estimation of the pharmacologically active concentration–time profile in humans should be based on a set of dedicated PK/PD studies with quantitative data analysis and evaluation. These should include studies (1) to identify or confirm the PK/PD driver of efficacy of the development candidate in relevant PD and/or efficacy models and (2) to quantitatively describe the relationship between the plasma concentration–time profile and the level of efficacy seen for different doses and dosing schedules in models which are relevant for the intended patient population. The predictive power and hence the confidence in PK/PD model increases by incorporating the PK/PD understanding which has evolved in the LO phase not only on the candidate compound but also on other compounds in the project. If available, front-runner compounds or competitor drugs can be very valuable to develop and validate prediction algorithms. In oncology, for example, the translatability of the efficacious exposure in tumour xenograft studies in mice to the efficacious exposure in human patients has been demonstrated using a set of antitumour drugs used in human patients (Simeoni et al. 2013). This approach quantitatively links the ability of different drugs to inhibit tumour growth to the exposure in a set of dose–response studies based on an in vivo potency measure which directly correlates with the human efficacious AUC of the antitumour drugs tested (Rocchetti et al. 2007). Capturing the tumour growth rates following different dosing regimens also allows the estimation of experiment-independent parameters, thereby increasing the confidence in the model parameters. PK/PD models to estimate the efficacious exposure can be further enhanced by integration of target engagement markers or subsequent MoA readouts and by attempting to establish a temporal link between the expression of pharmacological activity upon target binding with more “distant” efficacy endpoints such as tumour stasis or regression (Yamazaki 2013; Yamazaki et al. 2015; Venkatakrishnan et al. 2015).

Besides the confidence in the relevance and translatability of the results from animal models to the human situation, the relevance of unbound plasma concentrations for the concentrations in the effect compartment needs to be understood. For targets residing in the body’s periphery with unrestricted proximity to the blood circulation, the time course of the unbound plasma concentrations most likely will correspond to the unbound concentration at the target site. This can be demonstrated experimentally by plotting the time course of total plasma vs. total target tissue concentrations. If they run in parallel, the unbound concentrations will be the same in both compartments, i.e. the unbound plasma concentration fully represents the free concentration of the compound at the target site. This information on plasma-target site equilibrium is very important for clinical studies where generally only plasma concentrations are accessible. If the target site in the body is not in direct correspondence with the plasma compartment, a possible exposure difference and/or time delay should be evaluated (Gabrielsson et al. 2009, 2011). For targets which are not in proximity to the general blood circulation or even locate behind a physiological barrier, e.g. CNS, additional data are needed to translate unbound plasma concentrations into relevant effect compartment concentrations (Reichel 2009, 2014, 2015).

PK/PD modelling based on popPK approaches combining different, independent experiments and using the data points from individual animals is very powerful to reduce noise and to extract experiment-independent parameters for the predictions.

The predicted therapeutic exposure in humans has also to be consistent with the existing knowledge of the target biology. Predictions carry particularly high confidence if they are based on experimental evidence for target engagement and expression of pharmacological activity (MoA) in an animal model which is relevant for the intended indication and patient population.

2.4.3 Prediction of Human Therapeutic Dose

Conceptually, human dose estimation is simple (Fig. 7): The predicted PK, i.e. dose–exposure relationship in humans from above, is used to estimate a dose/dosing schedule for humans which is anticipated to mimic the unbound concentration–time profile that is expected to be efficacious as well as safe in humans (Heimbach et al. 2009; Venkatakrishnan et al. 2015).
Fig. 7

Schematic illustration of the prediction of the anticipated efficacious therapeutic dose in human. The approach integrates information on (1) the therapeutic exposure that is expected to be pharmacologically active in patients (left panel), (2) the predicted PK parameters and plasma concentration–time profile in humans (middle panel) and (3) the estimated therapeutic window based on the no adverse effect levels (NOAELs) in animal toxicology and safety pharmacology studies (right panel)

The accuracy of the predicted dose depends as much on the prediction of the human PK as it does on the estimated therapeutic exposure. The estimation of the human therapeutic dose often results in a range originating from different scenarios that reflect the most relevant uncertainties in the current understanding and set of data available. These may relate to both the prediction of the human PK and the estimation of the human therapeutic exposure. The use of PD or disease models in different preclinical species helps in eliminating any species differences between pharmacokinetics and pharmacodynamics, thereby refining the extrapolation to the human situation. For best-in-class projects, the confidence in the dose estimations may be further increased by applying the same prediction algorithm to competitor compounds with proven clinical efficacy and using possible differences as correction factors (Lowe et al. 2007).

The design of PK/PD studies which specifically elude on possibilities for intermittent dosing schedules along with information on the mode-of-action in target cells and/or in surrogate cells (e.g. in the circulation) as well as relevant PD readouts will be informative for alternative scheduling options. In vitro washout experiments where the time course of the cellular activity is followed after removal of the compound may provide important input to model such alternative intermittent dosing regimens with high probability for efficacy. Such studies are also helpful to delineate any temporal disconnects between pharmacokinetic and pharmacodynamics as a consequence of time events along the downstream pharmacological cascade. In oncology, intermittent dosing schedules are particularly attractive as they may increase the therapeutic window of the antitumour agents.

Predictions on alternative treatment schedules may also support the strategic positioning of the drug for clinical use and/or to support treatment schedule or dosing related competitive advantages. The prediction of the therapeutic dose when the drug candidate is going to be used in combination with other drugs is an important consideration for the predictions in oncology (Simeoni et al. 2013). For drug combinations, it is important to examine whether the interaction is additive or synergistic (Schmieder et al. 2013) and to exclude that the additional benefit is due to pharmacokinetic drug–drug interactions, e.g. via CYP inhibition. For compounds where pharmacologically active metabolites are expected to significantly contribute to efficacy, these need to be incorporated quantitatively into the PK/PD models used for the prediction of the human therapeutic dose of the proposed drug candidate.

2.4.4 Estimation of the Therapeutic Window in Human

The therapeutic window is the ratio between the unbound exposure which does not yet show first signs of toxicity and the unbound exposure that is needed for efficacy. A first estimate of the therapeutic window derives from the comparison of the unbound efficacious AUC (from animal PD/efficacy studies) with the unbound AUC in toxicology studies/species which did not elicit adverse effects, i.e. NOAELs. In addition, a range of safety pharmacology studies is performed to estimate plasma concentrations which affect key physiological functions. As conservative estimates, the maximal unbound plasma concentrations are used to evaluate safety risks, i.e. Cmax,u, which are in turn compared to unbound Cmax concentrations predicted for the highest efficacious dose in humans.

Whereas regulatory safety pharmacology and toxicology studies are being performed prior to entry into human according to authorities’ guidelines, potential development candidates are evaluated in non-GLP pilot toxicology studies. They encompass, for instance, 2-week toxicology studies with three dose levels in rodents starting from the dose corresponding to the efficacious unbound AUC with the other two doses representing different multiples thereof. The study design is project and compound specific, depending on prior knowledge from in vivo PK and PD studies, the indication space and the intended patient population.

The selection of the species for the toxicological examination of the compound depends on the metabolic characterisation of the compound, i.e. species similarities in the metabolite pattern with that generated by human hepatocytes. If a human-specific metabolite is not generated in the tox species, it may be synthesised by medicinal chemistry and applied directly in additional study arms. Also important is the achievability of multiples of exposures using microcrystalline material and formulations which can be used later by regulatory toxicology studies. In cases where this poses difficulties, special activities from formulation development are required to progress the project.

The estimation of the therapeutic window also affects the evaluation of potential drug–drug interaction (DDI) risks, depending on whether potential AUC increases would stay within a large therapeutic window or go beyond, thereby putting patients on co-medications at risk. DDI risks may arise from CYP inhibition and/or induction (Prueksaritanont et al. 2013) but also from interactions with transporter proteins at the level of absorption, distribution or elimination (Giacomini et al. 2010).

While current estimations of the therapeutic window are often static, i.e. based on AUC or Cmax values, dynamic modelling approaches are used more often as they better reflect the dynamics of the concentration effect relations both for efficacy and safety readouts (Muller and Milton 2012; Parkinson et al. 2013).

2.4.5 Identification DMPK-Related Development Risks

Experience shows that there is hardly any project without issues. The better and the earlier these issues – or risks – are known, the better one can react on them through dedicated de-risking measures. DMPK-related development risks may relate to a number of aspects such as:
  1. 1.
    Safe use in humans, e.g.:
    • Potential for drug–drug interactions as perpetrator and/or as victim, occurrence of toxic or reactive metabolites

    • Nonlinear PK in the therapeutic dose range

    • Potential for large interindividual variability due to involvement of polymorphic drug-metabolising enzymes

    • Drug accumulation due to very long half-life

    • Clearance pathway not compatible with special patient populations, e.g. renal insufficiency

     
  2. 2.
    Efficacious use in humans, e.g.:
    • Intestinal absorption issues

    • Formulation-related issues

    • CYP induction/autoinduction

    • Potential for large interindividual variability due to involvement of polymorphic drug-metabolising enzymes

     
  3. 3.

    Difficulty to achieve multiples of exposure of the parent drug and relevant human metabolites in the species used for in vivo toxicology and safety assessment, thereby limiting the exposure range in regulatory toxicology studies and hence complicating or even restraining dose escalations during clinical development

     

In addition to many other aspects which are not the subject of this chapter, the decision to start preclinical development takes into account the complete ADME and DMPK characterisation of the compound, the anticipated PK behaviour of the compound in humans, the expected therapeutic exposure and predicted dose regimen in patients, the estimated safety margins and the development risks identified. Any risk that has been identified is a strength rather than a weakness of the project as only the issues which have been identified yet allow proposals of appropriate risk mitigation steps at the various phases of preclinical and clinical development.

A problem-solving attitude to design and subsequently test project-specific in vivo study protocols integrating PK, PD, BM, Tox endpoints can provide extremely useful information for the understanding of and confidence in the compound and its subsequent progression into drug development. Working in cross-functional teams with preclinical and clinical experts allows tailoring of a clear path forward to resolve any issues identified and to carve out an early clinical development plan that permits an explicit testing strategy of the disease hypothesis in POC study designs that are powerful and conclusive and unbiased by exposure concerns.

It has to be noted that the handover of a project from discovery to development is not a point transition. Even though the responsibility during development is taken on by development functions, discovery PK continues to accompany the progression of the project and to contribute with their knowledge and expertise to resolving any issues that are coming up. Once human PK data from phase I studies are available, these are used by discovery PK to compare them with the predictions made and to explore any deviations observed. These comparisons are very rewarding and an important source to sharpen the tools and approaches to predict the PK behaviour in humans. Similarly, human data related to pharmacodynamic or efficacy readouts will be used to revisit the PK/PD relationships made in order to derive lessons learned and to grow confidence in the approach for future programmes.

3 Summary and Outlook

Since its foundation, discovery pharmacokinetics has moved a long way from starting off as service function which served to improve basic PK parameters such as clearance, half-life and bioavailability in potency-optimised compounds in response to the excessively high attrition of 40% due to poor pharmacokinetic properties of drug candidates during clinical testing in the 1980s and early 1990s (Kennedy 1997). Even in its rather service-oriented role, the impact of addressing ADME issues from early on in drug discovery programmes was astonishing, reducing PK-related failure rates down to 10% in about a decade (Kola and Landis 2004). The impressive effect has owned discovery DMPK not only full acceptance; it is now seen at the core of drug discovery being an integral part of every drug discovery project (Smith 2011).

Still, beyond designing of adequate DMPK properties into the new drug candidates, there is a great deal of contribution left to help reducing the unacceptably high attrition due to insufficient efficacy and unacceptable safety (Kola and Landis 2004; Morgan et al. 2012; Cook et al. 2014). A rigorous application of pharmacokinetic principles in order to gain a quantitative understanding of the role of new drug targets in proposed disease mechanisms and subsequently in the patients by means of rigorous PK/exposure and PD/efficacy modelling and simulation approaches will be very powerful to increase the success rate of drug discovery and development. Optimising ADMET properties of lead compounds during the LO phase in conjunction with building up of a growing understanding of those PK/PD properties that are required to elicit efficacy should be a key guiding principle during the LO phase. The generation of quantitative PK/PD models is a powerful approach to support the selection of adequate candidate drugs for preclinical and clinical development. The availability of such models will facilitate understanding of the relationship between target exposure and target engagement, the subsequent expression of pharmacological activity along with the proposed mode-of-action as well as pharmacodynamic and efficacy endpoints first in animal models and ultimately in patients (Danhof et al. 2007, 2008). Supporting drug discovery and development by exposure-based modelling and simulation (M&S) activities along the various phases of the project has been proposed to be a very powerful means to increase the success of clinical candidates (Morgan et al. 2012; Visser et al. 2013; Cook et al. 2014; Jones et al. 2015).

In oncology, for example, empirical and mechanistic PK/PD modelling and simulations of new antitumour agents has been demonstrated to add significant value to (1) the understanding of the target biology, (2) the confidence in a new disease hypothesis, (3) the identification of optimal doses and dosing schedules and (4) the selection of responsive patient groups (Venkatakrishnan et al. 2015). Translational modelling of preclinical tumour growth inhibition data from animal models to the effects in patients with regard to surrogate response markers, tumour growth inhibition and progression-free survival is an area which can benefit strongly from back translation of incoming clinical data to the discovery teams. Currently, M&S activities which aim at establishing a quantitative link between PK/exposure and mechanistic or semi-mechanistic models of biological pathway modulation are rapidly growing (Gabrielsson and Hjorth 2012; Yamazaki 2013; Venkatakrishnan et al. 2015).

Ultimately, such quantitative pharmacology models may not only be predictive of the effect of the inhibition of one target; they may also allow to simulate the interference of a drug or a combination of drugs with a network of targets, thereby giving rise to the discovery and development of new generations of medicines for treating diseases beyond the classical one drug–one target–one disease paradigm.

Although, today, we are still a long way from understanding and translating cellular effects to pharmacodynamic and safety-related effects in tissues and further to clinical readouts, the exposure-based PK support is a powerful approach that helps advancing (1) the discovery, optimisation, selection and characterisation of high-quality drug candidates with PK/PD properties that carry sufficient potential to show efficacy and safety in patients at the predicted dose and dosing scheduling and (2) the development through ultimately allowing a more rigorous testing of new disease hypotheses in patients without bias due to uncertainty in target exposure, making clinical proof-of-concept studies much more powerful than in the past.

References

  1. Ballard P, Brassil P, Bui KH, Dolgos H, Petersson C, Tunek A, Webborn PJ (2012) The right compound in the right assay at the right time: an integrated discovery DMPK strategy. Drug Metab Rev 44(3):224–252CrossRefPubMedGoogle Scholar
  2. Beaumont K, Smith DA (2009) Does human pharmacokinetic prediction add significant value to compound selection in drug discovery research? Curr Opin Drug Discov Dev 12(1):61–71Google Scholar
  3. Bueters T, Ploeger BA, Visser SA (2013) The virtue of translational PKPD modeling in drug discovery: selecting the right clinical candidate while sparing animal lives. Drug Discov Today 18(17–18):853–862CrossRefPubMedGoogle Scholar
  4. Chen Y, Jin JY, Mukadam S, Malhi V, Kenny JR (2012) Application of IVIVE and PBPK modeling in prospective prediction of clinical pharmacokinetics: strategy and approach during the drug discovery phase with four case studies. Biopharm Drug Dispos 33(2):85–98CrossRefPubMedGoogle Scholar
  5. Cook D, Brown D, Alexander R, March R, Morgan P, Satterthwaite G, Pangalos MN (2014) Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nat Rev Drug Discov 13(6):419–431CrossRefPubMedGoogle Scholar
  6. Danhof M, de Jongh J, De Lange EC, Della Pasqua O, Ploeger BA, Voskuyl RA (2007) Mechanism-based pharmacokinetic-pharmacodynamic modeling: biophase distribution, receptor theory, and dynamical systems analysis. Annu Rev Pharmacol Toxicol 47:357–400CrossRefPubMedGoogle Scholar
  7. Danhof M, de Lange EC, Della Pasqua OE, Ploeger BA, Voskuyl RA (2008) Mechanism-based pharmacokinetic-pharmacodynamic (PK-PD) modeling in translational drug research. Trends Pharmacol Sci 29(4):186–191CrossRefPubMedGoogle Scholar
  8. DiMasi JA, Feldman L, Seckler A, Wilson A (2010) Trends in risks associated with new drug development: success rates for investigational drugs. Clin Pharmacol Ther 87(3):272–277CrossRefPubMedGoogle Scholar
  9. DiMasi JA, Reichert JM, Feldman L, Malins A (2013) Clinical approval success rates for investigational cancer drugs. Clin Pharmacol Ther 94(3):329–335CrossRefPubMedGoogle Scholar
  10. Empfield JR, Leeson PD (2010) Lessons learned from candidate drug attrition. IDrugs 13(12):869–873PubMedGoogle Scholar
  11. Frank R, Hargreaves R (2003) Clinical biomarkers in drug discovery and development. Nat Rev Drug Discov 2(7):566–580CrossRefPubMedGoogle Scholar
  12. Gabrielsson J, Hjorth S (2012) Quantitative pharmacology: an introduction to integrative pharmacokinetic-pharmacodynamic analysis. Apotekarsocieteten, Stockholm, p 262Google Scholar
  13. Gabrielsson J, Dolgos H, Gillberg PG, Bredberg U, Benthem B, Duker G (2009) Early integration of pharmacokinetic and dynamic reasoning is essential for optimal development of lead compounds: strategic considerations. Drug Discov Today 14:358–372CrossRefPubMedGoogle Scholar
  14. Gabrielsson J, Green AR, Van der Graaf PH (2010) Optimising in vivo pharmacology studies–practical PKPD considerations. J Pharmacol Toxicol Methods 61(2):146–156CrossRefPubMedGoogle Scholar
  15. Gabrielsson J, Fjellström O, Ulander J, Rowley M, Van Der Graaf PH (2011) Pharmacodynamic-pharmacokinetic integration as a guide to medicinal chemistry. Curr Top Med Chem 11(4):404–418CrossRefPubMedGoogle Scholar
  16. Giacomini KM, Huang SM, Tweedie DJ, Benet LZ, Brouwer KL, Chu X, Dahlin A, Evers R, Fischer V, Hillgren KM, Hoffmaster KA, Ishikawa T, Keppler D, Kim RB, Lee CA, Niemi M, Polli JW, Sugiyama Y, Swaan PW, Ware JA, Wright SH, Yee SW, Zamek-Gliszczynski MJ, Zhang L (2010) Membrane transporters in drug development. Nat Rev Drug Discov 9(3):215–236CrossRefPubMedGoogle Scholar
  17. Grime KH, Barton P, McGinnity DF (2013) Application of in silico, in vitro and preclinical pharmacokinetic data for the effective and efficient prediction of human pharmacokinetics. Mol Pharm 10(4):1191–1206CrossRefPubMedGoogle Scholar
  18. Hartung IV, Hitchcock M, Pühler F, Neuhaus R, Scholz A, Hammer S, Petersen K, Siemeister G, Brittain D, Hillig RC (2013) Optimization of allosteric MEK inhibitors. Part 1: venturing into underexplored SAR territories. Bioorg Med Chem Lett 23(8):2384–2390CrossRefPubMedGoogle Scholar
  19. Heimbach T, Lakshminarayana SB, Hu W, He H (2009) Practical anticipation of human efficacious doses and pharmacokinetics using in vitro and preclinical in vivo data. AAPS J 11(3):602–614CrossRefPubMedPubMedCentralGoogle Scholar
  20. Hopkins AL, Keserü GM, Leeson PD, Rees DC, Reynolds CH (2014) The role of ligand efficiency metrics in drug discovery. Nat Rev Drug Discov 13(2):105–121CrossRefPubMedGoogle Scholar
  21. Jones HM, Chen Y, Gibson C, Heimbach T, Parrott N, Peters SA, Snoeys J, Upreti VV, Zheng M, Hall SD (2015) Physiologically based pharmacokinetic modeling in drug discovery and development: a pharmaceutical industry perspective. Clin Pharmacol Ther 97(3):247–262CrossRefPubMedGoogle Scholar
  22. Kennedy T (1997) Managing the drug discovery/development interface. Drug Discov Today 2(10):436–444CrossRefGoogle Scholar
  23. Kerns E, Di L (2008) Drug-like properties: concepts, structure design and methods: from ADME to toxicity optimization. Academic, Amsterdam, p 552Google Scholar
  24. Kola I, Landis J (2004) Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov 3(8):711–715CrossRefPubMedGoogle Scholar
  25. Li C, Liu B, Chang J, Groessl T, Zimmerman M, He YQ, Isbell J, Tuntland T (2013) A modern in vivo pharmacokinetic paradigm: combining snapshot, rapid and full PK approaches to optimize and expedite early drug discovery. Drug Discov Today 18(1–2):71–78CrossRefPubMedGoogle Scholar
  26. Lowe PJ, Hijazi Y, Luttringer O, Yin H, Sarangapani R, Howard D (2007) On the anticipation of the human dose in first-in-man trials from preclinical and prior clinical information in early drug development. Xenobiotica 37:1331–1354CrossRefPubMedGoogle Scholar
  27. Lücking U, Jautelat R, Krüger M, Brumby T, Lienau P, Schäfer M, Briem H, Schulze J, Hillisch A, Reichel A, Wengner AM, Siemeister G (2013) The lab oddity prevails: discovery of pan-CDK inhibitor (R)-S-cyclopropyl-S-(4-{[4-{[(1R,2R)-2-hydroxy-1-methylpropyl]oxy}-5-(trifluoromethyl) pyrimidin-2-yl]amino}phenyl)sulfoximide (BAY 1000394) for the treatment of cancer. ChemMedChem 8(7):1067–1085CrossRefPubMedGoogle Scholar
  28. Lüpfert C, Reichel A (2005) Development and application of physiologically based pharmacokinetic-modeling tools to support drug discovery. Chem Biodivers 2(11):1462–1486CrossRefPubMedGoogle Scholar
  29. Mahmood I (2005) Interspecies pharmacokinetic scaling: principles and application of allometric scaling. Pine House Publishers, Rockville, p 393Google Scholar
  30. Meanwell NA (2011) Improving drug candidates by design: a focus on physicochemical properties as a means of improving compound disposition and safety. Chem Res Toxicol 24(9):1420–1456CrossRefPubMedGoogle Scholar
  31. Morgan P, Van Der Graaf PH, Arrowsmith J, Feltner DE, Drummond KS, Wegner CD, Street SD (2012) Can the flow of medicines be improved? Fundamental pharmacokinetic and pharmacological principles toward improving Phase II survival. Drug Discov Today 17(9–10):419–424CrossRefPubMedGoogle Scholar
  32. Muenster U, Pelzetter C, Backensfeld T, Ohm A, Kuhlmann T, Mueller H, Lustig K, Keldenich J, Greschat S, Göller AH, Gnoth MJ (2011) Volume to dissolve applied dose (VDAD) and apparent dissolution rate (ADR): tools to predict in vivo bioavailability from orally applied drug suspensions. Eur J Pharm Biopharm 78(3):522–530CrossRefPubMedGoogle Scholar
  33. Muller PY, Milton MN (2012) The determination and interpretation of the therapeutic index in drug development. Nat Rev Drug Discov 11(10):751–761CrossRefPubMedGoogle Scholar
  34. Parkinson J, Visser SA, Jarvis P, Pollard C, Valentin JP, Yates JW, Ewart L (2013) Translational pharmacokinetic-pharmacodynamic modeling of QTc effects in dog and human. J Pharmacol Toxicol Methods 68(3):357–366CrossRefPubMedGoogle Scholar
  35. Parrott N, Paquereau N, Coassolo P, Lavé T (2005) An evaluation of the utility of physiologically based models of pharmacokinetics in early drug discovery. J Pharm Sci 94(10):2327–2343CrossRefPubMedGoogle Scholar
  36. Pelkonen O, Baumann A, Reichel A (2002) Pharmacokinetic challenges in drug discovery. Ernst Schering research foundation workshop, vol 37. Springer, Heidelberg, p 306Google Scholar
  37. Peters SA, Ungell AL, Dolgos H (2009) Physiologically based pharmacokinetic (PBPK) modeling and simulation: applications in lead optimization. Curr Opin Drug Discov Dev 12(4):509–518Google Scholar
  38. Plowright AT, Johnstone C, Kihlberg J, Pettersson J, Robb G, Thompson RA (2012) Hypothesis driven drug design: improving quality and effectiveness of the design-make-test-analyse cycle. Drug Discov Today 17(1–2):56–62CrossRefPubMedGoogle Scholar
  39. Poulin P, Jones RD, Jones HM, Gibson CR, Rowland M, Chien JY, Ring BJ, Adkison KK, Ku MS, He H, Vuppugalla R, Marathe P, Fischer V, Dutta S, Sinha VK, Bjornsson T, Lave T, Yates JW (2011) PHRMA CPCDC initiative on predictive models of human pharmacokinetics. Part 5: prediction of plasma concentration-time profiles in human by using the physiologically-based pharmacokinetic modeling approach. J Pharm Sci 100(10):4127–4157CrossRefPubMedGoogle Scholar
  40. Prueksaritanont T, Chu X, Gibson C, Cui D, Yee KL, Ballard J, Cabalu T, Hochman J (2013) Drug-drug interaction studies: regulatory guidance and an industry perspective. AAPS J 15(3):629–645CrossRefPubMedPubMedCentralGoogle Scholar
  41. Reichel A (2009) Addressing central nervous system (CNS) penetration in drug discovery: basics and implications of the evolving new concept. Chem Biodivers 6(11):2030–2049CrossRefPubMedGoogle Scholar
  42. Reichel A (2014) Integrated approach to optimizing CNS penetration in drug discovery: from the old to the new paradigm and assessment of drug–transporter interactions. In: Hammarlund-Udenaes M, de Lange E (Autor, Herausgeber), Thorne RG (eds) Drug delivery to the brain. AAPS Advances in the pharmaceutical sciences series. Springer, New York, pp 339–374Google Scholar
  43. Reichel A (2015) Pharmacokinetics of CNS penetration. In: Di L, Kerns EH (eds) Blood-brain barrier in drug discovery: optimizing brain exposure of CNS drugs and minimizing brain side effects for peripheral drugs. Wiley, New Jersey, pp 7–41Google Scholar
  44. Reynolds CH, Tounge BA, Bembenek SD (2008) Ligand binding efficiency: trends, physical basis, and implications. J Med Chem 51(8):2432–2438CrossRefPubMedGoogle Scholar
  45. Roberts SA (2001) High-throughput screening approaches for investigating drug metabolism and pharmacokinetics. Xenobiotica 31(8–9):557–589CrossRefPubMedGoogle Scholar
  46. Roberts SA (2003) Drug metabolism and pharmacokinetics in drug discovery. Curr Opin Drug Discov Dev 6(1):66–80Google Scholar
  47. Rocchetti M, Simeoni M, Pesenti E, De Nicolao G, Poggesi I (2007) Predicting the active doses in humans from animal studies: a novel approach in oncology. Eur J Cancer 43(12):1862–1868CrossRefPubMedGoogle Scholar
  48. Rostami-Hodjegan A (2012) Physiologically based pharmacokinetics joined with in vitro-in vivo extrapolation of ADME: a marriage under the arch of systems pharmacology. Clin Pharmacol Ther 92(1):50–61CrossRefPubMedGoogle Scholar
  49. Schmieder R, Puehler F, Neuhaus R, Kissel M, Adjei AA, Miner JN, Mumberg D, Ziegelbauer K, Scholz A (2013) Allosteric MEK1/2 inhibitor refametinib (BAY 86-9766) in combination with sorafenib exhibits antitumor activity in preclinical murine and rat models of hepatocellular carcinoma. Neoplasia 15(10):1161–1171CrossRefPubMedPubMedCentralGoogle Scholar
  50. Simeoni M, De Nicolao G, Magni P, Rocchetti M, Poggesi I (2013) Modeling of human tumor xenografts and dose rationale in oncology. Drug Discov Today Technol 10(3):e365–e372CrossRefPubMedGoogle Scholar
  51. Smith DA (2011) Discovery and ADMET: where are we now. Curr Top Med Chem 11(4):467–481CrossRefPubMedGoogle Scholar
  52. Smith DA, Di L, Kerns EH (2010) The effect of plasma protein binding on in vivo efficacy: misconceptions in drug discovery. Nat Rev Drug Discov 9(12):929–939CrossRefPubMedGoogle Scholar
  53. Smith DA, Allerton C, Kalgutkar A, van de Waterbeemd H, Walker DK (2012) Pharmacokinetics and metabolism in drug design. Wiley-VCH, Weinheim, p 268CrossRefGoogle Scholar
  54. Tsaioun K, Kates SA (2011) ADMET for medicinal chemists: a practical guide. Wiley, New Jersey, p 512CrossRefGoogle Scholar
  55. Tuntland T, Ethell B, Kosaka T, Blasco F, Zang RX, Jain M, Gould T, Hoffmaster K (2014) Implementation of pharmacokinetic and pharmacodynamic strategies in early research phases of drug discovery and development at Novartis Institute of Biomedical Research. Front Pharmacol 5:174CrossRefPubMedPubMedCentralGoogle Scholar
  56. van den Bergh A, Sinha V, Gilissen R, Straetemans R, Wuyts K, Morrison D, Bijnens L, Mackie C (2011) Prediction of human oral plasma concentration-time profiles using preclinical data: comparative evaluation of prediction approaches in early pharmaceutical discovery. Clin Pharmacokinet 50(8):505–517CrossRefPubMedGoogle Scholar
  57. Venkatakrishnan K, Friberg L, Ouellet D, Mettetal J, Stein A, Trocóniz I, Bruno R, Mehrotra N, Gobburu J, Mould D (2015) Optimizing oncology therapeutics through quantitative translational and clinical pharmacology: challenges and opportunities. Clin Pharmacol Ther 97(1):37–54CrossRefPubMedGoogle Scholar
  58. Visser SA, Aurell M, Jones RD, Schuck VJ, Egnell AC, Peters SA, Brynne L, Yates JW, Jansson-Löfmark R, Tan B, Cooke M, Barry ST, Hughes A, Bredberg U (2013) Model-based drug discovery: implementation and impact. Drug Discov Today 18(15–16):764–775CrossRefPubMedGoogle Scholar
  59. Wang J, Urban L (2014) PREDICTIVE ADMET integrative approaches in drug discovery and development. Wiley, New Jersey, p 616CrossRefGoogle Scholar
  60. Yamazaki S (2013) Translational pharmacokinetic-pharmacodynamic modeling from nonclinical to clinical development: a case study of anticancer drug, crizotinib. AAPS J 15(2):354–366CrossRefPubMedPubMedCentralGoogle Scholar
  61. Yamazaki S, Lam JL, Zou HY, Wang H, Smeal T, Vicini P (2015) Mechanistic understanding of translational pharmacokinetic-pharmacodynamic relationships in nonclinical tumor models: a case study of orally available novel inhibitors of anaplastic lymphoma kinase. Drug Metab Dispos 43(1):54–62CrossRefPubMedGoogle Scholar
  62. Zhang D, Surapaneni S (2012) ADME-enabling technologies in drug design and development. Wiley, New Jersey, p 622CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  1. 1.Research Pharmacokinetics, Global Drug Discovery, Bayer PharmaBerlinGermany

Personalised recommendations