Editorial

From in vivo to in vitro/in silico

1. Evaluation of research so far:
ADME-Tox screening in drug
discovery
Expert Opin. Drug Metab. Toxicol. Downloaded from informahealthcare.com by Memorial University of Newfoundland on 08/01/14
ADME: progress and challenges
Han Van de Waterbeemd
Pharmaceutical Consultant in In Vitro and In Silico ADME Technologies,
han.vdwaterbeemd@btinternet.com
High-throughput screening technologies in biological sciences of large

libraries of compounds obtained via combinatorial or parallel chemistry

2. Future prospects: challenges in
approaches, as well as the application of design rules for drug-likeness, have
drug metabolism,
resulted in more hits to be evaluated with respect to their ADME or drug
pharmacokinetics and
metabolism and pharmacokinetic properties. The traditional in vivo meth-
toxicology
ods using preclinical species, such as rat, dog or monkey, are no longer suffi-
3. Expert opinion: ADME studies in
cient to cope with this demand. This editorial discusses the changes towards
the future
medium- to high-throughput in vitro and in silico ADME screening. In addi-
tion, much more attention is now put on early safety and risk assessment of
promising lead series and potential clinical candidates.

Keywords: ADME, DMPK, high-throughput screening, in combo screening, in silico ADME,

lead profiling, toxicology

Expert Opin. Drug Metab. Toxicol. (2005) 1(1):1-4

For personal use only.

1.Evaluation of research so far: ADME-Tox screening in

drug discovery

Only recently have pharmacokinetics, drug metabolism and toxicology of selected

clinical candidates been studied during clinical development. Analysing the root
cases of attrition demonstrated that lack of efficacy and toxicity, along with inappro-
priate ADME, are among the major determinants for failure [1,2]. Lack of efficacy, in
addition to poor target binding may be caused by poor absorption and distribution,
and rapid metabolism, leading to drug concentrations at the target that are too low
[3]. As toxicity is a major factor in causing attrition, the development of compounds
is often halted even before a detailed human pharmacokinetics or efficacy study is
performed. Hence, the real impact of ADME processes remains somewhat hidden in
(incomplete) attrition data. However, it became good practice in the 1990s to
collect ADME and toxicity data during the drug discovery stage so that it can be
used in decision making to select the best possible clinical candidates [4]. However,
the success of this move to early involvement has been questioned [5]. The main
contribution of a discovery (research) department of drug metabolism and pharma-
cokinetics has been to enable the design of pharmacokinetically adequate rather
than optimal molecules and, thus, to make it possible to work on difficult targets [5].
Despite all of the preclinical efforts, there will be a need for extensive clinical
pharmacokinetics to provide a safe prescription to patients [5].
The number of compounds synthesised has increased dramatically through
combinatorial library strategies in chemistry, and is driven by high-throughput screen-
ing (HTS) in biology. Much attention is also paid to the design criteria for lead- and
drug-like compounds [6], which further increased the output of compounds with
appropriate ADME properties. The concept of property-based design [7] is now com-
monly used to address ADME issues as early as possible. Thus, the former in vivo ani-
mal ADME evaluation could no longer cope with the demand, and in vitro ADME
Ashley Publications screens became widely introduced. Despite rapid advances in the use of automation
www.ashley-pub.com and robotics to increase the throughput of the in vitro ADME screens, it became clear
that screening all available compounds is not necessarily most efficient and

10.1517/17425255.1.1.1 © 2005 Ashley Publications Ltd ISSN 1742-5255 1

 From in vivo to in vitro/in silico ADME: progress and challenges
cost-effective. It is, therefore, sensible and vital to develop
in silico tools to predict and simulate various ADME properties,
and use these in decision-making in a well-balanced fashion
together with in vivo and in vitro approaches.
ADME studies are aimed at getting an early estimate of
human pharmacokinetics (PK) and metabolism [8]. These stud-
ies include: estimates of dose, dose interval, absorption, bioa-
vailability, clearance, volume of distribution, half-life, clearance
mechanism, potential for drug–drug interactions, involvement
Expert Opin. Drug Metab. Toxicol. Downloaded from informahealthcare.com by Memorial University of Newfoundland on 08/01/14
of major metabolising enzymes and metabolite formation.
Early toxicology and safety studies should eliminate
compounds before they enter lengthy and costly clinical trials.
As a result of the recent withdrawal of a number of marketed
drugs, more pressure is now put on pharmaceutical companies
regarding safety evaluations, including pharmacological and
toxicological safety. Investigation of the potential to cause QT
prolongation is now routine. However, the interpretation of
the data in not straightforward, as many marketed drugs can
prolong the QT interval [9,101].
2. Future
prospects: challenges in drug
metabolism, pharmacokinetics and toxicology
2.1 In vitro screening
For personal use only.
Based on the knowledge gained from high-throughput screen-
ing in biology, many in vitro ADME screens can now run in
medium- or high-throughput using automation, robotics and
miniaturisation [10-12]. However, due to the nature of the
assays, the typical analytical end point is often liquid chroma-
tography/mass spectrometry. Cell-based assays, such as the
Caco-2 screen for permeability/absorption are, therefore,
quite expensive, particularly when run in screening mode on
many compounds. The trend is to investigate either cheaper
in vitro alternatives such as the parallel artificial membrane
permeability assay (PAMPA) method or to move to in silico
approaches. A proper synergistic hybrid combination of
in vitro and in silico methods [13], which has been called the
in combo approach [14], may be the most cost-effective
approach to ADME screening in drug discovery.
2.2 Drug–drug interactions
Regulatory authorities require information to be submitted
on the potential for interactions to cause adverse effects.
With the availability of in vitro systems this aspect is, there-
fore, often considered at early stages of discovery, including
hit evaluation. Oxidative metabolism by cytochrome P450
(CYP) enzymes is, for most drugs, the major route of elimi-
nation. As P450s are also capable to metabolise multiple
substrates, their inhibition is the major focus of drug–drug
interaction (DDI) studies. CYP3A4 is not only the most
abundant in the liver, but is also present in the gut wall, and
is responsible for the metabolism of 50 – 60% of all drugs.
This enzyme is, therefore, highly susceptible to both reversi-
ble and irreversible (mechanism-based) inhibition [15]. Most
CYP3A4 substrates or inhibitors are also P-glycoprotein
2 Expert Opin. Drug Metab. Toxicol. (2005) 1(1)
(P-gp) substrate or inhibitor. It is believed that CYP3A4 and
P-gp in the gastrointestinal tract work in concert to limit
uptake of xenobiotics, including drugs [16]. Current inhibi-
tion studies are based on Ki or mean inhibitory concentra-
tion values, but more quantitative approaches would be
beneficial [17]. Great progress has been made in the reliable
simulation of DDIs, even taking into account the variability
in the population [18]. A further question is how metabolites
contribute to DDIs and a better understanding of allosteric
kinetics of P450s may be required [19].
2.3 Enzyme induction
Although clinically less important than enzyme inhibition,
protocols are being developed to estimate enzyme induction.
It needs to be seen where to place such assay or screen in the
discovery process [20].
Transporter proteins constitute a significant fraction of
membrane-bound proteins [21,22]. They are typically expressed
in all organs involved in uptake and elimination of drugs,
including the gastrointestinal tract, blood–brain barrier, liver
and kidney. It is hoped that in the near future three-dimen-
sional structures of the key transporters might be obtained.
Interaction of drugs with transporters can alter their behav-
iour in membrane transport, which may result in active
uptake, efflux and rapid elimination, for example. In other
words, the pharmacokinetics of a drug may be influenced by
transporters. Drug–drug interactions may have their cause
apart from a metabolic component, in the involvement of
transporters [20,23]. However, although the basic knowledge of
transporters is growing rapidly [21], the clinical significance is
open to much debate, but some examples have been presented
[24]. A number of P-gp assays have been developed, as well as
some double and triple transfected assays. More transporter
assays will soon be available. The challenge will be to translate
the flood of experimental data into relevant information for
drug design projects.
2.4 In silico ADME
Prediction and simulation of various ADME properties is con-
siderably less expensive than in vitro screening. Therefore, great
efforts have been made to turn all available data into predictive
computational models using quantitative structure–activity
relationship methods and molecular modelling [14,25]. In vitro
data are now generated for many ADME and physicochemical
end points and can be used to build more robust models.
Model updating will need to be automated and fitted in the
data generation cycle. There is a need for both local (project-
specific) as well as global (general) models. Unfortunately, there
is still a paucity of human in vivo data, and, thus, models based
on these will need to be handled with care. Interindividual vari-
ability within the population is another key factor to take into
account for human predictions. Predictions will be ranges
rather than hard numbers. For the development of potential
drugs, predictions may contribute to high-throughput
pharmaceutics and rational drug delivery [26].
 Van de Waterbeemd

2.5 Simulations: PBPK and PK/PD 2.7 Metabolite identification

Physiologically-based pharmacokinetic (PBPK) models are
based on well-known principles in chemical engineering,
and describe the human or animal body as a series of pipes
and tanks [27,28]. Convenient software has now become
available and makes PBPK modelling more accessible to
drug discovery and development to predict various PK
parameters and concentration–time profiles in the body.
Population variability [29] and specific groups such as chil-
Expert Opin. Drug Metab. Toxicol. Downloaded from informahealthcare.com by Memorial University of Newfoundland on 08/01/14
With increasing resolution of mass spectrometry and nuclear
magnetic resonance it becomes feasible to detect small fractions
of metabolites. Debate is ongoing to define major versus minor
metabolites. Some metabolites may be pharmacologically active
and contribute to the overall PK. Reactive metabolites may
bind to apoproteins and cause idiosyncratic reactions. The chal-
lenge remains to detect these as early as possible. It has been
suggested that time-dependent inhibition should be in standard

dren and the elderly can also be taken into account. The
more experimental data are available, the better the simula-
tions, making these approaches of interest in the discovery
and clinical development settings. Some commercial
programmes have put considerable effort in simulating the
absorption process, which is of interest in optimising
pharmaceutical formulations.
Pharmacokinetic/pharmacodynamic (PK/PD) modelling
links dose–concentration relationships (PK) to concentra-
tion–effect relationships (PD). This approach helps to simu-
late the time course of drug effects depending on the dose
regimen [30]. Specialised software and better understanding
of the applications will help to speed up the clinical
development.
For personal use only.
in vitro screening protocols [20]. Good progress has been made
in metabolite prediction [33].
3. Expert opinion: ADME studies in the future
The pharmaceutical world is rapidly changing. The output of
new chemical entities is low despite the enormous investments
in screening technology. The increasing demands of the regula-
tory bodies have put more pressure on the delivery of high qual-
ity clinical candidates and ultimately at the end of the pipeline,
thus resulting in safe and efficacious new drugs. As in most
other industries, the pharmaceutical industry will invest more
in computer predictions and simulations. This seems to be
taking off only slowly, and much is still driven on experimental
data. The future no doubt will bring about a move from ‘wet’ to

2.6 Toxicology and safety prediction
Drug safety is a great concern to patients and, therefore, to the
regulatory bodies. Early prediction gets more and more atten-
tion in any drug discovery programme. Simple in vitro screen-
ing assays for, for example, hERG and other cardiac ion
channels, genetic toxicology and cytotoxicity, are now
routinely added to the growing battery of biology, ADME
and toxicity/safety screens. Predictive toxicology its still in its
early days [31]. A promising tool is the integration of ADME,
toxicology and pharmacology data to predict side effects, as in
the BioPrint approach [32].
‘web’ profiling. This pharmaceutical ‘model’ may change, with
large pharmaceutical companies increasingly outsourcing to
smaller biotechnology companies. With a better understanding
of the underlying principles in drug metabolism and pharma-
cokinetics, the field will evolve from an empirical one to a truly
rational, in cerebro approach to drug design [4]. In this future
new world based on automated screening, model building and
decision-making, the acronym ADME may well be redefined as
‘automated decision-making engine’. In any case, the field of
ADME (DMPK) is more crucial than ever to drug discovery
and development.

Bibliography 5. SMITH D, SCHMID E, JONES B: • Introduction to the concept of property-

Papers of special note have been highlighted as of
interest (•) to readers.
1. KENNEDY T: Managing the drug
discovery/development interface. DDT
(1997) 2:436-444.
2. KOLA I, LANDIS J: Can the
pharmaceutical industry reduce attrition
rates? Nature Revs. Drug Disc. (2004)
3:711-713.
3. KUBINYI H: Drug research: myths, hype
and reality. Nature Revs. Drug Disc. (2003)
2:665-668.
4. EDDERSHAW P, BERESFORD AP,
BAYLISS MK: ADME/PK as part of a
rational approach to drug discovery. DDT
(2000) 5:409-414.
Do drug metabolism and pharmacokinetic
departments make any contributions to
drug discovery? Clin. Pharmacokinet. (2002)
41:1005-1019.
6. LIPINSKI CA, LOMBARDO F,
DOMINY BW, FEENEY PJ: Experimental
and computational approaches to estimate
solubility and permeability in drug
discovery and development settings.
Adv. Drug Del. Revs. (1997) 23:3-25.
• Original rule-of-five paper.
7. VAN DE WATERBEEMD H,
SMITH DA, BEAUMONT K,
WALKER DK: Property-based design:
optimization of drug absorption and
pharmacokinetics.
J. Med. Chem. (2001) 44:1313-1333.
based drug design.
8. PALMER AM: New horizons in drug
metabolism, pharmacokinetics and drug
discovery. Drug News Perspect. (2003)
16:57-62.
9. ABRIEL H, SCHLAEPFER J, KELLER DI
et al.: Molecular and clinical deteriminants
of drug-induced long QT syndrome: an
iatrogenic channelopathy. Swiss Med. Wkly
(2004) 134:685-694.
• Good list of marketed drugs with QT
prolongation.
10. SAUNDERS KC: Automation and robotics
in ADME screening. DDT Technologies
(2004) 1:373-380.
11. JENKINS KM, ANGELES R,
QUINTOS MT et al.: Automated high-

Expert Opin. Drug Metab. Toxicol. (2005) 1(1) 3

 From in vivo to in vitro/in silico ADME: progress and challenges

throughput ADME assays for metabolic
stability and cytochrome P450 inhibition
profiling of combinatorial libraries.
J. Pharm. Biomed. Anal. (2004)
34:989-1004.
12. WANG J, URBAN L: The impact of early
ADME profiling on drug discovery and
development strategy. Drug Disc. World
(2004):73-86.
Expert Opin. Drug Metab. Toxicol. Downloaded from informahealthcare.com by Memorial University of Newfoundland on 08/01/14
19. ATKINS WM: Implications of the allosteric
kinetics of cytochrome P450s. DDT (2004)
9:478-484.
20. TUCKER G, HOUSTON JB,
HUANG S-M: Optimising drug
development: strategies to assess drug
metabolism/transporter interaction
potential – towards a consensus. Eur. J.
Pharm. Sci. (2001) 13:417-428.
• List of probe substrates and inhibitors for
28. POGGESI I: Predicting human
pharmacokinetics from preclinical data.
Curr. Opin. Drug Disc. Dev. (2004)
7:100-111.
29. YANG J, ROSTAMI-HODJEGAN A,
TUCKER G: Prediction of fluconazole
interaction with midazolam and triazolam:
incorporating population variability. J. Clin.
Pharmacol. (2001) 52:465-486.

13. YU H, ADEDOYIN A: ADME-Tox in
drug discovery: integration of experimental
and computational technologies. DDT
(2003) 8:852-860.
14. DICKINS M, VAN DE
WATERBEEMD H: Simulation models for
drug disposition and drug interactions.
DDT Biosilico (2004) 2:38-45.
30.
CYPs.
21. HO RJ, SHEN DD: Research on the role of
membrane transporters in drug disposition:
present challenges and future directions.
Curr. Drug Metab. (2004) 5:i-ii.
22. DANTZIG AH, HILLGREN KM,
DE ALWIS DP: Drug transporters and
ZUIDEVELD KP, VAN DER GRAAF PH,
NEWGREEN D et al: Mechanism-based
pharmacokinetic–pharmcodynamic
modelling of 5-HT1A receptor agonists:
estimation of in vivo affinity and intrinsic
efficacy on body temperature in rats.
J. Pharmacol. Exp. Ther. (2004)
308:1012-1020.

15. ZHOU S, CHAN E, LIM LY et al.:
Therapeutic drugs that behave as
mechanism-based inhibitors of cytochrome
P450 3A4. Curr. Drug Metab. (2004)
5:415-442.
• List of CYP3A4 mechanism-based
inhibitors.
16. CUMMINS CL, JACOBSEN W,
For personal use only.
their role in tissue distribution. Ann. Rep.
Med. Chem. (2004) 39:279-291.
23. BALAYSSAC D, AUTHIER N,
COUDORE F: Drug–drug interactions
linked to a P-glycoprotein inhibition.
Lett. Pharmacol. (2004) 18:76-80.
24. AYRTON A, MORGAN P: The role of
transport proteins in drug absorption,
31. EGAN WJ, ZLOKARNIK G,
GROOTENHUIS PD: In silico prediction
of drug safety: despite progress there is
abundant room for improvement. DDT
Technologies (2004) 1:381-387.
32. KREJSA CM, HORVATH D,
ROGALSKI SL et al.: Predicting ADME
properties and side effects: the BioPrint

BENET LZ: Unmasking the dynamic
interplay between intestinal P-glycoprotein
and CYP3A4. J. Pharmacol. Exp. Ther.
(2002) 300:1036-1045.
17. BLANCHARD N, RICHERT L,
COASSOLO P, LAVE T: Qualitative and
quantitative assessment of drug–drug
interaction potential in man, based on Ki,
IC50 and inhibitor concentration. Curr.
Drug Metab. (2004) 5:147-156.
18. ROSTAMI-HODJEGAN A, TUCKER G:
In silico simulations to assess the in vivo
consequences of in vitro metabolic drug–
drug interactions. DDT Technologies (2004)
1:441-448.
distribution and disposition. Xenobiotica
(2001) 31:469-497.
25. VAN DE WATERBEEMD H,
GIFFORD E: ADMET in silico modelling:
towards prediction paradise? Nature Revs.
Drug Disc. (2003) 2:192-204.
• Recent review of ADME/toxicity
predictions.
26. VARMA MVS, KHANDAVILLI S,
ASHOKRAJ Y et al.: Biopharmaceutic
classification system: a scientific framework
for pharmacokinetic optimization in drug
research. Curr. Drug Metab. (2004)
5:375-388.
27. LEAHY DE: Progress in simulation
modelling for pharmacokinetics. Curr.
Topics Med. Chem. (2003) 3:1257-1268.
approach. Curr. Opin. Drug Disc. Dev.
(2003) 6:470-480.
33. LANGOWSKI J, LONG A: Computer
systems for the prediction of xenobiotic
metabolism. Adv. Drug Deliv. Revs. (2002)
54:407-415.
Websites
101. http://www.torsades.org
Affiliation
Han Van de Waterbeemd
Pharmaceutical Consultant in In Vitro and
In Silico ADME Technologies, UK
Tel: +44 (0) 1622 686 630;
E-mail: han.vdwaterbeemd@btinternet.com

4 Expert Opin. Drug Metab. Toxicol. (2005) 1(1)