Chem. Res. Toxicol.

2006, 19, 1393-1401 1393

PerspectiVe
Future of ToxicologysMechanisms of Toxicity and Drug Safety:
Where Do We Go from Here?
James L. Stevens
Toxicology and Drug Disposition, Lilly Research Laboratories, 2001 West Main Street,
Greenfield, Indiana 46140
ReceiVed August 27, 2006
Recent high-profile drug withdrawals increase the pressure on regulators and the pharmaceutical industry
to improve preclinical safety testing. Understanding mechanisms of drug toxicity is an essential step
toward improving drug safety testing by providing the basis for mechanism-based risk assessments.
Nonetheless, despite several decades of research on mechanisms of drug-induced toxicity and the
application of various new technologies to preclinical safety assessment, the overall impact on preclinical
safety testing has been modest. Assessing the risk of exposing humans to new drug candidates still depends
on preclinical testing in animals, which in many, but not all cases, predicts outcomes in humans accurately.
The following offers a perspective on the challenges and opportunities facing efforts to improve preclinical
safety testing and outlines gaps and needs that must be addressed. A case is built for focusing solutions
on defined problems within the current safety testing paradigm rather than imposing wholesale change.
Targets for application of new technologies, including in silico screening, biomarkers, surrogate assays
and ‘omic technologies, are outlined. Improving drug safety testing will depend on improving the
application of mechanism-based risk assessment but will also require improving public and private
collaborations in order to focus research regarding the mechanism of drug-induced toxicity on the most
important problems.

1. Introduction 1393 increase concern at a time when pharmaceutical industry

2. Background: Pharmaceutical Safety 1393 productivity is nearly flat (2). A Food and Drug Administration
Assessment (FDA) report (3) suggests that decreased pipeline productivity
2.1. Target Validation and Lead Generation 1394 creates a risk that “the biomedical revolution will not deliver
2.2. Lead Optimization 1395 on its promise of better health” and that there is “...an...urgent-
2.3. Preclinical Safety Assessment: From 1395 ...need for applying technologies such as genomics, proteomics,
Clinical Candidate to Registration bioinformatics systems, and new imaging technologies...to detect
3. Opportunities, Gaps, and Needs: Improving 1396 safety problems early...” Others point out that “...there is little
Preclinical Safety Testing evidence that [new technologies] have had a major or direct
3.1. Knowledge Management: How, When, and 1396 impact on the safety assessment which supports...first studies
Why Does Preclinical Safety Assessment
of new drugs in humans” (4). To improve drug safety and deliver
Fail?
better health outcomes to patients, it is critical for toxicology
3.1.1. Opportunities, Gaps, and Needs 1396
3.2. In Silico Approaches to Drug Safety 1396 to advance our understanding of mechanisms of ADRs, apply
Assessment new technologies to the practical challenge of predicting drug
3.2.1. Opportunities, Gaps, and Needs 1397 safety, and improve mechanism-based risk assessment to ensure
3.3. Surrogate Models for Safety Assessment 1397 the safe use of new drugs in man.
3.3.1. Cell- and Tissue-Based Surrogate 1397 Herein, I offer a perspective on the current state and future
Assays needs for mechanism-based risk assessment in drug safety.
3.3.2. Surrogate Animal Models 1397 Comments are limited to the development of small-molecule
3.3.3. Opportunities, Gaps, and Needs 1397 therapeutics. A brief summary of the preclinical safety assess-
3.4. Biomarkers 1398 ment paradigm is followed by suggestions for future research
3.4.1. Opportunities, Gaps, and Needs 1398 directions. The opinions expressed herein are mine alone, and
3.5. Systems Toxicology and Pharmacogenetics: 1398 I apologize in advance for any significant omissions.
Making Sense of High Content Information
3.5.1. Opportunities, Gaps, and Needs 1399
4. Summary: Improving Mechanism-Based Risk 1399 2. Background: Pharmaceutical Safety Assessment
Assessment
Achieving an acceptable efficacy and safety profile for a new
drug is a complex process, requiring optimization of many
1. Introduction variables within a single chemical structure (5). From a safety
Despite decades of research, adverse drug reactions (ADRs) perspective, the complexity lies largely in the quality and
remain a significant problem (1). Recent drug withdrawals interpretation of data from a discrete series of studies that define
10.1021/tx060213n CCC: $33.50 © 2006 American Chemical Society
Published on Web 10/07/2006
1394 Chem. Res. Toxicol., Vol. 19, No. 11, 2006 SteVens

Figure 1. Overview of preclinical safety assessment. The figure shows a schematic representation of the drug discovery and development process
from selection of a target to registration of a new drug entity (adapted from ref 14). Abbreviations used include the following: IND, investigative
new drug application; NDA, new drug application; CE, candidate evaluation; CIB, clinical investigator brochure; MRSD, maximum recommend
starting dose; and SAR, structure-activity relationship.

the preclinical safety assessment paradigm (6). The details and tion with organ toxicity. If the target in man has multiple forms,
regulatory expectations are described in detail elsewhere (http:// verifying the expression of similar isozymes and selectivity in
www.fda.gov/cder/guidance/; http://www.ich.org/); the steps are preclinical animal models is important. Mapping the tissue
summarized in Figure 1. distribution of the drug target is useful to determine if toxicity
tracks closely with target expression. Knockout mice can also
2.1. Target Validation and Lead Generation highlight the impact of target inhibition in normal tissues and
in disease pathogenesis. In general, multiple approaches are
Target selection can be based on prior clinical data with required to decipher the mechanism of target organ toxicitys
prototypical first-in-class molecules or mechanisms of disease Having the right tools and an effective risk management strategy
pathogenesis, or it can be inferred indirectly from genetic data are the keys to success.
in humans or preclinical species. Proactive safety assessment
at this stage is largely an in cerebro and/or in silico exercise. A recent example of this complexity is IKKβ, a drug target
Safety issues inherent in modulating a target can be anticipated implicated in diseases including oncology, asthma, bone loss,
from existing drug precedents; for example, agonists for arthritis, and diabetes (9, 10). IKKβ phosphorylates IκB, which
peroxisome proliferators activator receptors (PPAR) might be is degraded, releasing NFk-B to activate genes linked to
anticipated to be tumorigenic, increase heart weight, and produce apoptosis, inflammation, and, perhaps, cell division. Knockout
plasma volume expansion in preclinical studies (7). For novel mice deficient in IKKβ die in utero due to hepatocellular
targets, safety concerns must be inferred from literature on apoptosis and liver degeneration (11). At first glance, the genetic
genetic studies in humans and lower organisms or by mining data suggest that inhibiting IKKβ might result in liver injury,
pathways involved in a disease process. but targeted deletion of IKKβ in the liver did not result in
At the target-to-lead stage, the focus of safety assessment hepatocyte apoptosis or embryo lethality in utero (12). The risk
should be on strategies to assess any anticipated safety issues. that these observations represent to humans is not clear since
In silico analysis of chemical structure for safety end points, there are few published data from clinical trials with IKKβ
such as mutagenicity, are used prior to chemical synthesis (8). inhibitors (9, 10). Nonetheless, reports of clinical trials suggest
Prototypical small molecules or proteins (e.g., antibodies), even that margins of safety (MOS) sufficient to allow testing in
if not sufficiently “druglike” for clinical use, can be useful tools humans have been achieved. Additional data will be forthcoming
to investigate safety issues that may emerge during lead in this rapidly developing field. IKKβ is an important drug
optimization. Pharmacodynamic (PD)-based biomarkers of target; publication of preclinical and clinical data will clarify
efficacy, such as antibodies that recognize post-translational whether or not the benefits warrant any risks.
modifications associated with target modulation; for example, As toxicities emerge during development, one of the most
phosphorylation, ubiquitination, and acetylation states of pro- important questions to address is whether the mechanism is
teins, help link dose-response relationships for target modula- related to the intended pharmacology (on-target toxicity) or to
PerspectiVe Chem. Res. Toxicol., Vol. 19, No. 11, 2006 1395

MOS are confined by the dose-response curve. Designing

partial agonists/antagonists or less potent molecules with a
“shallow” slope to the dose-response curve may help. Defining
the human relevance of any pharmacologically linked toxicity
is also critical. If the toxicity is due to off-target effects, the
SAR can be guided away from offending chemical features
linked to the dose-limiting toxicity by shifting the dose-
response curve for toxicity to higher doses (Figure 2, right
panel). In either case, a mechanism-based screen reduces the
time and amount of compound necessary to find an optimized
drug candidate. Establishing the mechanism and validating the
Figure 2. Dose-response relationships and on- vs off-target toxicity. assay format are the primary goals of mechanistic investigations
A schematic (adapted from ref 14) representation of the relationships
between the no observed adverse effect level (NOAEL) marking the
during lead optimization.
threshold for toxicity (T) and an efficacy dose (E) producing an 80% An example where strategies to optimize and manage an on-
of maximum biological response (ED80). Note that the ED80 occurs as target toxicity are in place is hemorrhage associated with
about 30-40% target modulation on the dose-response curve. On- anticoagulant therapy (references in ref 15). Species differences
target toxicity, on the left-hand side, is shown for two drugs that differ in coagulation parameters are easily measured with standard
in the MOS based on the slope and maximum response values of the
two drugs. Off-target toxicity, on the right-hand side, is depicted by biochemical assays. When blood levels of anticoagulant are
two separate dose-response curves, one for efficacy (solid line) and maintained in an acceptable range, a therapeutic end point is
one for toxicity (dotted line); MOS is defined by the difference in the achieved, reduced clot formation without excessive bleeding,
ED80 and NOAEL from the two separate curves. allowing careful use of anticoagulants in the clinic. Anticoagu-
lants are not ideal drugs, but clinical management of dose-
the chemical structure itself (off-target toxicity). Differentiating limiting bleeding episodes is common.
between these two potential mechanisms is crucial to designing An example where mechanistic information hampers progress
strategies to increase safety; understanding the basic biology is the development of selective serotonin (5-hydroxytryptamine,
and pharmacology of the target is the first step. Drug develop- 5-HT) receptor agonists, specifically 5HT2c, in obesity (16).
ment is a high-risk business. As Pasteur noted, “Chance favors Valvulopathies occur in humans taking a combination of
the well-prepared mind.” When resolving safety issues in drug fenfluramine and phentermine for weight loss (17). Despite
development, chance favors the well-prepared plan; to assess literature reports of preclinical models and suggestions that the
safety later in development starts at target selection. 5-HT2B receptor subtype is responsible (18), the relationship
of a particular 5-HT receptor and valvulopathy in patients has
2.2. Lead Optimization not been proven. Difficulty in clinical monitoring and the fact
that the condition may not be reversible if the drug is withdrawn
During lead optimization, the characteristics of a “good drug”
add to concerns. Because alternate therapies, such as diet and
are built into potential clinical candidates (5). From a safety
exercise, are available, the unknown mechanism and high level
perspective, the goal is to identify the dose-limiting toxicities,
of concern leave a difficult path to testing this clinical hypothesis
the most sensitive species, and set doses for later definitive
for the treatment of obesity.
safety studies. Potential candidates are often screened for safety
using pilot repeat dose studies (13). Toxicity with one molecule
can be easily circumvented by testing another, but recurring 2.3. Preclinical Safety Assessment: From Clinical
toxicity within the lead series can stall progress. However, Candidate to Registration
without understanding the mechanism and relevance to humans, Once a single clinical candidate is selected, the chemical die
abandoning a novel target or lead series based on toxicity in is cast, and the biological properties are locked in the structure.
short-term studies may be premature. However, to proceed, a The emphasis for preclinical testing shifts from candidate
strategy to design safer compounds must be in place. identification to candidate evaluation (Figure 1); the focus of
The MOS is defined in preclinical models by the separation mechanism-based risk assessment shifts as well to the patient.
between parameters defining the efficacy and safety; in general, Before submitting an investigational new drug application
they are based on exposure rather than administered dose (14). (IND), safety is assessed in rodent and nonrodent species
Two things are required to widen the MOS during lead according to regulatory expectations (6, 14). A primary goal of
optimization: knowledge of mechanism (on- vs off-target) and these safety studies is to identify a maximum recommended
screening tools to inform the structure-activity relationship safe starting dose (MRSD) for the clinic (19). A maximum
(SAR). If the mechanisms cannot be differentiated and a screen tolerated dose defines the target organs at the highest dose
cannot be implemented, finding better compounds may be without mortality. The no observed adverse effect level (NOA-
reduced to empirical screening rather than rational drug design. EL) defines the highest dose at which adverse effects are not
Establishing the PD relationships among exposure (Cmax and seen preclinically (14). A human equivalency dose, extrapolated
area under the curve), target inhibition, efficacy, and toxicity from the NOAEL and divided by a safety factor, establishes
in vivo is a first step to differentiating mechanisms. Testing the MRSD, the point where dose escalation begins in the clinic,
compounds in knockout mice can also help differentiate on- vs and a key parameter in setting the dose range used to test the
off-target contributions if the pathology is replicated in the clinical hypothesis.
mouse. Inactive compounds within the series can be used to Mechanism-based risk assessment impacts the MRSD and
differentiate on-target toxicity from off-target mechanisms the safety factor used to set the MRSD in three ways: (i)
related to the chemical scaffold. understanding mechanisms of species sensitivity, (ii) establishing
If an on- vs off-target mechanism is established, the strategies methods to monitor the risk, and (iii) defining the reversibility
to increase the MOS will differ. As noted on the left in Figure of target organ toxicity. In the absence of data on human
2, with an on-target mechanism, boundaries for increasing the relevance, the default position is to estimate the MRSD using
1396 Chem. Res. Toxicol., Vol. 19, No. 11, 2006 SteVens

the NOAEL from the most sensitive preclinical test species. is difficult to address since most preclinical safety data are not
However, if the mechanism of toxicity in the most sensitive publicly available. Nonetheless, a recent reanalysis of available
species is not relevant to humans, the context can change. For data suggests that target organ toxicity for cardiovascular,
example, the renal toxicity of efavirenz, a reverse transcriptase hematopoietic, and gastrointestinal toxicities is predicted with
inhibitor, in rats is not relevant to the human due to differences an efficiency of ∼80%; for the urinary tract, it is predicted at
in routes of metabolism (20). This was a critical decision point about 70% followed by hepatic, skin, and neurological toxicities
in development and illustrated an effective use of mechanistic at 50% or less (4). The same studies point out that when other
studies (Miwa, G. Personal communication). Clinical monitoring species predict human toxicity, 90% of the issues are identified
and reversibility are also critical. If the toxicity can be monitored by current practices. This suggests that target organ toxicity
with a clinical biomarker and is reversible in preclinical species, not replicated by preclinical species contributes disproportion-
the default MRSD for phase I trials may 1/10 the HED based ately to failure and should be a focus for improVement.
on a safety factor of 10 (19). If toxicity is severe, not easily
monitored, or reversible, the allowable MRSD may be much 3.1.1. Opportunities, Gaps, and Needs
lower. Limiting the clinical dosing range due to preclinical safety
Although there are a number of public and private knowledge
concerns can be a major cause of trial failure.
bases available (8, 25, 26) and the landscape is improving, the
targets for improvement outlined above are based on incomplete
3. Opportunities, Gaps, and Needs: Improving safety data. Improving the content of preclinical safety knowl-
Preclinical Safety Testing edge databases is essential to more precisely define how new
Understanding the current practice of preclinical safety approaches and animal models can provide solutions to specific
assessment creates the context for three areas that toxicology problems in two ways: first, by directing technical solutions
must address in the future: (i) improving safety predictions for toward pathologies not modeled by preclinical test species, and
man, (ii) addressing the “pipeline problem” caused by high second, by focusing improvements to the most problematic
attrition due to preclinical toxicology, and (iii) defining mech- target organ toxicities. Improved information sharing will ensure
anisms of toxicity. Suggestions regarding opportunities, gaps, that knowledge management experiments identify the most
and needs are offered below. A first step is the effective use of important causes of attrition and key gaps in predicting human
knowledge management to highlight opportunities for improve- safety. This is an iterative process since the Bayesian nature of
ment. biology implies that new knowledge will reshape our interpreta-
tion of historical data (27).
This incomplete knowledge base also contributes to the
3.1. Knowledge Management: How, When, and Why
apparent failure of new technologies noted by Greaves et al.
Does Preclinical Safety Assessment Fail?
(4). In the author’s opinion, this is not a failure of the
Managing the current knowledge base on drug safety, e.g., technology. Rather, a combination of exaggerated expectations
literature, drug safety data, etc., to define the opportunities for and a failure to address the right questions, i.e., mechanism-
improvement is critical. Three related questions need to be based hypothesis testing, with technologies have contributed to
addressed. (i) How often do failures occur? (ii) When do failures a perception of failure. For example, should one expect to
occur? (iii) Also, why do failures occur? Recent literature improve predictions of safety by applying a new technology to
addresses these three questions to some extent. a preclinical model that is a poor surrogate for the biological
To understand how often safety predictions fail, Lazarou et outcome of interest? If species differences in basic pathophysi-
al. (21) estimated the frequency of ADRs based on the incidence ology hamper safety predictions for humans, the probability that
in hospitalized patients in the United States. In 1994, 2 million a new measurement will yield improved results is low. Ad-
ADRs and 106000 fatalities occurred in hospitalized patients dressing technical gaps in information and knowledge manage-
or 6.7 and 0.32% incidence, respectively. This may be an ment (28) will outline where technical solutions are needed to
overestimation since the incidence of underlying disease and target the right problem with a testable hypotheses in appropriate
multiple drug treatments that can contribute to ADR frequency models. This discussion frames specific opportunities for
(22, 23) is probably higher as compared to the general improving mechanism-based risk assessment in preclinical drug
population. Nonetheless, this study outlines a clear need to safety testing.
improve human safety; however, from a purely technical point
of view, a 6% failure rate suggests that the safety assessment 3.2. In Silico Approaches to Drug Safety Assessment
works reasonably well. Therefore, targeting specific issues, Avoiding an offending chemical feature prior to synthesis is
rather than reVamping the entire process, is more likely to the simplest way to avoid toxicity. Recent reviews outline the
tighten the safety net. state-of-the-art application of in silico models to predicting
Now the second questionsWhen do failures to detect human toxicity (8, 28, 29), including the systems approaches addressed
risk first occur? A retrospective analysis of clinical and later. At the earliest stages, e.g., target-to-lead (Figure 1), in
preclinical data suggests that 70% of ADRs were preceded by silico models offer the opportunity to interrogate chemical space
findings during preclinical testing and that the first observations prior to compound synthesis (13, 30). Either local or global in
were in studies of 30 days or less (24), i.e., the studies that silico models can be applied to predict the behavior of virtual
support first dose in humans (Figure 1). Internal industry data libraries (31). As synthesis progresses, local models, built rapidly
agree since nearly 80% of clinical candidate attrition due to in real time as a lead series expands, allow modeling of
preclinical safety occurs prior to first human dose (data not accumulating biological or biochemical data. Global models
shown). Therefore, to improVe risk assessment and reduce based on large training sets drawn from databases can be used
pipeline attrition, it makes sense to focus technical solutions at to profile the properties of any chemical structure against an
or before this piVotal decision point in drug deVelopment. end point (26). Current applications for predicting individual
Having considered how and where, why does preclinical toxicology issues in silico have been summarized (8), as are
safety assessment fail to detect some target organ toxicity? This recent efforts to model carcinogenic risk (8, 29).
PerspectiVe Chem. Res. Toxicol., Vol. 19, No. 11, 2006 1397

How a model is applied can be as important as the model the potentially fatal cardiac arythmia, Torsades de Pointe (32).
itself. Given the number of parameters that must be optimized Electrophysiological recording in cells or tissue explants and
in a drug molecule, it may be unrealistic to expect even a suite screening for binding to the human ether-a-go-go (hERG), the
of in silico models to predict the biological properties of a single potassium channel responsible for repolarization of the cardi-
clinical candidate. By analogy, the Heisenberg uncertainty omyocyte, are used to screen compounds prior to collecting
principle describes limitations to the accuracy in measuring only electrophysiological recordings in conscious animals (33).
two parameters, the position and momentum, of a particle. A Interpreting the liability for Torsade’s must be approached on
more practical approach is to “sieve” the many parameters that a case-by-case basis; negative results in surrogate assays do not
describe druggable chemical space in silico into “bins” of guarantee safety in vivo (32). Nonetheless, this suite of tools
predicting biological and biopharmaceutical properties. Biologi- has moved the issue from empirical screening to rational drug
cal tests on selected compounds check the accuracy of the seive design and hypothesis testing.
and enrichment of bins without the need to integrate the entire Organ slices and primary cell models can be used to address
chemical space. This iterative approach can also guide an a variety of target organ toxicities. Isolated hepatocytes have
effective “fit-for-purpose” plan to define safety issues (30). been used to screen for hepatotoxicity and drug-drug interac-
Using an in silico model to look for the metaphorical needle in tions. Screening for PPARR agonists in isolated hepatocytes is
a haystack may be less practical than determining if the haystack routine. Isolated hepatocytes from rodents, nonrodents, and
is rich in needles before one starts looking. humans are useful to screen compounds for induction of drug-
metabolizing enzymes and drug-drug interactions and to
3.2.1. Opportunities, Gaps, and Needs establish human relevance for metabolism (34, 35). Organ slices
are useful models when the tissue architecture is important for
Although in silico models exist for some applications, there toxicity, e.g., if toxicity requires interactions between different
are many gaps where in silico models either underperform or cell types, and they provide opportunities for species compari-
cannot be built due to a lack of information. In addition, the sons (36). Regardless of the assay format, knowledge of the
application of in silico models is an information technology- mechanism is necessary to link results from a screen with
intensive and model-dependent exercise (8, 26, 28). Barriers to pathophysiology to ensure that the data will translate to an
effective implementation include (i) a lack of model compounds improved safety profile in vivo.
(training sets) that exemplify the toxicity, (ii) access to in vivo
studies data in a common format, and (iii) availability of relevant 3.3.2. Surrogate Animal Models
higher throughput assays to extend the chemical space available
Lower organisms with shorter life cycles are amenable to
to the model. Training set compounds are necessary to train
higher throughput testing. Developmental and behavioral tox-
the model, but chemical structures and associated study data
icities are only two examples where disrupting complex
are often intellectual property and unavailable. When available,
interactions among many cell types results in toxicity. The value
in vivo data may be fragmented and incomplete or generated
of surrogate animal models seems clear; for example, knowledge
in different species and strains, adding further variation (25).
of apoptosis can be traced to basic mechanisms defined in
Structures in public databases (26) can help inform models but
Caenorhabditis elegans (37). Recent reviews suggest that
may have little relevance to druggable chemical space. Finally,
zebrafish show considerable promise as surrogate animal models
a lack of physiologically based higher throughput surrogate
for cardiovascular toxicity (38), C. elegans models offer insights
assays applicable to issue-driven toxicity assessment hampers
into neurotoxicity (39), and amphibian models have been used
the ability to rapidly expand the chemical space of interest within
for developmental toxicity screening (40). Both public and
the model. These, and other gaps, can limit the impact of in
private databases, too numerous to reference, offer a wealth of
silico models for predicting drug toxicity. Progress using
genetic, biochemical, and cell biology contents for eukaryotic
predictive ‘omic databases for in silico predictions will be
organisms from yeasts to mice.
addressed later.
3.3.3. Opportunities, Gaps, and Needs
3.3. Surrogate Models for Safety Assessment
Lead optimization is the final opportunity to design properties
During lead optimization (Figure 1), the lead series narrows into a clinical candidate (Figure 1). There is a critical need for
to a subset of druggable compounds (5). If unacceptable toxicity, mechanism-based surrogate models to address specific issues
e.g., low MOS, emerges from pilot animal studies, defining an during lead optimization, accelerate candidate identification, and
on-target vs off-target mechanism is critical, but assay formats reduce later pipeline attrition. The importance of this juncture
that rapidly and reliably inform the SAR are also essential. In in the development pipeline cannot be overemphasized. The
vivo studies, the current gold standard for predicting safety, are safety studies supporting first human exposure are the points
poor SAR tools since differences in absorption and disposition where safety issues for man are first noted (24); the quality of
confuse compound comparisons and throughput is low. Sur- the candidates emerging from lead optimization will impact
rogate assays have a higher capacity, and compounds can be safety and pipeline productivity directly.
compared at equivalent concentrations, but results are only An important gap to implementing mechanism-based sur-
meaningful if connected to pathophysiology in vivo. A review rogate assays is a poor understanding of the cellular physiology
of surrogate assays is beyond the scope of this discussion, but in vivo and how physiological pathways adapt at the cellular
useful links offer entry to the literature (http://ecvam.jrc.it/ level in vitro. A focus on developing primary cell culture
index.htm). Illustrative examples are outlined below. models, tissue explants, and even a return to more classical
isolated organs provides opportunities for species comparisons.
3.3.1. Cell- and Tissue-Based Surrogate Assays Where primary tissues are not available, stem cells may provide
populations of differentiated cells that can be used as safety
A battery of surrogate assays has been deployed to detect screens (41). Additional work will be necessary for wider
delayed ventricular repolarization (Qt prolongation) linked to applications of these technologies.
1398 Chem. Res. Toxicol., Vol. 19, No. 11, 2006 SteVens

Defining the relationships among surrogates and the complex safety. Simply associating dose and time dependence with
physiology of mammalian species represents an opportunity for toxicity is insufficient; mechanism-based biomarkers are neces-
systems biology approaches (see below). Global transcript sary to provide accurate risk assessment and facilitate clinical
profiling can be an effective first step to defining similarities monitoring. Additional mechanism-based PD and validated
and differences in the basic physiology of preclinical test species premonitory safety biomarkers are critical to improving pre-
vs surrogate models, even in the absence of toxicant treatment. clinical safety assessment, pipeline productivity, and monitoring
Progress has been made with hepatocyte profiling (42-44), and human safety.
hepatocyte gene expression profiles have been used to build Species differences are an important issue. Lack of cross-
predictive toxicogenomic databases (45). Combinatorial ap- reactivity for reagents, such as antibodies, is a significant hurdle
proaches to defining conditions that support physiological to applying biomarkers to preclinical safety assessment. Mining
functions of primary cells and technical advances in multiplexing the genome and tissue expression profiles are useful for
biological measurements of cellular function in higher through- gathering candidate biomarkers, but functional annotation of the
put formats are also essential (46). New sources of physiologi- rat, dog, and monkey genome, typical rodent and nonrodent
cally based in vitro models, e.g., surrogate animals and stem species used in preclinical testing, is incomplete. Species
cells, to interrogate liabilities for specific target organ toxicities, differences in the activity of small molecules or therapeutic
are sorely needed. Effective deployment of batteries of assays proteins targeting important classes of receptors, such as
that can be “fit-for-purpose” will allow the flexibility to address G-protein-coupled receptors or kinases, hamper interpretation
safety issues early in development (30). of safety data and create uncertainty in assessing on-target safety
issues with biologics. Increased attention on the need for
3.4. Biomarkers biomarkers in preclinical safety assessment is necessary to solve
these problems.
The need for biomarkers in preclinical safety assessment is Streamlining validation criteria is also critical. This will
highlighted in a follow-up report to the FDA Critical Path require increased cooperation among academic, regulatory, and
document (47): “New biomarker development has stalled....- industrial scientists. In addition, a single marker will not predict
[P]otential new biomarkers have been proposed, but the essential multiple pathologies in a single organ, e.g., ALT and hepato-
work needed to evaluate their utility...has not been carried out.” toxicity, while combinatorial biomarkers approaches, although
Preclinical studies help validate a biomarker as a surrogate or more complex and costly to develop, provide deeper insight
clinical end point to diagnose or predict clinical outcomes. (49) and may be an opportunity for systems toxicology
However, preclinical application of biomarkers can also provide approaches.
earlier signals of developing safety issues in vivo. Preclinical
biomarkers important for mechanism-based risk assessment 3.5. Systems Toxicology and Pharmacogenetics:
include PD markers and safety markers. Safety markers can be Making Sense of High Content Information
further subdivided into premonitory and surrogate markers for
injury. Pharmacogenetics (often used interchangeably with pharma-
PD biomarkers measure the response of the target and relate cogenomics) investigates the interactions between a patient or
drug exposure to target modulation and biological outcome, animal genome and drug responses (15). Systems toxicology
necessary information to differentiate on- vs off-target mech- “represents an analytical approach to the relationships among
anisms of toxicity. Any measurement technology that monitors elements of a system” as applied to toxicology (28, 50). Both
a process linked mechanistically to target modulation can be require integration of information from multiple sources to
useful for developing PD biomarkers. Premonitory and surrogate describe and predict the behavior of biological systems; sum-
safety biomarkers measure changes that occur prior to, or maries of technologies and computational models are available
concurrent with, morphological evidence of pathology, respec- (8, 28, 45, 50). Two aspects will be addressed briefly, predicting
tively. For example, dogs exhibit focal myocardial necrosis after toxicity using ‘omic approaches and predicting individual patient
isoproterenol treatment (48). An increase in heart rate may be safety through application of pharmacogenetic information.
premonitory for cardiac injury while cardiac troponins and/or Available public and commercial and databases and clas-
creatine kinase are released coincident with the rupture of sification algorithms (25, 28, 45, 51, 52) establish that global
cardiomyocytes and serve as surrogate markers of cell death. transcript and metabolite profiles can be used to classify toxicity
Likewise, the release of ALT and AST with concurrent from unknown compounds. It is noteworthy that the classifica-
hyperbilirubinemia may be a surrogate for hepatic injury, but tions appear robust despite the small numbers of structurally
increases in ALT and AST, with or without bilirubinemia, may dissimilar compounds (e.g., ∼10 for single pathology) used to
occur in the absence of morphological evidence of injury. Early train the models (53, 54). For example, a phospholipidosis
signals might indicate that injury will progress or may accom- classifier was built using 12 model compounds and a signature
modate without an increase in severity or collapse of the MOS. set of less than 20 genes (55). This suggests that biological
Predicting whether or not an early signal is a transient and classifiers can be built based on the complexity inherent in the
reversible effect, or a harbinger of progressive and irreversible biological data relating to pathogenesis as opposed to an
injury, describes the essence of the predictive toxicology extensive SAR based on similar structures. Although progress
dilemma. may seem slow relative to expectations, results suggest that the
predictive toxicology database can assist in safety predictions
3.4.1. Opportunities, Gaps, and Needs and increase knowledge of the mechanism.
Although many genetic polymorphisms may contribute to
Mechanism-based PD and safety biomarkers allow safety disease, variability in the phenotype and genotype of the
issues to be monitored and species differences to be defined as enzymes and transporters that govern the metabolism and
a candidate approaches the clinic. Increased focus in this area disposition of a drug offers the greatest opportunity to under-
represents an opportunity to improve risk assessment, relate stand and eliminate ADRs. The Paracelsus quotation, “All things
findings in preclinical species to humans, and increase clinical are poison...the dosis (sic) determines that a thing is not a
PerspectiVe Chem. Res. Toxicol., Vol. 19, No. 11, 2006 1399

poison”, is central to toxicology (reference in ref 37), but a of basic mechanisms of drug-induced target organ toxicity in
modern Paracelsus, schooled in pharmacogenetics and epide- preclinical animal models and humans is also essential to
miology, might have said, “The dose determines the poison, improve mechanism-based risk assessment for new and existing
but the patient and environment determine the dose that drugs. Progress will require both effective translation of basic
poisons.” Because toxicity is directly related to exposure, it is research to practical applications and a clear focus on the right
not surprising that variation in metabolism and disposition technical problems. A few final points frame the opportunities
contributes to drug toxicity (15). Accordingly, differences in and challenges already enumerated.
metabolism and drug-drug interactions may contribute to ADRs Improving the availability of safety data to fuel public and
from a wide variety of drugs (23). Current preclinical safety private collaboration is essential. The NIH Roadmap and FDA
assessment practices are designed to reduce variability and yield Critical Path documents call for more focused application of
statistically robust results. This experimental necessity under- basic science in toxicology. Toxicologists must embrace the
estimates contributions from genetic variability to toxicity. intent of these documents and work to refine and steer the
Efforts to understand and apply murine (the Mouse Phenome applications to the most productive outcomes. Regulatory
Project; www.jax.org) and rat (www.physiogenix.com) genetic agencies must enable the development of new technologies and
variability to understand drug responses may offer opportunities processes by providing incentives to bring forward new safety
to improve our understanding of ADRs using preclinical models. data. Industry must enter into open collaborations and facilitate
Additional work will be necessary to establish the relevance of the use of relevant compounds and safety data to test novel
genetic diversity in drug responses with preclinical species and methods to improve performance and metric progress on safety
human. assessment. All sectors must participate in educating the public
to the risk:benefit considerations for any drug (57).
3.5.1. Opportunities, Gaps, and Needs The basic science of toxicology must advance to meet these
challenges. As noted by Liebler (58), major advances often come
Recent advances in the tools and application of systems from outside the field of toxicology. Changes to study sections
toxicology and pharmacogenetics offer the opportunity to structure at the NIH raise fear in the academic community that
improve preclinical safety testing. Effective information and the focus of funding for research in toxicology will be diffused
knowledge management apply here as well but are discussed across interdisciplinary study sections. Although there is risk,
above with regard to in silico applications. opportunity and change require risk taking; toxicologists must
A mechanistic understanding of species differences is another adapt and compete based on the clinical impact and application
key area of need. Phylogenetics defines the genetic basis for of their research.
differences among species and relatedness within lineages. The The high-content nature of systems approaches offers great
fact that species differences are at the heart of preclinical safety opportunity and significant barriers that must be addressed.
assessment calls for a “phylogenetic systems toxicology” Systems approaches as currently applied often describe what
approach to safety assessment. Investigation of species differ- “might happen” or what “did happen” but do not predict the
ences must progress from descriptive to mechanism-based probability of an outcome. Risk assessment is neither a purely
systems biology approaches. This requires consistent functional qualitative nor a retrospective exercise. Risk assessment, as
annotation of the various genomes and mapping differences in practiced in pharmaceutical development, attempts to quantify
physiology and pathophysiology across species including sur- risk (e.g., MOS) in advance to enable good risk:benefit decisions
rogate animal models. Understanding the genetic basis of drug regarding therapeutic outcomes and patient safety. Systems
toxicity will require systems approaches to integrate a “physi- approaches are complex, but must, nonetheless, achieve a level
omic” view for preclinical species and humans. of quantitation and develop capabilities for intuitive visualization
The low incidence of ADRs and “curse of dimensionality” of results to allow machine learning to be incorporated into
are barriers that also confound predicting responses for indi- human learning and decision making. Understanding the mech-
vidual patients (15). The well-recognized statin-induced rhab- anism is a key to defining these relationships.
domyolysis incidence is only 3.4 per 100000 and depends on To comment on this or other Future of Toxicology perspec-
multiple factors (23). Low frequency of ADRs with complex tives, please visit our Perspectives Open Forum at http://
genetic and environmental etiology may only appear after a drug pubs.acs.org/journals/crtoec/openforum.
has been on the market for some time (1); understanding genetic
and environmental contributions is difficult without the clarity
Acknowledgment. I thank Drs. Lorrene Buckley, Myrtle
of 20:20 hindsight. However, considerable progress will be Davis, Michael Dorato, Thomas Jones, Derek Lieshman, Armen
required to improve pharmacovigilance and streamline phylo- Tashjian, Craig Thomas, John Vahle, David Watson, and Daniel
genetic systems toxicology approaches to help elucidate the roles Wierda for critical comments and Dr. Gerald Miwa for personal
of the environment and genetics in mechanisms of ADRs. communications.

4. Summary: Improving Mechanism-Based Risk References

Assessment
(1) Lasser, K. E., Allen, P. D., Woolhandler, S. J., Himmelstein, D. U.,
The preceding discussion provides a perspective and sug- Wolfe, S. M., and Bor, D. H. (2002) Timing of new black box
gestions on directions for future mechanistic research on drug warnings and withdrawals for prescription medications. J. Am. Med.
Assoc. 287 (17), 2215-2220.
safety as well as practical applications to mechanism-based risk (2) Kola, I., and Landis, J. (2004) Can the pharmaceutical industry reduce
assessment. Toxicology is poised to deliver safer drugs to attrition rates? Nat. ReV. Drug DiscoVery 3 (8), 711-715.
patients at lower cost, but technical and nontechnical challenges (3) FDA (2004) Innovation and Stagnation: Challenges and opportunities
must be met to realize this vision. Recent advances in on the critical path to new medical products. http://www.fda.gov/oc/
initiatives/criticalpath/whitepaper. html.
understanding the mechanism of metabolic activation of a drug
(4) Greaves, P., Williams, A., and Eve, M. (2004) First dose of potential
to reactive species (56) and the basic mechanism of cell death new medicines to humans: How animals help. Nat. ReV. Drug
(37) offer new insights, but a focus on improved understanding DiscoVery 3 (3), 226-236.
1400 Chem. Res. Toxicol., Vol. 19, No. 11, 2006 SteVens

(5) Lipinski, C., and Hopkins, A. (2004) Navigating chemical space for (28) Waters, M. D., and Fostel, J. M. (2004) Toxicogenomics and systems
biology and medicine. Nature 432 (7019), 855-861. toxicology: Aims and prospects. Nat. ReV. Genet. 5 (12), 936-948.
(6) Dorato, M. A., and Vodicnik, M. J. (2001) The toxicological (29) Contrera, J. F., Maclaughlin, P., Hall, L. H., and Kier, L. B. (2005)
assessment of pharmaceutical and biotechnology products. In Prin- QSAR modeling of carcinogenic risk using discriminant analysis and
ciples and Methods of Toxicology (Hayes, W. A., Ed.) pp 243-283, topological molecular descriptors. Curr. Drug DiscoVery Technol. 2
Taylor and Francis, Philadelphia. (2), 55-67.
(7) El-Hage, J. (2004) Preclinical and Clinical Safety Assessment (30) Mayne, J. T., Ku, W. W., and Kennedy, S. P. (2006) Informed toxicity
for PPAR Agonists. http://www.fda.gov/CDER/present/DIA2004/ assessment in drug discovery: Systems-based toxicology. Curr. Opin.
Elhage.ppt. Drug DiscoVery DeV. 9 (1), 75-83.
(8) Ekins, S., Nikolsky, Y., and Nikolskaya, T. (2005) Techniques: (31) Wilson, A. G., White, A. C., and Mueller, R. A. (2003) Role of
Application of systems biology to absorption, distribution, metabolism, predictive metabolism and toxicity modeling in drug discoverysA
excretion and toxicity. Trends Pharmacol. Sci. 26 (4), 202-209. summary of some recent advancements. Curr. Opin. Drug DiscoVery
(9) Karin, M., Yamamoto, Y., and Wang, Q. M. (2004) The IKK NF- DeV. 6 (1), 123-128.
kappa B system: A treasure trove for drug development. Nat. ReV. (32) Lawrence, C. L., Pollard, C. E., Hammond, T. G., and Valentin, J. P.
Drug DiscoVery 3 (1), 17-26. (2005) Nonclinical proarrhythmia models: Predicting Torsades de
(10) O’Neill, L. A. J. (2006) Targeting signal transduction as a strategy to Pointes. J. Pharmacol. Toxicol. Methods 52 (1), 46-59.
treat inflammatory disease. Nat. ReV. Drug DiscoVery 5 (7), 549- (33) Chiang, A. Y., Holdsworth, D. L., and Leishman, D. J. (2006) A one-
563. step approach to the analysis of the QT interval in conscious
(11) Li, Z. W., Chu, W., Hu, Y., Delhase, M., Deerinck, T., Ellisman, M., telemetrized dogs. J. Pharmacol. Toxicol. Methods 54 (2), 183-188.
Johnson, R., and Karin, M. (1999) The IKKbeta subunit of IkappaB (34) O’Brien, P. J., and Siraki, A. G. (2005) Accelerated cytotoxicity
kinase (IKK) is essential for nuclear factor kappaB activation and mechanism screening using drug metabolising enzyme modulators.
prevention of apoptosis. J. Exp. Med. 189 (11), 1839-1845. Curr. Drug Metab. 6 (2), 101-109.
(12) Arkan, M. C., Hevener, A. L., Greten, F. R., Maeda, S., Li, Z. W., (35) Sivaraman, A., Leach, J. K., Townsend, S., Iida, T., Hogan, B. J.,
Long, J. M., Wynshaw-Boris, A., Poli, G., Olefsky, J., and Karin, M. Stolz, D. B., Fry, R., Samson, L. D., Tannenbaum, S. R., and Griffith,
(2005) IKK-beta links inflammation to obesity-induced insulin resis- L. G. (2005) A microscale in vitro physiological model of the liver:
tance. Nat. Med. 11 (2), 191-198. Predictive screens for drug metabolism and enzyme induction. Curr.
(13) Sasseville, V. G., Lane, J. H., Kadambi, V. J., Bouchard, P., Lee, F. Drug Metab. 6 (6), 569-591.
W., Balani, S. K., Miwa, G. T., Smith, P. F., and Alden, C. L. (2004) (36) Vickers, A. E., and Fisher, R. L. (2005) Precision-cut organ slices to
Testing paradigm for prediction of development-limiting barriers and investigate target organ injury. Exp. Opin. Drug Metab. Toxicol. 1
human drug toxicity. Chem.-Biol. Interact. 150 (1), 9-25. (4), 687-699.
(14) Dorato, M. A., and Engelhardt, J. A. (2005) The no-observed-adverse- (37) Orrenius, S., and Zhivotovsky, B. (2006) The future of toxicology-
effect-level in drug safety evaluations: Use, issues, and definition(s). does it matter how cells die? Chem. Res. Toxicol. 19 (6), 729-733.
Regul. Toxicol. Pharmacol. 42 (3), 265-274. (38) Zon, L. I., and Peterson, R. T. (2005) In vivo drug discovery in the
(15) Wilke, R. A., Reif, D. M., and Moore, J. H. (2005) Combinatorial zebrafish. Nat. ReV. Drug DiscoVery 4 (1), 35-44.
pharmacogenetics. Nat. ReV. Drug DiscoVery 4 (11), 911-918. (39) Driscoll, M., and Gerstbrein, B. (2003) Dying for a cause: Invertebrate
(16) Halford, J. C. (2006) Obesity drugs in clinical development. Curr. genetics takes on human neurodegeneration. Nat. ReV. Genet. 4 (3),
Opin. InVest. Drugs 7 (4), 312-318. 181-194.
(17) Gardin, J. M., Schumacher, D., Constantine, G., Davis, K. D., Leung, (40) Song, M. O., Fort, D. J., McLaughlin, D. L., Rogers, R. L., Thomas,
C., and Reid, C. L. (2000) Valvular abnormalities and cardiovascular J. H., Buzzard, B. O., Noll, A. M., and Myers, N. K. (2003) Evaluation
status following exposure to dexfenfluramine or phentermine/fenflu- of Xenopus tropicalis as an alternative test organism for frog embryo
ramine. J. Am. Med. Assoc. 283 (13), 1703-1709. teratogenesis assaysXenopus (FETAX). Drug Chem. Toxicol. 26 (3),
(18) Mekontso-Dessap, A., Brouri, F., Pascal, O., Lechat, P., Hanoun, N., 177-189.
Lanfumey, L., Seif, I., Haiem-Sigaux, N., Kirsch, M., Hamon, M., (41) Chaudhary, K. W., Barrezueta, N. X., Bauchmann, M. B., Milici, A.
Adnot, S., and Eddahibi, S. (2006) Deficiency of the 5-hydroxy- J., Beckius, G., Stedman, D. B., Hambor, J. E., Blake, W. L., McNeish,
tryptamine transporter gene leads to cardiac fibrosis and valvulopathy J. D., Bahinski, A., and Cezar, G. G. (2006) Embryonic stem cells in
in mice. Circulation 113 (1), 81-89. predictive cardiotoxicity: Laser capture microscopy enables assay
(19) FDA (2005) Estimating the maximum safe starting dose in initial development. Toxicol. Sci. 90 (1), 149-158.
clinical trials for therapeutics in adult healthy volunteers. http:// (42) Baker, T. K., Carfagna, M. A., Gao, H., Dow, E. R., Li, Q., Searfoss,
www.fda.gov/CDER/guidance/5541fnl.doc. G. H., and Ryan, T. P. (2001) Temporal gene expression analysis of
(20) Mutlib, A. E., Gerson, R. J., Meunier, P. C., Haley, P. J., Chen, H., monolayer cultured rat hepatocytes. Chem. Res. Toxicol. 14 (9), 1218-
Gan, L. S., Davies, M. H., Gemzik, B., Christ, D. D., Krahn, D. F., 1231.
Markwalder, J. A., Seitz, S. P., Robertson, R. T., and Miwa, G. T. (43) Waring, J. F., Ciurlionis, R., Jolly, R. A., Heindel, M., Gagne, G.,
(2000) The species-dependent metabolism of efavirenz produces a Fagerland, J. A., and Ulrich, R. G. (2003) Isolated human hepatocytes
nephrotoxic glutathione conjugate in rats. Toxicol. Appl. Pharmacol. in culture display markedly different gene expression patterns depend-
169 (1), 102-113. ing on attachment status. Toxicol. in Vitro 17 (5-6), 693-701.
(21) Lazarou, J., Pomeranz, B. H., and Corey, P. N. (1998) Incidence of (44) Boess, F., Kamber, M., Romer, S., Gasser, R., Muller, D., Albertini,
adverse drug reactions in hospitalized patients: A meta-analysis of S., and Suter, L. (2003) Gene expression in two hepatic cell lines,
prospective studies. J. Am. Med. Assoc. 279 (15), 1200-1205. cultured primary hepatocytes, and liver slices compared to the in vivo
(22) Uetrecht, J. P. (1997) Current trends in drug-induced autoimmunity. liver gene expression in rats: Possible implications for toxicogenomics
Toxicology 119 (1), 37-43. use of in vitro systems. Toxicol. Sci. 73 (2), 386-402.
(23) Law, M., and Rudnicka, A. R. (2006) Statin safety: A systematic (45) Fielden, M. R., and Kolaja, K. L. (2006) The state-of-the-art in
review. Am. J. Cardiol. 97 (8A), 52C-60C. predictive toxicogenomics. Curr. Opin. Drug DiscoVery DeV. 9 (1),
(24) Olson, H., Betton, G., Robinson, D., Thomas, K., Monro, A., Kolaja, 84-91.
G., Lilly, P., Sanders, J., Sipes, G., Bracken, W., Dorato, M., Van, D. (46) Flaim, C. J., Chien, S., and Bhatia, S. N. (2005) An extracellular matrix
K., Smith, P., Berger, B., and Heller, A. (2000) Concordance of the microarray for probing cellular differentiation. Nat. Methods 2 (2),
toxicity of pharmaceuticals in humans and in animals. Regul. Toxicol. 119-125.
Pharmacol. 32 (1), 56-67. (47) Critical Path Opportunities Report (2006) http://www.fda.gov/oc/
(25) Fostel, J., Choi, D., Zwickl, C., Morrison, N., Rashid, A., Hasan, A., initiatives/criticalpath/reports/opp_report.pdf.
Bao, W., Richard, A., Tong, W., Bushel, P. R., Brown, R., Bruno, (48) Sandusky, G. E., Means, J. R., and Todd, G. C. (1990) Comparative
M., Cunningham, M. L., Dix, D., Eastin, W., Frade, C., Garcia, A., cardiovascular toxicity in dogs given inotropic agents by continuous
Heinloth, A., Irwin, R., Madenspacher, J., Merrick, B. A., Papoian, intravenous infusion. Toxicol. Pathol. 18 (2), 268-278.
T., Paules, R., Rocca-Serra, P., Sansone, A. S., Stevens, J., Tomer, (49) Koop, R. (2005) Combinatorial biomarkers: From early toxicology
K., Yang, C., and Waters, M. (2005) Chemical effects in biological assays to patient population profiling. Drug DiscoVery Today 10 (11),
systems-data dictionary (CEBS-DD): A compendium of terms for 781-788.
the capture and integration of biological study design description, (50) Hood, L., and Perlmutter, R. M. (2004) The impact of systems
conventional phenotypes, and ’omics data. Toxicol. Sci. 88 (2), 585- approaches on biological problems in drug discovery. Nat. Biotechnol.
601. 22 (10), 1215-1217.
(26) Yang, C., Benz, R. D., and Cheeseman, M. A. (2006) Landscape of (51) Hayes, K. R., and Bradfield, C. A. (2005) Advances in toxicogenomics.
current toxicity databases and database standards. Curr. Opin. Drug Chem. Res. Toxicol. 18 (3), 403-414.
DiscoVery DeV. 9 (1), 124-133. (52) Lindon, J. C. H. E. N. J. K. (2006) Metabonomics techniques and
(27) Eddy, S. R. (2004) What is Bayesian statistics? Nat. Biotechnol. 22 applications to pharmaceutical research and development. Pharm. Res.
(9), 1177-1178. 23 (6), 1075-1088.
PerspectiVe Chem. Res. Toxicol., Vol. 19, No. 11, 2006 1401

(53) Fielden, M. R., and Zacharewski, T. R. (2001) Challenges and (56) Baillie, T. A. (2006) Future of toxicology-metabolic activation and
limitations of gene expression profiling in mechanistic and predictive drug design: Challenges and opportunities in chemical toxicology.
toxicology. Toxicol. Sci. 60 (1), 6-10. Chem. Res. Toxicol. 19 (7), 889-893.
(54) Nicholson, J. K., Holmes, E., Lindon, J. C., and Wilson, I. D. (2004) (57) Smith, D. A., Johnson, D. E., and Park, B. K. (2006) Editorial
The challenges of modeling mammalian biocomplexity. Nat. Biotech- overview: Safety of drugs can never be absolute. Curr. Opin. Drug
nol. 22 (10), 1268-1274. DiscoVery DeV. 9 (1), 26-28.
(58) Liebler, D. C. (2006) The poisons within: Application of toxicity
(55) Sawada, H., Takami, K., and Asahi, S. (2005) A toxicogenomic mechanisms to fundamental disease processes. Chem. Res. Toxicol.
approach to drug-induced phospholipidosis: Analysis of its induction 19 (5), 610-613.
mechanism and establishment of a novel in vitro screening system.
Toxicol. Sci. 83 (2), 282-292. TX060213N