This tutorial is an introduction aimed at academics and other researchers who are
new to the field of drug discovery. We outline here the fundamental concepts and
processes of drug discovery. Our goal is to guide researchers toward the steps necessary
to translate benchside findings into bedside applications, and to locate resources
that can help provide reagents and services needed in this process. The views presented
here are based on pharmaceutical industry experiences, but are by no means the only
perspective on the highly complex and diverse field of drug discovery and development.
For more comprehensive textbooks and reviews on this topic, please refer to our
list of references.
We represent here the process of drug discovery and development in a flow diagram
that serves as a navigation bar throughout the tutorial. This flow diagram implies
a sequential process, but in reality a large number of iterative cycles are involved
in which a project will spin until, finally, a winner is determined that fits the
profile needed to proceed to the next step. Thus, the flow chart should be seen
as the overall sequence of steps that needs to be taken, but not as a literal depiction
of any given drug discovery program.
Navigation: The top navigation bar serves as entry into the tutorial. Submenus and
tutorial text open by clicking on links.
Drug discovery and development has its own vocabulary, which we attempt to define
in the glossary of terms. The
references encompasses ample reading material for the interested and motivated
reader; however, we gladly accept recommendations
of additional citations.
Drug discovery and development can broadly follow two different paradigms—physiology-based
drug discovery and target-based discovery. The main difference between these two
paradigms lies in the time point at which the
drug target is actually identified. Physiology-based drug discovery follows
physiological readouts, for example, the amelioration of a disease phenotype in
an animal model or cell-based assay. Compounds are screened and profiled based on
this readout. A purely physiology-based approach would initially forgo target identification/validation
and instead jump right into screening. Identification of the drug target and the
mechanism of action would follow in later stages of the process by deduction based
on the specific
pharmacological properties of
By contrast, the road of target-based drug discovery begins with identifying the
function of a possible therapeutic target and its role in disease. Given the thousands
of human or pathogen genes and the variety of their respective
gene products, this can be a difficult task. Furthermore, insight into the
"normal" or "native"
function of a gene or gene product does not necessarily connect the gene
or gene product to disease.
The two paradigms are not mutually exclusive, and drug discovery projects can employ
a two-pronged approach. The genomics revolution has been the main driver of the
target-based paradigm over the last decade.
Currently, all existing therapies together hit only about 400 different drug targets,
according to a Science review article (Drews, 2000
). The same review estimates that there are at least 10 times as many potential
drug targets that could be exploited for future drug therapy.
- The disease mechanism defines the possible cause or causes of a particular disorder,
as well as the path or
phenotype of the disease. Understanding the disease mechanism directs research
and formulates a possible treatment to slow or reverse the disease process. It also
predicts a change of the disease pattern and its implications.
- Disease mechanisms can be broadly classified into the following groups
- Defects in distinct genes—genetic disorders
- Infection by bacteria, fungi, or viruses
- Immune/autoimmune disease
- Trauma and acute disease based on injury or organ failure
- Multicausal disease
- Disease genes have been identified based on hereditary patterns even before the
knowledge of the
DNA sequences of the human genome. Following an original founder mutation,
these genetically inherited diseases run in families; examples include phenylketonuria,
cystic fibrosis, Huntington disease, Fanconi's anemia, and autosomal-dominant familial
- The specific gene defects or
mutations that bring about a hereditary disorder have been identified for a
number of diseases. Progress in DNA
sequencing technology has enabled rapid identification of disease genes
screening. Early intervention is possible for a limited number of hereditary
- A large fraction of disease, however, is not based on the mutation of a single gene,
but rather on a number of genes that together determine a person's risk of developing
a particular disease. For example, certain mutations in the BRCA gene family raise
the risk for cancer. However, this risk does not always equal 100 percent certainty,
and individuals bearing certain BRCA mutations may never develop cancer. Certain
allelic variants can increase susceptibility for diseases, such as the ApoE4 allele
does for Alzheimer's.
- Environmental factors such as diet, toxic exposures, trauma, stress, and other life
experiences are assumed to interact with genetic
susceptibility factors to result in disease. Thus, drug targets may include
pathways related to environmental factors.
- Target Type
- Targets for therapeutic intervention can be broadly classified into these categories:
- The "drugability" of a given target is defined either by how well a therapeutic,
small molecule drugs or
antibodies, can access the target, or by the
efficacy a therapeutic can actually achieve. A long list a parameters
influences drugability of a given target; these include cellular location, development
resistance, transport mechanisms such as export pumps, side effects, toxicity,
- Some target
classes, for example, the G-protein coupled receptors (GPCRs), have been successfully
targeted, and a sizable number of drugs prescribed today hit this particular class.
Therefore, the GPCR target type is considered drugable.
genomics can be broadly defined as the systematic analysis of gene activity
in healthy versus diseased organisms/organs/tissues/cells.
- Specifically, functional genomics employs the large-scale exploration of gene
function that includes the analysis of regulatory networks, biochemical
protein-protein interactions, the effects of gene
knockouts or gene upregulation or
gain-of-function, and the results of functional complementation
- Functional genomics aims to determine disease mechanisms and to identify disease
disease markers. It also aims to guide the understanding of
signal transduction pathways that either lead to disease or indicate
therapeutic strategies for the development of novel therapeutics.
- Functional genomics relies heavily on disease models that are based on the high
homology of genes and their function in a variety of organisms ranging from nematodes
- Functional genomics employs high-throughput sequencing and high-density arraying
of gene expression and activity of gene products. The information content of functional
genomics experiments is exceedingly large; it requires sophisticated statistical
analysis, which has accelerated the discipline of bioinformatics.
Target validation requires a demonstration that a
molecular target is critically involved in a disease process, and that modulation
of the target is likely to have a therapeutic effect.
The validation of a molecular target in vitro usually precedes the validation of
the therapeutic concept in vivo; together this defines its clinical potential. Validation
involves studies in intact animals or disease-related cell-based models that can
provide information about the integrative response of an organism to a
pharmacological intervention and thereby help to predict the possible profile
of new drugs in patients.
Transgenic animals where the target gene is
knocked out have become an important experimental approach for the determination
of the function of targets (genes) in a whole organism.
- Knockouts of genes that are essential in development are usually lethal. Inducible
knockouts, i.e., transgenic models where the transgene can be switched on or off
at will in the adult animal, can be used to study the function of such essential
- Disease models are transgenic animal models which present a
phenotype that bears the hallmarks of a certain disease. These can be combined
with knockouts to study the effect of modulating or inhibiting the function of the
gain-of-function models reactivate
gene expression of the target gene, and often ameliorate or even reverse
the disease phenotype. Knockins also are used to create disease models.
- Knockins or gain-of-function can also be lethal. For example, switching on or restoring
the function of cell cycle genes in post-mitotic cells often leads to cell death.
Selective switching-on of genes might present a therapeutic strategy if such restoration
of gene function can be engineered in a tissue- or organ-specific fashion.
- Most neurodegenerative disease models have been generated by introducing mutant
genes that cause autosomal-dominant forms of the disease in humans. In these models,
the mutant gene (such as APP, presenilin, tau, superoxide dismutase-1, expanded
huntingtin) is assumed to result in a toxic gain of function, but the actual mechanisms
by which the mutations cause disease phenotypes remain debated.
- The action and interaction of genes and their
gene products is complex. Research aimed to define
pathways that control and regulate processes in living organisms provides
valuable information for drug discovery.
- The knowledge of a pathway allows definition and separate targeting of upstream
or downstream targets.
Inhibition or modulation of selected targets could lead to the same therapeutic
side effects or better drugability.
- Knowledge of pathways and their relation to each other helps researchers understand
side effect profiles.
- Identification of one disease target can lead to a number of alternative
drug targets in the same pathway and increase the possibilities for a novel
therapeutic. Examples include the drugs acting on the cholesterol synthesis pathway.
- The best validation of a target is clinical
efficacy and safety data.
- Second- and third-generation therapeutics often have better efficacy and
side effect profiles based on the clinical trials and track record of first-generation
- Efficacy in clinical trials, i.e., amelioration or reversal of disease in human
patients, is the ultimate validation of a target.
- Efficacy in animal disease models does not always predict the outcome in patients.
The reliability of disease models for the prediction of human clinical trials varies
widely among diseases and needs to be assessed on a case-by-case basis.
DNA/RNA and RNAi
- Antisense DNA/RNA are
analogs thereof that are complementary to a specific sequence of
DNA. The underlying concept of antisense therapeutics is that the
antisense compound binds to the native target to form a double-stranded sequence
and thus inhibits its normal function. An antisense drug for viral retinitis has
- Interfering RNA or RNAi is a
gene silencing phenomenon, whereby specific double-stranded RNAs (dsRNAs) trigger
the degradation of homologous messenger RNA (mRNA). The specific dsRNAs are processed
into small interfering RNA (siRNA), which initiates the cleavage of the homologous
mRNA in a complex named the RNA-induced silencing complex (RISC).
- Introduction of either dsRNA or siRNA into cells leads to inhibition of the biological
function encoded in the targeted mRNA-the underlying concept of RNAi therapeutics.
For example, this approach is being investigated to silence mutant alleles of tau,
APP, ataxin, and SOD1.
Knock-out and Chemical Biology
- The prevailing approach for target validation involves the study of the biology
of a disease. Knowledge of the disease mechanism and the underlying biological
pathways leads to the identification and characterization of
- A fundamentally different approach involves using
compound collections to
phenotypes generated by exposure to molecules; this connects chemical
structures to biological effects from the start irrespective of the
molecular targets/pathways that are hit.
- This "chemical biology" takes a holistic and random approach to drug discovery.
It may complement traditional, deductive approaches.
- In chemical biology, chemical
knockouts are a new method whereby the effect of a chemical
compound, not a genetic manipulation, knocks out the function of a gene
and thus leads to a readable phenotype. Chemical biology and chemical knockouts
rely on the creation of diverse
chemical libraries of many thousands of compounds.
- Cornerstone technologies for this new way of drug discovery are combinatorial chemistry
and genetic manipulation of biosynthetic pathways in microbes for the production
of new compounds.
The key to drug discovery is an assay that fulfills several important criteria:
- Relevance: Does the readout unequivocally relate to the target?
- Reliability/Robustness: Are results reproducible and statistically significant?
- Practicality: Do time, reagents, and effort correlate with quality and quantity
- Feasibility: Can assay be run with resources at hand?
- Automation: In order to screen large numbers of compounds, can assay be automated
and run in highly parallel format?
- Cost: Does cost of the assay permit scale-up for high-throughput screening?
The quality of an assay determines the quality of data, i.e., compromising on assay
development can have substantial downstream consequences.
- In vitro/Cell-based
- In-vitro assays monitor a surrogate readout. Examples for such a readout are the
catalytic action of an isolated
enzyme, the binding of an
antibody to a defined
antigen, or the growth of an engineered
- An in-vitro assay system can be designed using only
recombinant reagents, reagents that were isolated from
lysates, whole crude lysates, or intact cells.
- Cell-based assays range in their complexity from simple
cytotoxicity assays or cell growth to reporter gene assays that monitor activation
or upregulation of certain genes or their gene products.
functional assays are usually more complex. They combine several molecular components
to mimic the function of a biological process, such as activation of a
pathway. Biological processes that can be monitored in cell-based functional
assays include changes in cell
cell migration, or
apoptosis are the catalytic action of an isolated enzyme, the binding
of an antibody to a defined antigen, or the growth of an engineered cell line.
In general, in-vitro assays are more robust and cost-effective, and have fewer ethical
implications than whole-animal experiments. For these reasons they are usually chosen
for high-throughput screening, where tens of thousands of data points are generated
in the hunt for novel drug molecules.
- In vivo/Animal
- In-vivo testing involves whole organisms. It assesses both
pharmacology and biological
efficacy in parallel.
- Animal models have specific characteristics that mimic human diseases. The technologies
for the creation of
transgenic animals, where certain genes are either deleted, modulated, or
added, have progressed tremendously in the last decade. As a consequence, the predictive
power of animal models for human disease and pharmacology is improving. Even so,
human biology and disease is so complex that for many diseases or pharmacological
parameters, the human remains the definitive model. For some disease, e.g., hepatitis
C, adequate models still do not exist.
- It is important to note that some experts in the pharmaceutical industry and the
U.S. Food and Drug Administration (FDA) believe that inadequate animal models, or
the lack of animal models altogether, are a major hurdle in drug discovery and development.
- Pharmaceutical companies have long used model organisms in preclinical efficacy
and safety studies. With the emerging knowledge of whole genomes, researchers are
now increasingly seeking animal models not only of specific diseases, but also of
their underlying particular
pathways to broaden assays from pharmacology to include mechanism of action.
- Regarding current animal models of Alzheimer disease, scientists debate whether
they adequately model the disease. Amyloid-depositing models, for example, have
scant, if any, cell loss. Disease models are usually incomplete models of pathology
or mechanism, and their utility in drug
is limited by the validity of the pathway in human disease pathogenesis.
- High-throughput screening (HTS) aims to rapidly assess the activity of a large number
of compounds or extracts on a given target. The term HTS is used when assays are
run in a parallel fashion using multi-well
assay plates (96-, 384-, 1536-well).
- Assays run in 1536-well plates with minuscule volumes (single-digit microliter to
nanoliter scale) are sometimes referred to as ultra high-throughput screening or
- Today, HTS/UHTS commonly involves semi-automation or full automation for
liquid handling, sample preparation, running of the actual assays, as well as
data analysis. HTS laboratories frequently employ robots and the latest detection
technologies for assay readouts.
- Assay development for HTS/UHTS faces formidable challenges in terms of reagent stability
and cost, environmental robustness (temperature, oxidation, agitation), and statistics
(signal-to-noise ratios, Z and Z' quality measures). Therefore, the ultimate design
of a HTS/UHTS assay often differs from its respective lower- throughput format.
- Large HTS/UHTS operations are significant investments. Optimal and cost-effective
use, as well as minimization of down-time, are important issues in today's drug
- It is common in today's HTS environment to run a primary screen through a 1,000,000
in a matter of days. However, while the actual screen may only take a few days,
assay development usually involves weeks of engineering and fine-tuning to achieve
sufficient speed and robustness, as well as cost-effectiveness.
Hits to Lead—Overview
After successful development of an assay,
compound library follows. Primary
screens will identify
hits. Subsequently, confirmation screens and counter screens
leads out of the pool of hits. This winnowing process is commonly referred
to as "hits-to-leads."
The success of screening depends on the availability of
compounds, as well as their quality and diversity. Efforts to synthesize,
collect, and characterize compounds are an essential and costly part of drug discovery.
Compound libraries are the "bread and butter" of
screening. There are several sources for compounds:
- Natural products (NPs) from microbes, plants, or animals. NPs are usually tested
as crude extracts first, followed by isolation and identification of active compounds.
- (Random) collections of discreetly synthesized compounds.
- Random libraries exploring "chemical space."
- The total number of possible small organic molecules with a molecular mass of less
than 500 that populate "chemical space" is estimated to exceed 1060-vastly more
than were ever made or indeed will ever be made.
- Given this near -infinite number of theoretical compounds, one can either focus
the search around known molecules or
pharmacophores with biological activity, or sample the chemical compound
universe with a random selection of diverse representatives. Both approaches are
used, and complement each other, in today's drug discovery efforts.
- In contrast to the theoretical small-molecule universe, the idea of "privileged"
structures has been advanced. Such structures represent a discreet selection of
compounds with the highest probability of having biological activity, i.e., of interacting
with the universe of biological diversity that has developed on Earth. Likewise,
this biological diversity can be viewed as privileged, because all organisms on
Earth together do not contain anywhere near the theoretical number for 300 amino
acid proteins, 10390.
An important practical measure for the value of a random library is chemical diversity,
which analyzes how similar one compound in the library is to one other.
- In silico/CADD
- Advances in computing power and in structure determination by x-ray crystallography
and NMR have made computer-aided drug design (CADD)
and structure-based drug design (SBDD) essential tools for drug discovery.
- Elucidation of
structures has been industrialized in recent years, such that structural information
about a given drug target, or the binding conformation of a drug, are available
to the scientist at earlier stages of drug discovery. HIV protease inhibitor drugs
are a prominent success story for SBDD.
- Virtual (in-silico)
screening sifts through large numbers of compounds based on a user-defined set
of selection criteria. Selection criteria can be as simple as a physical molecular
property such as molecular weight or charge, a chemical property such as number
heteroatoms, number of hydrogen-bond acceptors or donors. Selection criteria
can be as complex as a three-dimensional description of a
binding pocket of the target protein, including chemical functionality and
- In-silico screening can involve simple filtering based on static selection criteria
(i.e., molecular weight). Alternatively, it can involve actual docking of
ligands to a target site, which requires computer-intensive algorithms for
conformational analysis, as well as binding energies.
- Selection criteria are often combined, either in Boolean fashion or otherwise, to
generate complex queries which, for example, describe a
SAR established from experimental data. Scoring functions are used to rank
compounds that meet selection criteria.
- Initially, in-silico screening was intended to filter out the majority of
compounds that have little chance of hitting a
target. In this way, one can either reduce the actual number of compounds
being screened in a benchtop assay, or enrich a yet-to-be-screened library with
compounds that have a chance of hitting the target.
- With increasingly sophisticated algorithms describing the interaction of ligands
and receptors, in-silico screening is more commonly being implemented in drug discovery.
In-silico screening has been particularly helpful in projects where a wide-ranging
SAR around a discreet
is known (QSAR), or where high-resolution three-dimensional structural information
is available (SBDD).
and Combinatorial Chemistry
Screening relies on the availability and chemical synthesis of
- Today, a chemist typically supplies new compounds to the screener in milligram or
even sub-milligram amounts. Compound synthesis often involves the synthesis of precursors,
which can serve as the starting point for a
compound series. Such precursors tend to involve scale-up procedures, since
larger amounts are needed for subsequent
- By rule of thumb, one chemist synthesizes, purifies, and characterizes about 100
novel compounds per year, fewer if the task is complex. It takes approximately 10,000
different compounds to develop a drug that will make it to market.
- The large capacity and appetite of screening operations has motivated chemists to
develop new approaches involving parallel synthesis of many compounds. Such parallel
synthesis is called fast analoging when chemical space is explored around a defined
pharmacophore, or combinatorial chemistry when compounds are created by
combining arrays of building blocks employing the same underlying chemistry. Both
technologies have led to large
of synthetic compounds that are used for screening.
- A primary
screen is designed to rapidly identify
- The goals are to minimize the number of false positives and to maximize the number
of confirmed hits. One philosophy often quoted by people in screening operations,
especially HTS environments, is not to fret about compounds that were missed but
to really care about the quality of data for the compounds that repeat.
- Depending on the assay, hit rates typically range between 0.1 percent and 5 percent.
This number also depends on the cutoff parameters set by the researchers, as well
dynamic range of a given assay.
- Typically, primary screens are initially run in multiplets (i.e., two, three, or
more assay data points) of single
compound concentrations. Readouts are expressed as percent activity in comparison
to a positive (100 percent) and a negative (0 percent) control.
Hits are then retested a second time (or more often, depending on the assays' robustness).
The retest is usually run independently of the first assay, on a different day.
If a compound exhibits the same activity within a statistically significant range,
it is termed a confirmed hit, which can proceed to dose-response screening.
- Initial potencies of hits are either reported in milligrams per milliliter (mgs/mL),
where the molecular weight of
compounds is not weighed in, or in micromolar (uM), which takes into account
the different molecular weights of compounds.
hits have potencies between 1 and 100 uM, somewhat dependent on the
dynamic range and cutoff of assays.
- Hits with potencies in the nanomolar (nM) range are rare.
- Establishing a dose-response relationship is an important step in hit verification.
It typically involves a so-called secondary
screen. In the secondary screen, a range of compound concentrations usually
serial dilution is tested in an assay to assess the concentration or dose
dependence of the assay's readout.
- Typically, this dose-response is expressed as an
antibody-, or cell-based assays, or as an
EC50 in in-vivo experiments.
The shape of a dose-response curve, where drug concentration is recorded on the
x-axis and drug effect on the y-axis, often provides information about the mechanism
of action (MOA).
hits proceed to a series of counterscreens. These assays usually include drug
targets of the same
receptor family, for example, panels of GPCRs or
kinases. In cases where selectivity between subtypes is important,
counterscreens might include a panel of
protein complexes, or heterooligomers. Counterscreens profile the
action of a confirmed hit on a defined spectrum of biological target
- Selectivity toward a
drug target decreases the risk of so-called off-target
side effects. Selectivity and potency are often coupled, i.e., selectivity
increases with better potency.
- Counterscreens are also used to confirm the mechanism of action. For example, if
a drug molecule is believed to interfere with a particular amino acid side-chain
in a protein, it will not affect a mutant protein where that residue is changed
to a different amino acid. If a drug molecule is interacting with target class-specific
residues involved in catalysis, it will not affect a different target
The number and stringency of counterscreens can vary widely and depend on the drug
of Action (MOA)
- One of the goals throughout the discovery of novel drugs is to establish and confirm
the mechanism of action. In an ideal scenario, the MOA remains consistent from the
level of molecular interaction of a drug molecule at the target site through the
physiological response in a disease model, and ultimately in the patient.
- As an example, let's assume the
drug target is a protein
kinase. A confirmed
hit inhibits the in-vitro
catalytic activity of the kinase in the primary
screen, where a surrogate or known physiological
phosphorylated. In the next step, whole cells are exposed
to the same
inhibitor, the cells are
lysed, and the physiological or native substrate is isolated and its
phosphorylation state determined. Next, in a disease model dependent on a
pathway regulated by the target kinase, one assesses the effect of the inhibitor
on the pathway and the
. If the drug action in all three steps is consistent, an MOA is established.
Lead optimization is the complex, non-linear process of refining the chemical
structure of a confirmed
hit to improve its drug characteristics with the goal of producing a preclinical
drug candidate. This stage frequently represents the bottleneck of a drug discovery
Lead optimization employs a combination of empirical, combinatorial, and rational
approaches that optimize leads through a continuous, multi-step process based on
knowledge gained at each stage. Typically, one or more confirmed hits are evaluated
in secondary assays, and a set of related
analogs, are synthesized and
The testing of analog series results in quantitative information that correlates
changes in chemical structure to biological and pharmacological data generated to
establish structure-activity relationships (SAR).
The lead optimization process is highly iterative. Leads are assessed in
assays for their "druglikeness." Medicinal chemists change the lead molecules based
on these results in order to optimize pharmacological properties such as bioavailability
or stability. At that point the new analogs feed back into the screening hierarchy
for the determination of potency, selectivity, and MOA. These data then feed into
the next optimization cycle. The lead optimization process continues for as long
as it takes to achieve a defined drug profile that warrants testing of the new drug
- Medicinal chemistry blends synthetic chemistry, molecular modeling, computational
pharmacology to discover and design new drugs, and investigate their
interaction at the molecular, cellular, and whole-animal level.
- Medicinal chemistry combines empirical knowledge from the structure-function relationships
of known drugs with rational designs optimizing the physicochemical properties of
- For example, medicinal chemists improve drug
efficacy, particularly with respect to stability and bioavailability, by developing
pro-drugs. Pro-drugs are engineered in such a way that they undergo chemical
transformation either in the bloodstream or specific tissues such as the liver.
Upon transformation, biologically active
are released, which are the actual drugs.
- Animal PK/PD/ADME
pharmacodynamics (PD), and absorption, distribution, metabolism, and excretion
(ADME) assess the general pharmacology and mechanisms of action of drugs.
Lead molecules are administered via different routes: intravenous (iv), intraperitoneal
(ip), subcutaneous (sc), intramuscular (im), rectal, intranasal (IN), inhalational,
oral (po), transdermal, topical, etc. The main models used are rodents including
mouse and rat, but larger animals such as dogs, pigs, and, more rarely, monkeys,
are also used under certain circumstances. The main objective is to understand the
effects on the whole organism of
exposure to a novel chemical entity, and to predict the new drug's behavior
- PK/PD/ADME studies are an integral part of lead optimization. They feed back into
the medicinal chemistry effort aiming to optimize the physicochemical properties
of new leads in terms of minimal toxicity and
side effects, as well as of maximum
efficacy toward disease.
- PK/PD/ADME studies are expensive and usually have limited throughput. Some PK/PD
studies require specific
radioisotope labeling of lead molecules, all of which tend to draw
heavily on medicinal chemistry resources.
- PK/PD/ADME studies rely heavily on analytical methods and instrumentation. The recent
innovation and progress in mass
spectroscopy, (whole-body) imaging, and
chromatography technology (HPLC, LC-MS, LC-MS-MS) have tremendously
increased the quantity and quality of data generated in PK/PD experiments.
- A large number of parameters is assessed. Here is a partial list: (ADME); bioavailability
(F) and protein binding; stability and half-life (t1/2); maximum serum concentration
(Cmax); total exposure or
area under the curve (AUC)
; clearance (Cl); volume of distribution (Vd); drug-drug interactions; onset of
drug action; multicompartmental analysis of blood, liver, and other tissues.
- The definition of toxicity is the degree to which a substance or mixture of substances
can harm humans or animals. Acute toxicity involves harmful effects in an organism
through a single or short-term exposure. Chronic toxicity is the ability of a substance
or mixture of substances to cause harmful effects over an extended period, usually
upon repeated or continuous exposure that can last for the entire life of the exposed
organism. This may well apply to many Alzheimer's drugs.
- These days, the screening process includes a series of standard assays early on:
inhibition (using either
recombinant cytochrome P450
enzymes or liver microsome), MTT-like
cytotoxicity assays, effects on cardiac HERG channels. Toxicity
in these relatively simple in- vitro assays flags
leads and goes into the risk-benefit evaluation of which lead series
can advance into preclinical studies.
- Animal models are used for escalating dose studies aimed at determining a maximum
tolerated dose (MTD).
This step involves monitoring a series of parameters, such as body weight, food
intake, blood chemistry (BUN),
and liver activity. Biopsies are usually stored in freezers for subsequent pathological
- Animal toxicity studies require relatively large amounts of
compound. The purity of the compound needs to be very high in order to exclude
toxicities stemming from impurities. The norm for short-term animal toxicity is
one- or two-week studies. Long-term testing in animals ranges in duration from several
weeks to several years. Some animal testing continues after human tests have begun
in order to learn whether long-term use of a drug may cause cancer or birth defects.
Empirically, medicinal chemists find it difficult to "engineer away" existing toxicity.
Hence, time and money is spent instead on lead series that come without early liabilities.
formulation and delivery of drugs is an integral part of the drug discovery
and development process. Indeed, formulation problems and solutions influence the
design of the
lead molecules; they feed back into the iterative lead optimization cycle,
as well as the preclinical and clinical evaluations.
- In turn, formulation and delivery are closely linked. For example, intravenous delivery
of a novel drug might call for a different formulation than oral delivery, because
parameters such as metabolic stability or solubility can differ significantly.
- If formulation substances are not generally recognized as safe (GRAS),
they become part of the safety assessment and their
behavior, as well as toxicity profile, needs to be documented in the IND (investigational
new drug) application. In fact, side effects such as local irritation or allergic
reactions are often attributable to drug formulation, not the active pharmaceutical
- Formulation substances might exhibit different biological activity than the actual
drug. For example, certain formulations enhance absorption through their interaction
with the cell membrane of the gastrointestinal tract.
- Formulation and delivery are highly specialized fields of research, and formulation
scientists are now part of serious drug discovery and development programs from
the early stages.
- Indeed, a sizable number of drug discovery and development programs in the pharmaceutical
and biotech industry are centered around new ways of formulating already known and
even marketed drugs to increase their
or safety profiles.
The decision to take a new drug candidate into the development phase entails a significant
commitment in terms of money, resources, and time.
attrition rate for making it to market is a disheartening nine in 10
compounds, and development costs per approved drug amount to $800 million,
according to a study by the Tufts CSDD (xyz, 2003, PoW link.) The average time to
develop a new drug was 12 years and 10 months in 2002.
The number of new chemical entities (NCEs) gaining market approval has decreased
over the last decade down to 20 per year. At the same time, the estimated average
NCEs needed for the pharmaceutical and biotech industry to sustain a 5 percent
growth rate is 50.
In terms of standards, drug development requires attention to the following:
GLP-Good Laboratory Practice refers to nonclinical laboratory studies that support
or are intended to support applications for research or marketing permits;
GMP-Good Manufacturing Practice, also known as cGMP ("current" GMP), is a set
of regulations requiring that quality, safety, and effectiveness be built into foods,
drugs, medical devices, and biological products.
21 CFR-describing the code of regulations for food and drugs. Part 11 has become
particularly relevant describing the standards and regulations on electronic data
and electronic signatures.
FDA requirements, a sponsor must first submit data showing that the drug is
reasonably safe for use in initial, small-scale clinical studies. Depending on whether
compound has been studied or marketed previously, the sponsor may have several
options for fulfilling this requirement: (1) compiling existing nonclinical data
from past in-vitro laboratory or animal studies of the compound; (2) compiling data
from previous clinical testing or marketing of the drug in the United States or
another country whose population is comparable to the U.S. population; or (3) undertaking
new preclinical studies designed to provide the evidence necessary to support the
safety of administering the compound to humans.
- During preclinical drug development, a sponsor evaluates the drug's toxic and
pharmacologic effects through in-vitro and in-vivo laboratory animal testing.
Genotoxicity screening is performed, as well as investigations on drug absorption
metabolism, the toxicity of the drug's
, and the speed with which the drug and its metabolites are excreted from the body.
At the preclinical stage, the FDA will generally ask, at a minimum, that sponsors:
(1) develop a pharmacological profile of the drug; (2) determine the acute toxicity
of the drug in at least two species of animals, and (3) conduct short-term toxicity
studies ranging from two weeks to three months, depending on the proposed duration
of use of the substance in the proposed clinical studies.
- Upon nomination of a development candidate, it becomes imperative that the highest-quality
compound can be provided for preclinical and clinical development repeatedly and
consistently at reasonable cost and in a timely manner.
- The initial synthetic route will be revised and optimized to achieve:
- Accessibility of readily available and cost-effective starting material
- Minimization of synthetic and purification steps
- Feasibility of scale-up from microgram to gram, and possibly to kilogram scale.
- Reduced cost of goods (COGS
- The FDA will require a Chemistry, Manufacturing and Controls (CMC)
documentation package for any drug entering clinical trials.
- Active Pharmaceutical Ingredients (API)
- Description and characterization
- Synthesis/method of manufacture
- Process controls
- Specifications (list of tests, methods and acceptance criteria)
- Purity profiles
- Container/closure system for drug substance (DS) storage
- Container/closure system for drug product (DP) shelf life
Examples abound where quality or cost of compounds led to delays or failure of clinical
trials, underscoring the importance of process development for the overall success
of a new drug.
- IND Application
- In many ways, the investigational new drug (IND) application is the result of a
successful preclinical development program. The
IND is also the vehicle through which a sponsor advances to the next stage
of drug development known as clinical trials (human trials).
- Generally, this includes data and information in three broad areas:
Pharmacology and Toxicology Studies: Preclinical data to permit an assessment
of whether the product is reasonably safe for initial testing in humans.
- Manufacturing Information: Information pertaining to the composition, manufacture,
stability, and controls used for manufacturing the drug substance and the drug product.
This information is assessed to ensure the company can adequately produce and supply
consistent batches of the drug.
Clinical Protocols and Investigator Information: Detailed protocols for proposed
clinical studies to assess whether the initial-phase trials will expose subjects
to unnecessary risks. Also, information on the qualifications of clinical investigators
to assess whether they are qualified to fulfill their clinical trial duties. Clinical
investigators are professionals, generally physicians, who oversee administration
of the experimental compound.
- Types of
: "Commercial INDs" are applications that are submitted primarily by companies whose
ultimate goal is to obtain marketing approval for a new product. However, there
is another class of filings broadly known as "non-commercial" INDs, which, in fact,
account for the vast majority of INDs filed. Submitted by NIH and other sponsors,
these INDs include "Investigator INDs," "Emergency Use INDs," and "Treatment INDs."
Clinical trials are a peculiar hybrid between a formalized and strictly regulated
process on the one hand and a sophisticated stratagem on the other, particularly
when it comes to patient selection, statistical methodology, disease markers, and
endpoints employing cutting-edge research. They are also expensive, accounting for
50 to 70 percent of the drug discovery and development cost. They can be very long,
lasting many years depending on therapeutic area.
Ninety percent of
NCEs entering clinical trials fail. Forty percent of compounds fail in Phase
1, 62 percent of successful Phase 1 compounds fail in phase 2, 40 percent of successful
Phase 2 compounds fail in Phase 3, and a surprising 23 percent of successful Phase
3 compounds fail at the registration stage, when the FDA denies approval for a completed
New Drug Application (NDA.)
In 1991, the main reason for failure was problems in
PK/bioavailability (40 percent) followed by lack of efficacy (30 percent)
and toxicology (12 percent). In 2000, the main reason for failure was lack of
efficacy (27 percent), followed by commercial and market reasons (21 percent)
and toxicology (20 percent).
Success rates vary with therapeutic area: Cardiovascular (20 percent), arthritis/pain
(17 percent) and infectious disease drugs (16 percent) fare better than drugs for
CNS diseases (8 percent), oncology (5 percent,) or women's health (4 percent).
All current clinical trials registered with the FDA are listed at this website:
- Phase I—Overview
- Phase 1 includes the initial introduction of an investigational new drug into humans.
These studies are closely monitored and may be conducted in patients, but are usually
conducted in healthy volunteer subjects. These studies are designed to determine
the metabolic and pharmacologic actions of the drug in humans, the side effects
associated with increasing doses, and, if possible, to gain early evidence on
efficacy. During Phase 1, sufficient information about the drug's pharmacokinetics
pharmacological effects should be obtained to permit the design of well-controlled,
scientifically valid Phase 2 studies.
- Phase 1 studies also evaluate drug
metabolism, structure-activity relationships (SAR),
and the mechanism of action (MOA) in humans. These studies also determine which
investigational drugs are used as research tools to explore biological phenomena
or disease processes. The total number of subjects included in Phase 1 studies varies
with the drug, but is generally in the range of 20 to 80.
In Phase 1 studies, CDER (Center for Drug Evaluation and Research) can impose a
clinical hold (i.e., prohibit the study from proceeding or stop a trial that has
started) for reasons of safety, or because of a sponsor's failure to accurately
disclose the risk of study to investigators. Although CDER routinely provides advice
in such cases, investigators may choose to ignore any advice regarding the design
of Phase 1 studies in areas other than patient safety.
- Safety and
- The first Phase 1 study is usually a single-dose study where healthy volunteers
receive a range of single doses of the investigational drug. The design and determination
of the dose range relies on data such as the maximum tolerated dose (MTD) determined
in preclinical animals studies. Vital signs and physiological parameters, such as
blood chemistry, are closely monitored in the volunteers and the
PK parameters in humans are determined for a single dose.
- The safety and PK data from the single-dose study serve as guide posts for a subsequent
multiple-dose study in healthy volunteers, where indicated.
- Occasionally, Phase 1 testing is divided into two steps known as Phase 1a and Phase
1b. Phase 1a studies normally are conducted as a short-term study to ensure safety
before embarking on a longer and more comprehensive Phase 1b study. Phase 1b studies
can include actual patients and might provide first indications about drug
efficacy against disease.
Establishing the safety of a new drug molecule is paramount in Phase 1. Also essential
is the determination of the best dosage or dosage regimen for subsequent, larger
phase 2 trial(s), where the assessment of drug effectiveness in patients moves to
- Phase II
- Phase 2 includes early
controlled clinical studies conducted to obtain some preliminary data on the
efficacy of the drug for a particular indication (or indications) in patients with
the disease. This testing phase also helps determine common short-term side effects
and risks associated with the drug.
- Decisive or pivotal trials are usually run as randomized controlled trials (RCT).
Randomization introduces a deliberate element of chance into the assignment of treatments
to trial patients.
- Phase 2a: Pilot trials to evaluate efficacy and safety in selected populations of
about 100 to 300 patients who have the condition to be treated, diagnosed, or prevented.
They often involve hospitalized patients who can be closely monitored. Objectives
may focus on dose-response, type of patient, frequency of dosing, or any of a number
of other issues involved in safety and efficacy.
Phase 2b: Well-controlled trials to evaluate safety and efficacy in patients who
have the condition to be treated, diagnosed, or prevented. These trials usually
represent the most rigorous demonstration of a medicine's efficacy.
- Phase III
- Phase 3 studies are expanded,
controlled, and uncontrolled trials. They are performed after preliminary evidence
effectiveness has been obtained in Phase 2, and are intended to gather the
additional information about safety and effectiveness needed to evaluate the overall
benefit-risk relationship of the drug. Phase 3 trials should provide an adequate
basis for extrapolating the results to the general population and conveying that
information in the physician labeling. These studies usually include several hundred
to several thousand people.
- In both Phase 2 and 3, the Center for Drug Evaluation and Research (CDER), a branch
FDA, can impose a clinical hold if a study is unsafe or if the protocol
design is deficient in meeting its stated objectives. The FDA aims to ensure that
this determination reflects current scientific knowledge, agency experience with
clinical trial design, and experience with the class of drugs under investigation.
- FDA approval/disapproval decisions are based on the results of pivotal studies.
To be considered pivotal, a study must meet at least these 4 criteria:
- Be controlled using placebo or a standard therapy.
- Have a double-blinded design when such a design is practical and ethical.
- Be randomized.
Be of adequate size.
- Although the amount of information and data submitted in
NDAs varies, the components of NDAs are uniform. The components of any NDA are,
in part, a function of the nature of the subject drug and the information available
to the applicant at the time of submission. As outlined in Form FDA-356h (Application
to Market a New Drug for Human Use Or As An Antibiotic Drug For Human Use) NDAs
can consist of as many as 15 different sections:
- Chemistry, Manufacturing, and Control (CMC);
- Samples, Methods Validation Package, and Labeling;
- Nonclinical Pharmacology and Toxicology;
- Human Pharmacokinetics and Bioavailability;
- Microbiology (for anti-microbial drugs only);
- Clinical Data;
- Safety Update Report (typically submitted 120 days after the NDA's submission);
- Case Report Tabulations;
- Case Report Forms;
- Patent Information;
- Patent Certification; and
- In the primary review process, reviewers attempt to confirm and validate the sponsor's
conclusion that a drug is safe and effective for its proposed use. The review is
likely to involve a reanalysis or an extension of the analyses presented in the
NDA. For example, the medical reviewer may seek to reanalyze a drug's effectiveness
in a particular patient subpopulation not analyzed in the original submission. Similarly,
the reviewer may disagree with the sponsor's assessment of evaluable patients and
seek to retest effectiveness claims based on the patient populations defined by
- Review team members communicate extensively with each other. If a medical reviewer's
reanalysis of clinical data produces results different from those of the sponsor,
the reviewer will forward this information to the statistical reviewer with a request
for a statistical reanalysis of the data. Likewise, the pharmacology reviewer may
work with the statistical reviewer in evaluating the statistical significance of
potential side effects in long-term animal studies.
When the technical reviews are complete, each reviewer develops a written evaluation
of the NDA that presents their conclusions and their recommendations on the application.
The division director or office director then evaluates the reviews and recommendations
and decides the action that the division will take on the application. The result
is an action letter that provides an approval, approvable, or non-approvable decision
and a justification for that recommendation.
- Phase IV—Overview
- Phase 4 trials are done after a drug has received a market approval. These trials
are monitoring drugs that are available for doctors to prescribe, rather than experimental
drugs that are still being developed. Pharmaceutical companies run Phase 4 trials
to find out:
- More about safety and side effects of the drug.
- What the long-term risks and benefits are.
How well the drug works when it is used more widely than in clinical trials.
Below are some current examples of approved drugs that were retracted:
Reason Drug Is Pulled
Fenfluramine (weight loss)
Pulmonary hypertension, heart valve disease
Posicor (hypertension, angina)
Reduced liver enzymes
Heart problem when taken with other drugs
Hoescht Marion Roussel
Propulsid (nocturnal heartbeat)
Heart attack, stroke
Severe cardiovascular problems
Raplon (airway muscle relaxant)
Hepatitis, liver failure
Ischemic colitis, constipation