Cato - Clinical Drug Trials and Tribulations PDF
Cato - Clinical Drug Trials and Tribulations PDF
and Tribulations
Second Edition, Revised and Expanded
edited by
Allen Cato
Lynda Sutton
Cato Research Ltd.
Durham, North Carolina
Headquarters
Marcel Dekker, Inc.
270 Madison Avenue, New York, NY 10016
tel: 212-696-9000; fax: 212-685-4540
The publisher offers discounts on this book when ordered in bulk quantities. For more
information, write to Special Sales/Professional Marketing at the headquarters address
above.
Neither this book nor any part may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopying, microfilming, and recording,
or by any information storage and retrieval system, without permission in writing from
the publisher.
Executive Editor
James Swarbrick
PharmaceuTech, Inc.
Pinehurst, North Carolina
Advisory Board
Peter G. Welling
Institut de Recherche Jouveinal
Fresnes, France
Allen Cato
Lynda Sutton
Allen Cato III
Preface
Contributors
David F. Bernstein, Ph.D. Cato Research Ltd., Corona del Mar, California
Allen Cato, M.D., Ph.D. Cato Research Ltd., Durham, North Carolina
Allen Cato III, Ph.D. Cato Research Ltd., San Diego, California
Joseph A. DiMasi, Ph.D. Tufts Center for the Study of Drug Development,
Tufts University, Boston, Massachusetts
Marlene E. Haffner, M.D., M.P.H. United States Food and Drug Administra-
tion, Rockville, Maryland
Louis A. Morris, Ph.D. Louis A. Morris & Associates, Dix Hills, New York
Paul J. Reitemeier, Ph.D. National Center for Ethics, Veterans Health Admin-
istration, White River Junction, Vermont, and Department of Medicine, Dart-
mouth Medical School, Hanover, New Hampshire
† Deceased
Jay Philip Siegel, M.D., F.A.C.P. Center for Biologics Evaluation and Re-
search, United States Food and Drug Administration, Rockville, Maryland
W. Leigh Thompson, M.D., Ph.D., Sc.D., F.A.C.P., F.C.C.M.* Eli Lilly and
Company, Indianapolis, Indiana
Karen D. Weiss, M.D. Office of Therapeutics Research and Review, Center for
Biologics Evaluation and Research, United States Food and Drug Administration,
Rockville, Maryland
Michael G. Wilson, M.S. Michael G. Wilson and Company, Inc., New Pales-
tine, Indiana
* Retired
Allen Cato
Cato Research Ltd., Durham, North Carolina
Lynda Sutton
Cato Research Ltd., Durham, North Carolina
It may be easier for a camel to pass through the eye of a needle than it is for a
new chemical entity to reach the marketplace. Drug development is a long and
costly process fraught with tribulation. The tortuous pathway traveled by a new
drug from synthesis to sale requires the constant percolation of data through rigor-
ous clinical and regulatory filters. This process is complex, and success cannot
be guaranteed. The ability to always predict which drug will have all the qualities
necessary to gain regulatory approval and to be marketed remains as elusive as
a camel in a needle’s eye.
New drugs do make it from discovery to the market, but only at the approxi-
mate rate of one in every 10,000 new molecules synthesized. It is a long, costly,
and extremely risky process involving a steady progression through multiple
stages, with treacherous decision points along the way. Most of all, it is a process
involving the constant percolating of data through rigorous filters strewn with
tribulations and complicated by the difficulty of making decisions that affect
human health when all the facts are not known.
Despite the daunting challenge of bringing a new drug from discovery to
market, new medicines continue to be developed that may have a significant
effect on our health. You may wonder, ‘‘What has medicine done for man-
kind lately?’’ In the United States, the adult life expectancy increased by nearly
Smallpox killed an average of more than 1500 people per year between
1900 and 1904; it is now eradicated worldwide, and children are no
longer vaccinated against the disease.
Polio struck more than 16,000 people annually in the early 1950s; today,
it has been eliminated from the Western Hemisphere.
During the past 50 years, vaccines also have been responsible for drastically
reducing the morbidity and mortality from measles, Haemophilus influenzae
type b (Hib), diphtheria, pertussis, tetanus (DPT, typically administered together),
hepatitis B, and chicken pox. In addition to improved health, substantial economic
benefits have been realized. For example, the CDC estimates that the United
States recoups its investment in the eradication of smallpox every 26 days. De-
spite the obvious advances of modern medicine, many patients still have infec-
tious, chronic, or genetic diseases and will benefit from the research of today
finding the effective treatments of tomorrow. Pharmaceutical research targeting
the top 12 major medical needs exceeds $645 billion annually in direct medical
expense and lost productivity. The diseases included in this figure are Alzhei-
mer’s disease, arthritis, asthma, cancer, congestive heart failure, coronary heart
disease, depression, diabetes, hypertensive disease, osteoporosis, schizophrenia,
and stroke (2). Just as it did 50 years ago, innovation continues today to bring
us new knowledge through genetic research, molecular biology, and enhanced
computer technology. This is the promising future of drug development.
However, developing the vaccines or any of the drugs potentially used to
treat the indications listed above is a substantial undertaking. To comprehend
clearly the magnitude of the drug-development process, it is useful to consider
the many different areas involved. Figure 1 depicts some of the key disciplines
contributing to the process. Information from each of these areas feeds into a
common funnel with a filter, where multiple decisions must be made progres-
sively regarding the compound’s survival, or lack thereof.
Figures 2 and 3 illustrate the process broken down into preclinical and
clinical segments. Keep in mind, however, that the process is a dynamic one.
The various disciplines listed are constantly interacting, and the entire flow of
data requires constant feedback and fine tuning. For example, a compound’s tox-
icity, however slight, may be considered to outweigh its pharmacological effect.
This information would be given by the toxicologist to the chemist, who would
make other compounds with slight modifications, attempting to retain the pharma-
cological effect while decreasing or eliminating the toxic effect.
Once a compound has been synthesized in the lab and tested in animals,
an Investigational New Drug (IND) application is submitted to the U.S. Food
and Drug Administration (FDA), requesting permission to initiate clinical studies
of the drug in humans. The IND summarizes the preclinical work and includes
the first clinical protocol. It is not until the drug has been experimentally tested
in humans under controlled conditions (after Phase III) that the company may
file an application to market the drug (a New Drug Application [NDA] if filed
with the FDA, or a Marketing Authorization Application [MAA] if filed in Eu-
rope). The application summarizes all preclinical (safety and efficacy in animals),
clinical (safety and efficacy in humans), and manufacturing data known about
the drug, and requests permission to market this new drug.
Figure 4 illustrates the attrition ratio of a new chemical entity as it works
its way from synthesis through preclinical development to IND, and subsequently
through clinical development to NDA. An attrition ratio of 10,000:1 (not consid-
ered good betting odds by most people) is the bad news. The good news is that
95% of all drugs for which an NDA is submitted are ultimately approved for
marketing.
Dose Time
Species Drug (mg/kg/day) (months) Effect
In animals
Imipramine (Troframil) Antidepressant
Fenfluramine (Pondimin) Anorectic
Thioridazine (Mellaril) Antipsychotic
Chlorcyclizine (Fedrazil) Antihistamine
Zimelidine Antidepressant
In humans and animals
Chloroquine (Plaquenil) Antimalarial
The case described represents just one of many decisions that must be made
before beginning clinical trials. Clinical drug trials are described as Phases I–V.
The first trials in humans that test the drug for safety are considered Phase I.
These studies usually employ normal volunteers, and may expose about 50 indi-
viduals to the drug. For known toxic compounds such as anticancer agents, only
patients with the targeted illness would be used.
The first studies to define efficacy are considered Phase II. These studies
are typically conducted to determine the best dosage regimen for the Phase III
efficacy studies. In general, 100–300 patients would be entered into various con-
Before discussing some of the problems that can arise during clinical trials, a
brief review of some of the basic components constituting a clinical trial is in
order. Figure 7 illustrates the study periods providing the framework for any
clinical trial.
Prestudy activities include design and setup of the study, and poststudy
activities include data entry, analysis, and report generation. Inclusion and exclu-
sion criteria are determined early in the clinical development process, during the
screening period. Before entry into the study, baseline determinations are made
to which all subsequent changes will be compared. The heart of a trial is the
treatment phase, which consists of drug safety modules and auxiliary modules
(Fig. 8), many of which will repeat measurements made at the time of the initial
screening or baseline. For Phase II and Phase III trials, specific parameters of
efficacy will be assessed. The posttreatment period is the stage at which final
measurements are made for safety; it is also the time to assess the effect of with-
drawal of drug relative to elimination of the disease state or a return toward the
baseline state.
Complex problems reach the decision filter at every stage of drug develop-
ment, even in the early clinical pharmacology phase (Fig. 9). To delineate the
pharmacokinetics of the compound in humans, it is common during Phase I to
situation above in the Phase I trial describes a tribulation that can occur at any
point in clinical drug development. However, many issues specific to Phases II
and III also must be anticipated. As seen in Fig. 10, different types of efficacy
studies may be undertaken (see Chapter 6). Special studies, such as tests for
addictive potential or studies allowing compassionate use of the drug (see Chapter
9), occur during these phases.
As already shown in Fig. 8, information regarding safety is collected in
every study. An attempt is always made to determine any adverse events that
may be caused by the drug. The process is especially difficult in patients, because
illness itself is defined by a grouping of adverse events. The critical question
when any adverse event occurs during a clinical trial is, ‘‘Why did it occur?’’
Did the event occur spontaneously, or as a result of an underlying disease, or as
a result of a procedure conducted? Or was it caused by the drug?
What data are needed to answer those questions? Figure 11 depicts points
along the course of a clinical trial at which data must be gathered to make an
assessment. Figure 12 lists some of the numerous information points required
before an accurate judgment can be made.
If Figures 11 and 12 seem unnecessarily complex and unduly detailed rela-
tive to the assessment of causality for an adverse event, consider the study of an
antidepressant. A probe was made at baseline just before initiation of treatment
(see Fig. 11) to determine the clinical status of the depressed individuals who
were about to enter into the study. As seen in Table 4, an impressive background
of complaints existed before any drug medication. In a 6-week study, multiple
probes will be performed to detect adverse events. Consider, then, if headache
is reported as an episode during treatment (Fig. 11), it will be extremely difficult
to assign causality relative to baseline when more than half the patients reported
headache at baseline.
The symptoms listed in Table 4 afflict all of us from time to time, but
assessments of causality for more serious events should not require the detailed
data reporting depicted in Fig. 12, right? Wrong! Table 5 lists serious adverse
events not present at baseline but occurring during placebo treatment. If these
events had taken place during active therapy, it would have been very difficult
to avoid assigning causality to the drug (see Chapter 14).
Any type of adverse event must then flow through the decision filter. The
tribulations associated with assessing causality can be multiplied if case report
forms (CRFs) are improperly designed. Poorly designed CRFs during Phase II
will compound and multiply the problems encountered in Phase III. Proper design
of CRFs at the start of clinical trials will create a firm foundation for passing
through the multiple decision filters on the way to new drug approval.
A type of tribulation that occurs more in Phase II, and particularly in Phase
III trials, involves adherence to the drug regimen. Drug adherence is loosely
described as the number of dosages actually taken by a patient compared with
the number prescribed. Alas, as with most things in life, further reflection reveals
a far more complex subject. Were dosage administrations properly spaced, were
they taken with meals (if required) or without food (if required), were they taken
with forbidden concomitant medications? With larger Phase III outpatient studies,
the variability of adherence is exaggerated. Adherence is further hindered by
prolonged or complex prescriptions. It can destroy the statistical validity of an
otherwise carefully controlled trial.
Consider the extreme example of a 71-year-old patient who was admitted
to the intensive care unit after being found unconscious at home. Because of his
Insomnia 92
Tiredness/fatigue 74
Anorexia 59
Headache 54
Although the search for new chemical entities or natural product extracts to treat
diseases is likely to continue for years to come, changes are underway that will
have a profound impact on how we diagnose and treat disease. The identification
of various active entities such as cytokines and delineation of their functions has
already led to a new class of molecular therapies. This process is going to take
a quantum leap forward now that the entire genetic code of a human being is
accessible on the Internet.
One new area generating interest and huge investment is called pharmaco-
genomics—an attempt to identify therapy targeted to an individual’s specific
genetic composition. The desired result would enhance efficacy, or minimize
toxicity, or both. A consequence of deciphering the human genetic code is the
increasing number of blood tests that can reveal disease-gene mutations and pre-
Guiding a new chemical entity through the tribulations involved in the drug test-
ing and approval process is a task that is exciting and rewarding as well as long
and complex. Advances in gene transfer, electronic data capture, and real-time
data analysis all promise to increase our chances of success. Ultimately, however,
it is through dedication, skill, and lots of luck that the drug development process
is successful and we can provide a new medication to the people who need it to
fight the pain and suffering of illness. At those times, the camel truly has made
its way home through the eye of a needle.
REFERENCES
David S. Duch
Cato Research Ltd., Durham, North Carolina
Robert M. Ferris
Retired, Charlotte, North Carolina
I. INTRODUCTION
The major change that has occurred in the drug development process over the
last 15–20 years has been the introduction of significant advances in new technol-
ogies that expedite the design, screening, and identification of new chemical enti-
ties. A brief review of these technologies, together with a review of the various
approaches used in the discovery of new chemical entities, is presented to update
the readers and encourage their deeper involvement in areas pertinent to their
interests. The use of these technologies has forced the drug discovery process to
evolve into a rapid, integrated, and usually very targeted process. In addition,
changes in regulatory requirements, as well as the introduction of International
Conference on Harmonization (ICH) Guidelines, have led to new concepts for
the timely and cost-efficient development of drugs for registration in world mar-
kets. The authors’ experiences in interpreting these guidelines and applying them
to the drug development process are presented for the readers’ consideration.
of drugs have come from plants and their analogs, microbial broths containing
various metabolites of microorganisms, animal cells and their extracts, animal
and marine toxins, and, more recently, genetic engineering (Fig. 1).
Recent advances in technologies such as combinatorial chemistry and com-
puter-assisted design have markedly increased the ability of the chemist to supply
new chemical entities for study. However, diversity of molecules is not easily
obtainable, even with the advent of these new advances in technology. In fact,
high-throughput screening techniques have made the process of screening mole-
cules so rapid that the diversity of structures available for screening has dwindled
to the point that the search for natural products with their inherent diversity is
now taking on additional importance.
2. Targeted Screening
Many pharmaceutical companies test compounds in assays specifically selected
to reveal the therapeutic activity of interest. Such a process is usually referred
to as targeted screening. For example, large-scale, cell-based assays have been
developed to screen for potential antitumor agents (1). As an outgrowth of earlier
antitumor screening methods, the National Cancer Institute has developed a
disease-oriented approach to drug discovery that uses a total of 60 human tumor
cell lines derived from eight cancer types (lung, colon, breast, melanoma, kidney,
ovary, brain, and leukemia). The initial design proposed that leads demonstrating
disease specificity would be selected for further testing based on disease type
specificity in the assay, unique structure, potency, and demonstration of a unique
pattern of cellular cytotoxicity or cytostasis, since this pattern could indicate a
unique mechanism of action or intracellular target. Computerized programs have
been developed to prioritize and enhance the diversity of compounds entered into
the screen as well as to analyze the data obtained in relation to the other com-
pounds in the database. Results obtained using this cell-based assay have recently
been reviewed (2).
An alternative approach to the evaluation and selection of antitumor agents
based on targeted screening has been developed by Von Hoff (3). This assay
determines the response of the individual patient’s tumor to specific antitumor
agents and has been used both to screen for new agents, as well as to determine
tumor types against which an agent will be active.
Evaluation of the activity of different molecules on a specific enzyme or
receptor, for example, on dopaminergic D2 receptors, in search of a novel anti-
psychotic agent is an additional example of targeted screening. This approach is
generally more cost-effective and more rapid than a general screening program
(4). Targeted screening can improve specificity, efficacy, and duration of action,
while minimizing side effects. However, compounds having activity against other
targets would not be selected using this limited approach. Since compounds made
for one purpose frequently demonstrate pharmacological properties in other areas
3. Molecular Modification
Identification of a lead structure rarely yields a compound that possesses all the
properties needed for full clinical development. Most of the time, the compound
has to be modified to improve potency, reduce side effects, increase bioavail-
ablilty, decrease metabolism, decrease toxicity, or alter its properties in some
other favorable way to make the compound a viable therapeutic agent. This pro-
cess is called lead structure optimization. Maxwell has divided this process into
two classes, enlightened or unenlightened opportunism (5). An example of en-
lightened opportunism is that of discovering new pharmacological properties at
an early stage and developing better agents than the original with this activity
(Fig. 2).
For example, cisplatin, an important chemotherapeutic agent, has dose-
limiting nephrotoxicity and neurotoxicity. Additional studies yielded an analog,
carboplatin, that had mylosuppression as the limiting toxicity. Further synthesis
yielded oxaliplatin and later DWA 2114R, which have sensory neuropathy and
neutropenia, respectively, as the limiting toxicities. Additional examples of en-
lightened opportunism can be found in a review article by Maxwell (5).
Enlightened opportunism is in contrast to unenlightened opportunism,
where, at a late stage in the development of a compound, one attempts to make
5. Clinical Observations
Many innovative drugs have originated from the astute observations of physicians
who recognized that what appeared to be apparent side effects of drugs were
actually novel therapeutic properties. The discovery of the diuretic and glucose-
lowering activities of the antibacterial sulfonilamides, which were developed into
three distinct classes of sulfonamide drugs (i.e., antibacterials, diuretics, and hy-
poglycemics), is often cited as an example (6). Consider as well, the antidepres-
sant effects of the tuberculostat, iproniazid; the anxiolytic properties of the neuro-
leptic, buspiron; and the antirheumatic effects of the antibacterial, penicillamine
(5,6). This source of drug discovery has slowed markedly in recent years, possibly
because of the imposition of regulatory guidelines needed to ensure the safety
of drug development (5).
An interesting case description and analysis of research that led to innova-
tive therapeutic agents since World War II can be found in the book, Drug Dis-
covery, A Casebook and Analysis, by Maxwell and Eckhardt (8).
3. Bioinformatics
The immense growth in biotechnology, together with the marked expansion of
information technology, has lead to the formation of a new concept called bioin-
formatics (21). The impetus for this birth probably came from the development
4. Chirality
Today, preclinical development of new chemical entities must take into account,
at very early stages, the issue of whether the molecule possesses any chiral cen-
ters. If so, the enantiomers should be resolved, and their efficacy, toxicity, and
safety should be assessed. The results of these studies will help determine whether
to develop an individual enantiomer or the racemate. The decision to develop a
racemate or an enantiomer should be made only after a thorough understanding
of the pharmacological, toxicological, and pharmacokinetic properties of the sub-
stance. Although not all-inclusive, the following examples illustrate typical situa-
tions in which a racemate might be developed (23):
The enantiomers have been shown to have pharmacological and toxicologi-
cal profiles similar to the racemate.
The enantiomers are rapidly interconverted in vitro and/or in vivo so that
administration of a single enantiomer offers no advantage.
One enantiomer of the racemate is shown to be pharmacologically inactive,
and the racemate is demonstrated to be safe and effective.
Synthesis or isolation of the preferred enantiomer is not practical.
Individual enantiomers exhibit different pharmacological profiles, and the
racemate produces a superior therapeutic effect relative to either enan-
tiomer alone.
The decision to market one enantiomer or the racemate should be made on a
case-by-case basis after considering all available data.
5. X-Ray Crystallography
The goal of rational drug discovery and development is the enhancement of the
activity of a ligand to obtain a clinically useful agent. The development of thera-
7. Monoclonal Antibodies
The pioneering work of Milstein (31) initiated the development of monoclonal
antibodies as a therapeutic entity. Normal immune responses of B lymphocytes
are polyclonal in nature and yield a heterogeneous mixture of antibodies. In order
to produce monoclonal antibodies, murine antibody-producing cells were immor-
talized by fusion with plasma cell tumors incapable of producing immunoglobulin
but capable of supporting antibody synthesis and secretion. Subsequent cloning
of the hybridomas yielded clones that are capable of producing large amounts
of homogeneous murine antibody that react with a single epitope.
However, the use of murine antibodies in humans is restricted because of
the immunogenic nature of the immunoglobulins. Repeated use of murine anti-
bodies elicits an anti-immunoglobulin response known as the human antimouse
antibody (HAMA) response. In addition, other limitations to the use of antibodies
have been observed and have been overcome in part through the use of genetic
engineering to humanize the murine antibodies (32). The redesign of murine anti-
bodies using these techniques has resulted in the formation of antibodies that
elicit a considerably reduced immune response relative to the murine antibody.
Although monoclonal antibodies have been most widely used for the diag-
nosis and treatment of cancer, they are also being evaluated as therapies in other
areas as well. For example, abciximab (c7E3 Fab) is a chimeric human-murine
monoclonal antibody Fab fragment that binds to the platelet glycoprotein IIb/
IIIa receptor and inhibits platelet aggregation. The addition of abciximab to stan-
dard aspirin and heparin therapy reduced the incidence of ischemic complications
during the initial postoperative period in high-risk patients who were undergoing
percutaneous coronary angioplasty or directional atherectomy. Abciximab also
reduced the incidence of clinical restenosis when compared with placebo during
a 6-month follow-up of these patients (33).
Monoclonal antibodies have been widely used for the diagnosis, localiza-
tion, and treatment of cancer (34). Their advantages include a relative selectivity
for tumor tissue coupled with a relative lack of toxicity. However, their ability
to affect tumors is minimal unless aided by other mechanisms, the amount of
antibody delivered to tumors is low, and the diffusion of antibody through tumors
is often poor. The development of a HAMA response is also limiting. To aid in
the antitumor effectiveness of monoclonal antibodies, antitumor drugs such as
8. Gene Therapy
Human gene transfer is a therapeutic approach in which the genome of human
somatic cells, but not germline cells, is modified for the purpose of treating dis-
ease. It has also been used to generate a population of marked cells for the purpose
of tracing the origins of recurrent tumors. A large number of gene therapy proto-
cols have been designed to treat cancer, HIV infection, and diseases caused by
a gene defect, such as cystic fibrosis, familial hypercholesterolemia, and severe
combined immunodeficiency caused by adenosine deaminase deficiency (36,37).
The transfer of genes to the target cells has been accomplished through the
use of viral vectors or nonviral delivery. Nonviral delivery systems include in-
vivo delivery using plasmid–liposome complexes, direct injection of naked DNA,
and transfection of target cells ex vivo. The viral vectors that have been the most
widely studied include retroviruses, adenoviruses and adeno-associated viruses;
however, the use of herpes virus for delivery to the central nervous system (CNS)
and vaccinia virus vectors have also been investigated (38,39).
Each of these means of transfer has advantages and disadvantages. Gene
delivery using either liposomes or naked DNA has no involvement of viruses
and thus could not replicate or recombine to form infectious agents, Moreover,
use of these methods of delivery would cause fewer inflammatory or immune
responses and can be used to transfer genetic material of unlimited size. These
advantages are offset by the inefficiency of transfer and the temporary expression
of the transferred gene. The viral vectors have a greater efficiency of gene trans-
fer. Retroviral vectors require dividing cells for integration of the genetic material
whereas adenoviruses and adeno-associated viruses do not require cellular prolif-
eration. Retroviruses and adeno-associated viruses, but not adenoviruses, result
in a stable incorporation of the genetic material into the genome of the target
cell. With adenoviruses or liposomes as vectors, the transferred genetic material
remains epichromosomal, and consequently, there is only short-term expression
of the transferred gene. Therefore, to maintain persistent expression, frequent
administration would be required. Since adenovirus is immunogenic, repeated
use of this vector could be limited by an immune response. A potential advantage
9. Genomics
One of the goals of the Human Genome Project is to determine the sequences
of the 100,000 genes in the human genome. In general, the approach has been
to sequence sections of cDNA from libraries created from RNA of various tissues
to obtain ‘‘expressed sequence tags.’’ Since the source cDNA libraries from vari-
ous human tissues are normalized to remove housekeeping genes, the frequency
that certain cDNAs appear in the various tissue-specific databases will be a re-
flection of their expression levels. Using the present techniques, the sequencing
rate is relatively slow. However, new sequencing techniques are being developed
that will be more rapid, efficient, accurate, and cost-effective (40).
High-throughput methods using DNA arrays for the monitoring of gene
expression patterns in a highly parallel fashion have been developed (41,42).
DNA arrays of genomic fragments, cDNA clones, or oligonucleotides are also
allowing genome-wide genetic mapping, the cloning of members of gene families
within and across species, the scanning for mutations in genes, and the definitions
of networks of genes controlled by particular transcription factors (43).
Genomics will provide a large number of potential targets. However, one
of the challenges inherent in the use of these targets for drug development is the
identification of targets that cause a specific disease rather than those that just
correlate with the disease. It has been reported (44) that the most important dis-
eases for which treatment is needed number between 100 and 150. Many of these
diseases are caused, at least in part, by genetic factors. On the basis of multigene
involvement in many genetic diseases as well as other assumptions, it was con-
cluded that there was a potential for 3000 to 10,000 new and interesting drug
targets. Since targets for which therapies presently exist number only slightly
over 400, the potential targets for which there are no drugs far outnumbers those
for which there are (44).
Once a compound has met the criteria to justify calling it the ‘‘lead molecule’’
of choice, it undergoes more advanced studies aimed at determining its potency
A. Pharmacology
At the present time, regulatory agencies do not define precise requirements for
submission of pharmacological data for an IND, New Drug Application (NDA),
CTX, or Marketing Authorization Application (MAA). The following recommen-
dations have been suggested as appropriate procedures to be followed for submis-
sions to the United States, United Kingdom, and European authorities. In the
1. Primary Pharmacology
The primary pharmacology is concerned with defining the pharmacological ac-
tions relevant to the proposed therapeutic use. The term pharmacological actions
encompasses all potential therapeutic actions as well as conventional pharmaco-
logical activities.
Listed below are the actual guidelines that deal with the requirements nec-
essary for adequate presentation of the primary pharmacology of a new chemical
entity (NCE).
Where possible, it is desirable to present data that establish the mechanism
of the principal pharmacological action.
Frequently, it is not possible to fully explain the mechanism of action of
a particular compound. This is particularly true for CNS-active compounds. At
other times, it may be simple to explain the mechanism of action of a compound
(i.e., antidepressant activity achieved through inhibition of monoamine oxidase,
or antihypertensive activity resulting from beta-receptor blockade). It should be
kept in mind that this guideline is not meant to delay the filing of an IND while
the researcher attempts to elucidate the mechanism of action of a compound. It
is simply a request to do what is reasonable without undue delays in filing.
The validity of models used should be established where practicable.
Validity means that accepted models, either so established in the literature
or through appropriate defensible studies conducted in one’s own laboratories,
should be used. Nevertheless, there are always inherent dangers in being totally
confident that any animal model will be predictive of a drug’s efficacy in humans.
For example, the reversal of tetrabenazine-induced sedation and ptosis in mice
has been routinely used in some laboratories as a reliable and predictive test for
antidepressant activity in humans. However, it was soon discovered that, while
These data will ultimately have to be correlated with pharmacokinetic and toxico-
logical data. In fact, it is advisable to do joint pharmacodynamic and pharmaco-
kinetic studies when feasible.
3. Drug Interactions
Interaction of the drug substance with other compounds, when relevant to the
proposed therapeutic usage, should be investigated. These interactions are primar-
ily concerned with drug–drug interactions, but could also include interaction with
excipients. Recently, the Food and Drug Administration (FDA) has frequently
required metabolism studies (P450 studies) to be conducted as part of the submis-
sion package. For additional information, see Sec. III. C.
B. Toxicology Studies
1. In Vivo Versus In Vitro Studies
Toxicological evaluation of a drug candidate has traditionally been carried out
in-vivo, usually in mice, rats, dogs, or monkeys. However, there has been an
increased effort to develop and use in vitro models for toxicological, as well as
pharmacokinetic, evaluation during the development process. The use of in vitro
systems has been stimulated by the rapid scientific advances that have been made
IV. CONCLUSION
V. REFERENCES
1. Grever MR, Schepartz SA, Chabner BA. The National Cancer Institute: cancer drug
discovery and development program. Semin Oncol 1992; 19(6):622–638.
2. Weinstein JN, Myers TG, O’Connor PM, et al. An information-intensive approach
to the molecular pharmacology of cancer. Science 1997; 275:343-347.
3. Von Hoff DD. Human tumor cloning assays: applications in clinical oncology and
new antineoplastic agent development. Cancer Metastasis Rev 1988; 7:357–371.
4. Algate DR. Application of the pharmacological screening process. Drug Metab Rev
1990; 22(6–8):809–820.
5. Maxwell RA. The state of the art of the science of drug discovery—an opinion.
Drug Dev Res 1984; 4:375–389.
6. Kubinyi H. Strategies and recent technologies in drug discovery. Pharmazie 1995;
50:647–662.
David M. Cocchetto
GlaxoSmithKline, Research Triangle Park, North Carolina
Proof of safety
Substantial evidence of efficacy
Informative labeling for the product
Demonstration of manufacturing of the product to the desired strength,
quality, purity, and identity
The FD&C Act requires that all drugs distributed in interstate commerce
in the United States have proof of safety and substantial evidence of efficacy;
an investigational drug (which by definition lacks such demonstration of safety
and efficacy) may be administered to patients in the United States in order to
gather evidence of safety and efficacy after such a drug obtains an exemption
(i.e., an IND) for conduct of specific clinical studies for specific indications by
specific clinical investigators in the presence of ongoing sponsor commitments
to monitor these studies and provide certain information to the FDA.
The FD&C Act itself defines the requirement for an IND, but it does not
provide more detailed information on the procedures and operational approaches
to be used in drug development in order to satisfy these legal requirements. In
order to provide operational details, subsequent to passage of the FD&C Act, the
FDA developed the implementing regulations governing drug development. Most
of these regulations are found in Title 21 of the Code of Federal Regulations.
Part number
in Title 21 Regulation
50 Informed Consent
56 Institutional Review Boards
201 Prescription Drug Labeling
202 Prescription Drug Advertising
211 Current GMPs
312 Investigational New Drug Application
314 New Drug Applications
320 Bioavailability and Bioequivalence Requirements
Beyond these regulations, which are enforceable as law, the FDA provides
guidance documents on the drug development process. These guidance docu-
ments take the form of guidelines, ‘‘Points to Consider’’ documents, and ‘‘Infor-
mation Sheets.’’ Such guidances are informal communications from the FDA in
that they reflect the FDA’s best judgment at the time of their preparation, but
neither the FDA nor sponsors are legally obligated to adhere to the provisions
of guidances.
The FDA uses an introductory part of its IND regulations (21 CFR 312.22) to
present several important principles of the IND submission. These four principles
should be carefully heeded since they provide insight into the key concepts used
by the FDA in reviewing INDs and they can help the sponsor prepare IND docu-
ments that meet the FDA’s expectations.
The first important principle is that the IND must present adequate informa-
tion to permit the FDA to evaluate the drug’s suitability for use in the proposed
clinical study. An application that provides insufficient information to enable a
specific and detailed review of all elements relevant to the proposed clinical study
is unlikely to be allowed to proceed. Similarly, as clinical studies progress over
the years under an IND, subsequent submissions to the IND must continue to
present an adequate set of information to enable appropriate evaluation by the
FDA of proposed clinical studies. For a clinical study in an IND to be acceptable
to the FDA, it must meet the safety standard (i.e., human subjects will not be
exposed to an unreasonable and significant risk); further, the FDA must be con-
Figure 1 Six technical disciplines that comprise all drug regulatory applications in the
United States.
tion needed for an initial IND. Given the youth of this new guidance, its imple-
mentation bears close observation.
Acute and repeat-dose toxicity studies provide a screen for initial detection
and characterization of the spectrum of the drug’s toxicity on all major
organ systems. Note that the classical single-dose LD50 study (i.e., deter-
Route of
Type of study Species or test administration
Each toxicology study should be done with drug administration via the intended
clinical route. Each study should elicit sufficient toxicity findings to enable char-
acterization of the toxicity profile of the drug, within the constraints of the studies
completed to date. For drugs with limited bioavailability by the intended clinical
route of administration, adequate characterization of the toxicity profile may be
facilitated by conduct of some repeat-dose toxicology studies using parenteral
administration of the drug. Typically, the duration of human use in Phases I and
II may be as long as the duration of animal studies. However, in Phase III, longer-
term animal studies must be complete, since the duration of human use cannot (in
general) exceed one-half the duration of treatment in animal toxicology studies.
The FDA must have time to review an initial IND before the investigational drug
is shipped and the first clinical investigation is initiated. This review time is
essential to protect the public health because the FDA must assure that, based
on the evidence presented in the initial IND, it is reasonable to proceed with the
first proposed clinical investigation. The more specific nature of the FDA’s re-
view is described in a separate section below. The 30-day review clock begins
Figure 3 Members of the FDA’s review team for a drug regulatory application.
After submission of an initial IND and its acceptance by the FDA, the sponsor’s
effort to develop the drug continues, leading to collection of new information
and the need to submit this information to the FDA. The FDA must be kept
informed of the evolving, increasing scientific knowledge of the drug. The objec-
tives of subsequent submissions to the IND are to maintain the IND in ongoing
good regulatory standing, as well as to assure that new information is submitted
to the FDA before action is initiated by the company. For example, implementa-
tion of a new facility for manufacturing drug product should be adequately docu-
mented and submitted to the FDA before initiation of use of resulting drug prod-
uct in human studies governed by the IND.
The types of submissions to an established IND are as follows:
Reference in Title 21
These submissions are required by the IND regulations in order to assure that
the sponsor provides the FDA, on an ongoing basis with diligence, with new
information as it becomes available on the investigational drug. The FDA’s crite-
ria for review of these subsequent submissions are essentially the same criteria
as applied to the initial IND, i.e.:
Submissions continue to build logically on the base of scientific information
about the drug.
Submissions support the conclusion that humans will not be exposed to
unreasonable or significant risk.
The evolving clinical plan or new protocol must be adequate in design to
meet its stated objectives.
IX. SUMMARY
The IND process for novel investigational drugs gives the sponsor the opportunity
to prepare a logical and integrated collection of scientific information to support
REFERENCES
1. Final rule. New drug, antibiotic, and biologic drug product regulations (‘‘IND re-
write’’). Fed Reg 1987; 52(53):8798–8857.
2. Food and Drug Administration. Guidance for industry: content and format of investi-
gational new drug applications (INDs) for Phase 1 studies of drugs, including well-
characterized, therapeutic, biotechnology-derived products. Washington, DC: Food
and Drug Administration, U.S. Department of Health, Education, and Welfare, No-
vember 1995.
3. Goldenthal EI. Current views on safety evaluation of drugs. FDA Papers 1968
(May):13–19.
4. General Considerations for the Clinical Evaluation of Drugs. Washington, DC: Food
and Drug Administration, U.S. Department of Health, Education and Welfare, Sep-
tember 1977.
5. General Considerations for the Clinical Evaluation of Drugs in Infants and Children.
Washington, DC: Food and Drug Administration, U.S. Department of Health, Edu-
cation and Welfare, September 1977.
6. Traina VM. The role of toxicology in drug research and development. Med Res Rev
1983; 3:43–72.
7. Hoyle PC, Cooper EC. Nonclinical toxicity studies of antiviral drugs indicated for
the treatment of non-life-threatening diseases: evaluation of drug toxicity prior to
Phase I clinical studies. Regulatory Toxicol Pharmacol 1990; 11:81–89.
8. Scales MDC, Mahoney K. Animal toxicology studies on new medicines and their
relationship to clinical exposure: a review of international recommendations. Ad-
verse Drug React Toxicol Rev 1991; 10:155–168.
9. Cartwright AC, Matthews BR, eds. International Pharmaceutical Product Registra-
tion: Aspects of Quality, Safety and Efficacy. London: Ellis Horwood, 1994:411–
552.
10. LD50 test policy: notice. Fed Reg 1988; 53:39650–39651.
W. Leigh Thompson*
Eli Lilly and Company, Indianapolis, Indiana
Rocco L. Brunelle
Eli Lilly and Company, Indianapolis, Indiana
Michael G. Wilson
Michael G. Wilson and Company, Inc., New Palestine, Indiana
Symptoms, signs, and laboratory tests are the inputs to health decisions about
prognosis, diagnosis, and treatment. The quality with which they are elicited and
interpreted limits the quality of health decisions.
Symptoms are God’s reminders to visit your physician or modify your be-
havior. Symptoms are elicited from patients or their companions, and a skilled
practitioner knows not only what questions to ask, but in what order, with what
flavoring, and with careful observation of the nonverbal and verbal responses.
Signs are detected by the health practitioner using her or his five primary
senses: seeing a rash, hearing a murmur, feeling the pulse, smelling the fetor
hepaticus of liver failure, and tasting the sweet urine of a patient with diabetes
mellitus. The best practitioners have a sixth sense, poorly defined, that the patient
is sick or dying, that the patient is recovering, or that the patient is faking. When
a nurse with this sixth sense tells a physician that the patient is going sour, a
bright physician hastens to patientside.
God provided us with a few sensors for a narrow bandwidth of energy
spectra more suitable for living than diagnosing. Our elegant eye–brain visual
system, which can detect the subtle yellowing of jaundice under the tongue, has
been supplemented with shorter-wavelength detectors of Röntgen rays, magnetic
resonance imaging (MRI) detectors of electron spin in radio frequency fields,
* Retired
* Subsequent to our initial publications, approved guidelines for the definition and determination of
reference ranges were published by the National Committee for Clinical Laboratory Standards in
1995 (11). That work of the EPTRV proved to be a most useful basis for the development of the
NCCLS guidelines. Basically, the a-posteriori method as described by the NCCLS protocol guide-
line was the methodology employed to construct our ranges.
ethanol and 460 U/L in similar patients who are non-Caucasian and admit to
imbibing ethanol. Similar creatinine kinase activities upper tails vary from 218
U/L in young Caucasian women who smoke but do not drink to 2770 U/L in
young non-Caucasian men who smoke and drink.
Consider the distribution of blood hemoglobin concentrations. This illustra-
tion shows a Gaussian probability distribution on the abscissa. On this scale, a
tors in defining notable versus expected analyte values. The first and 99th percen-
tile values for the 32 demographic groups are reprinted in Appendixes B to E.
Almost all of these tests were performed in adults. The difficulty of establishing
pediatric reference ranges has been emphasized (13,14).
Are 1% tails suitable? Remember, you are trying to identify notable results.
In this reference population we can predict what proportion of patients will have
notable results using the 1% and 99% tails. If the analyte results are not correlated,
we would identify 1 ⫺ (.98)36 ⫽ 52% of patients as having one or more notable
values. If we repeated the observations at five intervals, and the results were not
correlated, we would identify 97% of patients as having one or more notable
values.
Are the first and 99th percentiles appropriate for panic values that lead to
immediate notification of the investigator and sponsor? Probably not. Kost (15)
surveyed U.S. medical centers to establish a consensus of critical values for ur-
gent clinician notification.
Reference ranges describe one measurement in a defined population. But
what of intraindividual variation between repeated measurements? To estimate
this, we sampled 3556 patients from the population above who had duplicate
samples drawn at intervals of 1–2 weeks (1). The largest 1% of decrements and
increments between paired samples were called delta limits. These were not found
to be related to the patient characteristics. The delta limits (2) are reprinted in
the appendixes. You may find them a useful standard against which to compare
intraindividual variation in your clinical trial patients.
At the start of a trial, one might exclude patients with underlying liver
REFERENCES
1. Thompson WL, Brunelle RL, Enas GG, Simpson PJ. Routine laboratory tests
in clinical trials: interpretation of results. J Clin Res Drug Devel 1987; 1:95–119.
I. INTRODUCTION
The goal of any clinical trial is to yield valuable and reliable scientific information
that can ultimately guide medical practice. To that end, appropriate attention to
study design, including the selection of reasonable and relevant trial endpoints,
is essential for success. Trial endpoints are events or measurements that reflect
the effects of an intervention on the study participants. Most trials are designed
to collect data on many different endpoints, the nature and number of which will
depend on the condition or disease of the individuals who are in the study, the
anticipated pharmacological and toxic effects of the intervention, and the objec-
tive of the clinical investigation. In the initial studies, where safety and dosing
information is of prime interest, study endpoints usually are measurements of
pharmacokinetics and safety. In later clinical development, there is growing em-
phasis on evaluation of activity and, ultimately, efficacy. Efficacy is generally
evaluated by using adequate and well-controlled clinical trials designed to show
that the intervention meaningfully affects the trial’s efficacy endpoints.
Efficacy endpoints either directly characterize the clinical benefits of the
intervention or are surrogates believed to predict clinical benefit. The choice of
efficacy endpoints involves selection of an outcome measure, determination of
when and how to measure it, and determination of how to analyze the results.
These choices have profound effects on the success of a trial.
Missing data may substantially reduce the ability to draw valid inferences from
a trial; appropriate selection of an endpoint can reduce the amount of missing
data and its effect on the study.
An endpoint should be sensitive to the desired effects of the interventions
under study. Selection of an endpoint sensitive to the hypothesized differences
in treatment effects between the experimental and control arms will increase the
likelihood of demonstrating a difference (i.e., power) or decrease the size of study
required to demonstrate a difference.
An endpoint should be clinically meaningful. In general, an endpoint
should be a direct measure of an important benefit to patients, such as longer
survival, less pain, or improved physical functioning. Endpoints based on scoring
systems should be validated to ensure they reflect true clinical benefit. Endpoints
that do not directly measure benefit (e.g., laboratory tests, x-rays) are of value
as primary efficacy measures only when they correlate with and predict clinical
benefit.
It is desirable that an endpoint be one whose meaning can be readily com-
municated to those who will receive the results of the trial. Communication of
the results of a trial and the nature and magnitude of a drug effect, whether in
a manuscript, a presentation, or a package label, is easier if endpoints have a
readily understood clinical meaning (e.g., percentage survival at 1 year), than if
alternative endpoints are used (e.g., an unfamiliar scoring system). Selection of
endpoints that are readily communicated potentially can enhance the impact of
a trial on medical practice.
It is desirable that an endpoint be objective, particularly in trials in which
subjects or investigators may be unblinded or in which the study treatment has
unblinding effects. Many factors may introduce bias into the assessment of an
endpoint that is subjectively assessed by patients or assessors; objective endpoints
reduce bias.
IV. CUTPOINTS
In clinical trials, measurements are sometimes turned into binary, eventlike end-
points through the establishment of one or more cutpoints. For example, rather
than assessing the change in measurements of CD4⫹ cell count, weight, or visual
acuity, one can assess the prevalence of a drop in CD4⫹ cell count below 200
cells/µL, a loss of at least 5 kg body weight, or improvement of vision by two
lines on an eye chart. Such approaches generally sacrifice sensitivity because
changes that do not involve crossing the cutpoint are ignored in the analysis.
Despite such loss of sensitivity, cutpoints are often used in an attempt to
ensure that only clinically meaningful changes affect the endpoint. In determining
what size of change to consider meaningful when creating a cutpoint, two sepa-
rate considerations are relevant. First, because of the variability inherent in all
measurements, one may wish to choose a size of change that has a high probabil-
ity of reflecting real change rather than measurement variation. Second, one may
wish to choose a size of change that indicates clear and meaningful clinical ben-
efit.
With regard to the first consideration, that is, distinguishing real changes
from measurement variation, it is important to note that a sufficiently large clini-
cal trial may be able to detect real differences between treatment groups of a
magnitude within the range of error or variation for individual measurements.
For example, suppose that in a clinical trial of an antiretroviral agent for HIV
infection, the plasma viral load, on average, remained unchanged in the control
group and showed a twofold decrease in the treatment group. If a twofold change
For diseases manifested by recurring events, such as asthma with periodic attacks
or multiple sclerosis with periodic relapses, and for treatments intended to reduce
VII. MULTIPLICITY
Use of a single primary efficacy endpoint is generally desirable in Phase III trials.
With increasing numbers of endpoints comes increasing concern regarding Type
1 error, that is, erroneously declaring that an endpoint was affected by the inter-
vention. Type 1 error is particularly of concern in Phase III efficacy trial(s) in
which the outcomes are intended to form the basis for establishing the product’s
efficacy and safety. A study with one primary endpoint is more likely to generate
a straightforward analysis and therefore minimize ambiguity in interpreting the
results. This type of study also often allows the trial to be more focused, simplifies
the trial design, and improves the power of the study.
In cases in which it is not known which efficacy outcome is most likely
to reflect the drug effect or which efficacy effect is most clinically relevant for
a new therapy in a specific disease setting, it is often useful to evaluate multiple
efficacy endpoints in a Phase II trial. Although such a study may not give defini-
tive results about the product’s effect on each endpoint, the results may be used
to identify the endpoint(s) most appropriate for Phase III testing.
If, in planning a Phase III trial, more than one clinical outcome is thought
to be critical to the primary evaluation, several approaches are available to limit
Type 1 error. Some approaches retain multiple primary endpoints with correct
application of statistical tests that adjust for multiplicity and preserve overall
Type 1 error rates (4–6). As a result of these adjustments, the effect on any one
of the endpoints must be more definitive (i.e., larger in magnitude and associated
with a lower nominal p value) to ensure statistical significance. Thus, when com-
pared with selection of a single endpoint, these approaches to correcting for multi-
plicity generally reduce the power to observe a drug effect on each of the indi-
vidual endpoints. Alternative approaches involve assessing multiple outcome
X. SURROGATE ENDPOINTS
XI. SUMMARY
The choice and the analysis of endpoints is a critical issue in design of a clinical
trial. Desirable characteristics of an endpoint include feasibility of measurement,
sensitivity to drug effects, meaningfulness, objectivity of measurement, and use-
fulness in communicating results. These characteristics will depend, in part, on
the type of endpoint chosen (e.g., events, measures, composites, scores, surro-
gates). Careful consideration of the characteristics of potential endpoints will
generally facilitate the development and selection of an appropriate endpoint.
REFERENCES
1. Friedman LM, Furberg CD, DeMets DL. Fundamentals of Clinical Trials. 3d ed. St
Louis: Mosby, 1996:223–245.
2. Jacobs LD, Cookfair DL, Rudick RA, et al. Intramuscular interferon beta-1a for
disease progression in relapsing multiple sclerosis. MSCRG. Ann Neurol 1996; 39:
285–294.
3. Collett D. Modelling Survival Data in Medical Research. London: Chapman & Hall,
1994:15–51.
4. Westfall PH, Tobias RD, Rom D, et al. Multiple comparisons and multiple tests.
SAS Institute, 1999:1–11.
5. Proschan MA, Waclawiw MA. Practical guidelines for multiplicity adjustment in
clinical trials. Control Clin Trials 2000; 21:527–539.
6. Gong J, Pinhero JC, DeMets DL. Estimating significance level and power compari-
sons for testing multiple endpoints in clinical trials. Control Clin Trials 2000; 21:
313–329.
W. Leigh Thompson*
Eli Lilly and Company, Indianapolis, Indiana
I. INTRODUCTION
The product decision to invest enormous sums in pivotal studies of safety and
efficacy of a new treatment should be made when the probability of success has
grown to 90% from the 10% on choosing a new molecule. This point can be
reached in less than 1000 days after 100 patients for $10 million. For optimal
speed, minimal waste resources, enhanced ethics, and maximal information from
each subject, the chemistry, toxicology, and clinical studies should not be done
en bloc in series, but rather elegantly braided together in a stronger, shorter con-
tinuum of data capture, analysis, and utilization to plan the next dose. A clear
focus on the goal—a better-quality investment decision—will aim this braid
straight to that target with no unraveling of unnecessary studies, with conserva-
tion of living subjects, and with maximum decision-critical information per min-
ute. Only three things are needed to achieve this new paradigm of efficiency:
clear thinking, an elegant informatics system for all data and users, and a plan
based on the relevance diagram of the product decision.
II. GOAL
* Retired
Note: See Appendices on pp. 439–449 for Routine Clinical Analyte Test Results.
Contrast the plodding en block linear ooze described above with smoothly inte-
grated, seamless streams of data from chemistry, toxicology, and clinical studies
deftly interwoven into a comprehensive knowledge braid that reaches a high prob-
ability of future success with the least time, money, effort, and expertise. To
accomplish this elegant braiding requires only a global focus on two critical ele-
ments—the ethics of human exposure to innovative molecules and a single pur-
pose of predicting success and failure with the least effort. Nothing should be
done that is not entirely ethical, but also, nothing should be done that is not
essential to reach a product decision at which you have a 90% probability of
regulatory, reimbursement, and pricing approvals in the desired markets. This
is no time for hidden agendas, meanders unraveling the braid, or ‘‘feel good’’
studies.
Why should we change the successful paradigms of the past? They served
us well in an era of high profitability, low competition, leisurely research and
marketing, and risk-averse avoidance of change. We might not have liked the
lengthening times and escalating costs, but we could then afford them and the
devil of change was ominous. Today, faster, more efficient decisions are impera-
tive. A new molecule in 2001 requires $1 billion investment to pay for the failures
and for global support of the launch. The Centre for Medicines Research esti-
mates that about 40 new molecules are launched worldwide annually at a current
investment of $50 billion. If research investments are 16% of sales, the average
molecule must achieve $6 billion in sales over its lifetime. If 1 million patients
use the treatment each year for 10 years of exclusivity, each must pay $600
(wholesale) to achieve this reimbursement of the investment. Currently one must
invest for 15 years to reach launch in the United States and about 12.8 years
worldwide. About 11% of new drugs that begin development, with teams focused
on them, will be launched. Perhaps one in four that is launched will earn a true
Before Gutenberg abolished illiteracy with cheap linear codes for thoughts, peo-
ple could recite oral histories in different verse orders or study stained glass
windows and statues in any order. After Gutenberg we learned the first page
before the second, courses proceeded from front cover to back, and you didn’t
spoil the mystery by peeking at the last page.
Henry Ford developed the mindless linear assembly line. The chassis was
built first, then the axles were applied before the wheels went on. Workers from
different countries, speaking different languages, could be minimally trained to
perform a single mindless function, such as adding the left rear wheel. No com-
munication was needed, and all the cars were identical, black, with interchange-
able parts—not the perfectly fitted, hand-crafted parts of a skilled artisan, but
inexpensively standardized modules linearly assembled.
Linearity was embellished during World War II by multiple assembly lines
making an airplane or submarine with product streams, each from its own assem-
bly line, arriving in proper order. Today we have refined this to just-in-time re-
sourcing, but the linear thinking hasn’t changed—in fact, it has spawned the
mathematical/management discipline of linear programming.
A. Hyperlinking Media
Your eye–brain isn’t linear. It soaks up all the data available, processes it in
priority order, and acts upon it before you consciously think chasm, tiger, or stop
sign. Speech leaps from one thought to the next. Transcripts of my lectures are
ungrammatical nonsense as I leap among media and thoughts.
Hyperlinked multimedia may restore in one generation the sentient capaci-
ties of homo sapiens, and make her more worthy of her species appellation. How
do these new tools simulate our natural eye–brain? The MIT Computer Sciences
Laboratory ‘‘smart room’’ sucks up images from CNN, print from The Wall
Street Journal, references from NLM, audioclips from your message machine,
and your own musings. It then braids these data for you by concept, learned from
observation of how you use the materials, the order you prefer to use it, and how
long you spend with each subject. It then presents to you new material in your
priority order—leaping from image to text to sounds almost as effortlessly as
your eye–brain to create a knowledge braid optimal for your interests. As these
tools become more user-seductive and we learn to communicate through virtual
reality and real-time language translators that seamlessly assume our individual
B. Hyperlinking Research
In rethinking decision research we need to be facile in braiding chemistry, toxi-
cology, and clinical studies into a tight synergistic web of data, all of which are
quickly validated, organized, analyzed, and displayed before the next step.
When climbing a mountain you seamlessly integrate data from positional
sensors, tension sensors in muscles, vision, wind sensors, etc., before each step.
Similarly, we need to acquire all the essential data, just in time, valid, analyzed,
displayed, and integrated into our decision for our next step. And each step, even
if it seemingly veers off to a ridge, should be the optimal path to the goal—the
best possible product decision.
What chemistry tasks are essential for your next toxicology study? What
toxicology tasks are essential for your next clinical study? Must the formulation
used in the first human be that which you will market worldwide? Of course not.
Consider what must be, what is essential, what is sine qua non for each step of
toxicology and clinical studies. Make each task modular and complete the mod-
ules just in time—no module before its time means no wasted resources.
If your goal is predicting market acceptance of your product from a few human
studies, should you study healthy, young, nonobese, Caucasian, nonsmoking
males taking no drugs or ethanol, eating a prescribed diet, and having no sex or
exercise while incarcerated in a closed ward for weeks? By definition these are
the subjects least likely to be taking your drug when you market it. What will
they tell you? They will have very little variability in the kinetics of your drug.
The halftimes will be almost identical and the confidence limits about the plasma
concentration–time curves will be very narrow. The graphs in the publication
will be beautiful.
Unfortunately, these data will not predict the responses to the drug of chil-
dren, fat old French women, Native Americans athletes, patients with multiorgan
failure taking many other medicines, and patients with aberrant drug metabolizing
enzymes.
Should the first human be healthy? Why? Won’t you learn more from patients?
If the target disease makes the patient so fragile that he is at special risk for an
unexpected serious adverse reaction, begin with a healthy volunteer but move
quickly to patients. Phase I trials can include both volunteers and patients braided
together.
There are not many such diseases. Life-threatening hepatitis B severely
compromises patients. Fialuridine, a nucleoside that seemed very promising in
a National Institutes of Health (NIH) trial in hepatitis B, killed five patients de-
spite liver transplants through very delayed unexpected toxicity. This toxicity
was not observed in extensive animal and prior clinical studies in several patient
populations. Perhaps the toxicity is only manifest in humans with fatally progres-
B. First Dose
Examining the distribution of test substance among my blood cells, proteins, and
water, one can calculate what dose, if fully absorbed and distributed only in my
blood, would achieve a plasma water concentration one-half to one-third that
which produced any detectable change in any other species. Perhaps that is an
appropriate first dose. For many molecules, the first humans can be dosed for 1
day based on acute toxicology in two species. That day of dosing can begin with
C. Intravenous Dosing
Even more appropriate, in my opinion, would be to give the first doses by intrave-
nous infusion. One can start with a very very small dose and gradually increase
the infusion rate over several hours, perhaps sampling plasma concentrations as
a guide. One can mimic any pattern of absorption from other administration sites
using a programmed pump and intravenous infusions. One can also give unla-
beled material by mouth and labeled material intravenously to measure absolute
bioavailability and define gastrointestinal absorption and presystemic elimina-
tion.
All the important data from the first patient should be available in 48 h. Consider
the toxicokinetic data in several species. At what blood concentrations have the
animals begun to manifest effects? What were the concentrations in the first hu-
man? Considering the acute toxicity in animals, should you next give a dose to
the second human that will achieve one-third or two-thirds the minimal effective
plasma water concentration in animals? Give the second patient labeled material
and complete the studies in 48 hours. From the first patient you have begun to
recognize which samples are the most important to analyze immediately. As you
approach the minimally effective dose, institute efficacy and surrogate efficacy
assessments.
A. Blinding
If you wish to believe data, especially subjective data about adverse events, blind
the patient and investigators and determine why the sponsor should not also be
blinded. Placebos are inexpensive, although they may be toxic. They have saved
B. Dose Escalation
For each new patient, use all the data at hand from chemistry, toxicology, and
prior patients to choose the dosage regimen. Control each patient through rapid
analysis of plasma drug concentrations. If a patient has unexpected slow elimina-
tion, perhaps a multiple dose should be reduced. If a patient has unexpected low
plasma concentrations, quickly study the patient and determine why. If the patient
has an adverse event, does it correlate with plasma concentrations? Will sensitive
psychomotor tests predict subtle brain effects that indicate drug passage through
the blood–brain barrier?
This Bayesian approach of using all data was employed in the transition
from traditional LD 50 testing to more modern methods. In olden LD 50 testing,
large batches of animals were each given one dose and observed for death. Per-
haps seven doses would be tested at once, with 10 animals of each gender per
dose. Perhaps the survivors were 20/20, 19/20, 17/20, 13/20, 8/20, 3/20, 0/20.
One could calculate that the LD 50 was between the fourth and fifth dose and you
could estimate the most likely value and its confidence limits. But this would
use 140 animals.
Instead, dose the first animal with your best estimate of the LD 50. If that
animal lives, dose the next animal with a larger dose. If that animal dies, dose
the next with a smaller dose. Continue a march through a few animals, converging
rapidly on the domain of the LD 50. If you model this you will find there to be
great efficiencies in using all data in planning the next dose. If you were paid to
minimize the number of rodents employed in such testing, I am sure you would
use an efficient Bayesian dosing regimen.
D. Toxicology
What toxicology do you need for the first tiny labeled dose in a human? I would
ask about the kinetics, tissue distributions, behavioral signs, laboratory tests, and
autopsy results in two appropriate species given very large single doses. De-
pending on these findings, I might also ask about the effects of multiple doses
given over 2 weeks, but in many cases I would be interested only in the single
large doses.
E. Chemistry
What chemical tests are needed? I would want to have reasonably well character-
ized material used in the toxicology studies and in me to ensure that no new
toxic contaminant has appeared in my dose, but if we are doing only single-dose
or short-term studies in animals there will be little need for stability studies,
testing animal feeds containing the test drug, etc. Using labeled material early
can be a major help before sophisticated assays are perfected.
If metabolism in the first human is unexpected, and significant metabolites
are seen that have not been observed in animals, one might wish to return to
other animal species and strains to find one in which the human metabolites can
be evaluated. Perhaps this metabolite will cause dose-limiting toxicity. Perhaps
knowing the toxicity of this metabolite in animals will make one wish to abort
the project or alter plans about the tolerated dose.
F. Effects
As the plasma concentration is increased, with larger doses one may begin to
see effects in the patients. Hopefully there is an acute surrogate that will predict
efficacy. Relate these effects to plasma concentration profiles. Soon you will be
ready for multiple doses.
The first multiple-dose patient might be given just a few doses. Shouldn’t
this be one of the single-dose patients in whom you can predict precisely the
plasma concentration profiles after multiple doses from studying the single dose?
Later patients might be treated for as long as test animals have indicated repetitive
dosing to be safe.
In every patient, dosing should be based on all the knowledge gained from
all prior animal and clinical studies, from everything known about that patient,
and from the best prediction of what plasma concentrations will be achieved after
the dose.
G. Protocols
Using all information before each dose is safer and more ethical, but it makes
it impossible to write a rigid protocol in advance. In some cases doses will esca-
late rapidly as plasma concentrations in humans are less than expected. In other
H. Fail Fast/Smart
Failure lurks with every dose. Suppose the first dose is eliminated at startling
speed by a unique human enzyme. Suppose the first dose is not absorbed from
the gut. Suppose the first dose has nearly complete presystemic elimination in
the liver. These may abort the project after one or a very few patients. We should
always prepare for failure and plan our studies to minimize the number of patient
exposures and overall patient risk as well as time and effort to failure. Tom Wat-
son of IBM always admonished his scientists to fail fast. I would say fail smart.
Science respects failures that inform and denigrates stupid successes.
J. Rechallenge
The single-dose patients are excellent choices for the first multiple-dose trials.
Single-dose patients also can be rechallenged for specific studies of kinetic vari-
ables. Study the same patient given a test dose together with drugs that will alter
plasma protein association or urine pH. To define oral absorption, give a second
test dose together with divalent cation-containing antacids or after a proton pump
inhibitor or during diarrhea, which can be induced with oral hypertonic mannitol
or sorbitol. Give a test dose together with activated charcoal to see if this is
effective therapy if a child overdoses on a dose taken home. Give a dose into
the duodenum, jejunum, ileum, cecum, or rectum, or sample the succus entericus
to see if drug appears in the bile or intestinal secretions.
These are the patients in whom a second dose can define absorption from
nasal installation, a skin patch, sublingual administration, rectal suppositories,
enemas, enteric coated capsules timed to disintegrate in the cecum, or regular
enteric coating. If the material is not irritating, and you have tolerated one dose,
why would you want special toxicity studies for these unusual routes of adminis-
tration? I wouldn’t.
L. Braiding
Some of the long stream of chemical testing must be done before the first toxicol-
ogy study. Some will be needed for the first human. If the project has not failed,
other chemistry studies will be needed before multiple doses or before using solid
dosage forms or inhalation techniques. Unexpected human metabolites or new
contaminants will necessitate further chemistry studies. Certainly much will be
required before the pivotal trials. But if 9 of 10 projects fail, think how many
chemistry studies can be saved by doing them just in time, as they are needed.
Some toxicology will be needed before the first human dose. The findings
in that first patient will guide additional toxicology. New rodent strains or other
species might be needed to simulate the kinetics and metabolites in humans. If
a key metabolite is observed only in humans and ferrets, perhaps ferrets would
be a good model to test.
We need more diversity in test animals. We have stuck to a few strains of mice
and rats because they are predictable and cheap, but are they as helpful as they
might be? Wouldn’t one animal metabolizing a new drug exactly the way it is
metabolized in humans be far more useful than thousands of animals that don’t
metabolize it? Why don’t we use rodents expressing selected human drug metabo-
lizing enzymes? Would you be interested in cancer promotion in transgenic ani-
mals with knocked-out tumor suppressor genes or promoted oncogenes? If
plasma protein association is a dominant feature, why don’t we have animals
expressing human plasma albumin? Would it be more informative to do our stud-
ies on variegated herds of mice and rats from many different strains, of different
ages, levels of obesity, etc? If you knew that certain of these diverse creatures
had unusual responses, wouldn’t that be interesting? Is it better to be ignorant
of these observations? Do we pursue ostrich research in the same sand box over
and over?
IX. INFORMATICS
A. Decision Trees
Decision research should follow a decision tree through a series of decisions.
Most decisions can be mapped in advance. Many variables that will help you
make the decisions can be mapped. Each of these can be estimated, with the
information you have at the beginning, and then the estimates can be refined as
you proceed through the decision tree. Some variables will be defined precisely
and become deterministic. In many cases you will be surprised to find that only
a few observations of the variable will refine your estimates enough to make
further refinement irrelevant for the following decisions. This is one of the most
important features of decision analysis in research—you can measure precisely
the value of information and pay for information only if it will influence a subse-
quent decision.
At the beginning of decision research, write the product label. This exercise
will expose a number of decisions that must be made and the knowledge that
will help you make better-quality decisions. From this, design the decision tree
of your decision-phase research. Estimate all the variables from the information
you have. Then see how the variables will alter subsequent decisions. For the
most critical variables, see if they can be studied first. Don’t follow a rubber
stamp, but design the studies dynamically, the order in which they will be com-
pleted, as well as the extent of data to be collected. If the next datum will not
influence a decision, save it until an affirmative product decision is made. No
datum before its time—all data just in time.
B. Relevance Diagrams
Ron Howard, ScD, invented the relevance diagram as a linear transform of the
exponential decision tree. On one page you can represent 10–20 decisions and
50 or more nodes representing variables. The relationships are shown by arrows
that indicate which variables are relevant (influenced by) other variables. If you
were to estimate the halftime of a new drug in the first human, you might predi-
cate your estimate on the halftime in animals, the age of the patient, the liver
and kidney function of the patient, etc.
A well-constructed relevance diagram is the map to research. The decisions
are highlighted. Before each decision there are variables you should estimate
X. RESULTS
A. Savings
Designing elegantly braided decision research from a relevance diagram should
save half or two-thirds of traditional time, effort, and expertise. A few drugs
might not reach 90% probability of success at product decision if they have very
delayed effects. You may have to make the product decision based on surrogates
such as plasma cholesterol concentration or markers of bone metabolism without
knowing for sure that the drug prevents heart attacks or vertebral fractures until
you have treated thousands of patients for years. But your investment decision
will be based on decision-phase studies of surrogates designed in elegant braided
fashion.
B. Value
What value is created in discovery, in decision-phase research, or in registration
research leading to launch? If the probability of launch is only 0.1, then your
expected value is 10% of the product’s value at launch. When you reach first
human dose, and an industry-average 0.22 probability of regulatory approval,
your expected value has increased 120% from the 0.1 probability point. At prod-
uct decision, if your probability of successful launch is 0.9, you have increased
your expected value 900% from the start of decision research. These expected
values might be the maximum you should expect a partner to pay for your project.
If the partner is risk-averse, he will pay even less until you have eliminated much
of the risk. At what point should a small inventor firm try to sell its product?
C. Value/Assets
Examine the assets employed to move the project from milestone to milestone.
From molecule selection (start) to completion of single-dose animal studies in-
More expected value per asset employed is generated as research proceeds toward
the product decision. After the product decision, however, there may be the need
to spend $100M to eliminate that last 0.1 of uncertainty and harvest the value
in the marketplace. This expense can be predicted accurately and there is little
associated risk after the product decision.
If the firm undertaking the pivotal studies is successful and spends
$100M to reap the equivalent of $300M, that would seem to be a good
investment. How much should be shared with the originator who has spent
perhaps $30M on discovery and $10M on development? Perhaps $100M to
the originator will provide a 250% profit and leave a 200% profit to the new
owner.
D. Failure
Unfortunately, the above summary neglects failure. If nine discovery projects
fail for this one to succeed, and if each costs $30M, who will have paid the
$270M? In decision-phase research the developer will be careful to evaluate the
project at each milestone and use an options model to decide whether to make
the next investment. Thus failure in development will usually cost less than $10M
of development investments, but may cost all of the $30M of discovery invest-
ments unless a second molecule, with greater probability of success, can emerge
from the same discovery efforts.
E. Synergies
A small firm may have its future fixed to one product or a cluster of related
products and either thrives or dies on its success. A large firm diffuses the risk
and funds failures from successes. As long as a small firm can raise funds, it will
continue decision-phase research to reap the greatest value when it strikes a deal
XI. CONCLUSION
Why haven’t we always done research this way? We treat patients one at a time,
using all available knowledge. The early statisticians and trial design specialists
were explorers who valiantly sifted through data to generate hypotheses and test
them. They used all the data they could find. Then frequentist statistical thinking
began to dominate. It arose from agriculture research in which large fields could
be planted in various ways, manured, and, much later, the crops harvested and
assessed. The frequentist would then make inferences about the population from
which the samples were drawn and about future samples that could be drawn
from the same population. These slow, plodding, smelly trials may be suitable
for crop cycles, but patients are at your door needing care. You can’t send them
away until the study is harvested. You must treat them using all your knowledge
now.
The second problem with traditional frequentist statistics is that it focuses
on the central tendency—the mean. For approval, the U.S. FDA demands two
studies that have shown the means of two samples, for example, your test drug
versus placebo—to vary enough that the alpha error is .05 or less. But we don’t
treat patients in herds. Unfortunately, the U.S. FDA thinks of patients in just that
way. What is the proper dose of digoxin if you treat 100 patients each with the
same dose? The answer might be 0.5 mg per day orally, but some with hyperthy-
roidism and atrial fibrillation will be undertreated and others with hypomagnese-
mia or hypokalemia or renal failure will be overdosed.
This is what is so malignant about the U.S. FDA restricting access to drugs
until their frequentist statistical studies of herds has resulted in data that pleases
them, rather than focusing on the patients in need now. It also leads the U.S.
Allen Cato
Cato Research Ltd., Durham, North Carolina
Myron B. Peterson
Cato Research Ltd., Washington, D.C.
I. INTRODUCTION
Because of unique physiological, ethical, legal, and economic factors, drug devel-
opment in the pediatric population (patients aged 16 years or younger) is a com-
plex process. Approximately 80% of the prescription drugs approved by the U.S.
Food and Drug Administration (FDA) and marketed in the United States are not
approved for use in children or are restricted to older age groups because of
limited clinical trials. Thus, children have been denied the benefit of many ad-
vances in therapeutics and remain, as Dr. Harry Shirkey (1) described them some
33 years ago, therapeutic orphans.
The failure to conduct pediatric drug trials for the proper labeling and for-
mulation of drugs is directly related to the following six factors:
To illustrate how the factors outlined in the Introduction may be used in practical
applications (i.e., before submitting a New Drug Application to the FDA), con-
sider the following two examples.
A. Example 1: Tracrium
Tracrium (atracurium besylate), is a nondepolarizing neuromuscular blocking
agent with an intermediate duration of action (45 min).
1. Severity of the disease. This compound was intended for use during
surgery, and hence would be considered a clinically significant agent.
2. Availability of alternative therapy. No other agents of intermediate
When the New Drug Application (NDA) for Tracrium was approved by
the FDA, adequate dosing instructions for pediatric patients were included in the
approved package insert.
The fact that, for years, very few drugs were labeled safe and effective for pediat-
ric use led the FDA to issue labeling requirements in 1979 (8) to address this
problem. These regulations required that specific pediatric indications be de-
scribed in the Indications and Usage section and the recommended pediatric doses
in the Dosage and Administration section of the drug label. Recommendations
for pediatric use had to be based on substantial evidence derived from adequate
and well-controlled trials in the pediatric population, unless the requirement was
waived. If safety and efficacy could not be proved, or if the drug posed specific
hazards to children, specific comments were to be included in the Pediatric Use
section of the label.
Although the intent of these new regulations was to promote pediatric label-
ing, it had the opposite effect because it called for enhanced data collection in
children, which was often difficult or impractical to gather. During a 5-year period
(1984–1989), the American Academy of Pediatrics found that about 80% of all
drugs approved had no pediatric use information.
Because of concerns related to inappropriate use or nonuse of these drugs,
the FDA wrote a new pediatric rule in 1994 (9). This rule was an attempt to both
clarify the FDA position vis-à-vis the 1979 requirements, and at the same time
simplify the requirements for industry to provide adequate pediatric information.
The requirement that pediatric labeling must be based on adequate and well-
V. CONCLUSIONS
Daniel C. Cato
Cato Research Ltd., Durham, North Carolina
David B. Thomas
Roche Molecular Systems, Inc., Pleasanton, California
I. INTRODUCTION
The drug development process has matured into a system used primarily for cap-
turing information on paper. Although this process is entrenched in industry pro-
cedures and is well understood, documented, and accepted, it is also recognized
as somewhat inefficient and costly. To date, the greatest progress in moving clini-
cal trial information by electronic modes has been in the transfer of patient testing
results from centralized laboratories to trial sponsors’ databases. This technical
innovation was made by the central laboratories that had the electronic technolo-
gies available through their own laboratory information systems and saw an op-
portunity to attract business from clinical trial sponsors by offering efficient trans-
fer of testing information to the sponsor.
For more than 20 years, various proposals have been made for expanding
electronic data capture to encompass an increased portion of trial information.
The motivation behind these proposals has been the quest to (a) reduce the time
it takes to collect and verify trial information, (b) simplify the collection process
of trial information in hopes of increasing the accuracy of study data, and (c)
reduce the resources (costs) involved in conducting trials. Only in the last few
years has the technology been available to help us see how these goals might be
achieved.
B. Bar Coding
Another time-consuming yet critical aspect of paper-based systems is tracking
the source documentation pages of original data or data changes. Traditionally,
this task has been done by some system of numbering CRF pages and other study
documentation. These numbers must be accurately applied to each document and
then placed in the study database as indication of the source of information en-
tered. A filing system must then be maintained that allows retrieval of each page.
The delay between the patient visit and the time the data are entered into the spon-
sor’s database is a problem inherent in paper-based study data collection systems.
If CRFs are collected by the sponsor when the patient completes the study, then,
in a year-long study, at least a year will have passed between the patient’s first
visit and the time the data are entered into the database. Most sponsors increase
the efficiency of their process by requiring that CRF data be submitted to a central
data management center at specified intervals during a patient’s progress through
the trial. The advantage of such a procedure is that questionable or invalid data
can be identified and data issues can be resolved on an ongoing basis. However,
in practice, it is often difficult for clinical sites to keep up with the sponsor’s expec-
tations for the prompt and accurate completion and submission of CRF data. In
addition, when the sponsor receives the CRF data, it takes some time before it can
be screened and entered and items can be identified that need to be returned to the
site for resolution. During this time, the site personnel may be unaware that the
sponsor has issues with the site’s interpretation of the study protocol or completion
of the CRFs, and the particular problem may continue unchecked.
One way industry has tried to decrease the time between the investigator’s
evaluation of the patient and entry of that data into the database is by implement-
ing remote data entry (RDE) systems. In an RDE system, data entry and capture
capabilities are provided at each study site. Personnel at each site are trained
to enter the data upon completion of the CRF or after each patient visit. In this
way, the data, although not necessarily checked by the clinical monitor, are avail-
able to the sponsor in a more timely manner. After entry, the data can be sent
to the sponsor by upload over a modem connection or by mailing removable
storage devices (e.g., floppy disks, CDs). Therefore, data questions from the spon-
sor can be posed earlier in the process, when the site is still familiar with the
patient.
Although RDE systems have been used successfully, many problems exist
with the overall strategy. RDE requires additional work at the site because data
Until recently, the implementation of remote data capture systems was sig-
nificantly hindered by issues related to the regulatory standard that all information
from trial sites be attributed to an individual qualified to certify the accuracy of
the data. Traditionally this standard had been fulfilled by a qualified investigator
or designee who signed and dated a form. Because there was no alternative to
the written signature, data submitted electronically by the site at some point still
had to be printed and sent back to the investigator for review and signature. In
effect, the signature process negated some of the efficiency gained through remote
data entry. However, in 1997, a Food and Drug Administration ruling (1) recog-
nized the validity of electronic records and electronic signatures. Title 21 of the
Code of Federal Regulations, Part 11, represents a significant step forward in
electronic data capture and submission of clinical data.
With the range of technologies available today, a variety of systems exists for
the remote electronic data capture of trial data into the sponsor’s central database,
and the workflow from EDC systems can be fairly uncomplicated. Site personnel
enter study data directly into the EDC system instead of first transcribing the
information onto paper CRFs. To reduce data queries, logic checks can be built
into systems so that inconsistent or out-of-range values can be flagged at the time
they are entered. This allows erroneous data to be corrected before it is committed
to the trial database, and it prompts for additional explanatory information that
may assist in interpretation of the data submitted.
Alternatively, the data may be entered into a locally stored database and
then transmitted to the sponsor’s database, where it is edited. At some point
during the entry process, the system should provide for site review and certifica-
tion that the information being submitted is consistent with the source documenta-
tion (e.g., patient records). For information derived from source documentation at
the site, monitoring will still be necessary to verify that the submitted information
matches the source database and to reconcile any discrepancies—the same as in
paper-based systems.
EDC systems for clinical studies can be categorized as offline or online,
differentiated by whether or not an active connection to the study’s central data-
base is required. Each type of EDC system is discussed below.
V. REVIEW
VI. CONCLUSIONS
The technologies employed in EDC systems for clinical trials are now fairly well
established and are evolving rapidly. The business argument for improved data
quality and the associated reduction of study cycle time using these technologies
seems both rational and compelling. The difficult decisions for trial sponsors
relate to technology choices and how to integrate new study management and
data capture technology into ongoing clinical trial programs. It is fair to say that,
to date, the early adopters of these technologies have not gained an obvious ad-
vantage over those organizations that stayed with traditional paper-based CRF
systems. However, going forward, it is difficult to imagine that EDC will not
dominate the clinical trial data management process.
REFERENCES
1. 62 Fed Reg 13429 (1997) (Electronic Records; Electronic Signatures; Final Rule.
Codified at 21 CFR § 11).
2. U.S. Food and Drug Administration. Guidance for Industry. Computerized Systems
Used in Clinical Trials. April 1999.
3. Feller RE. Electronic data capture: survey 2000. Association of Clinical Research
Professional publication. The Monitor, Spring 2001; 14:37–43.
Peggy J. Berry
Dey Laboratories, Napa, California
Allen Cato
Cato Research Ltd., Durham, North Carolina
I. INTRODUCTION
Until recently, the use of investigational drugs outside the scope of well-con-
trolled clinical studies was almost nonexistent. Patients who were not eligible
for ongoing clinical studies were not given access to some potentially promising
drugs that might have made a positive difference in their quality or length of
life.
As a result, awareness began to emerge that patients with certain diseases
or conditions who were afforded a limited number of, or often no, treatment
options or who were unresponsive to all available treatments, would likely be
willing to accept much greater risks when being treated if it was possible that
they would receive some benefit from the treatment. A recognized need led to the
development of a mechanism that would allow these patients access to otherwise
unavailable investigational drugs.
In developing a mechanism for patient access to therapy, many issues had
to be considered. First, it would be imperative for a physician to recognize the
need for treatment with the investigational drug. Second, the physician would
have to be willing to accept the responsibility of using an investigational drug
and monitoring the patient at closer and more controlled intervals than might be
necessary with marketed drugs. Third, to avoid compromising the scientific study
of the safety and effectiveness of investigational drugs through well-planned,
well-controlled clinical studies, the patient must be deemed ineligible for all on-
going clinical studies with the drug. The result was a compassionate-use excep-
tion, granted upon documented justification, for investigational drugs to be used
to treat an individual patient. By identifying an individual patient’s need, con-
tacting a drug company to confirm availability of the drug, and submitting a case
study and other documents to and receiving approval from the Food and Drug
Administration (FDA) and the Institutional Review Board (IRB), the physician
could treat the patient with the investigational drug. The physician was obligated
to become familiar with the investigator’s brochure for the product and to report
back to the FDA and the drug company sponsor on the results of the treatment.
A few years’ experience reviewing and granting compassionate-use excep-
tions made clear the need for more definitive regulations. It was believed that
not all patients were being given the opportunity to receive promising new thera-
pies when their condition could benefit from early treatment. In addition, it was
noted that in some cases, after a drug had been demonstrated to be effective and
safe in at least one clinical study but was not yet approved by the FDA or in
wide distribution, patients with severe diseases who could benefit from treatment
Within 1 week after the emergency use, the investigator must file documentation
to the FDA along with all completed regulatory forms. Emergency-use treatments
with an investigational drug must be documented fully and completely and are
subject to Good Clinical Practice (GCP) regulations.
These expanded-use protocols not only afford the patient access to promis-
ing investigational drugs, they also provide the sponsor with an opportunity to
study the drug in a broader population. These studies provide useful data for
premarketing surveillance and may serve as a natural transition to the observa-
tional surveys that characterize Phase III and Phase IV development.
Along with the decision to proceed with the expanded use of the investiga-
tional drug, numerous important issues must be considered, including the issues
resulting from the decision itself.
A. Legal Issues
Can a patient sue the FDA or the drug company for access to an investigational
or nonmarketed drug? Once given access, can the patient sue the investigator,
the drug company, or the FDA for damages sustained while taking the drug under
expanded-use protocols? The answer to both questions is yes. The patient has
the right to sue or to pursue other available avenues (such as petitioning Congress
or the FDA) to demand access to a drug that he believes he needs. The patient
does not give up any legal rights at any time before, during, or after participation
in a typical or expanded-use clinical study and therefore may seek legal remedy
to any damages he alleges.
Patients have applied public pressure through visible campaigns to gain
access to drugs whose distribution may have been discontinued or strictly limited.
One such example, pimozide (Orap), was provided to patients with Tourette’s
syndrome (an orphan disease). Public pressure from orphan-disease groups and
a visible public education campaign coerced McNeil into continuing its supply
of the drug after the company had decided to quit producing it (1).
Political and public pressure can influence decisions of the FDA and of
independent advisory committees. A recent example was spurred by a debate
over experimental gene therapy. A politically well-connected 51-year-old female
presented with Grade 4 glioblastoma (2). After the patient underwent brain sur-
gery twice, radiotherapy, chemotherapy, and an experimental radioactive mono-
clonal antibody treatment, a San Diego physician proposed, as a last recourse,
providing her with an experimental vaccine treatment with interleukin-2. The
treatment was presented for compassionate approval to the National Institutes of
Health’s (NIH) Recombinant DNA Advisory Committee (RAC) and to the FDA.
The RAC did not have policies in place to review expeditiously the treatment
and safety issues and to respond to requests for compassionate use of gene ther-
apy. Lacking substantial information on the use of gene therapy and lacking for-
mal procedures for review of such requests, the RAC determined that the experi-
mental gene therapy treatment was not yet developed enough to be used in
compassionate requests. The request for use was denied at least until the RAC
could discuss its policy for handling gene therapy requests, scheduled for the
next meeting—3 months later.
After the denial, Senator Tom Harkin (D, Iowa) wrote a letter persuading
the RAC chairperson to seek a temporary solution to the problem of policy until
the issue could be further discussed and to give timely consideration to individual
requests for compassionate use from terminally ill patients. Review by the FDA
was occurring in parallel to review by the RAC. The FDA’s review of the gene
B. Regulatory Issues
1. Drug Availability
The availability of the drug may lag far behind demand for the drug, and this
factor can temper the motivation of the FDA to approve its use. The lack of
commercial amounts of drug supply or the expense of obtaining raw materials
to make the drug could also lead to use of the drug under conditions similar to
expanded use.
The synthesis of a compound may entail many steps in a small laboratory
and be quite time-consuming and expensive. The initial scale-up to commercial
production may leave impurities and, because the process changes, stability data
may not be available. Lack of supply poses a difficult problem. If the drug is for
use by patients who have incurable diseases and no alternative treatment exists,
the sponsoring drug company must decide whether to institute an expanded-use
protocol. If a protocol is employed, a decision must be made to determine how
patients will be selected, because insufficient supply exists for all patients with the
disease or condition. This was an especially troublesome problem in the trial of
potential medications for AIDS (4). When treatments are offered that have even
3. Package Insert
Marketing approval may be facilitated by the increased number of patients ex-
posed to the investigational drug under an expanded-use protocol. Labeling of
new drugs is typically conservative until the drug has been marketed and used
in larger numbers of patients and on a long-term basis (if required by the indica-
tion). By employing the expanded-use protocol, which would include broader
patient populations and variations of the indication, the sponsor may gain more
liberal labeling from the increased exposure. However, by attempting to achieve
more liberal labeling, the sponsor runs the risk that a delay in marketing approval
could occur if difficulties develop in the interpretation of the data from an ex-
panded-use study.
Side effects and drug interactions observed during the expanded-use study
are required to be included in the package insert (which will generally increase
the incidence of the side effect). By admitting patients who would be ex-
cluded from the controlled clinical study, the chance also exists that side effects
and drug interactions that were not observed during the tightly controlled pre-
marketing studies will be reported. These effects and interactions will require
further characterization by the drug company and may need to be included in
the package insert. In a severely ill population where the course of the disease
is poorly understood, it may be difficult to ascribe causality to either drug or
disease. A well-designed expanded-use protocol and study controls help to mini-
mize the speculation that accompanies causality of adverse events during clinical
trials.
During an expanded-use study (5) of FK 506 in liver transplant patients
presenting with dysfunction who had failed previous treatment, it was concluded
that the incidence of side effects observed during the expanded-use study was
The introduction of a new chemical entity into the New Drug Application (NDA)
process often heralds the formulation of a unique mode of therapy for patients
with a specific diagnosis. During clinical development, the patient populations
identified in protocols are rigorously defined and screened under stringent
inclusion/exclusion criteria. This is done to minimize ‘‘background noise’’ in
clinical studies, making the results easier to interpret and side effects easier to
For the expanded-use study to be cost-effective for the sponsor, little or no fund-
ing should be expended for patient expense reimbursement or for investigators
who desire to prescribe the investigational drug to their patients. However, it is
the sponsor’s responsibility to provide clinical trial material (investigational study
drug) with appropriate labeling and instructions for use (including but not limited
to the investigator brochure), appropriate recording instruments, and the medical,
scientific, and monitoring expertise to conduct the trials in accordance with the
expanded-use protocol and the current FDA regulations and guidelines.
C. Monitoring
These nonpivotal studies provide a base from which research personnel can be
trained and educated. Because few formal training programs in clinical research
are available, more direct exposure and interaction can be established between
the sponsor and investigators interested in drug development. This exposure helps
prepare both investigators and industry personnel in clinical development, provid-
ing more cost-effective expertise in the development of future NDAs while im-
proving the quality of scientific interactions.
Monitoring is especially tenuous under these expanded-use protocols, and
the data obtained can often be difficult to interpret. No matter how simplistic a
As safety is the primary concern in the introduction and maintenance of any drug
in the commercial marketplace, the expanded-use protocol provides an opportu-
nity for obtaining long-term data not captured in the shorter and more expensive
efficacy and safety trials. Most important, the opportunity to gain insight into
any possible unpredicted toxicity not evident in the controlled clinical trial pro-
gram could emerge early to enhance acceptable balances of benefit and risk. Side
effects can also reveal themselves that could lead to a new indication for the
compound or modification of its use. Although the number of patients participat-
ing in an expanded-use protocol would not approximate the larger number of
patients involved in the Phase IV epidemiological studies, information on previ-
ously unseen risks can be extrapolated so that the safety parameters that protect
the consumer can be anticipated.
It is the rare patient who suffers from an isolated disorder and takes only
a single medication. In the expanded-use protocol, much information can be
gained by evaluating potential drug interactions with the current medications the
patient may be taking. Frequently, this information has tremendous impact on
the overall therapeutic efficacy of the investigational drug and its risk and benefits
when given with other forms of therapy.
It is a distinct advantage to the sponsor to limit the number of centers to
include only those investigators who have had previous experience with the drug.
Such action may succeed in carefully controlling the information and data being
obtained during the monitoring of these studies, but it does not fulfill the overall
obligation of implementing this type of protocol.
With the initiation of the expanded-use protocol, the sponsor must construct
a safety data management program within the home facility or with a contract
organization so that the sponsor can fulfill its reporting obligations to the FDA.
In data management, because of the expanded-use, uncontrolled nature of the
studies, efficacy data may be of academic value only, but all information (both
safety and efficacy data) must be appropriately tabulated for reference. Negative
The lag in NDA approval has long been known to be an aggravating and expen-
sive proposition for both the FDA and sponsors. These investigators’ IND re-
quests can be eliminated by referring the investigator to the sponsoring company
for the supply of clinical material and an appropriate protocol under which to
conduct their clinical studies, which are filed under the sponsor’s IND. Thus,
investigators who would normally apply for their own IND would save the FDA,
sponsor, and themselves excess paperwork and would still be able to conduct
the investigational study. Although a company may not wish to support a full
expanded-use program, it may consider supplying the drug to investigators who
wish to pursue their own IND. Submission of numerous INDs of this type, how-
ever, may drain the FDA’s and the sponsor’s resources.
Also, when side effects do occur with the investigational drug, it allows
the sponsor to effectively channel this information to the FDA rather than having
individual reports trickle into the regulatory agency. Furthermore, these studies
allow the sponsor to obtain early cost-effective safety data (while substantiating
efficacy data) that will ultimately benefit the patient. This type of collaborative
effort, coordinated by the sponsor, could foster a closer network among the FDA,
investigator, scientists, and patients. This strategy could substantially ease the
FDA workload and expedite approval of the NDA for the benefit of everyone
involved.
The FDA has developed regulations for the treatment, use, and sale of
investigational drugs (7). The regulation states that an investigational drug may
be used to treat patients outside a controlled clinical trial if the disease is ‘‘seri-
ous’’ or ‘‘immediately life-threatening,’’ if no satisfactory alternative drug/ther-
apy is available, if the drug is currently being investigated in a controlled clinical
trial under an IND, and if the sponsor is actively pursuing marketing approval
of the drug. The FDA commissioner can deny a request for use if insufficient
evidence exists of safety and effectiveness to support its use. The regulations
further provide safeguards for placing a treatment IND on clinical hold and a
mechanism for obtaining prior FDA approval for the sale of an investigational
drug during a clinical trial. Many companies, rather than filing a formal treatment
protocol, will conduct an expanded-use, single-patient-use, or ‘‘continuation’’ or
‘‘extension’’ protocol under their IND. Although these other types of protocols
are not defined in the regulations, a provision exists for the single-patient-use
protocol (which is the same as the expanded-use protocol). If the company de-
cides to allow expanded-use type of enrollment only to those patients who have
VIII. CONCLUSIONS
The expanded-use protocol, when initiated with careful scientific thought and
when appropriately managed, can be an extremely important instrument for ob-
taining cost-effective information for the FDA, the scientific/medical/academic
communities, and the patient. It can provide a training ground for enhancing
the quality and cost-effectiveness of future clinical trials. Most important, those
patients who have been nontolerators or nonresponders to alternative treatments,
those who have responded well in the controlled clinical trials of the investiga-
tional drug, and those who are in life-threatening situations can benefit tremen-
dously and are the primary motivation for this protocol. During the course of the
drug development of any compound, a balance must be struck between the risks
and benefits of initiating the expanded-use protocol. The high standards applied
to the early development of the clinical study must not be compromised during
the conduct of the larger, open study. It is the responsibility of the investigator,
the sponsor, and the FDA to maintain a collaborative environment so that these
studies can serve a patient population in need of the drug before its approval and
marketing.
REFERENCES
1. Myers AS. Orphan drugs and orphan diseases: the consumer’s viewpoint. In: Brewer
GJ, ed., Orphan Drugs and Orphan Diseases: Clinical Realities and Public Policy.
New York: Liss, 1983:147–157.
2. Thompson L. Gene therapy. Harkin seeks compassionate use of unproven treatments.
Science 1992; 258(5089):1728.
3. Healy B. Remarks for the RAC Committee meeting of January 14, 1993, regarding
compassionate use exemption. Hum Gene Ther 1993; 4:195–197.
4. Berry WR, House KW, Lee JT, Plagge PB, Meshad MW, Grapski R. Results of a
compassionate-use program using intravenous ondansetron to prevent nausea and
vomiting in patients receiving emetogenic cancer chemotherapy. Semin Oncol 1992
Dec; 19(6 Suppl 15):33–37.
5. Murio JE, Balsells, J Lazaro JL, Charco R, Margarit C. Compassionate use of FK
506 in liver transplantation. Transplant Proc 1995; 27(4):2336.
6. Tastemain C. Confusion reigns over compassionate use of AIDS drug. Nat Med 1995
Oct; 1(10):986.
7. Department of Health and Human Services. Investigational New Drug, Antibiotic,
and Biological Product Regulations: Treatment Use and Sale. Fed Reg 1987; May
22, 52:19466.
Dale H. Cowan
Cleveland Clinic Foundation, Cleveland, Ohio
I. INTRODUCTION
Clinical drug trials represent research with human subjects. All research involv-
ing human subjects that is supported by the U.S. federal government or the results
of which are to be used in applications for drug or device approval must be
conducted in accordance with regulations promulgated by the U.S. Department
of Health and Human Services (HHS) (1) and the U.S. Food and Drug Adminis-
tration (FDA) (2). The regulations of both the HHS and the FDA require that an
Institutional Review Board (IRB) ‘‘shall review and have authority to approve,
require modifications in (to secure approval), or disapprove all research activities
covered by [the] regulations’’ (3).
The review of clinical drug trials by IRBs raises a number of interesting
and difficult issues. These relate to the origin and sponsor of the proposed trial,
the nature of the institution the IRB serves, and the manner in which the norms
for determining ethical conduct in clinical trials can be applied to specific trials.
This chapter will review the ethical principles underlying research involv-
ing human subjects, the legal authority for IRBs, and the regulatory requirements
affecting the operations of IRBs. It will then discuss the role of IRBs in reviewing
clinical trials by examining how IRBs can assess the scientific design of trials,
the competency of the investigator, the manner of selecting subjects for the trial,
the balance of risks and benefits, informed consent, and provisions for compensat-
ing for research-related injuries. The chapter will conclude with observations
regarding the role of the lay members of IRBs, problems of reviewing multi-
institutional trials, and how to monitor the conduct of trials that are approved.
Among the basic ethical principles that are generally accepted in our cultural
tradition, three are particularly relevant to the ethics of research involving human
subjects (4). These are the principles of respect for persons, beneficence, and
justice.
B. Beneficence
The principle of beneficence states that ‘‘we . . . have a moral duty to weigh and
balance possible benefits against possible harms in order to maximize benefits
C. Justice
The principle of justice represents the concept of fairness. It requires that the
benefits and burdens of any activity, such as participation in clinical trials, be
distributed equitably across the population. In short, it states that equals should
be treated equally.
The Belmont report identifies five widely accepted formulations of just
ways to distribute benefits and burdens. These are (a) to each person an equal
share, (b) to each person according to individual need, (c) to each person ac-
cording to individual effort, (d) to each person according to societal contribution,
and (e) to each person according to merit.
Historically, the burdens of serving as research subjects fell largely upon
socially and economically disadvantaged individuals, whereas the benefits tended
to accrue to the more affluent members of society. One of the most flagrant exam-
ples of this inequity was the Tuskegee syphilis study, in which disadvantaged,
The legal authority for IRBs derives from two parallel sets of federal regulations.
One set of regulations was promulgated by the Department of Health and Human
Services and implements the 1974 amendments to the Public Health Service Act
(12). These regulations are codified in Title 45 of the Code of Federal Regulations
(CFR), Part 46. The second set of regulations was promulgated by the FDA under
the Federal Food, Drug, and Cosmetic Act (13). These regulations are codified
in Title 21 of the CFR; regulations pertaining to IRBs are in Part 50 and those
pertaining to informed consent are in Part 56.
A. HHS Regulations
The HHS regulations cover research supported or conducted by the Department
of Health and Human Services. They represent the basic policy of HHS for pro-
tecting human research subjects. The applicability of the regulations is restricted
to those activities that meet the definition of research: ‘‘a systematic investigation
designed to develop or contribute to generalizable knowledge’’ (14). Although
the regulations technically apply only to those research projects that are con-
ducted or funded in whole or in part by HHS, virtually all public and private
granting agencies require that all research they sponsor that involves human sub-
jects be conducted in accordance with the HHS requirements.
Subpart A of the regulations specifies that each institution ‘‘covered
by th[e] regulations shall provide written assurance to the Secretary (of HHS)
that it will comply with the requirements set forth in th[e] regulations’’ (15).
The regulations further specify the minimum elements for such assurances. Addi-
tionally, they specify requirements for IRB membership, function, and operation,
B. FDA Regulations
The FDA has the legal authority to regulate clinical investigations in the United
States when the investigational products move across state or national boundaries.
Under FDA regulations, review and approval by an IRB is required for any ex-
periment that involves a test article and one or more human subjects, either pa-
tients or healthy persons, and that is subject to the requirements for prior submis-
sion to the FDA (16). Such review is also required for any experiment the results
of which are intended to be submitted later to, or held for inspection by, the
FDA.
The regulations of the FDA are identical or similar to those of HHS in
nearly all essential respects. Such differences as do exist reflect the different
statutory authority under which the separate sets of regulations were promulgated
and the difference in mission between the FDA and the National Institutes of
Health (NIH), the agency within HHS charged with overseeing the implementa-
tion and enforcement of the HHS regulations. The difference in mission between
the FDA and the NIH is reflected in the FDA’s approach to compliance with its
regulations utilizing its traditional tools of inspections and audits. The NIH, as
noted previously, relies on the assurance mechanism.
The FDA regulations, like the HHS regulations, specify requirements for
IRB membership, function, and operation, and the criteria according to which
approval may be given for conducting research. Since these requirements are
similar, a single committee can be established to undertake the activities required
by both sets of regulations. Additionally, the FDA regulations allow a wide vari-
ety of ways in which private practitioners not affiliated with an institution can
obtain necessary IRB review of their clinical research activities. These include
review by an institutional IRB that agrees to assume this additional function or
by IRBs formed by a local or state health agency, a medical school, a medical
society, a state licensing board, or a nonprofit or for-profit independent group.
All IRBs, regardless of sponsorship, that are assuming responsibilities for re-
viewing and approving clinical research protocols subject to FDA authority must
comply with the IRB regulations set out by the FDA.
The FDA and HHS regulations specify the duties and membership requirements
of IRBs and the criteria for approving research. Since the two sets of regulations
are similar, these will be reviewed by reference to the HHS regulations.
A. Duties of IRBs
IRBs are required to review and have the authority to approve, require modifica-
tions in, or disapprove all research activities covered by the regulations (17).
They must require that information given to subjects as part of informed consent
is in accordance with the general requirements for informed consent that are set
out in the regulations. Additionally, they may require that other information be
given to subjects when they judge that such information would further protect
the rights and welfare of the subjects (18).
IRBs must require documentation of informed consent in all studies except
those specified in the regulations in which documentation may be waived (19).
Clinical drug trials are not among the classes of studies in which documentation
of informed consent may be waived.
IRBs must provide written notification to investigators and institutions of
their decisions to approve, require modifications in, or disapprove proposed re-
search activities (20). Decisions to disapprove proposed research proposal must
be accompanied by a statement of reasons for the decision and provide the investi-
gator an opportunity to respond in person or in writing (21).
IRBs must conduct continuing reviews of research they approve at least
once each year. More frequent reviews may be required if the risk of a particular
research project so warrants (22). IRBs have the authority to suspend or terminate
approval of research that is not being conducted in accordance with their require-
ments or that has been associated with unexpected serious harm to subjects. Such
action must be accompanied by a statement of reasons for it and be communicated
to the investigator, appropriate institutional officials, and the Secretary of HHS
(23).
The regulations require that IRBs must follow the written procedures that
are set out in the assurances they have filed with HHS (24), review proposed
research at convened meetings at which a majority of IRB members are present,
vote approval by a majority of members present at the meeting (25), and be
responsible for reporting to the appropriate institutional officials and the Secretary
of HHS ‘‘any serious or continuing noncompliance by investigators with the
requirements and determination of the IRB’’ (26).
Institutions that are cooperating in multiinstitutional studies, such as clini-
cal drug trials, must each review and approve the proposed studies. Such institu-
B. IRB Membership
The regulations require that ‘‘[e]ach IRB shall have at least five members, with
varying backgrounds to promote complete and adequate review of research activi-
ties commonly conducted by the institution. The IRB shall be sufficiently quali-
fied through the experience and expertise of its members, and the diversity of
the members’ backgrounds . . . to promote respect for its advice and counsel in
safeguarding the rights and welfare of human subjects. In addition to possessing
the professional competence necessary to review specific research activities, the
IRB shall be able to ascertain the acceptability of proposed research in terms
of institutional commitments and regulations, applicable law, and standards of
professional conduct and practice. The IRB shall therefore include persons
knowledgeable in these areas’’ (28).
Additional requirements specify that IRBs may not consist entirely of men
or women or members of one profession (29), must include at least one person
who is a nonscientist (30), and must include at least one individual who is other-
wise not affiliated with the institution (31). Finally, the regulations expressly
forbid members participating in an initial or continuing review of a project in
which the member has a conflicting interest (32).
The requirements for IRB membership have particular relevance to the
manner in which they carry out their assigned functions.
Having reviewed the ethical principles underlying research with human subjects,
the legal authority for IRBs, and the regulatory requirements affecting IRB activi-
ties, it is useful to consider the manner in which IRBs can carry out their responsi-
bilities in reviewing clinical drug trials.
C. Selection of Subjects
The norm that subjects shall be selected equitably is based on the principle of
justice. The principle of justice requires that the benefits and burdens of research
be distributed fairly. It is inappropriate for subjects of clinical trials to be drawn
primarily from members of a specific subpopulation of the community unless the
disease being treated affects only members of that group. For example, it would
be inappropriate for a clinical trial of a new antiarthritic drug to be undertaken
solely with individuals from lower socioeconomic groups who receive their care
E. Informed Consent
Assuring that adequate provisions exist for securing informed consent is a central
duty of IRBs. The requirements for informed consent are specified in the federal
regulations. These require that investigators ‘‘shall seek such consent only under
circumstances that provide the prospective subject . . . sufficient opportunity to
Five special issues arise with respect to the role of IRBs in reviewing clinical
trials. These are the role of lay members of IRBs, the review of multi-institutional
trials sponsored by national cooperative groups, the duty to monitor the trials,
the financial risk assumed by patients entering clinical trials, and adherence to
new policies promulgated by the FDA or HHS.
(1) The Office of Protection from Research Risks (OPRR) now requires
that each local IRB receive a copy of the NIH-approved sample consent
document and the full NIH-approved protocol as a condition for review
and approval of the local informed consent documents.
(2) Any deletion or substantive modification of information concerning
risks or alternative procedures contained in the sample informed con-
The directive quoted above, issued by the OPRR, asserts that the policy
contained in it does not reflect any change in the OPRR’s policy concerning the
importance of local IRB review. Rather, it reiterates the requirement that ‘‘[e]ach
IRB must continue to review all protocol and informed consent documents with
the greatest of care, regardless of any prior review at the national level’’ (61).
These two examples illustrate that IRBs must remain current with federal
guidelines and regulations and be prepared to implement them when they become
effective.
VII. CONCLUSIONS
The review of clinical drug trials by Institutional Review Boards can be difficult
and challenging. Adequate review by IRBs requires a firm understanding of the
ethical principles that underlie the participation of human subjects in research,
the legal authority under which IRBs operate, and the regulatory requirements
that inform IRB activities. Resolution of specific issues requires careful balancing
of competing considerations. Judicious decision making on the part of IRBs can
enable important research into the treatment of disease to proceed while assuring
that adequate safeguards exist to protect the rights, safety, and well-being of the
subjects of the research.
ACKNOWLEDGMENT
Portions of this chapter were drawn from material previously prepared by the
author that appeared in Human Subjects Research (Plenum Press, New York,
1982), and are used with the permission of the publisher.
REFERENCES
1. The regulations of HHS that provide for protection of human research subjects are
set out at 45 CFR 46.
Paul J. Reitemeier
National Center for Ethics, Veterans Health Administration, White River
Junction, Vermont, and Department of Medicine, Dartmouth Medical
School, Hanover, New Hampshire
The moral convictions of thoughtful and well-educated people are the data
of ethics, just as sense perceptions are the data of science.
W. D. Ross, The Right and the Good, 1930
The opinions of the author expressed in this essay are personal and do not reflect those of the National
Center for Ethics, the Veterans Health Administration or the federal government.
A. Research Design
Two significant obstacles to comparing medical therapies are the extensive time
and the large number of subjects required to complete such studies with sufficient
accuracy for the results to be reliable. To minimize both the time required and the
number of subjects needed, research designs have been developed that emphasize
statistical power and validity. For more than 40 years, blinded, randomized clini-
cal trials (BRCTs) have been the ‘‘gold standard’’ of study designs because they
emphasize both statistical power and validity and provide for economy of re-
sources. BRCTs have definitively established some treatments as effective and
eliminated others as ineffective (1). Large, multicenter trials often use data moni-
toring boards to provide an added degree of regulatory oversight so that the safety
and interests of research subjects are maximally protected (2,3). This is most
common when one of the therapies is experimentally new. Therefore, the need
to identify the most effective therapies as quickly as possible can best be met if
researchers conduct BRCTs.
But ethical concerns in BRCT research have been raised by several com-
mentators (4–10). Physicians who conduct BRCT research must adjust their med-
ical treatment of enrolled patients so it accords with the research protocol require-
ments. Adherence to these requirements is necessary to produce therapy
comparison results that will be most compelling from a scientific point of view,
so clinicians are persuaded by the results and use them in selecting therapies to
offer to patients in the future. Conducting well-designed comparison trials on a
few patients will enable many more future patients to benefit by avoiding what
the trial reveals to be the inferior therapy. The comparison process is driven
and success is measured by the objective endpoints specified in the trial design.
Endpoints typically include either time-to-death (survival) of subjects or a speci-
fied physiological change such as tumor shrinkage, blood-count level, or other
objective measures. Trial endpoints are specified by the study investigator so
analysis of aggregate subject outcomes can be measured independently of the
* The notion of what constitutes ‘‘information’’ in this context is a long-debated issue and essentially
divides along frequentist versus Bayesian lines. Clinical investigators begin by assuming protocol
A is equivalent to protocol B (on whatever metric is important). Random variability, especially in
the earliest stages of the study, will produce interim outcomes that look different simply by chance.
However, investigators choose to assume the null hypothesis (that there is no difference) continues
to be true until such time as the available data provide sufficient information to conclude they are
no different. In this regard, all information short of that allowing a choice to be made is considered
noninformative. [I owe this important clarification to Professor James Anderson, University of Ne-
braska.]
1. Prerandomization
Zelen (20) has suggested that investigators who are in clinical equipoise with
respect to multiple therapy options privately use a randomization mechanism
to select the treatment option for their patients, and then inform the patient
which treatment will be used if the patient consents to participate in the study.
This technique is known as prerandomization, and two versions of it have
been developed. In both versions one-half of the patients are secretly prerandom-
ized to the standard care arm A and followed without their knowledge (or con-
sent) of participating in a trial.* The difference between the two schemes is
how the second half, those secretly prerandomized to the experimental arm B,
are treated.
In version 1, the patients prerandomized to experimental arm B are in-
formed of the treatment and then asked for consent to receive the experimental
treatment B. If the patient refuses treatment B, he or she is offered the standard
treatment arm A and asked to consent to be followed as part of the research
study. In version 2, patients prerandomized to the experimental arm B are asked
for their preference of A or B, and that preference determines which treatment
they receive. To make the relevant statistical comparisons among all participants,
all the group 2 patients (those prerandomized to the B arm and then asked to
choose between A and B) have to be compared with all the group 1 patients
(those prerandomized to A, but not asked to choose). As the group 2 patients
express an aggregate preference for one arm over the other (either A ⬎ B or
B ⬎ A), the total number of patients enrolled in the study must enlarge to main-
tain statistical validity.
The main advantage of Zelen’s two prerandomization designs is that they
allow investigators to accrue subjects into protocols more effectively because
one-half of the subjects are ignorant of their participation (21).
* It may be the case that in actual clinical practice, all of these studies using prerandomization obtain
consent from subjects randomized to all of the treatment arms. Accordingly, it is unclear whether
the single-arm design is used, or ever has been used. [I owe this empirical caution to Professor
Donald Marquis, University of Kansas.]
2. Expert Selection
Before conducting a course of clinical therapy under BRCT conditions, inves-
tigators can identify an initial probability distribution for subject outcomes
based on the reported prior clinical experience of patients similarly situated
and treated. Kadane (25) has suggested that, as trial data accumulate, this ini-
tial probability distribution should be updated and made more accurate through
regular review by a panel of experts. Over time, as improved therapies are devel-
oped, research goals can be refined and the detailed prognostic variables and
determinants better identified and firmly established. As the process continues
further, the experts’ opinions about the relative merit of each treatment option
become further refined, and computer models of the opinions are generated. Even-
tually, individual prognostic variables can be entered into a computer program,
and a model of all the experts’ opinions concerning treatment selection for a new
patient can be consulted before entering the trial. At this point, a treatment may
be offered to the patient if, and only if, at least one expert finds it is the best
treatment for that patient. Patients are intentionally not informed of the existence
of the experts’ opinions for fear that patients may (erroneously) favor one expert’s
opinions over another’s, or make their participation decision based on idiosyn-
cratic personality factors that may influence the statistical validity of the trial
results.
Critique. Failure to provide patients who are prospective research subjects
information concerning the opinion of experts regarding the therapy offered to
them diminishes the patient’s autonomy by withholding information potentially
relevant to the patient’s perspective (26). The experts’ opinions are directly
sought by the study designers for their own purposes, so it is reasonable to assume
that at least some patients also may feel these opinions are of material interest
3. Uninformed Consent
A third approach is to either autocratically decide not to inform prospective sub-
jects about the interim results of the different arms of the research study at all,
or to mention that the data will exist, but then ask them voluntarily to waive their
rights to lucidity and informed consent when they enroll in a clinical trial (27,28).
This approach eliminates the need for the patient and investigator to make several
important choices by eliminating all but one option for treatment, namely, the
one selected by random assignment. Interpretation of interim study data about
the relative efficacy of the two arms is prohibited, as is looking at interim study
data about reported side effects. All subjects enrolled in the trial are followed to
completion. Investigators using this approach must ask potential research sub-
jects, in effect, whether they wish to waive their rights to informed consent and
to give uninformed (or at least underinformed) consent to participating in a
BRCT. In dialogue, it might sound like the following:
We cannot conduct the best possible research if we have to tailor the therapy
according to individual patients’ personal preferences. Therefore, we would
like to ask you to agree to be treated according to the research protocol we
have designed and which you can review, but to await the results of the
completed study before making any treatment decisions that are in conflict
with the study being conducted.
As a result of this tension between serving the patient’s needs for personal
care (the PCI) and the study design’s needs for careful protocol adherence (the
SI), changes in the physician’s intention to treat the patient frequently arise after
the occurrence of certain events. Such events include partial adherence to proto-
col, withdrawal of subject participation, errors in following the protocol, changes
in the subject’s clinical state, changes in dosages in response to hematological
or other changes in the subject, and discontinuation of the study as a whole.
Rabeneck (1, p. 509) argues that changes in the physician’s intention to treat a
study subject is a problem in clinical research is because:
. . . [a]ny treatment changes that occur after randomization are ignored and
all outcomes are charged to the originally assigned therapy. This intention
to treat policy (ITT) protects the validity of the answer because it is assumed
that no adequate adjustment for bias is possible after the subjects have been
randomized to the treatment groups.
This [removal] policy attempts to answer the question of the effect of treat-
ment as taken [italics added). However, as with the ITT policy, it fails to
reflect the entire clinical experience in all patients and again limits the scope
of the trial’s results. . . . [But] when there are significant modifications of
the randomly assigned treatments, clinical treatment of patients is not ad-
vanced by either the policy of entirely excluding treatment changes from
This dilemma between the all-inclusive versus the all-removal policies ap-
pears to create an ineliminable conflict of interest for the BRCT physician-
researcher, and a conflict of interest, by its very nature, threatens the professional
integrity of the physician-researcher. Purtillo (30, p. 4) has noted that
At what cost to current research subjects’ welfare can scientific certainty and the
benefit to future patients be pursued ethically? Can we continue to conduct
BRCTs in the face of their assault on subject autonomy in deference to the scien-
tific imperative? On the other hand, can we afford not to conduct BRCTs in light
of their superior economic ability to establish statistical power and validity in
comparing different therapies’ benefits to subjects with the least overall risk to
current subjects?
Attempts to resolve the conflict between the obligations of the PCI and
those of the SI have been made from three different approaches, each without
success. First, resolving the conflict at the level of ethical theory is not possible
unless and until general agreement is reached on which overarching fundamental
ethical theory is best for purposes of clinical research: a duty- and rights-based,
patient-centered theory or a goal-based, outcome-centered theory. Prospects are
not promising for such an agreement to emerge in the near future. The second
approach, exemplified by Zelen’s prerandomization alternatives and Kadane’s
Bayesian study designs, do not resolve the conflict, but only supplant it with
other difficulties, most notably deception of subjects through nondisclosure of
research participation and the risk of misrepresentation of scientific uncertainty
to patients. The third approach, asking subjects to waive informed consent or to
provide underinformed consent to the risks inherent in BRCT participation, may
provide a solution at the theoretical level, but in practice, patients and physicians
do not behave as purely rational, intentionally uninformed agents. They do not
select medical treatment, especially in cases of serious illness, from behind a veil
Many more patients are treated off research protocols than are treated on them.
Taylor et al. (31) reported a 1993 study in which 80% of the study subjects
were enrolled by 10% of the more than 1700 physician members of an oncology
cooperative study group. All respondents of the study indicated they had a ‘‘sys-
tematic pattern of patient preselection for entry onto trials beyond the formal
inclusion/exclusion trial criteria.’’ Eighty-three percent defined randomization
and adherence to trial protocol as serious challenges to their ability to make indi-
vidualized treatment decisions.
The traditional view of the relationship between individual physicians and their
patients has been characterized by Katz (35) as follows:
Physicians have esoteric knowledge . . . which patients are incompetent to
understand . . . so patients must trust physicians’ altruism, and . . . allow
physicians authority over them.
1. Rabeneck L, Viscoli CM, Horwitz RI. Problems in the conduct and analysis of ran-
domized clinical trials: are we getting the right answers to the wrong questions?
Arch Intern Med 1992; 152:507–512.
2. Smith MA, Ungerleider RS, Korn EL, Rubenstein L, Simon R. Role of independent
data-monitoring committees in randomized clinical trials sponsored by the National
Cancer Institute. J Clin Oncol 1997; 15(7):2736–2743.
3. Walters L. Data monitoring committees: the moral case for maximum feasible inde-
pendence, Stat Med 1993; 12(5–6):575-580.
4. Koppelman L. Randomized clinical trials and the therapeutic relationship. Clin Res
1983; 31(1):1–25.
5. Miller B. Experimentation on human subjects: the ethics of randomized clinical tri-
als. In: VanDeVeer D, Regan T, eds. Health Care Ethics: An Introduction. Philadel-
phia: Temple University Press, 1987:127–159.
6. Beecher HK. Ethics and clinical research. N Engl J Med 1966; 274:1354–1360.
7. Passamani E. Clinical trials: are they ethical? N Engl J Med 1991; 324(22):1589–
1592.
8. Schafer A. The ethics of randomized clinical trials. N Engl J Med 1982; 307:719–
724.
9. Hellman S, Hellman DS. Of mice but not men: problems of the randomized clinical
trial. N Engl J Med 1991; 324(22):1585–1589.
10. Applebaum PS, Roth LR, Lidz CW, Benson P, Winslade W. False hopes and best
data: consent to research and the therapeutic misconception. Hastings Cent Rep
1987; 17:20–24.
11. Friedman B. Equipoise and the ethics of clinical research. N Engl J Med 1987; 317:
141–145.
12. Gifford F. Community equipoise and the ethics of randomized clinical trials. Bio-
ethics 1995; 9(2):127–148.
13. Gifford F. The conflict between randomized clinical trials and the therapeutic obliga-
tion. J Med Philos 1986; 11:347–366.
14. Friedman B. A response to a purported ethical difficulty with randomized clinical
trials involving cancer patients. J Clin Ethics 1992; 3(3):231–234.
15. Fried C. Medical Experimentation: Personal Integrity and Social Policy. New York:
American Elesevier, 1974.
16. Ellenberg S. Randomization designs in comparative trials. N Engl J Med 1984; 310:
1401–1408.
17. Smith M, Simon R, Cain D, Ungerleider RS. Children and cancer: a perspective
from the Cancer Therapy Evaluation Program, National Cancer Institute. Cancer
1993; 71:3422–3428.
18. Marquis D. Leaving therapy to chance. Hastings Cent Rep 1983; 6:40–47.
19. Marquis D. An argument that all prerandomized clinical trials are unethical. J Med
Philos 1986; 11:367.
20. Zelen M. A new design for randomized clinical trials. N Engl J Med 1979; 300:
1242.
Cheryl K. Fiedler
SCIREX Corporation, Hartford, Connecticut
H. Russell Searight
St. Louis University School of Medicine, St. Louis, Missouri
I. INTRODUCTION
The federal requirements for the type of information that must be present in an
informed consent document are presented above. While there are specific require-
ments regarding the content of the informed consent document, there is no accom-
panying mandate to ensure patient comprehension of the information. If the goal
of the informed consent process is to allow the patient to make a truly ‘‘in-
formed’’ choice, every effort must be made to assure patient understanding of
the information presented to him or her. Sorrell (6). has suggested that, to promote
subject comprehension, the informed consent document should be as brief as
possible, direct, not complex, and written at a grade level commensurate with
the subject’s reading ability (generally eighth grade or less). The information
must be presented, preferably by a professional directly involved in the research,
in a nonthreatening manner; the ideal is to present the information verbally to
supplement the written document. Patients must also be allowed to ask questions
(1,7).
Even when all the above measures are undertaken (which may be seldom),
research has shown that subjects’ memory of information provided to them during
the informed consent process is fragmentary; some information is retained while
other material is either never comprehended or forgotten, either completely or
‘‘selectively.’’ When this occurs, the question of whether informed consent has
been obtained is often raised. Ethicists would argue that, to be truly informed,
subjects must not only be given information at the outset of the study, but the
information must be retained and used on an ongoing basis for decision making
(7). Historically, research in the area of retention of information during the in-
formed consent process has focused on the subjects’ ability to recall or recognize
specific study information given to them during the consent process (7–11). This
research is typically conducted by a test, written or otherwise, that is given to
the subject at a predesignated time after signing the informed consent document.
While results may vary among studies, specific patterns have emerged.
When interviewing a large number of cancer patients, Cassileth et al. (12)
found the amount of information (as perceived by the study subject) contained
in the informed consent document was positively associated with scores on a
recall-type quiz. This test was given to the participants a short time after the
informed consent document was signed, and then patients were asked to rate the
amount of information as ‘‘just right,’’ ‘‘too little,’’ or ‘‘too much.’’ Patients
who perceived that the amount of information presented to them was ‘‘just right’’
had significantly higher scores on the test than did patients who perceived the
amount of information was either ‘‘too little’’ or ‘‘too much.’’
The degree of complexity of an informed consent document with respect
to reading difficulty level may also influence comprehension or understanding.
Consent form readability has been the subject of empirical study. Investigators
Institutional Review Boards focus their attention almost exclusively on the con-
sent form itself. Relatively little attention is devoted to whether subject recruit-
ment is actually carried out as described and whether protection of participant’s
rights follows established guidelines (1). For practical purposes, the informed-
consent document has become the focus for external review of a study’s ethical
appropriateness.
Oral presentations are a helpful adjunct to the written form but may not
be required. More brief consent documents with accompanying oral presentations
may be permitted under certain conditions such as when obtaining consent from
Our research suggests that well-educated clinical trial participants appear to com-
prehend about 70% of the standard consent material. Among the basic informed
consent dimensions, risks and possible adverse events may be less well retained
or comprehended (23). One strategy to improve subjects’ understanding would be
repeated testing with a structured interview or written test such as the Deaconess
Informed Consent Comprehension Test (DICCT). The DICCT is a 14-item struc-
tured interview test that assesses subjects’ comprehension of the informed con-
sent material. The questions are based on the informed consent elements included
in the Belmont Report (14). Subjects are instructed to respond in their own words.
Each DICCT question is scored on a 0, 1, 2 scale depending on the quality of
the reasoning demonstrated. Our early studies with the DICCT indicated that the
VIII. CONCLUSION
There is a growing body of research indicating that the ethical and educational
objectives of informed-consent are incompletely realized. Subjects do not appear
to retain and/or comprehend important aspects of the information that they are
given at a study’s outset. Our research with the DICCT suggests that about 30%
of relevant study information is not comprehended and/or retained over a 1-hour
period. More disturbingly, under current consent procedures, subjects do not
make the important distinction between personal medical care and their research
participation.
These findings leave a number of potentially disturbing implications for
the clinical investigator. How much knowledge about a study should a potential
research study be able to demonstrate? Is 70% adequate? 60%? 90%? Even if
screening tests were employed as a prerequisite to enrolling study participants,
the absence of a clear quantitative standard for adequate performance prevents
these measures from having practical utility. A process-oriented approach to in-
formed consent could address this issue. Repeated testing about a study’s pur-
poses, etc., with further review of failed areas for retesting until a specified crite-
rion is reached could better assure subject comprehension. For studies involving
multiple contacts, investigators, together with IRBs, could agree on a core body
of information that subjects must know at enrollment, with additional information
presented at later visits. This sequential presentation would likely optimize acqui-
sition of information.
From the perspective of enrolling the maximum number of subjects, inves-
tigators have probably benefited from the therapeutic misconception. If subjects
REFERENCES
1. Appelbaum PS, Lidz CW, Meidel A. Informed consent: legal theory and clinical
practice. New York Oxford University Press, 1987.
2. Rothrnan D. Strangers at the bedside. New York: Basic Books, 1991.
3. Beecher H. Ethics and Clinical Research. N Engl J Med 1966; 274:1354–1360.
4. Department of Health Education & Welfare: The Belmont Report. Washington, DC:
U.S. Government Printing Office, 1979.
5. Santelli JS, Rosenfeld WD, DuRant RH, Dubler N, Morreale M, English A, Rogers
AS. Guidelines for adolescent health research: a position paper for the Society for
Adolescent Medicine. J Adolesc Health 1995; 17:270–276.
6. Sorrell JM. Effects of writing/speaking on comprehension of information for in-
formed consent. Western J Nurs Res 1991; 13:110–122.
7. Hamilton MP. Role of an ethicist in the conduct of clinical trials in the United States.
Control Clin Trials 1981; 1:411–420.
8. Silva MD, Sorrell JM. Enhancing comprehension of information for informed con-
sent: a review of empirical research. IRB 1988; 10:1–6.
9. Taub HA, Baker MT. The effect of repeated testing upon comprehension of informed
consent materials by elderly volunteers. Exp Aging Res 1983; 9:135–
138.
10. Taub HA. Comprehension of informed consent for research: issues and directions
for future study. IRB 1986; 8:7–10.
11. Hassar M, Weintraub M. ‘‘Uninformed’’ consent and the wealthy volunteer: an anal-
ysis of patient volunteers in a clinical trial of a new anti-inflammatory drug. Clin
Pharmacol Ther 1976; 20:379–386.
12. Cassileth BR, Zupkis RV, Smith KS, March V. Informed consent—why are its goals
imperfectly realized? N Engl J Med 1980; 302:896–900.
13. Fry EA. A readability formula that saves time. J Reading 1968; 11:513–516.
14. Flesch R. A new readability yardstick. J Appl Psychol 1948; 32:221–233.
15. Grundner TM. On the readability of surgical consent forms. N Engl J Med 1980;
302:900–902.
16. Riecken HW, Ravich R. Informed consent to biomedical research in Veterans Ad-
ministration hospitals. JAMA 1982; 248:344–348.
Allen Cato
Cato Research Ltd., Durham, North Carolina
Susan L. Watts
Family Health International, Research Triangle Park, North Carolina
Lynda Sutton
Cato Research Ltd., Durham, North Carolina
Marlene E. Haffner
United States Food and Drug Administration, Rockville, Maryland
I. INTRODUCTION
Orphan products, which include drugs and biologics, are an often-neglected area
of clinical research. Orphan products are used to treat rare diseases or conditions
that by definition, affect fewer than 200,000 people (or up to 1 in 1300) in the
United States. Approximately 5000 orphan diseases have been identified, most
of which lack effective drug treatment (1).
The names of some rare diseases have become familiar. For instance, amyo-
trophic lateral sclerosis (ALS), commonly known as Lou Gehrig disease, is an
orphan disease affecting between 20,000 and 40,000 people in the United States.
ALS is one of the 40 neuromuscular diseases supported by the Muscular Dystro-
phy Association, the highly visible national organization behind the annual Labor
Day telethon for muscular dystrophies. Then there is Huntington’s chorea, a rare
disease that affects a population of 14,000 to 20,000 and became widely recog-
nized after claiming the lives of folksinger Woody Guthrie and boxer Izzard
Charles. Other well-known orphan diseases include Tourette’s syndrome (affecting
A. Varying Perspectives
The government, academicians, individual patients, and industry present different
perspectives regarding orphan drug development. The government—with a re-
sponsibility to maintain the public’s health and welfare—is ethically bound to
mandate that the patients with rare and debilitating diseases have access to effec-
B. Special Groups
Although orphan drugs conjure up images of isolated populations of suffering
patients overlooked by mainstream medical science, there are also three rather
large groups of patients for whom few drugs are being specifically developed:
children, the elderly, and pregnant women. Most prescription (and nonprescrip-
tion) drugs contain warnings against prescribing them for children; nonetheless,
many of these drugs are used to treat children. Such warnings—usually found
in the official labeling or package insert of the drug—are necessary because insuf-
ficient clinical study data in pediatric use prevent the sponsor from making appro-
priate recommendations (see Chapter 7).
Elderly patients also metabolize drugs differently than younger adults—
something for which drug formulations do not always account. Although children
and elderly patients are routinely treated with nonrecommended drugs, most preg-
nant women are advised not to take any drugs at all for the duration of their
pregnancy because of unknown risks of drug teratogenicity.
It is curious to consider that, in the face of medical and drug therapy ad-
vances, pregnant women have lost medications once employed to help them with
some of the more unpleasant and health-threatening side effects of their condition.
Heightened coverage by the media and the ensuing liability trials buried Bendec-
tin and thalidomide. Bendectin (doxylamine and vitamin B6) fell victim to nonsci-
entist jurors who had to make decisions that extended beyond their understanding
II. HISTORY
The first major attempt to address the special needs and inadequate resources for
particular patient diseases was the 1972, voluntarily initiated Department of
Health, Education and Welfare Interagency Committee on Drugs of ‘‘Limited
Commercial Value.’’ This committee outlined all the legal and insurance road-
blocks and recommended removing them. Furthermore, it identified the availabil-
ity of governmental and industry support and the need for incentives to stimulate
interest. It also recommended further study.
A new task force was convened in 1978 with broader representation and
considerable input from special-interest groups. The Office of Technology As-
sessment of Congress and a special commission appointed by the Secretary of
Health, Education and Welfare contributed additional data and clarification of
issues; in addition, professional staff in the executive office of the White House
participated.
These earlier efforts culminated in the Orphan Drug Act, which was initially
introduced in the House in 1981 by Representative Henry Waxman, then chair
of the Subcommittee on Health and the Environment of the House Committee
on Energy and Resources. After several amendments by both Houses, the bill
finally passed in December 1982 and was signed into law by President Ronald
Reagan in February 1983.
The images of philanthropy and the political advantages of offering orphan com-
pounds are often motivation enough for a company to pursue orphan drug re-
search and development. Because of the tremendous resources and expenditures
necessary for drug research, pharmaceutical companies tend to be the only facili-
ties capable of developing drugs. Therefore, these companies have an ethical
obligation to promote and support the research necessary for the development of
these compounds. Industry, unlike academia or private enterprise, has the unified
communications network crucial for the investigation, data collection, interpreta-
tion, and reporting of the information gathered during the characterization of a
new chemical entity (see Chapter 2). For industry to ignore this obligation, seek-
ing only the most profitable ‘‘Goliath’’ drugs, is unacceptable indeed, and poten-
tially shortsighted, as discussed previously.
Considering the incentives provided by the Orphan Drug Act to pharmaceu-
tical companies, it is surprising that more companies have not aggressively pur-
sued orphan drug development. This lack of activity may stem from a fear of
stigmatizing a drug as an orphan, thereby limiting its sales potential.
Orphan drugs, in theory, undergo the same rigors of testing and develop-
ment to which drugs with more widespread application are subjected. Pharmaco-
logical, safety, and efficacy parameters all undergo similar scrutiny, although
differences arise in determining therapeutic categories and dosing administration.
Because many orphan diseases are serious or life-threatening, development is
expedited (as is efficacy testing and data collection) and costs may be minimized;
as a result, the reduction in time and the tax credits afforded orphan drug research
make the development of drugs with orphan status appealing.
* The following Internet site provides a list of products currently designated as orphans by the Office
of Orphan Products Development: http://www.fda.gov/orphan/designat/list.htm.
The crucial problems in orphan drug development center on the quality of the
clinical research. Often too few patients exist to demonstrate efficacy statistically.
Patients are often dispersed throughout the country, and great efforts (and money)
VI. CONCLUSIONS
Orphan drug development is an emotional issue that attempts to even the odds
for the underdog. The success of the Orphan Drug Act has been far greater than
anticipated and continues to encourage manufacturers to develop such drugs.
Government incentives to stimulate research have been successful in afford-
ing those rare patients the hope many have sought for their disease. In the first
10 years after enactment of orphan drug legislation, the pharmaceutical industry
produced many new orphan drugs that helped more than 2 million Americans
suffering from rare diseases (11). Today, that number is more than 10 million.
Some few orphan drugs may be very profitable. Whether the motivations are
philanthropic, egalitarian, or profit-based, all sectors appear to have reaped re-
wards from orphan drug legislation, and the future should bring even greater
orphan drug availability and access to therapies. David now has a much better
chance of slaying Goliath.
REFERENCES
ADDITIONAL READING
Haffner ME, Kelsey JV. Evaluation of orphan products by the U.S. Food and Drug Admin-
istration. Int J Technol Assess Health Care 1992; 8(4):647–657.
Henry V. 1992 Le Tourneau Award. Problems with pharmaceutical regulation in the
United States. J Leg Med 1993; 14:617–639.
Nelson S. Irey†
Armed Forces Institute of Pathology, Washington, D.C.
I. INTRODUCTION
The analysis and evaluation of adverse drug reactions (ADRs) is a major problem
in the clinical practice of medicine, the development of new drugs, controlled
clinical trials, and the postmarketing surveillance period. Just as there are stan-
dards and requirements established as guidelines in the chemical, pharmacologi-
cal, and toxicological phases preceding the marketing of a new drug, there are
also guidelines for causality assessment of individual human cases in both the
premarketing and the postmarketing phases of new drug development.
This chapter presents an algorithm that is applicable to the individual: the
single-event ADR case. This methodology, relating to the use of therapeutic,
diagnostic, and prophylactic-type drugs in a clinical setting, should permit the
diagnostician to make one of three responses after an assessment of an ADR
case: an assured yes, a firm no, or a reasoned admission of uncertainty.
* The opinions or assertions contained herein are the private views of the author and are not to be
construed as official or as reflecting the views of the Department of Defense, the Department of
the Army, or the Armed Forces Institute of Pathology. The author is grateful to Phyllis S. Bojnowski
for her expert typographical support.
†
Deceased.
1. The drug ‘‘explosion’’ that has taken place in recent decades has added
a new dimension to the practice of modern medicine: an iatrogenic
category-the ADR.
2. The possibility of the occurrence and presence of an ADR should be
a constant component of the modern physician’s differential diagnosis,
along with the already well-established infectious, neoplastic, meta-
bolic, degenerative, and vascular groups of disease.
3. The clinicopathological picture presented by the ADR case is often not
readily distinguishable from non-drug-induced diseases. The clinical
and morphological findings of ADRs have the same limited number
of final common paths that characterize these other (non-drug) human
illnesses.
There are three major requirements for establishing the occurrence of an ADR:
3. Temporal Eligibility
The time factor in assessment of ADRs is a very important one and in some
cases is of critical diagnostic importance. This is true not only in establishing
‘‘eligibility’’ of the drug, but also in linking the drug to the reaction. On the
other hand, the time element may be equally important in denying eligibility and
also may make linkage of the drug with the clinicopathological picture a most
unlikely possibility.
It is quite apparent that a drug cannot be responsible for an ADR if the latter
is already in progress before the drug is first administered. This dys-synchrony is
illustrated in the following cases.
Example 1: A 50-year-old white male with a long history of rheumatoid ar-
thritis had a terminal illness of 5 weeks’ duration that was dominated by
cardiac and cerebral symptoms and signs. He had been receiving indometha-
cin, and a low platelet count had been one of the features of the laboratory
findings during this final 5 weeks. The case was diagnosed an indomethacin-
induced thrombotic thrombocytopenic purpura.
At autopsy, there were thrombotic vascular lesions of the small vessels
of the heart and brain that were compatible with this morphological diagno-
sis. However, the medication data in the patient’s chart revealed that the
indomethacin was not started until the beginning of the fourth week of his
terminal 5-week illness. Indomethacin was therefore ineligible to have initi-
ated his terminal illness and could not have been responsible for the throm-
botic lesions found at autopsy in the heart and brain.
4. Latent Period
Latent period refers to the time interval between the initial administration of the
drug and the onset of the ADR (in Fig. 1, it is the interval T0 to T1). The latent
period is not rigidly fixed or exactly predictable, but it tends to fall within certain
limits.
Characteristically, cyanide deaths occur in seconds to minutes. Most ana-
phylactic deaths occur within 20–30 min after contact with the lethal antigen,
while jaundice associated with chlorpromazine has its onset within 3 days to 3
weeks after the beginning of therapy. The fatal pancytopenia following chloram-
phenicol appears in 1–3 months, while hepatic angiosarcoma related to thorium
dioxide has a latent period of one to several decades. The ultimate in length of
latency is one to several generations from a drug-induced mutational germ cell
change to its manifestation in a conceptus.
Consideration of the latent period in an ADR is of use in an ADR assess-
ment in one of two ways: the latent period may be too long, or it may be too
short.
Example 1: An overlong latent period was noted in the case of an inadvertent
overdose of meprobamate given to a 4-year-old male child, who suffered
cardiorespiratory arrest 16 h later. During the intervening 16 h following the
400-mg dose there were no signs or symptoms indicating any problem. The
demise of the patient was associated with a sudden cardiorespiratory arrest,
with no premonitory symptoms or signs. Although meprobamate was initially
blamed for this cardiorespiratory death, it was thought to be an unlikely can-
didate because of the prolonged latent period: this drug usually reaches a
peak blood level within 2 h, and slowly declines over the next 10 h.
Further clinical information and the autopsy findings provided a more
likely cause of death. There had been a number of prior episodes of non-
drug-related cardiorespiratory arrests. Further, the autopsy revealed multiple
congenital anomalies of the central nervous system: hypoplasia of the cere-
bellum and spinal cord, polymicrogyria and pachygyria, and hydrocephalus
ex vacuo.
While the clinical history and autopsy findings showed a non-drug-
related cause for this death, the prolonged latent period alone would have
essentially disqualified meprobamate as the responsible agent.
1. Exclusion
Exclusion consists of selecting one drug from a group of drug candidates by the
use of the time flow chart.
Figure 3 Time flow chart demonstrating the temporal eligibility of stilbamidine to have
produced the elevation of the BUN in the case described. (From Ref. 7, by permission
of the publisher.)
Figure 4 Time flow chart illustrating the temporal eligibility of tetracycline in relation
to the subsequent renal failure (creatinine, dotted line; BUN, solid line). (From Ref. 7,
by permission of the publisher.)
This case met the ‘‘eligibility’’ criteria but failed to meet the ‘‘exclusion’’
part of the linkage criteria. Drug causation was eliminated because of the finding
of a concurrent co-morbid state (bilateral renal artery thromboses) substantiated
by a special radiological procedure and by the morphologic findings at autopsy.
The value of a broadened database is illustrated in this case and is generally
a requirement when attempting to analyze the frequently complex problems pre-
sented by ADR cases.
2. Dechallenge
The principle involved in the dechallenge method of linkage is that if there is a
reversible effect present, then removing the cause will eliminate the effect.
Example: A 56-year-old hypertensive and diabetic white male was treated
with 11 drugs for more than 5 years. About midway in this interval, he devel-
oped a progressive anemia and leukopenia.
Figure 5 presents curves on the disease markers (hemoglobin [Hgb]
and white blood cells [WBC]) and time-related drug administration informa-
tion on the 11 drugs with which he was being treated. The horizontal lines
indicate the dates the drugs were started, duration of administration, and
when they were discontinued.
Assessment was made in this case in the spring of 1977 (at the junction
of the solid lines with the dashed lines on the Hgb and WBC markers). At
that time, hydralazine was judged to be the most likely drug candidate, if
indeed the hematological changes were drug-related. The other 10 drugs were
eliminated for the following reasons: they were given for too short a time
(Lasix, Pro-Banthine, and Indocin), they were temporally ineligible (quinine,
Elavil, and chlorthalidone), or they were given continuously and over too
long a time with no evident ADR (chlorpropamide, propranolol, hydrochlo-
rothiazide, and Dimetapp). The one drug that was the most likely candidate,
from the time flow chart, was hydralazine, which showed temporal eligibility
and was given continuously during the drop in the hemogram elements.
On the basis of this graphical analysis, a presumptive diagnosis of
hydralazine-induced anemia and leukopenia was made, and it was recom-
mended that this drug be discontinued.
During the next 3 months, rises in the Hgb and WBC curves (shown
in the dashed lines of the graph) were noted. The final diagnosis of a
hydralazine-induced dyscrasia was based on the return toward normalcy of
the hemogram and on the failure to find a non-drug-related cause for these
hematological abnormalities.
3. Rechallenge
The principle involved in the rechallenge method of linkage of a drug to an ADR
is implied in the phrase post hoc, ergo propter hoc (after this, therefore because
of this). As applied, if a drug has been incriminated with a reaction, and the
ADR disappeared when the drug was discontinued, a rechallenge with this drug
followed by a return of the ADR would increase the probability that the drug
and the ADR were empiric correlates.
While intentional challenge is not often done, such a rechallenge may occur
inadvertently.
Example: A 64-year-old woman with rheumatoid arthritis, hypothyroidism,
and Raynaud’s disease was treated for 6 years with chlorambucil and predni-
sone. She died of progressive liver failure, pancytopenia, and necrotizing
bronchitis.
Figure 6 is a time flow chart on this case. The markers of the ADR
in the liver are serum glutamic oxaloautic transaminase (SGOT), lactic dehy-
While this case was chosen to illustrate the method of linkage by rechal-
lenge, it also illustrates two episodes of dechallenge. Further, since there is litera-
ture precedent for an association between liver damage and chlorambucil, this
case also demonstrates the fifth method of linkage (pattern) as listed in Fig. 2.
5. Pattern
The pattern method of linking a drug with an ADR shifts the focus of attention
to the clinicopathological findings in an ADR, and away from the identification
of the causative drug. This shift of emphasis is necessary when detailed time-
related drug and disease marker data are denied to the evaluator of the case. The
site-process profile may than be used as a guideline for searching past experience
and the literature for cases that have matching features. Matching features found
in the literature may include associations with certain drugs or chemicals, which
serve as a guideline for a focused examination of the patient’s history for the
causative agent.
Example: A 36-year-old white male was hospitalized after he developed jaun-
dice. A liver biopsy revealed a generalized centrilobular zonal loss of hepato-
cytes (zonal necrosis), central collapse of the lobular reticulum, and promi-
nence of Kupffer’s cells containing both lipofuscin and hemosiderin.
The morphological changes seen in this liver biopsy are consistent
This ‘‘pattern’’ method may also be used in excluding drugs. If the drugs
or chemicals suggested by the morphological findings are not identified or dis-
closed by historical or toxicological efforts, then the morphological changes ap-
pear to remain non-drug-related or non-chemical-related.
This qualitative procedure identified what was there; how much was there
was immaterial.
1. Liver dysfunction (as monitored by liver function tests) was evident fol-
lowing the second exposure to this anesthetic (multiple exposures to hal-
othane often precede liver damage).
2. Unexplained fever occurred on the third day after this second exposure
(a part of the profile of halothane-liver damage).
3. The centrilobular and universal zonal necrosis is consistent with the
findings in the liver associated with drug hepatotoxicity.
The liver damage and subsequent hepatic failure in this case was
judged as probably due to the halothane anesthetic. It was possible to arrive
at a reasoned choice of one of the three candidates discussed above because
there were adequate, detailed, and time-related data on the therapeutic agents
used, the adverse clinical circumstances, the liver function tests, and the tis-
sue changes reflecting the liver damage. The multiple methods utilized in-
clude latent period, temporal eligibility, exclusion, and morphological/
clinicopathological patterns.
C. Degree of Certainty
The third major task in analyzing and assessing ADRs is determining the degree
of certainty one has as to the causal relationship between the drug and the clinico-
pathological findings. Interposed between the definite causative and negative cat-
egories are three shades of certainty (probable, possible, and coincidental) that
titrate between these two extremes. These degrees of certainty are defined as
follows.
1. Causative: Cases in this class are those in which there is no doubt that
a drug has caused the reaction. This category is essentially limited to drug over-
dose cases or those cases in which the causative agent can be objectively identi-
fied (asbestos bodies; alpha tracks of thorium dioxide, etc.).
Parenthetically, the overdose cases with drug levels in lethal ranges
should have an important negative findings: no anatomical cause of death at au-
topsy.
2. Probable: This term is equivalent to the phrase ‘‘consistent with,’’ and
cases in this category of certainty fall short of the ‘‘causative’’ designation be-
cause they lack an objective and quantitative laboratory finding that is the sine
qua non of the causative category. Cases placed in this category have the follow-
ing characteristics.
3. Possible: Cases are put in this category when the relationship between
the drug and the clinicopathological findings can be neither confirmed nor denied.
There are three subdivisions in this category.
(a) Cases with potential causes other than the drug in question. The clini-
copathological picture could have been produced by the basic disease,
a co-morbid state, or by some other modality of therapy.
(b) Cases in which some of the criteria for eligibility and linkage have
been met, but some have not because of lack of adequate information.
Such a case could be put in this category temporarily while awaiting
more information or placed here permanently if it were evident that
further data would not be forthcoming.
(c) Cases that have met all the criteria of eligibility and linkage but for
which there is no known precedent literature. Such a case might be
a new and emergent ADR. It could be placed in the ‘‘possible’’ group,
awaiting the appearance of similar cases for cluster studies at a later
time.
lack of adequate information and the resultant inadequate data base for the assess-
ment of ADR cases.
III. CONCLUSIONS
REFERENCES
Allen Cato
Cato Research Ltd., Durham, North Carolina
Lynda Sutton
Cato Research Ltd., Durham, North Carolina
I. INTRODUCTION
Clinical trials, regardless of how well controlled and well designed, always face
unavoidable potential risks and unexpected events. Perhaps the most challenging
event is the unexpected death of a subject—the ultimate adverse effect (1). In
spite of such events, investigational drug trials must be performed, for no alterna-
tive way exists to test new drugs safely and effectively before they are brought
to the attention of the medical community. As stated in 1865 by Claude Bernard
(2), a nineteenth-century physiologist:
The principle of medical and surgical morality, therefore, consists in never
performing on man an experiment which might be harmful to him to any
extent, even though the results might be highly advantageous to science, i.e.,
to the health of others. . . . So, among the experiments that may be tried on
man, those that can only harm are forbidden, those that are innocent are
permissible, and those that may do good are obligatory.
A. Stage 1 Investigation
Stage 1 of the investigation begins by obtaining a preliminary history of events
surrounding the death. A telephone call between the medical monitor and the
investigator is usually conducted before any other actions are taken. From prelim-
inary data obtained in this telephone call, the medical monitor and the clinical
team can more readily initiate the specific tasks that must begin as soon as possi-
ble. The medical monitor may then ask the investigator to request a complete
autopsy, which can be very valuable in understanding the cause of death. Unless
the case is deemed a medical examiner’s case, however, autopsies, although often
requested, are difficult to obtain because the wishes of the deceased’s family
must be respected. Samples of blood and cerebrospinal fluid should be obtained
as soon as possible so that levels of drugs and metabolites can be measured.
Cerebrospinal fluid is important to examine because some drugs cross the blood–
brain barrier and some (e.g., bromide) do not. In addition, the investigator should
expedite a faxed or e-mail written summary of the death to the medical monitor.
If the test agent is still in Phase I clinical trials, all dosing of the test agent
at study sites should be temporarily discontinued. Even if a test agent is not
known, as in a double-blind, controlled study, the blind should be broken immedi-
ately, but only for that subject. Even if the blind reveals that the medication was
placebo, the investigation should proceed as though the medication were active
drug. This practice will ensure that the appropriate information is collected,
whether or not the patient is on placebo. It could be found later that an error in
packaging or dispensing of the medication, or related factors of the study itself,
including the study protocol, design, or site execution, are implicated in the death.
Also at this time, analysis of the clinical trial material should begin in order to
confirm the content (i.e., drug or placebo) of the clinical trial material taken by
the deceased. Therefore, clinical trial material should be immediately shipped
back to the sponsor or appropriate laboratory for analysis.
Having initiated the specific actions, the monitor or other qualified person
should go to the study site to obtain all available additional information by thor-
oughly investigating all of the records, including not only the medical chart, but
also nurses’ notes, pharmacy records, and clinic charts. Information from the
subject’s history, including peculiar events such as experimental therapy that oc-
curred before their entry, will often reveal facts that may now be relative to the
gators at other sites should be informed of the preliminary results of the investiga-
tion into the cause of death. Table 2 summarizes these actions.
In our anticonvulsant case example, all patients in this study were known
to be taking the test agent, so the cause of death had to be determined quickly.
All steps in Stage 1 of the investigation were performed, with the following
results:
Therefore, the preliminary decision was that the death was probably not drug
related. Thus dosing could continue, but no new patients could be entered. The
other study investigators were informed, and patients at the study site were told
of the death and given the option to discontinue. Of the patients hospitalized at
the site, only one chose to remain in the study.
B. Stage 2 Investigation
Stage 2 of the investigation lasts 1 week to 2 months and involves a thorough,
intensive review of all information. The following actions are taken:
Obtain and review final results of the autopsy, blood levels, and cerebrospi-
nal fluid.
Compile and review any available data from other study sites.
Conduct a more intensive review of the subject’s records and data collec-
tion forms.
Review the literature for any epidemiological factors that may be relevant
to the diseases or the population being studied.
Contact outside consultants.
Notify the Food and Drug Administration (FDA) of the death as soon as
possible, but within 3 days of the incident.
Table 3 summarizes the actions to be taken at this point in Stage 2 of the investi-
gation.
Having obtained all results, the determination can now be made as to the
relationship of the death to the test drug. Assuming the final decision is that the
death was not drug related, some specific actions must be taken. The investigator
can now be informed to allow an increase in dosing, permit new subjects to
enter the study, and continue the study according to protocol. Also, all other
investigators should be notified of the results and reminded to inform their Institu-
tional Review Boards. The FDA should be notified, and sent a complete written
report of the death. This report should include all available information on the
subject, including data collection forms, autopsy report, investigator’s report,
clinical trial material analysis report, pharmacist’s report, and monitor’s report
(Table 4).
V. PLACEBO-RELATED DEATHS
VII. CONCLUSIONS
REFERENCES
David F. Bernstein
Cato Research Ltd., Corona del Mar, California
I. INTRODUCTION
In the context of clinical research, the preparation of clinical trial material (CTM)
satisfies the short-term objective of providing the correct, blinded drug product
(including various doses of the investigational drug, matching placebos, and
matching competitive products) according to a randomized medication allocation
scheme for a specific clinical study. As such, investigational products and their
packaging, labeling, and assembly into patient kits are customized for each clini-
cal study.
In the context of overall drug development, the preparation of CTM contrib-
utes to the evolving scientific database that is the basis for a New Drug Applica-
tion (NDA). This satisfies a long-term objective of product approval from the
chemistry, manufacturing, and controls (CMC) perspective. Often, the prepara-
tion of ever-increasing amounts of a variety of investigational drug products can
identify manufacturing or processing problems that can be corrected in future
batches. Stability data on CTM (which is required to support the specific clinical
study) can be used to identify potential commercial packages and, in selected
cases, can be used as supportive stability data for registration.
Since both clinical research and CTM preparation require a lengthy, multi-
year development process, the technical issues surrounding the clinical supply
chain are themselves evolutionary. The greatest challenge in drug development
from the CMC perspective is that CTM preparation is a miniproduction operation
conducted in a research and development (R&D) environment. The principles
and practices for manufacturing, packaging, and labeling that are well established
The visible portion of the investigational supply process begins with the ordering
of drug supplies and ends with shipment of investigational products to the clinical
site. This is a sequential process that is not amenable to conducting operations
in parallel with the intent of saving time. These sequential processes require a
well-developed scenario, since each investigational supply project has a defined
cascade of activities. This cascade of activities needs to begin with excellent
communication of the overall clinical plan to clinical supply scientists and ends
with post-use accountability, reconciliation, return and destruction of CTM.
The preparation of CTM is customized in response to specific clinical pro-
tocols. The specialized nature of CTM operations does not lend itself to mass
production. Consequently, there is no clinical supply drug store that could imme-
diately supply the wide variety of investigational drug products in the wide vari-
ety of packages that are typically employed in clinical research. Packaged con-
tents (e.g., 30 capsules, 5 mL) are selected based on dosing frequencies and the
frequency of patient visits to clinical sites. Label copy, label text, and label sizes
are all customized and are subject to strict approval mechanisms, control and
separation of labels and inventory requirements. Global multicenter studies re-
quire language-specific labeling that addresses both domestic and international
regulatory requirements. Patient kits are not standard items; a 2-week, 6-week,
or 3-month supply of CTM would require three different-sized patient kits.
The investigational supply process is understandably time-consuming. It
would be nice if attempts to truncate time could be associated simply with in-
creased cost; however, projections of time frames are initially aggressive, and
attempts to shorten them often compromise quality. Any compromise of quality
during the preparation of lookalike drug products which are associated with spe-
cific patients can, and has, compromised an entire clinical study. Since it is diffi-
cult to detect mixups in even the best situation, sponsors often ask, ‘‘Are you
sure you packed my CTM correctly?’’ when the goals of a clinical study are not
Clinical supply projects are complex; they involve numerous separate functions
(manufacturing, packaging, labeling, testing), all of which usually follow differ-
ent departmental priorities. When multiple contractors are involved, the concept
of just-in-time clinical supplies is nice theoretically, but it is almost impossible
to achieve in practice.
Clinical supply projects that have had the benefit of parallel conceptualiza-
tion of the CMC program, the clinical program, and the resultant CTM program
have typically been completed in realistic, attainable, yet challenging time
frames. Defining any clinical supply project requires an early alert of the overall
clinical program so that bulk API or manufactured drug product requirements
and their lead times can be anticipated.
Since even draft or final clinical protocols often overlook many clinical
supply concerns (e.g., the frequency of CTM dispensing, overages needed to
account for missed clinical visits, packaging options to promote patient compli-
ance), these issues must be explored by alternative mechanisms. An early interac-
tive discussion among clinicians, statisticians, and clinical supply scientists can
identify and address the majority of issues that are inherent and unique to each
clinical study. In this way, trade-offs and options can be explored and the ‘‘best
way’’ and best time frame can be agreed to by all. Then, as the clinical protocol
is being finalized, clinical supply operations create a definition document. This
protocol covers all manufacturing, packaging, labeling, assembly, testing, and
distribution activities and identifies the responsibilities for and the details of each
of these activities.
Ordering CTM should occur in two stages, bulk and specific packaged product,
reflecting the lengthy lead times for manufacturing and the shorter lead times for
customized packaging activities.
Bulk supplies (the number of capsules, tablets, liters of liquid for an in-
jectable product) for the immediate clinical study, as well as the entire
clinical program, need to be estimated and manufactured. These projec-
tions need to include different dosage forms for bioavailability studies,
V. MANUFACTURE OF CTM
Investigational drug products are often manufactured to supply CTM for specific
clinical trials. The preparation of CTM is conducted under the umbrella of evolu-
tionary CMC activities that are directed at developing increasingly rugged and
sophisticated formulations and manufacturing processes. At some point in devel-
opment, CMC activities that have focused on CTM preparation need to be redi-
rected to supply the scientific and technical database for marketing registration
documents and commercial manufacturing.
An example of this evolutionary dosage form development process would
involve the development of a rudimentary, ‘‘simple’’ drug plus lactose hand-
filled capsule suitable for Phase 1 studies. Manual capsule operations can produce
500–1000 capsules per hour. As clinical requirements become larger, it will be
necessary to incorporate additional formulation ingredients (starch, lubricants)
that would be amenable to a semi-automatic encapsulation machine with its out-
put of 4000–8000 capsules per hour. The next variation suitable for expanded
Phase III studies and commercialization would involve high-speed encapsulation
(1000–1500 capsules per minute) and require the incorporation of additional for-
mulation ingredients (binders, disintegrants). If all three variations of this capsule
dosage form were to be developed, each would require preformulation studies
to determine drug–excipient compatibility and formulation studies to determine
VII. LABELING
Label is defined as the information and directions printed on the immediate con-
tainer; labeling refers to all other printed matter. In the United States, the only
legal requirement for label information is the ‘‘Caution—for investigational use’’
statement (21 CFR 312.6). However, other information is usually contained on
a label; some of this information is referred to in 21 CFR 201. The following
are usually included on CTM labels:
Name of sponsor.
Patient number or kit reference number.
Directions for use: For outpatient studies, specific patient directions are
recommended to promote patient compliance. Directions to refer to the
protocol are not helpful, since CTM dispensers and the pharmacy dis-
pensing the medication do not usually have access to the protocol.
Code number, which is a reference to the specific assembly operation. Since
this number must be the same for all products in a clinical trial, traceabil-
ity from an assembly operation to the specific batches of each product
used must be assured.
Interactive Voice Response System. The central office will then look up the drug
assignment for Patient 101 (which is active) and inform the site to dispense Kit
521. Similarly, the second patient at that site would be assigned Kit 244. Both
the clinical site, and central office document that Patient 101 received Kit 521.
This technique minimizes the CTM required at any clinical site, where space for
CTM is often at a premium, and reduces the potential for unused CTM due to
incomplete enrollment.
Assembly of CTM into patient kits is also known as secondary packaging, since
the actual drug product is not exposed to environmental conditions during this
operation. The use of patient kits facilitates dispensing by the clinical site, usage
by the patient, and return goods accountability by the clinical monitor.
The specific configuration of a patient kit depends on the frequency of
dispensing visits during the clinical trial. Examples of the contents of patient kits
are provided in Figs. 1 and 2.
Patient kits are often used even for ‘‘simple’’ Phase I studies involving
a less than simple dispensing scheme. In a placebo-controlled, ascending-dose
tolerance safety study, the no-effect dose of 10 mg was followed by dose escala-
tion to 20, 40, 80, 160, and 240 mg. Two cohorts of nine subjects (six on active,
three on placebo) would alternate doses so that each would take three doses.
Since it was impractical to develop seven different formulations (the six different
drug doses plus placebo), capsules containing placebo and 10 and 40 mg of drug
product were developed. Each subject took six capsules; depending on the dose
and the random allocation to active or placebo, the subject would need to take
Dose-level boxes that contained nine bottles of CTM, one bottle for each
patient could be prepared. In this option, six dose-level boxes, each con-
taining nine bottles and each bottle containing six capsules, would be
required. Label information would require that the dose level and patient
number be clearly highlighted. Since subjects for the initial 10-mg dose
level would be enrolled over a very short time period and all nine subjects
dosed on the same day, the dose-level box option facilitated the dispens-
ing process at the clinical site.
Subject kits that contained only the three bottles of CTM required for each
subject could be prepared. Eighteen subject kits, each containing three
bottles and each bottle containing six capsules, would be required. This
option provided greater control over specific subject CTM, but would
Number capsules/bottles
IX. TESTING
The release of CTM for distribution involves procedures that are similar to those
required for commercial products. However, basic differences between commer-
cial products and CTM require additional attention due to the following.
Once CTM is prepared into patient kits, shipment to clinical sites can be dictated
by patient enrollment. For multicenter studies, initial supplies can be used to open
a clinical site where resupply frequency is dependent on actual patient enrollment
at any given site.
Shipment of CTM should ensure that the drug products are protected
against exposure to undesirable environmental conditions. Often, early in the
development process, the absolute sensitivity of a drug product to excursions
outside acceptable ranges of temperature, humidity, or shear are unknown. In the
cases where drug is very expensive or in very limited supply, trial shipments to
a variety of dummy locations are made using placebo drug product where actual
environmental conditions are measured by markers, recorders, or high-tech de-
vices involving computer chips. These techniques are highly recommended for
biologics, proteins, or topical semisolid products, where freezing or refrigerated
conditions must be maintained or avoided or where freeze/thaw or refrigerated/
room-temperature cycling would be detrimental to product integrity.
Shipment by the sponsor is usually accompanied by a shipment/receipt
form intended to be completed by the clinical-site pharmacist or CTM study
coordinator and returned to the sponsor. This form represents the point in the
chain of custody of CTM which is at the interface of GMP (21 CFR 211.150)
and GCP (21 CFR 312.57, 312.59, and 312.62) requirements.
XI. RETURNS
In recent years, there has been a growing trend for outsourcing of various portions
of the investigational supply process. This need has resulted from the growth of
the large number of small, emerging firms that do not have facilities or the internal
expertise for manufacturing, packaging, or testing of CTM, coupled with the
trend for large pharmaceutical companies to out-source portions or all of their
clinical supply chain.
This trend can be seen by examining any list of contract services; there are
now over 12 formulation groups that focus on the preclinical development of
divergent dosage forms; over 25 analytical laboratories that can develop new
analytical methods, especially those for complex proteins; 7 manufacturing orga-
nizations devoted solely to CTM (not commercial) manufacturing; and 8 contract
packagers dedicated to clinical supply packaging and labeling.
Most contractors, especially those dedicated to clinical supply operations,
do a good job. Their effort can be universally measured by how well the project
is defined. Virtual companies that do not possess internal CMC/CTM expertise
are particularly vulnerable, since simply giving a contractor a draft clinical proto-
col and short time frame may result in a CTM configuration being barely adequate
and not meeting the overall project needs. Such firms should enlist the services
of competent clinical supply consultants who can assist them in the definition
process, suggesting the best approaches, and monitor contractor operations.
Since CTM operations are customized, contingency plans should be devel-
oped and changes should be anticipated because even the best definition docu-
ment may not address all the realities that appear during actual production opera-
tions. This is easily done by having a person in the plant during all critical CTM
activities; any deviations or unanticipated contingencies can then be addressed,
and client input and approval can be obtained immediately.
XIII. CONCLUSION
The preparation of CTM under GCP, GMP, GDP, and GSP represents good busi-
ness practices, since quality investigational materials need to meet both short-
term (i.e., specific clinical study) and long-term (i.e., NDA) objectives. As clinical
research progresses from Phase I to Phase III, early rudimentary investigational
drug products are replaced by sophisticated, fully evaluated formulations and
manufacturing procedures. Since changes to the formulation and manufacturing
Richard Granneman
Abbott Laboratories, Abbott Park, Illinois
I. INTRODUCTION
II. FUNDAMENTALS
V ⫽ 7.5 ⫹ 7.5 ⋅ f u ⫹ 27 ⋅
冢 冣
fu
f uR
(4)
D. Clearance
Expressed as volume/time, CL refers to the volume of plasma from which the
drug is totally removed over a specified period of time. The total CL of a drug
is a combination of individual clearances by any mechanism such as biotransfor-
mation, renal excretion, or biliary excretion. Clearance is one of the most impor-
tant of the pharmacokinetic parameters because it determines the AUC for a given
dose, and along with the distribution volume, determines the terminal-phase half-
life. In particular, clearance provides significant insight when compared to the
blood or plasma flow to the eliminating organs.
F. Pharmacodynamics
While pharmacokinetics describe the processes and rates of drug movement from
the site of absorption into the blood, distribution into the tissues, and elimination
by biotransformation or excretion, pharmacodynamics describes the drug effects
once it has reached the site of action (58). Typically, pharmacodynamics are
described by the parameters estimated from a mathematical model which may
have a biological or empirical basis. Pharmacokinetics and pharmacodynamics
are often modeled simultaneously because estimation of pharmacodynamic pa-
rameters using observations of drug effects in vivo requires a description of drug
concentrations at the time of the effect. A commonly used pharmacodynamic
model is the E max (or sigmoid E max) model. Concentration is the link between the
pharmacokinetic model (PK), relating dose to concentration, and the pharmaco-
dynamic model (PD), relating concentration to effect.
For drugs that cause an effect that is easily measured (e.g., change in blood
pressure), PK/PD modeling can provide insight to various factors influencing the
effects. Once the concentration–effect relationship has been defined, pharmacoki-
netic changes can be used to predict the change in intensity of a specific therapeu-
tic effect or adverse event. Predictions based on pharmacokinetic changes can
help optimize experimental designs and potentially can reduce the number of
studies conducted to improve the speed of drug development. In addition, the
results of PK/PD modeling can help to chose an appropriate dosage form and
regimen with respect to both efficacy and safety (59). Unfortunately, for many
drugs (e.g., anticancer drugs), PK/PD modeling is difficult at best because of the
nature of the efficacy data. However, PK/PD modeling of toxicity data may be
useful.
Metabolism
Protein binding
Formulation
In-vitro testing methodology
Analytical methodology
Bioavailability/bioequivalence information (including food effects)
Pharmacokinetic parameter estimates
Dose proportionality
Special populations
Drug–drug interactions
PK/PD relationships
Population PK/PD analyses
Support of labeling
A. Phase I
New drugs should be studied in people as early as safety allows. The initial
introduction of a new drug into subjects are the first Phase I studies. These studies
are closely monitored, typically conducted in healthy adult men, and are designed
to obtain information on the safety, pharmacokinetics, and pharmacological ef-
fects of the drug. In addition, the metabolic profile, adverse events associated
with increasing doses, and evidence of efficacy may be obtained. The primary
goal of Phase I studies is to demonstrate safety in humans while collecting suffi-
cient pharmacokinetic and pharmacological information to permit the determina-
tion of the dosage strength and regimen for Phase II studies. Because most com-
pounds are available for initial studies as an oral formulation, the initial
pharmacokinetic profile usually includes information about the absorption phase.
Additional studies, such as drug–drug interactions, assessment of bioequivalence
of various formulations, or other studies involving normal subjects are included
in Phase I.
Generally, the first study in humans is a rising, single-dose tolerance study.
The initial dose may be based on animal pharmacology data, such as 10% of the
no-effect dose. Doses are increased gradually according to a predetermined
B. Phase II
After the initial introduction of a new drug into humans, Phase II studies focus
on efficacy while using the pharmacokinetic information obtained in Phase I stud-
ies to optimize the dosage regimen. Phase II studies are not as closely monitored
as Phase I studies, and are conducted in patients. These studies are designed to
obtain information on the efficacy and pharmacological effects of the drug in
addition to the pharmacokinetics. Additional pharmacokinetic and pharmacologi-
cal information collected in Phase II studies may help to optimization the dosage
strength and regimen, and may provide additional information on the drug’s
safety profile (e.g., determine potential drug–drug interactions).
Efficacy trials of the drug should not to be initiated until the MTD has been
defined. In addition, the availability of pharmacokinetic information in healthy
volunteers is key to the design of successful efficacy trials. The clinical pharmaco-
A. Demographics
1. Body Size and Composition
Several factors relating to body size and composition often affect clearance and
apparent distribution volume (hence half-life) of a drug. Weight, gender, and age
are often confounded because early pharmacokinetic studies during drug develop-
ment generally require a narrow age range of subjects, exclude women, and stipu-
late that subjects have a body size within ⫾10% of their ideal body weight based
on their height. Population pharmacokinetic analysis of data from Phase III stud-
ies which include women can be a powerful source of information because the
number of patients studied is much greater and the ranges in body size and age
are larger than in previous studies.
Body size is often correlated with drug CL because CL is often correlated
with liver size, which is related to overall metabolic capacity. The relationship
2. Gender
Although gender-dependent differences in the pharmacokinetics of many drugs
have been demonstrated, these differences are often minor. The recent review by
Harris et al. provides comprehensive details (64). Apparent gender differences
in clearance are usually not as pronounced when corrected for weight or BSA
differences. However, for some drugs, relatively small mean differences in
plasma concentrations can mask large differences for some individuals, thus un-
derstating the individual effect on efficacy or toxicity. There is no a-priori basis
for the assumption that pharmacological effects would be the same in men and
women.
The sources of nonanthropometric differences due to gender are diverse,
ranging from absorptive effects attributed to differences in gastric secretion and
gastrointestinal motility (65,66) to effects of estrogen on metabolic activity (67–
70). In addition, the inhibitory effect of ethinylestradiol on CYP3A and CYP1A2
activity (17) demonstrate the potential importance of obtaining information about
the effects of menopause and the use of oral contraceptives for population phar-
macokinetic analysis. Thus, while early Phase I and sometimes Phase II studies
typically have been conducted with young healthy men, women are being in-
cluded in studies sooner for drugs intended for use in both men and women.
4. Age
In the past, new drugs were studied almost exclusively in adults due to safety
concerns. Today, there is increased emphasis by regulatory agencies on the col-
lection of pharmacokinetic information for children and elderly populations.
Again, due to safety concerns, it is vitally important to combine the pharmaco-
kinetic information obtained for adults with the knowledge of the potential
age-dependent changes in the pharmacokinetics of individual drugs to chose an
effective and safe dosage for children and elderly. Age-dependent changes in
pharmacokinetics may be the most complex source of variability. The functioning
of many different organs may change with age. Also, protein binding, percent
lean body mass, volume of extracellular fluid, and other characteristics of individ-
uals may change with age.
In general, renal function is lower in neonates and the elderly compared
with young adults. Similarly, hepatic function typically is lowest in newborns,
improves with age (most efficient in children and young adults), and declines
during aging (71). In addition, phase I reactions (e.g., oxidation) generally occur
at a slower rate in newborn as compared to adults; cytochrome P450, NADPH,
and cytochrome c reductase activities are approximately one-half the adult values
(72).
Unfortunately, the heterogeneity of the CYP system precludes the use of
total CYP enzyme content as a meaningful predictor of the rate at which an
individual substrate is metabolized. Both the absolute amount and the relative
proportion of different isozymes may be age-dependent. The fact that enzymatic
activities involving various substrates are influenced by gestational age to differ-
ent extents and follow different maturational patterns (73) may be due to age-
dependent changes in isozymes as opposed to total CYP content.
5. Disease States
The two standard approaches to evaluating the effects of disease states on the
pharmacokinetics of a new drug are controlled Phase I evaluations, and Phase II–
III population pharmacokinetic approaches. The latter has the benefit of greater
heterogeneity as well as reflecting actual usage in the target population. Drug
disposition may be altered in many disease states, particularly renal or hepatic
diseases for drugs that are primarily eliminated by those routes.
Dysfunction of the major organs of drug elimination may lead to reduced
drug clearance and increased t 1/2 . If clearance is reduced with no change in the
volume of distribution, the dose may be similar to that of a healthy person, but
the interval between doses may be increased to provide similar systemic exposure
while preventing accumulation of the drug to toxic concentrations. Thus, the
pharmacokinetics of new drugs often are determined in patients with renal or
hepatic impairment. The design of these studies is dependent on the sources of
clearance and routes of elimination of the drug.
The typical study conducted to assess the effects of renal impairment on
the pharmacokinetics of the drug includes subjects with end-stage renal disease
who require hemodialysis, subjects with various levels of renal function, and
subjects with normal renal function. Because subjects who require hemodialysis
have virtually no renal function, interpatient variability should be relatively low
and meaningful information can be obtained from a few (4–6) subjects. Informa-
B. Drug Interactions
In many disease states, the use of polytherapy is quite common, and the risk of
drug–drug interactions is high, both from pharmacokinetic and pharmacody-
namic perspectives. The likelihood of drug interactions and semiquantitative esti-
mates of magnitude may be predicted from in-vitro data (17). The potential for
interactions needs to be evaluated from two perspectives: (a) the potential that
the new drug may affect the kinetics of other drugs, and (b) the potential that
other drugs may affect the kinetics of the new drug. The former generally depends
on the ability of the new drug to affect various enzyme- and carrier-mediated
C. Environmental Components
Another factor contributing to intersubject pharmacokinetic variability is the en-
vironment, in particular, alcohol, tobacco, and diet. The effects of alcohol are
complicated because alcohol may either inhibit or induce drug metabolism. To-
D. Genetics
Genetic factors may have the most substantial impact on intersubject variability
of the hepatic CYP mono-oxygenases (97). The ability to predict the rate of drug
metabolism in individuals depends on the ability to gauge the activity of the CYP
enzymes in various human subpopulations. Special attention may be required for
drugs in which a polymorphic enzyme such as CYP2D6 or CYP2C19 is responsi-
ble for a large fraction of total clearance. In these cases, the magnitude of differ-
ence in clearance between PMs and EMs of probe substrates or genotyped indi-
viduals should be assessed. Inevitably, the question will arise as to whether PM
status confers greater efficacy or increased rates of adverse events under fixed-
dosage study designs.
The current technology allows genotyping for CYP mutations at a reason-
able cost. Proponents of genotyping claim that knowledge of genotype predicts
clearance which would allow prospective dosage adjustment for better control
over therapy. While this may be true for the ‘‘average’’ patient (i.e., population
4. Enterohepatic Recirculation
ER is a recurrent cycle in which compounds are excreted by the liver as compo-
nents of bile into the intestine, reabsorbed from the gastrointestinal tract into the
systemic circulation, and reexcreted after reuptake by the hepatic cells. For some
drugs and compounds such as bile acids, no change in the structure of the com-
pound occurs during recirculation. For drugs such as valproate, the glucuronide
metabolite is excreted in the bile, and the parent drug is regenerated by hydrolysis
of the glucuronide by β-glucuronidase before being reabsorbed. The process of
ER would be expected to increase the time during which the compound is present
in the systemic circulation and possibly increase the duration of effects. For com-
pounds that are toxic or are metabolized within hepatocytes to toxic metabolites,
enhanced toxicity could result from efficient ER.
Glucuronidation is one of several critical steps in the process, and a suitable
functional group (i.e., phenols, alcohols, and carboxylic acids) must be present
in order for glucuronidation to occur. As an example, glucuronidation of val-
proate occurs at the carboxylic acid group. Other compounds eliminated in the
bile as glucuronides that undergo ER include morphine, DDT, diphenylacetic
acid, mefanamic acid, and the nonsteroidal anti-inflammatory drug carprofen
(105).
Changes in the extent of ER (e.g., age-related changes in glucuronidation)
V. POPULATION ANALYSIS
REFERENCES
Angela Cahill
Bethesda, Maryland
I. INTRODUCTION
This chapter provides an overview of the process used in submitting New Drug
Applications (NDAs) and Biologics License Applications (BLAs) to the U.S.
Food and Drug Administration (FDA) and provides methods for organizing infor-
mation in the marketing application.
A. What Is an NDA/BLA?
A marketing application, as defined in Title 21 of the Code of Federal Regulations
(hereinafter referred to as 21 CFR) § 312.3(b), is an application for a new drug
submitted under section 505(b) of the act or a biologics license application for
a biological product under the Public Health Service Act. The ultimate goal of
clinical development of a new medicinal product is to obtain regulatory agency
authorization to sell and market the product. In the United States, the Food and
Drug Administration must approve the marketing application (i.e., the NDA or
BLA) before the new medicine can be marketed.
The marketing application is meant to provide evidence that the product
is both safe and effective when used under specified conditions and manufactured
appropriately and, therefore, provides all the information that supports each state-
ment in the proposed labeling for the product. Whether paper or electronic, from
the United States or of international origin, the marketing application consists of
many pages of documentation that support one to two pages of product labeling
(i.e., the Package Insert or the foreign Summary of Product Characteristics, and
labels on the containers) that the general public and physicians read and use.
* FDA classification policy [Center for Drug Evaluation and Research (CDER); http://www.fda.gov/
cder/da/da.htm#notes] includes:
* In accordance with the definition provided in 21 CFR § 314.3(b), listed drug status is ‘‘evidenced
by the drug product’s identification as a drug with an effective approval in the current edition
of FDA’s ‘Approved Drug Products with Therapeutic Equivalence Evaluations’ (the List) or any
supplement thereto, as a drug with an effective approval. A drug product is deemed to be a listed
drug on the date of effective approval of the application or abbreviated application for that drug
product.’’
†
In accordance with the definition provided in 21 CFR § 314.3(b), right of reference or use refers
to the ‘‘authority to rely upon, and otherwise use, an investigation for the purpose of obtaining
approval of an application, including the ability to make available the underlying raw data from
the investigation for FDA audit, if necessary.’’
* Monographs establish conditions under which OTC drugs are generally recognized as safe and
effective and are not misbranded.
1. Paper Copies
The paper review copies contain copies of Form FDA 356h, cover letter, letters
of authorization/certification items, comprehensive table of contents (index),
summary section, and particular technical sections. The review copies are distrib-
uted to the individual reviewers at the FDA, thus permitting concurrent reviews
to occur. The field copy, required for drug products as of 1993, is a copy of the
Chemistry review copy, and certified as such, that is sent to the applicant’s
‘‘home’’ FDA district office and used for preapproval inspections. The archival
copy is the FDA’s official copy (it includes the case report forms and tabulations)
and serves as a resource for reviewers.
The Guideline on Formatting, Assembling, and Submitting the New Drug
and Antibiotic Applications (1987) includes a tabular summary for the compila-
tion of the archival and review copies of the new drug application. It should be
noted and accounted for, however, that the tabular summary in this guidance is
not consistent with the updated Form FDA 356h.
2. Electronic Copies
The FDA plans to move to a paperless regulatory process by the year 2002. As
part of this goal, CDER and CBER have issued guidance documents for the
electronic submission of NDAs [Guidance for Industry: Providing Regulatory
Submissions in Electronic Format—NDAs (1999)] and BLAs [Guidance for In-
dustry: Providing Regulatory Submissions to CBER in Electronic Format—Bio-
logics Marketing Applications (1999)]. A general guidance document provides
a comprehensive review of the history of electronic submission guidances and
procedures [Guidance for Industry: Providing Regulatory Submissions in Elec-
tronic Format—General Considerations (1999)]. The advent of the electronic
submission should offer advantages to the FDA, the applicant, and the environ-
ment. An average archival submission is approximately 100–150 volumes (each
volume is 2 in. thick), with review copies of technical information adding many
more volumes to this count. The applicant has to coordinate this enormous effort
with the paper copy; the FDA has to coordinate the distribution, review, and
storage of the paper copy.
The guidance documents for electronic submissions provide detailed speci-
fications and information on format. Key factors in submitting the electronic doc-
ument are as follows.
F. Certification
Form FDA 356h includes a certification statement that must be signed by the
applicant’s representative agent or authorized official. The statement specifies
that there are no false statements in the application and that the applicant will
adhere to requirements for compliance with regulations for safety updates, GMPs,
labeling, prescription advertising, supplements, and annual reports.
sults, overall tabular summaries of clinical studies and nonclinical studies, and
chemistry documentation.
Although each application presents unique considerations, our experience
indicates that submission of a marketing application will be delayed significantly
if the applicable information is not incorporated immediately after completion
into a report, summary tables, and the proposed labeling. For an average market-
ing application, several months delay is often the difference between obtaining
exclusivity provisions or a marketing advantage or both over a competitor’s prod-
uct. In addition, sponsors must deal with decisions very late in development that
can greatly affect the speed at which a marketing application is compiled and
possibly reviewed. Examples of some of these decisions are whether to provide
a uniform format for study numbering, report synopses, tabular summaries of
safety and effectiveness, patient narratives (and corresponding case report forms),
and location of statements of compliance to GCPs, GLPs, or transfer of sponsor
obligations.
The pyramid concept for construction of the NDA/BLA can aid in devel-
oping a sense for the building block components and organization of a marketing
application. As shown in Fig. 1, the proposed labeling is developed at the IND
stage (i.e., the Investigator’s Brochure progresses to the Package Insert). The data
C. Summary of Studies
A summary table of studies for each technical discipline (i.e., nonclinical, clinical,
and chemistry) should be created to summarize the studies and data. This sum-
mary is then supported and cross-referenced to the full study reports.
Using the summary table of clinical studies as an example, this table pro-
vides an overview of all the clinical studies, portions of which are then provided
in the Clinical Pharmacology, Controlled, Uncontrolled, and Other studies sec-
tions according to the guideline for Item 8 of the NDA/BLA [Guideline for the
Format and Content of the Clinical and Statistical Sections of New Drug Applica-
tions (1988)]. When updated as a continuum throughout the development of the
product, this table can be used to present an overview of study information in
the end-of-Phase 2 and pre-NDA/BLA meetings with the FDA, in addition to
the final form appearing in the NDA/BLA. Information that this table should
include is the following.
1. Identifier, title, and site(s) where the study was performed. The study
identifier system should be clear to the reviewer. Status of the study
such as the start and completion dates, GCP status (a statement of
which is required), and the study category. It is important to note what
Does not contain a completed application form (i.e., Form FDA 356h)
Is not submitted in the form or does not contain the information required
Once accepted for filing, the FDA ‘‘review clock’’ then starts and the multi-
disciplinary review process begins. Safety updates are submitted by the applicant,
or a waiver requested, during the review process. The FDA ‘‘review clock,’’
according to 21 CFR § 314.101, is 180 days; review times, however, have histori-
cally always been longer. PDUFA was enacted (originally in 1992 and reautho-
rized in 1997) to permit collection of prescription drug applicant submission fees
to provide payment for additional reviewing staff at the FDA such that review
times can be more expeditious. Review time goals are now 10–12 months for
standard review and 4–6 months for priority review [PDUFA II Five-Year Plan
(1998)]. The FDA may have questions that require responses throughout the re-
view period. If the response to the request or of additional submitted information
is substantial, the information may be considered an amendment to the applica-
tion, and the review clock may be extended or restarted.
‘‘Rolling reviews’’ or ‘‘rolling applications’’ are attractive to applicants
because they represent a way in which to start the review process earlier and
potentially may allow for an earlier approval date. In accordance with 21 CFR
§ 314.50(d)(1)(iv), the Chemistry section of an application can be submitted 90–
120 days in advance of the remainder of the application, and review is dependent
on resources. The FDA Modernization Act of 1997 ‘‘Fast Track’’ initiative (Sec-
tion 112) for drugs intended to treat a serious or life-threatening conditions with
an unmet medical need allows for submission of components of the application
before the complete application, but stipulates that the application needs to pro-
vide a schedule for submission of sections that would make the application com-
plete. In this case, the review clock would begin once the application is complete.
The marketing application requires maintenance not only before submission, but
also after submission, during the review process, and after approval. Some of the
presubmission and postsubmission processes that effect the NDA/BLA content
are outlined below.
* Reproduced from D. Fordyce, ‘‘Establishing the Format and Content of an NDA/BLA’’ in S. Lin-
berg (ed.), Expediting Drugs and Biologics Development—A Strategic Approach, 2nd ed., with
permission from Parexel .
†
This information will be included in the request for the meeting and should be updated in the pre-
meeting information package.
‡
Although this information is not specifically cited in the aforementioned guidance document, it is
appropriate to include this information to obtain FDA feedback on the structure of critical elements
of labeling and the plan for organization and submission of the application (e.g., plans to submit
abbreviated reports versus electronic reports) and recommended in division-specific pre-NDA meet-
ing guidance. Regulatory information should include a summary of previous meetings, agreements,
and requests for information and subsequent responses, and should reference components of the
submission that address any outstanding issues.
B. Safety Update
In accordance with 21 CFR § 314.50(d)(5)(vi)(b) and 21 CFR § 601.2, safety
updates are required 4 months (120 days) after filing the application, after receiv-
ing an approvable letter, and at other times as requested by the FDA. The safety
updates are required to be submitted in the form of the integrated summary of
safety information. In addition, case report forms for all patients who died during
a clinical study* or who discontinued because of an adverse experience are re-
quired. According to the Guideline for Format and Content of the Clinical and
Statistical Sections of an Application, if the total exposure has changed substan-
tially, generally increased by 25% or more, the overall analysis of both new and
old data is required. Because a substantial increase in exposure could greatly
effect the conclusions of the application, it is appropriate to consider data cutoff/
lock dates such that the information is inclusive up to several months before the
submission.
* Generally, this includes information 4 weeks after a study is completed, or otherwise discontinuing
drug, or more if the product has a long half-life or has known late occurring effects [Draft Guid-
ance—Reviewer Guidance Document for Conducting a Clinical Safety Review of a New Product
Application and Preparing a Report on the Review (1996)].
ACKNOWLEDGMENTS
Special thanks to Gordon Johnson of Lachman Consultants and the Former Dep-
uty Director of the Office of Generic Drugs at the FDA for his input and expertise
regarding 505(b)(2) NDA applications.
REFERENCES
http:/ /www.fda.gov/cder/regulatory/applications/NDA.htm
http:/ /www.fda.gov/cder/da/da.htm#notes
Approved Drug Products with Therapeutic Equivalence Evaluations. 21st edition. ‘‘Or-
ange Book.’’ U.S. Department of Health and Human Services, Public Health Ser-
vice, FDA, CDER, Office of Information Technology, Division of Data Manage-
ment and Services, 2001 (http:/ /www.fda.gov/cder/ob/default.htm).
Federal Register. Abbreviated New Drug Application Regulations: Patent and Exclusivity
Provisions, Vol. 59, No. 190, October 3, 1994.
Federal Register. Biological Products Regulated Under Section 351 of the Public Health
Services Act; Implementation of the Biologics License; Elimination of Establish-
ment License and Product License, Proposed Rule, Vol. 63, No. 147, July 31, 1998.
Form FDA 356h, Application to Market a New Drug, Biologic, or Antibiotic Drug for
Human Use. http:/ /www.fda.gov/opacom/morechoices/fdaforms/cder html; http://
www.fda.gov/opacom/morechoices/fdaforms/cber html; http:/ /forms.psc.gov/
forms/FDA/fda.html.
PDUFA II Five-Year Plan, 01 July 1998. http:/ /www.fda.gov/oc/pdufa2/1998plan.html.
FDA. CDER. ODE IV Pilot. Targeted Product Information. http:/ /www.fda.gov/cder/tpi/
default.htm.
F-D-C Reports. The Pink Sheet, FDA ‘‘Targeted Product Information’’ Pilot Uses Label-
ing Goals at IND Stage, Vol. 61, Issue 11, March 15,1999.
F-D-C Reports. The Pink Sheet, Waxman/Hatch Should be Expanded to Cover Biologics,
Vol 60, Issue 46, November 16, 1998.
Food and Drug Modernization Act of 1997. Summary Prepared by Hyman, Phelps, and
McNamara, F-D-C Reports, Inc. Summary Prepared by the Biotechnology Industry
Organization. http:/ /www.biocentury.com/fdamod.html (by subscription only.)
Fordyce, D. Establishing the Format and Content of the NDA/BLA. In: S. Lindberg (ed.),
Expediting Drug and Biologics Development—A Strategic Approach, 2nd ed.,
chap. 9. Parexel , 1999.
U.S. Regulatory Reporter. Guide to 505(b)(2) Submissions. Vol. 11, No. 11, May 1995.
Washington Drug Letter. FDA to Issue ‘‘How-To’’ Guide for Naming Drugs. August 18,
1997.
Guidelines
Selected FDA Guidelines (http://www.fda.gov/cder/guidance/
index.htm;http://www.fda.gov/cber/guidelines.htm)
Guidance for Industry: Applications Covered by Section 505(b)(2), Octo-
ber 1999.
Guideline for Drug Master Files: September 1, 1989.
Guidance for Industry: Fast Track Drug Development Programs—Designa-
tion, Development, and Application Review, September 1998.
Formal Meetings with Sponsors and Applicants for PDUFA Products, Feb-
ruary 2000.
Guideline [for Item 3]—Format and Content of the Summary for New Drug
and Antibiotic Applications; February 1, 1987.
Guideline [for Item 4]—Format and Content of the Chemistry, Manufactur-
ing and Controls Section of an Application, February 1, 1987.
Guideline [for Item 5]—Format and Content of the Nonclinical
Pharmacology/Toxicology Section of an Application, February 1, 1987.
Guideline [for Item 6]—Format and Content of the Human Pharmacokinet-
ics and Bioavailability Section of an Application, February 1, 1987.
Guideline [for Item 7]—Format and Content of the Microbiology Section
of an Application, February 1, 1987.
Guideline [for Items 8 and 10]—Format and Content of the Clinical and
Statistical Sections of New Drug Applications, July 1, 1988.
Guideline on Formatting, Assembling and Submitting New Drug and Anti-
biotic Applications, February 1, 1987.
Guidance for Industry: Organization of an ANDA, February 1999.
Guidance for Industry: Providing Electronic Regulatory Submissions in
Electronic Format—General Considerations (IT 2), January 1999.
David L. Horwitz
LifeScan (a Johnson & Johnson Company), Milpitas, California
I. INTRODUCTION
Cultural differences
Patients
Investigators
Language differences
Regulatory differences
Communication logistics
Differences in clinical practice standards
Transportation of clinical trial drug supply and blood and tis-
sue samples
For a trial to be truly international, every country should be doing the same trial
in the same way. Otherwise, it is just a series of parallel single-country trials.
Doing the trial in the same way in each country implies more than just having
the same protocol. Trial monitoring, case report forms, and information flow
should all be uniform, so the data are easily merged. The proper way of conduct-
ing a clinical trial for the purpose of obtaining drug registration is generally called
Good Clinical Practice, or GCP. The initial emphasis of GCP regulations was to
assure the ethical conduct of a trial, but the concept has gradually expanded to
include issues of scientific conduct and data integrity. Guidelines for Good Clini-
cal Practice specify obligations of the investigator, sponsor, and Institutional Re-
view Board (IRB).
Ten years ago, there was sufficient diversity in GCP standards that it was
useful to have publications comparing various systems (2). At the present time,
considerable harmonization has been achieved (3). The ICH guidelines should
guide the conduct of all international trials. For the most part they do not conflict
with any national laws, but at times conflict is avoided by making the ICH guide-
line somewhat nonspecific and deferring to national regulatory requirements for
further definition. This is especially true in the areas defining institutional review
boards and informed consent. In the more technical areas such as protocol devel-
opment, study monitoring, quality assurance, accountability of clinical trial drug
supply, and documentation, it should be possible to use similar procedures in all
countries participating in a trial. The format of the Investigators Brochure is de-
fined somewhat differently by the ICH (3) and by the U.S. Food and Drug Admin-
istration (FDA) [21 CFR §312.23(a)(5)]. However, in the author’s experience it
has not been difficult to prepare an Investigators Brochure that successfully meets
the requirements of both formats.
As noted, the portions of the GCP guidelines that control the ethical conduct
of the study, namely, those relating to the IRB and to informed consent, are likely
to be most problematic. The need to do studies in an ethical manner is not in
question. However, due to legitimate cultural differences there is not uniform
agreement on, for instance, the proper composition of an IRB, how much needs
to be disclosed to consider a consent as ‘‘informed,’’ and the situations in which
The previous section considered some cultural considerations that can affect the
informed-consent process from a regulatory standpoint. Other factors to be con-
sidered are regional practices on what the physician traditionally tells the patient.
For instance, in some countries, especially in Asia, physicians often do not tell
patients that they have cancer. While the family is informed, both the family and
the physician are expected to ‘‘protect’’ the patient from knowing the diagnosis.
How, then, can such a patient give an informed consent to participate in a trial
of a new therapy for cancer? Like many such questions, there is no answer that
can be separated from one’s personal ethical background. A local IRB may be
very comfortable dealing with this issue, but a foreign regulatory authority may
find the consent process to be unacceptable. While there is not a solution to this
problem, it is critical that the sponsor understand in advance that such a situation
exists, so that an intelligent decision can be made on the feasibility of a multina-
tional trial in this circumstance.
Cultural considerations may also affect experimental design. In some coun-
tries, the concept of a placebo control is simply not acceptable. The norm is to
treat everyone, and so the control must be an alternative treatment or a lower
dose of the study drug. In other situations, an oral or topical placebo may be
acceptable, but an injectable or suppository placebo may not be because it is felt
that such intrusion without potential therapeutic benefit is not ethically accept-
able. The sponsor’s only alternatives may be to omit such locations from the
V. LANGUAGE CONSIDERATIONS
GCP standards require that clinical trials done for registration purposes be ade-
quately monitored. Monitoring is done to assure that the rights and well-being
of human subjects are protected; that the reported trial data are accurate, com-
plete, and verifiable from source documents; and that the conduct of the trial is
in compliance with the currently approved protocol, with GCP, and with the
applicable regulatory requirements. Generally, this monitoring is accomplished
by periodic visits to each clinical trial site by an appropriately trained monitor.
The monitor is often the primary contact between the investigator and the
sponsor, and so must play the difficult role of being both a public relations spokes-
person for the sponsor and also an authoritative ‘‘police’’ figure when necessary.
To do this well requires a person with exceptional interpersonal skills. In an
international trial, it is critical that the monitor be culturally sensitive to the cus-
toms of each country visited. Especially in the early stages of a trial, errors are
sometimes made in following the protocol or, more commonly, in completing
case report forms. The monitor is obligated to point out these errors and see that
they are corrected. However, in some cultures, acknowledging that one has made
an error results in ‘‘loss of face,’’ and many are unwilling to do it. If an error
Data collection in all trials is being increasingly automated. Through the use of
direct electronic data entry at the site, electronic transmission by modem or In-
ternet connection, or through faxed transmission of data on paper, collection of
data from multiple sites has become almost trivial. However, as the transmission
process becomes simpler, the efforts needed to assure the accuracy of the data
transmitted become more complex. For instance, if direct electronic entry of data
is done at each site, the computer screens used to enter the data must be designed
with the same attention to differences in language and specific use of terms and
abbreviations that was noted above when case report forms were discussed. Be-
cause predesigning computer screens is often more complex than redesigning
paper case report forms, this is sometimes more difficult to do. If data are trans-
mitted to a data processing center directly by the investigator, rather than being
collected on site by the study monitor, the data ‘‘clean-up’’ at the end of the
study is likely to take longer and, because it is done farther in time from the actual
creation of the data, may be more difficult to do—especially if investigators,
co-investigators, or their staff have changed or if source documents have been
misplaced. Most automated data analysis systems now being used to process
clinical trial data routinely generate a ‘‘query’’ when results appear to be missing
or inconsistent, or fall outside predetermined ranges. These queries often must
be brought to the investigators’ attention for resolution, and they frequently lack
the cultural sensitivity discussed above under monitoring.
A. Disease Prevalence
If one is expanding a trial internationally primarily to enhance patient recruitment,
knowledge of the demographics of the clinical condition in question is critical.
While no one is likely to go to northern Europe to study malaria, or central Africa
to study cystic fibrosis, the decisions are usually more subtle than this. Often,
market research will have already been done to determine the largest markets
for the disease, and these will frequently (but not always) be the best places to
go for clinical trials. They will also be the countries where the needs of local
populations, local thought leaders, and local regulatory bodies must also be con-
sidered, and so turn out to be the logical nations to consider first.
B. Subject Accessibility
A good country for a clinical trial must not only have a reasonably high preva-
lence of the condition being studied; the patient population must be accessible for
study. Countries with a strong national health care system may have centralized
registries of patients, allowing those with a given diagnosis to be readily found.
Countries with large tertiary referral centers, often associated with universities,
will be helpful if the disease is one which is often referred to such centers. On
the other hand, if the condition is generally managed by primary care providers,
the sponsor must either find investigators with a very large practice or be able
to attract patients to study centers. The local cultural climate regarding physician
referral of patients, patient self-referral for a study, and patient advertising to
recruit subjects must be carefully evaluated to see if adequate trial enrollment
will be possible. It is unwise to assume that patient recruitment techniques that
have proven successful in one’s own country will be equally successful in another
country. Indeed, they may prove to be illegal or ethically unacceptable to the
local medical community.
C. Investigator Qualifications
In addition to a suitable patient population, a good country for a clinical trial must
have a good investigator population. If the trial will be used for multinational
registrations, it is imperative that the investigator have a thorough understanding
of the principles of a randomized, controlled clinical trial. Investigators who are
unwilling to properly randomize patients or to follow a protocol rigorously will
D. Laboratory Support
Laboratory measurements are frequently primary end points in drug clinical trials.
Therefore, the availability of acceptable laboratories is critical. It is often desir-
able to use a single reference laboratory for all measurements in a multicenter
clinical trial, and this is generally accomplished by having each center ship its
samples to the reference lab or by using a service (often provided by commercial
reference laboratories) that picks up the samples from each site. When multiple
countries are involved, the logistics quickly become complex. If multiple labs
are used, each must be audited for compliance with acceptable laboratory prac-
tices, an acceptable quality assurance program, and timeliness of reporting. While
some laboratory proficiency testing programs are available internationally, not
all labs use them. Records may be kept in different languages, and different re-
porting units may be used. A common requirement for clinical trials is that each
center must have a certified laboratory, but not all countries have national certi-
fying bodies, and the standards of certification may be variable. Thus, particular
scrutiny is needed when multiple laboratories are to be used in a trial intended
for multinational drug registration.
Other types of complexities arise if it is decided to use a common reference
laboratory. The sponsor will almost always have to arrange for shipping of sam-
ples, as few, if any, commercial reference laboratories provide this service on
an international basis. Because of public health concerns, there is considerable
restriction on movement of medical samples that consist of body fluids. Some
countries require export approval from the Ministry of Health; others require
import approval. Packaging of samples must be done in ways that protect both
the sample and the carrier from accidental breakage. If samples must be sent at
controlled temperatures, delays in customs must be anticipated. Many air carriers
will not accept any shipments in Dry Ice on a passenger flight. Shipping costs
and the cost of customs brokers can add considerable expense to a trial. All of
these factors must be considered early in the trial planning, and a trial site (and
country) should be selected only after it is confirmed that provision can be made
for adequate handling of laboratory specimens.
E. IRB Reviews
Regulatory requirements vary from country to country, and must also be consid-
ered in the trial process. Some countries require government approval for any
trial on a new drug [similar to the U.S. Investigational New Drug (IND) applica-
tion]. In other countries, there is no formal government review and the govern-
ment will defer to the expertise of an institutional ethics committee. Virtually all
countries require that human studies be approved by an institutional review board
(IRB) or an ethics committee and, even if it is not required, the sponsor will
generally want such a review to be completed to assure that the trial will be
accepted internationally. Thus, in selecting trial sites the sponsor must inquire
about the availability of an IRB, how frequently it meets, and the timeliness of
its reviews. In some centers, an IRB may meet frequently and complete review
of all protocols submitted to it, and a review can be guaranteed within a specified
time period. In other jurisdictions, the IRB may meet only occasionally, meetings
may not be held during holiday periods (which may last a month in some coun-
tries), and the IRB will review only a fixed number of applications at each meet-
ing, resulting in a backlog if the protocols submitted exceed this number. All of
this needs to be determined in advance.
F. Government Regulation
When formal government review is required, the length of the process must be
considered. Some countries, like the United States, follow a relatively rapid pro-
cedure and allow a trial to begin 30 days after an application is received unless
the government acts to delay the start. In other countries, formal approval is
required before beginning. In some, the application to the government cannot be
made until after IRB approval is obtained, and the delays can be considerable.
In addition, for approval to begin the trial, permission must be obtained to export
the drug from the country of manufacture, and import it into the trial countries.
Export rules vary from country to country. In the United States, they have recently
been simplified, particularly for shipments to the more developed countries. Im-
portation, on the other hand, can be quite complex and may require multiple
approvals from several government jurisdictions. Generally, an in-country re-
source is needed to manage the process, and sponsors without an international
G. Trial Cost
Finally, the cost of the trial must be considered. Sometimes, the perceived ability
to reduce the cost of a trial is a primary reason for doing a trial in a specific
country. Sometimes the costs reduction is very real. Particularly when a study
involves hospitalization, hospital costs may be substantially less in some parts
of the world. Differences in stipends to investigators may also be quite different,
due to regional differences in cost of living. On the other hand, additional costs
are incurred, as noted above, due to shipping of drug supply, lab samples, and
case report forms. Also, cultural differences may make it difficult to settle on
what the true study budget really is. In the United States, for instance, there is
usually a research agreement that specifies a fixed budget that includes out-of-
pocket costs (such as local lab tests) and the portion of the investigator’s salary
that will be provided, as well as any institutional overhead. In other parts of the
world, less is committed to writing and it is assumed that the sponsor will provide
investigators with travel to international meetings and other incidental costs of
research. It can be very difficult, especially if one is not familiar with the local
language and does not understand the culture, to gain an understanding of this.
The first clue that one is not meeting the expected obligations may not come
until it is noticed that an investigator who appeared to be enthusiastic is spending
less time recruiting patients and may be devoting himself more to other trials. It
is critical that these issues be explored prior to beginning the trial. With good
communication, many potential misunderstandings may be avoided. Once again,
consultation with someone who is familiar with local customs, and who may
know what investigators have received in other clinical trials, may be instrumen-
tal in establishing a proper relationship with the investigator.
This chapter has limited its scope to those considerations that, in the author’s
experience, are important in assuring that an international clinical trial will go
forward with the greatest possible efficiency. Many other matters may come up
in the development of international trials that are similar to those seen in single-
country trials. Many of those are discussed in other chapters of this book. Some,
however, have additional complexities introduced when the intent of a trial is to
support multiple-country registrations. These are summarized in Table 5, which
may also serve as a final checklist for some key factors for every trial.
Will the clinical trial protocol, including entry criteria and end points, support the ulti-
mate labeled indication, package insert, or summary of product characteristics that is
desired in each country where registration will be sought?
Are there sufficient nonclinical data or preliminary clinical data to support commenc-
ing the trial in all countries of interest?
Is a mechanism in place to assure that all adverse events, regardless of where they oc-
cur, are collected on a timely basis and reported according to the regulatory require-
ments of all countries where the trial is being conducted, and that serious adverse
events are reported within statutory time limits?
X. CONCLUSIONS
REFERENCES
1. European Agency for the Evaluation of Medicinal Products Human Medicines Evalu-
ation Unit. ICH Topic E 5: Statistical Principles for Clinical Trials. Note for guidance
on ethnic factors in the acceptability of foreign clinical data, 1997.
2. Waxman RD, ed. GCPs in the U.S., E.C., and Nordic Council: an International Com-
parative Report. Cambridge, MA: Parexel International Corporation, 1991: The Bar-
nett International Education Program, Clinical Research Education Series.
Joseph A. DiMasi
Tufts Center for the Study of Drug Development, Tufts University,
Boston, Massachusetts
I. INTRODUCTION
Studies of clinical success rates for new drugs have focused on periods following
enactment of the 1962 Amendments to the federal Food Drug & Cosmetics Act.
This emphasis can be explained by at least two facts. Development data for pe-
riods prior to enactment of the Amendments are limited, and the addition of a
proof of efficacy requirement introduced by the Amendments substantially altered
the new drug development process.
The methods used to determine success rates have ranged from descriptive
statistics on the proportions of investigational drugs first tested during a given
period that have, up to some point in time, been approved for marketing, to infer-
ential statistical techniques designed to predict the proportion of a group of inves-
tigational drugs that will eventually be approved. The ultimate fate of the set of
investigational drugs under consideration is what we most want to know. How-
ever, useful information can be garnered from observing the pattern of approvals
up to the end of the study period, especially if nearly all of the drugs in the group
have reached their final outcomes.
A series of studies dating back to the 1970s has documented observed suc-
cess rates for investigational NCEs contained in a CSDD database (1–6). In par-
ticular, these studies use survey information on the development history of NCEs
taken into clinical testing in the United States by a large number of pharmaceuti-
cal firms. The group of firms included in each study varies somewhat from survey
2 25 0 75 25 0 0 0
4 23 2 50 25 .08 .02 .02
6 17 3 30 20 .15 .03 .05
8 6 4 20 10 .40 .04 .09
10 4 4 12 8 .50 .04 .13
12 3 4 5 7 .57 .04 .17
14 2 3 0 5 .60 .03 .20
number of years from the start of clinical testing. A binary regression technique
is then applied to the data to estimate the probability that the outcome will be a
success (i.e., marketing approval). Combining the estimates yields an estimate
of the probability that a drug will be approved a given number of years from the
start of clinical testing. A final success rate is then obtained by summing the
estimates for each year (more precisely, by integrating over time in the usual
continuous formulation of the estimation problem).
The basic approach can be illustrated with a simple example. Table 1 shows
the hypothetical outcomes of clinical testing on a group of 100 NCEs. Outcomes
are given at 2-year intervals from the date of IND filing. The fate of all drugs
in the portfolio is determined at 14 years from IND filing. In this case we know
what happens to all the drugs over time, but in a real success rate analysis there
would be some uncertainty because some of the data would be censored (i.e.,
some drugs would still be active at the time of the study). The fifth column of
the table (percent completed) gives the probabilities that an outcome will occur
in each of the 2-year intervals (survival analysis). The sixth column gives the
probability that the outcomes will be successes given that an outcome occurs
(binary regression). The seventh column (probability of approval) is the product
of the fifth and sixth columns; it gives the probabilities that a success will occur
in each of the 2-year intervals. In the last column the probabilities in the sixth
column are cumulated. The last element of this column is the final approval suc-
cess rate.
DiMasi (17) presents the most recent comprehensive results on clinical approval
success rates for NCEs. Both observed and predicted success rates were deter-
(b)
Figure 1 Cumulative approval success rates for self-originated NCEs with INDs filed
from 1964 to 1989. Success rates are shown for the IND filing periods, 1964–1969, 1970–
1974, and 1975–1979 in (a), and for the periods 1975–1979, 1980–1984, and 1985–1989
in (b). Approvals are counted through December 31, 1995.
(b)
Figure 2 Cumulative approval success rates for all NCEs with INDs filed from 1964
to 1989. Success rates are shown for the IND filing periods, 1964–1969, 1970–1974, and
1975–1979 in (a), and for the periods 1975–1979, 1980 to 1984, and 1985–1989 in Panel
(b). Approvals are counted through December 31, 1995.
be approved. Success rates for all NCEs are about 3%–4% higher than for self-
originated NCEs.
The same methodological approach to estimating final approval success
rates can also be applied to estimating the likelihood that a new drug application
(NDA) will be submitted to the FDA. Figure 4 shows predicted NDA submission
success rates for IND filing periods. The pattern of change over time is similar
to that for approval success rates. These results can be used with those in Fig.
3 to infer an NDA success rate (i.e., the probability that an NCE with an NDA
submitted will eventually obtain marketing approval). These calculations indicate
a weak downward trend in NDA success rates, ranging from 86.7% for INDs
filed from 1964 to 1969 to 82.2% for INDs filed from 1980 to 1984.
determine estimates for R&D costs and attrition rates for four categories. Statisti-
cal modeling was used to determine approval success rates for drugs entering
clinical testing, and the approach to determining phase transition probabilities
described above was used to determine attrition rates for clinical phases. The
proportion of drugs in the sample that were still in active testing was relatively
small (9.7%).
Figure 5 shows estimated probabilities of marketing approval by therapeu-
tic category for this sample of self-originated NCEs. The probabilities for Phase
I are approval success rates for drugs that enter clinical testing. The likelihood
of approval for drugs that enter Phase II and Phase III are also shown. The highest
overall clinical success rate is for the anti-infective category and the lowest is
for the neuropharmacological group. These results likely reflect, to some extent,
differences by category in the capacity for preclinical in-vitro and in-vivo testing
to predict human efficacy. The risks in neuropharmacological development re-
main substantial throughout the process. The results indicate that only about one-
half of the neuropharmacological drugs that enter Phase III testing will make it
to marketing approval. Overall, more than one-third of the NCEs that enter Phase
III testing will fail to make to the U.S. market.
The data for this study were also used to examine differences by therapeutic
category in dropout rates at various points in the development process. Figure 6
shows these dropout rates by clinical phase and therapeutic category. Relatively
few neuropharmacological failures are screened out in Phase I. Only 10.3% of
neuropharmacological NCEs are dropped in Phase I (12.9% of all neuropharma-
cological failures), compared to 25% for all NCEs (32.5% of all failures). Con-
versely, a relatively large number of cardiovascular NCEs are dropped in Phase
I (36.1%, or 48.9% of all cardiovascular failures). At the other end of the develop-
ment process, the performance of neuropharmacological NCEs is also notewor-
thy. Nearly one-fourth of all neuropharmacological failures occur after Phase III,
the most expensive of phases, is entered.
The biotechnology revolution has matured to the point where a large number of
drugs have entered clinical testing and some have reached the marketplace. Sev-
eral studies have been conducted in recent years to examine the success rates for
biotechnology-derived investigational drugs. Bienz-Tadmor et al. (10) examined
success rates for biopharmaceuticals using a database of therapeutic protein drugs
that entered clinical testing from 1980 to 1988. Biopharmaceuticals were defined
as drugs derived either through recombinant DNA technology or through hybrid-
oma technology (monoclonal antibodies). Given the limited amount of time avail-
able for these products to have proceeded through development and regulatory
review at the time of the study, Bienz-Tadmor et al. (10) defined success to
be the submission of a product license application (PLA) with the FDA. PLA
submission success rates were estimated by applying a nonstatistical mathemati-
cal approach that used the final outcomes that had occurred at the time of the
study to predict what would happen to drugs that were still in testing. Struck
(11) succeeded this study with an analysis of phase transition probabilities and
approval success rates for biopharmaceuticals that were in development from
1983 to 1991. The data included biologicals derived from nonrecombinant pro-
duction methods.
Both Bienz-Tadmor et al. (10) and Struck (11) found success rates that are
much higher than those that have been found for NCEs. Bienz-Tadmor et al. (10)
predicted that the probability of PLA submission for NBEs will be in the range
56%–64%. Multiplying estimated phase transition probabilities, Struck (11)
found an approval rate of 71% for biopharmaceuticals that enter clinical testing.
For the purpose of estimating success rates, however, both studies used datasets
that had strong limitations. For the Bienz-Tadmor et al. (10) analysis, 55% of
the drugs in the sample were still active at the time of the study (i.e., they had
neither been discontinued nor had a PLA submitted). In the case of Struck (11),
90% of the products were still in development or regulatory review. In addition,
using lengthy study periods for a new and evolving technology sector can disguise
important changes that occur during the period analyzed. Specifically, implicit
in these studies is an assumption that success and attrition rates for biopharmaceu-
ticals would be the same for the early 1980s, when the first investigational drug
candidates entered clinical testing, as they would be for the late 1980s or early
l990s.
Table 2 Approval and Discontinuance Rates for New Recombinant Entities (NREs)
and Therapeutic Monoclonal Antibodies (TMAbs) Entering Clinical Testing in the
United States, 1980–1994
Maximum
IND Number Number of Number of Percent percent
filing period of INDs approvals discontinuations approval approved
1980–1984
NRE 16 6 9 37.5 43.8
TMAbs 4 1 3 25.0 25.0
NREs ⫹ TMAbs 20 7 12 35.0 40.0
1985–1989
NRE 43 3 27 7.0 37.2
TMAs 45 1 32 2.2 28.9
NREs ⫹ TMAbs 88 4 59 4.5 33.0
1990–1994
NREs 65 1 16 1.5 75.4
TMAs 55 0 17 0 69.1
NREs ⫹ TMAs 120 1 33 0.8 72.5
Data source: Gosse et. al., Clin Pharmacol Ther, 1996; 60:608–618.
V. CONCLUSIONS
VI. REFERENCES
1. Wardell WM, Hassar M, Anavekar SN, and Lasagna L. The rate of development
of new drugs in the United States 1963 through 1975. Clin Pharmacol Ther 1978;
24:133–145.
2. Wardell WM, DiRaddo J, and Trimble AG. Development of new drugs originated
and acquired by United States-owned pharmaceutical firms, 1963–76. Clin Pharma-
col Ther 1980; 28:270–277.
3. Wardell WM, May MS, and Trimble AG. New drug development by United States
pharmaceutical firms with analyses of trends in the acquisition and origin of drug
candidates, 1963–79. Clin Pharmacol Ther 1982; 32:407–417.
4. Mattison N, Trimble AG, and Lasagna L. New drug development in the United
States, 1963 through 1984, Clin Pharmacol Ther 1988; 43:290–301.
F. Richard Nichol
Nichol Clinical Technologies Corporation, Newport Beach, California
I. OVERVIEW
The contract clinical research industry had its genesis in the biotechnical con-
tracting industry that was essentially established in the late 1940s, as a result of
the application of selected technologies developed during World War II, to the
life sciences industry. Organizations such as Battelle, Stanford Research Institute,
and others provided services to the pharmaceutical industry in the areas of chemi-
cal synthesis subcontracting and related technical services. At about the same
time, the contract toxicology laboratory industry was established, and govern-
ment regulations evolved during the 1950’s which aided the growth of that indus-
try. These contract providers grew rapidly after the passage of an updated version
of the Pure Food and Drug Act in 1962. This set of regulations required pharma-
ceutical manufacturers to demonstrate both safety and efficacy as a requirement
for approval of the sale of new medications in the United States This milestone
provides impetus for what ultimately evolved into the contract clinical research
industry, commonly referred to today as the ‘‘CRO industry.’’
During the balance of the 1960s, pharmaceutical companies established
medical research departments, biostatistical and regulatory affairs units, and re-
lated corporate functions to comply with the relatively new U.S. Food and Drug
Administration (FDA) regulations requiring safety and efficacy. By the mid-
1970s, it was obvious to some in the industry that additional resources might be
needed to complement pharmaceutical company resources, which could supply
additional capabilities relative to the requirements of preclinical and clinical de-
Today the CRO industry is robust, with annual growth rates in the 20% range
and total revenue in the $5.0 billion range. It is not surprising that this rapid
growth has attracted many additional service suppliers, many in the niche cate-
gory. The following are characteristics of the CRO industry.
From a financial perspective, there are approximately 10 CROs with annual
revenues of over $60 million, with the largest companies in the market having
clinical research revenues of over $6.3 billion per annum. In addition, large, full-
service, multinational CROs such as PPD, Quintiles, and Parexel, have expanded
via acquisition into vertically positioned service areas. For example, Quintiles
provides services in all disciplines required for product registration, including
preclinical pathology and toxicology, reference laboratory services, and outcomes
research services, and related functions, such as packaging, formulation, and con-
tract sales services. Large CROs will continue to make strategic acquisitions to
increase market share, which will be necessary for publicly traded CROs to sup-
port their market valuations.
Diversification into genomics, health care database management, and man-
aged care has occurred. Looking at the CRO market as a pyramid, the lower
segment is occupied by companies with specialized services. Companies with
strength in product registration, strategic planning, and regulatory consultation,
such as CATO Research, serve an important segment of the market, particularly
for biotech companies, and have reduced emphasis on capital-intensive elements
and related labor-intensive services common to large CROs. There are also a
The SMO industry began in early 1975, when the Institute for Biological Re-
search and Development (IBRD) began contracting with Multispecialty Group
Practices (MSGP) for exclusive rights for outpatient clinical trials. By 1978,
IBRD had exclusive, 5-year sales and clinical research management contracts
with 52 MSGPs, primarily in the Western, Midwestern, and Southwestern United
States. In 1983 the company began adding CRO services and evolved into the
first CRO/SMO in the United States.
During the 1990s, the number of SMOs increased rapidly. In early 1999,
nearly 60 commercial SMO organizations existed, from small two- and three-
person groups to companies with more than $20 million in annual revenues.
Many SMOs originated as networks of specialists and sub-specialists; the
majority of SMOs are in the United States. Many specialty-oriented networks
are now expanding into other areas in order to sustain growth, spread financial
risk, and benefit from economies of scale.
SMOs have two primary customers: large and middle market pharmaceuti-
cal and biotechnology companies (product sponsors), and large CROs. To date,
SMOs have avoided ownership or special relations with CROs, since such ar-
rangements may preempt them from being vendors to other large CROs.
In the future, large CROs may buy or affiliate with large national physi-
cians’ networks for the purpose of creating their own SMO. Some CROs have
begun to provide management services to physician groups, with the interest of
capturing exclusivity for their studies at that site. Eventually, SMOs may reach
sufficient size and diversity to create a buying opportunity for large CROs.
The CRO industry has grown rapidly since the mid-1980s, and several directory
services can provide useful information regarding the profiles of large and small
contractors. As mentioned, larger sponsors have ‘‘gatekeepers,’’ who can be very
helpful in matching internal needs with the right CRO in terms of size, price,
experience, geography, and interpersonal ‘‘chemistry’’ with the therapeutic team.
If the sponsor has contracted with the group previously, and had a positive experi-
ence in the same research area, requests for specific vendor personnel to be as-
signed to the new project may be accommodated by the CRO.
Absent a previous working relationship with the vendor, the following are
key elements of due diligence which can lead to a successful relationship.
Experience. The CRO should have experience in the disease area, with a
project of similar clinical research complexity. Programs with multiple
indications (i.e., duodenal and peptic ulcer disease at the same site), inpa-
tient versus outpatient (e.g., bone marrow transplant studies versus late
Phase III outpatient hypertension), and simultaneous multinational regis-
trations offer different challenges, and vendors need to have the capacity
and experience to accommodate these different sponsor needs.
Capacity. It is crucial to determine the capacity of the vendor to provide
quality services to the sponsor. This is one of the most difficult areas to
Once the selection process has been completed, it is essential that timelines with
appropriate deliverables be published to all members of the internal and external
development team. Assuming all clinical supplies are available and properly
packaged, the pre-study meeting represents one of the most important steps in
the entire development process. These meetings are usually held at pleasant and
relaxing sites with appropriate amenities, ensuring a positive kick-off to the regis-
tration process. This meeting will include investigators and their support person-
nel, members of the sponsor’s development team, and all managers from the
The evolution of the CRO industry over the last 25 years has been remarkable,
and the industry currently is going through significant metamorphosis as it re-
sponds to changes in the health care environment, customer requirements, and
the intense competition which exists in the market. Combined with CRO access to
public equity, which commenced in 1993, this will ultimately result in significant
improvements in quality. CRO businesses will expand and prosper. When com-
bined with progressive attitudes and creativity in the contractor market, which
is absolutely essential because of the high level of competition, this may result
in innovations and developments that may exceed the financial resources and the
core competency of sponsoring companies. This represents a major shift in spon-
sor philosophy, but in other industries, particularly telecommunications and infor-
mation technology, interlocking relationships, and alliances between vendors and
manufacturers, have been the key to success in highly competitive markets.
As the cost and risk of development of pharmaceutical products increases,
and as the impact of 21 CFR part 11 and HIPAA regulations emerge, careful
and appropriate reliance on contractors can be the difference between financial
success and failure, especially for small companies. Sponsors will learn to recog-
nize and utilize vendors who can produce higher-quality deliverables at a lower
cost, and much of this enhancement of the development process will be strongly
influenced by capital availability and the vision of management in the CRO
sector.
It is also likely that a trend will evolve in the middle-market segment of
the CRO industry. Also, niche providers will offer components of the develop-
ment process which can be integrated into a sponsor’s virtual development model,
and will probably compete on a very aggressive price and service basis, as large
multinational CROs, particularly those that are publicly traded, continue to evolve
toward larger multinational contracts with more desirable pricing structures.
Middle-market growth has occurred in retailing, information technology, and a
host of other highly competitive young but evolving industries, and it is expected
to occur in the CRO industry.
Taken together, the combination of highly qualified, experienced, and reli-
able niche providers in concert with the full-service large multinational CROs
gives customers a substantial choice, which will result in lower costs and higher
quality deliverables with improved time frames. Overall, the trends in the industry
appear to be quite favorable to sponsors, as the natural evolution of this service
industry moves toward maturation via investment in cost-effective technology.
Jean M. Findlay
Regional Pediatric Associates, Durham, North Carolina
To understand the primary care practitioner’s use of the package insert, it is first
useful to review the practitioner’s reasons for using the document, as summarized
in Table 1. To establish clearly the risk–benefit ratio of the drug, practitioners
may review the adverse reaction section of the package insert just before prescrib-
ing. At the same time as physicians decide whether the risk–benefit ratio is ac-
ceptable, they also decide what risks to describe to the patient. This is clearly
I. Prescribing a drug
A. To establish a risk–benefit ratio
B. To decide what information to convey to the patient
C. To establish necessary monitoring procedures for the new medication
1. Frequency of monitoring
2. Laboratory determinations
D. To evaluate potential drug interactions
II. Choosing between various drugs for the same indication
III. Searching for documentation that a given adverse effect is associated with the
drug in question
Package inserts are prepared by many different individuals and groups, and may
vary considerably in their content. The FDA requires that all labeling include
the following information:
1. Description
2. Clinical pharmacology
3. Indications and usage
4. Contraindications
5. Warnings
6. Precautions
7. Adverse reactions
8. Drug abuse and dependence
9. Overdosage
10. Dosage and administration
11. How supplied
12. Animal pharmacology or animal toxicology or both
13. Clinical studies
Considerable recent interest has arisen in providing drug labeling that is easier
for busy practicing physicians to use. Research has shown that most physicians
use the PDR or package insert primarily as a reference tool. The results of a
physician survey conducted by the FDA were presented at a public meeting on
October 30, 1995 (Federal Register, Vol. 65, No. 247, December 22, 2000,
p. 81084). In addition to presenting the survey data, information and input were
solicited from interested parties regarding professional labeling for practicing
physicians. Of 204 primary care physicians and 200 specialists who responded
to questionnaires, only 5% used the package insert directly, and 88% used the
I. Format suggestions.
A. Separate serious from nonserious adverse effects.
B. Highlight the adverse effect in the text.
C. Provide tables of dosages for specific indications.
II. Quantitation of adverse effects.
A. Provide the relative frequency of adverse effects, and explain how data was
gathered.
B. Compare the frequency and occurrence of adverse effects from long-term vs.
short-term therapy.
III. Additional sections or subsections to include.
A. Provide more information regarding use in specific populations.
B. List drug interactions.
C. Give information on drug abuse and dependence.
D. Supply patient counseling information.
E. Add a separate section specifying recommendations for baseline and periodic
laboratory work.
This summary is then followed by a table of contents for the package insert that
is organized by section and number (3.1, 9.4, etc.).
Although this format was greeted favorably by physician groups, represen-
tatives from the pharmaceutical industry raised objections based both on the extra
cost of production and the increased medication packaging size needed to enclose
the even greater bulk of the revised package insert.
Other package insert areas under review are the patient information leaflet
and changes in labeling for pediatric use. The first of these is still under consider-
ation, and although some companies have already initiated their own patient in-
formation package in easy-reading lay language, the issue of regulation versus
recommendation is still being studied. The second area was addressed in a final
rule by the FDA issued in May 1996, (Guidance for Industry, Content and Format
for Pediatric Use Supplements, May 1996. http:/ /www.fda.gov/cber/gdlns/
GDEPED.pdf). This rule revised the current ‘‘pediatric use’’ subsection. It con-
tinues to permit a pediatric indication for a particular drug (e.g., desmopressin
for childhood enuresis), but also allows labeling for pediatric use based on ade-
quate and well-controlled trials in adults, together with other information regard-
ing pediatric use. This is of particular importance to pediatricians who prescribe
medications that, although commonly used in the pediatric age group, have not
been officially approved for pediatric use. The rule also requires dosage instruc-
tions for each age group, including adolescents, who were previously grouped
with adults in dosage recommendations.
VI. CONCLUSIONS
Designing package inserts for use by busy practicing physicians involves several
conflicting issues. Physicians must rapidly make decisions regarding use of a
particular drug by gathering as much information as possible. The more easily
available the information, the better. Patients have a right to know what their
choices are when trying to make responsible decisions about their health. Pharma-
Louis A. Morris
Louis A. Morris & Associates, Dix Hills, New York
Peter H. Rheinstein
Cell Works Inc., Baltimore, Maryland
I. INTRODUCTION
Since the 1962 amendments to the Federal Food, Drug, and Cosmetic Act (the
Act), overseeing pharmaceutical promotion has been under the jurisdiction of
the Food and Drug Administration (FDA). Under the Act, the FDA regulates
pharmaceutical advertising and labeling to assure that promotional materials are
not false or misleading and that they provide adequate risk disclosures.
During the initial years of FDA regulation, pharmaceutical marketers pro-
vided a continuous flow of materials to health care providers. For the most part,
these materials described the results of scientific studies and the clinical impact
of pharmaceutical products. The materials were directed primarily at health care
providers, particularly prescribing professionals. FDA staff reviewed promotional
materials for compliance with the Act and regulations and provided feedback to
the industry, primarily in the form of letters noting marketing claims and practices
considered to be false or misleading, lacking in fair balance, or otherwise in
violation of the Act.
Over the past two decades, the FDA has continued to review promotional
materials and provide regulatory feedback regarding observed violations. How-
ever, in our view, much has changed in the nature and type of promotion. Innova-
The Act specifies that a drug is considered misbranded (a prohibited act) if its
labeling or advertising is false or misleading. Labeling is defined as written,
printed, or graphic material that accompanies the drug. Advertising is described
as being composed of various examples including print advertisements in journals
and broadcast through television and radio. As a general principle, labeling and
advertising material distributed by drug companies must be consistent with, and
not contrary to, the approved product labeling. In addition, there must be adequate
risk disclosures to assure that the audience is not misled and that recipients of
the promotional intervention have adequate access to prescribing information.
The false or misleading provisions of the Act not only cover what is stated, but
failure to reveal material facts in light of what is stated is also considered mis-
branding a drug.
To avoid being misbranded, both labeling and advertising pieces must con-
tain adequate information about drug usage, effects, and risks, to permit safe and
effective use of the product. This information is in the form of ‘‘full disclosure’’
(i.e., reprinting the entire package insert) for labeling pieces and ‘‘brief sum-
mary’’ disclosure (i.e., reprinting most of the major sections of the label except
indications, clinical trials, and dosage and administration) for advertisements.
For broadcast materials, advertisements must contain a brief summary, and
important product risks must be summarized in a ‘‘major statement’’ included
With changing themes and distributional vehicles, the industry presents promo-
tional information that constantly tests the limits of the FDA’s interpretations of
the Act and its regulations. For the remainder of this chapter, we review some
of these changes and the FDA’s regulatory responses. We start with some of the
newer messages that the industry seeks to deliver.
Prescription drugs are approved on the basis of their safety and effectiveness.
When innovative new products are approved in classes in which no competition
exists, communication of the mere presence of a product to treat an otherwise
untreated (or poorly treated) condition is often sufficient to market a product
successfully. However, when products are introduced into crowded markets, a
comparative advantage is important to differentiate the new product from the
competition.
Comparative claims representing direct, head-to-head studies of competing
products can provide audiences with highly informative and persuasive informa-
tion. Accordingly, the scientific basis for such claims must be firmly established.
FDA regulations require ‘‘substantial evidence’’ for comparative claims, gener-
ally defined as two adequate and well-controlled clinical trials.
Comparative claims denoting clinical improvement can be met by head-
to-head clinical trials. The FDA reviews such evidence to assure that the trials
V. NEW CHANNELS
Should products that are listed on the Internet as ‘‘in the pipeline’’ be con-
sidered institutional statements or preapproval promotion?
B. Formularies
Another new audience for pharmaceutical companies is the formulary committees
of managed care organizations (MCOs). Formularies have existed for many years
in hospitals and at MCOs. However, MCOs have recently begun to use formular-
ies more aggressively as a means of controlling the cost of operations. By limiting
the purchase of medicines to a few members of drugs in a class, the MCO can
engage in bulk purchasing and selection of less costly medications. Since formu-
lary members are usually charged with reviewing drugs to determine the most
cost-effective products in the therapeutic class, they have, in some instances, been
a prime target for pharmaceutical marketers.
Not only has the audience for pharmaceutical messages changed, the source of
those messages has also changed. Horizontal and vertical integration in the phar-
maceutical industry has changed the face of medical marketing. Drug companies
VIII. CONCLUSION
We would like to express appreciation to our colleagues at the Food and Drug
Administration who made helpful comments on earlier versions of this manu-
script. The views expressed are solely those of the authors and do not reflect the
policy of the Food and Drug Administration. The authors prepared this manu-
script in 1997.
Todd S. Keiller
Healthcare Business Development, Inc., Hopkinton, Massachusetts
I. INTRODUCTION
B. Financing Alternatives
Sometimes corporate sponsorship of academic research is not possible because
the academic discovery research is not at a stage to justify a company’s invest-
ment. Overall pressures to reduce health care spending have squeezed funding
for corporate research and development, which, in turn, has squeezed available
funding for academic research. However, from the company’s perspective, prop-
erly managed academic research may in fact be a less expensive alternative when
compared with the costs of the company’s own research. Industry often balks at
the overhead rates a university places on its direct costs. However, those in indus-
try who review these budgets often are not aware of their own company’s corpo-
rate overhead rate, which is often much higher than the academic rate. In many
instances, the company can collaborate with a university to create a research
program that suits the company’s needs and can pay for the academic scientists
to perform the targeted work. The company does not need to hire personnel on
a full-time basis, and the existing university laboratories can efficiently meet
the company’s needs. Once the research is over, the company has no further
responsibility for the personnel and the laboratories.
In addition, corporate sponsorship usually results in a license or an option
to license intellectual property either currently available or discovered during a
sponsored research program. To follow good business sense, a company must
consider numerous factors before licensing academic technology, including the
following:
In dealing with such financing alternatives, the university often must consider
timing and the size of available funds.
Establishing a joint venture between an academic institution and a commer-
cial entity can be useful for both parties. The company may need the academic
institution for the basic research function, and the academic institution may need
the company for other areas of expertise such as distribution, manufacturing,
development, or strategic planning. In a joint venture, both parties share the risks,
and the idea may be advanced to another level before achieving its full potential.
The venture capital community has invested close to $500 million in the
biotechnology industry. Academic-based intellectual property can be an impor-
tant part of this investment.
V. CONCLUSION
1. Campbell KD. Study examines impact of MIT biotech research. MIT Tech Talk. Janu-
ary 24, 1996. Available at: http:/ /web.mit.edu/newsoffice/tt/1996/jan24/41460.html.
Accessed January 2, 2001.
2. The Association of University Technology Managers. The AUTM Licensing Survey:
FY 1999. Available at: http:/ /www.autm.net/surveys/99/survey99A.pdf. Accessed
January 2, 2001.
3. The Association of University Technology Managers. The AUTM Licensing Survey:
FY 1998. Available at: http:/ /www.autm.net/pubs/surgey/menu.html. Accessed Oc-
tober 5, 2000.
4. Cho MK, Shohara R, Schissel A, Rennie D. Policies on faculty conflicts of interest
at US universities. JAMA 2000; 284:2203–2208.
Marion J. Finkel
Pharmaceutical and Regulatory Consultant, Morristown,
New Jersey
I. BACKGROUND
When the FDA initiates an action to remove a drug from the market for safety
concerns, it does so by proposing to withdraw approval of the company’s New
Drug Application. The Food, Drug and Cosmetic Act requires the FDA to permit
the holder of the application an opportunity to present its views at a hearing.
Such hearings may be public or may be held before an administrative law judge
in closed session. Often the FDA elects to present the matter to one of its advisory
committees with the dual purpose of (a) obtaining the opinion of a group of
experts following presentations of both sides of the issue and (b) hoping to obvi-
ate a formal hearing if the committee recommends withdrawal. In the latter case,
a company may still pursue the route of a formal hearing, but that route becomes
less desirable when an eminent group of experts has recommended removal of
a drug form the market, a recommendation not lost on the medical and lay press
and financial circles because of the open nature of the FDA’s advisory committee
meetings. As will be seen below, a recommendation by a committee against with-
drawal also carries great weight with the FDA.
This chapter is devoted to case histories of certain drug withdrawals or
withdrawal of one of the indications for a drug, and in one instance an aborted
withdrawal, that were based, with two exceptions, on scientifically documented
adverse clinical experience. The cases are illustrative of the diversity of forces
that play a role or come into play before or after a decision to withdraw a drug
has been made. These forces include, in addition to FDA staff, FDA advisory
committees and consultants, drug companies and their consultants, the medical
profession, the public, consumer organizations, the media, and Congress. In re-
cent years, litigation has played an increasing role in company decisions to re-
move a drug (or medical device) from the market, and in the future one can
expect that even the specter of such litigation may be enough to cause a precipitate
withdrawal.
IV. TICRYNAFEN
Ticrynafen (trade name Selacryn) was marketed in May 1979 as a diuretic and
antihypertensive agent by Smith Kline and French Laboratories (SKF). The drug
had, in addition, a uricosuric effect, a property that distinguished it from the
thiazides and that rendered it of particular interest for patients who developed
hyperuricemia from the thiazides or who had concomitant gout.
VI. PHENFORMIN
Phenformin (trade name DBI), an oral hypoglycemic agent for the treatment of
non-insulin-dependent adult-onset diabetes mellitus, was approved for marketing
in the United States in 1959. Shortly thereafter, reports began to appear in the
medical literature of lactic acidosis in association with use of phenformin. This
resulted in a precautionary statement in the labeling for the drug. Reports became
more numerous in the early 1970s. The high mortality rate (about 50%) in patients
who experience lactic acidosis resulted in considerable concern regarding the
phenformin-associated phenomenon.
In 1973, the FDA reviewed the subject with its Endocrinology and Metabo-
lism Advisory Committee. At that time, 4 million retail prescriptions of phenfor-
min per year were being filled in the United States. The committee concluded
that various predisposing factors, such as myocardial infarction or renal disease,
were required in order to precipitate lactic acidosis in phenformin users, and if
patients with such predisposing factors were eliminated from phenformin usage,
adequate benefit/risk ratio existed to support continued marketing of the drug
(16).
The labeling of the drug was strengthened and eventually included a black-
box warning concerning fatal metabolic acidosis. Continuing accumulation of
case reports of lactic acidosis, however, resulted in further consultation with the
Advisory Committee in October 1976. At that time, prescriptions were still nu-
merous, about 3.5 million per year. The risk of developing lactic acidosis could
VIII. NONIFENSINE
IX. BROMOCRIPTINE
X. CONCLUSIONS
XI. REFERENCES
1. Bakke OM, Wardell WM, and Lasagna L. Drug discontinuation in the United King-
dom and the United States, 1964–1983: Issues of safety. Clin Pharmacol Ther 1984;
35:559.
2. FDA. Minutes of meeting with Smith Kline and French Laboratories, January 15,
1980.
3. Recall letter issued by Smith Kline and French Laboratories, January 16, 1980.
4. FDA. Summary of supplement to NDA 10–588, July 9, 1968.
5. FDA. Pharmacology summary in NDA 10–588, October 24, 1968.
6. FDA. Pharmacology review in NDA 10–598, May 23, 1977.
7. FDA. Letter to Merrell-National Laboratories, December 27, 1977.
8. FDA. Talk Paper on Bendectin, October 1, 1979.
9. Summary of September 15 and 16, 1980, FDA Fertility and Maternal Health Drugs
Advisory Committee meeting.
10. FDA. Letter to Merrell-Dow, January 9, 1981.
11. FDA. Drug Bull 1981, (March):1–2.
12. FDA. Drug Bull 1982 (April):6.
13. FDA. Talk Paper on Bendectin, June 29, 1982.
14. FDA. Talk Paper, ‘‘Bendectin Production Ended,’’ June 9, 1983.
15. FDA Consumer, September 1983.
16. Phenformin hydrochloride: opportunity for hearing on proposal to withdraw ap-
proval of New Drug Application. Fed Reg 1977 (May 6); 42(88).
17. FDA. Drug Bull 1977 (August): 14–16.
18. Federal Food, Drug and Cosmetic Act, Sec. 505(e).
19. Ibid.
20. FDA. Talk Proper, ‘‘Phenformin,’’ May 18, 1977.
21. Neil Chalet, personal communication.
22. Phenformin hydrochloride; denial of petition for reconsideration. Fed Reg 1979
(April 6); 44(68).
23. FDA Physician Sponsor Application and Order Form for Phenformin Under IND
14,000.
24. FDA Instruction for Physician Sponsors Under IND 14,000.
25. Ciba-Geigy Pharmaceutical Co., personal communication.
26. FDA. Drug Bull 1979 (November): 26–27.
27. Summary of Ad Hoc Erythromycin Estolate Advisory Committee meeting, April 16
and 17, 1981.
28. FDA. Drug Bull 1982 (August): 14.
29. National Disease and Therapeutic Index Drug Book, Year Ending December 1981.
30. National Disease and Therapeutic Index Drug Book, Year ending March 1986.