The biopharmaceutical industry can bring new medicines
to market in a faster, safer, and less expensive way than
current government and industry policy allows. Recognizing
this reality, the U.S. Food and Drug Administration (FDA)
has taken a dramatic step to streamline drug development
by incorporating new technologies. The FDA's Critical Path
Initiative, announced in March 2004, has recommended evaluation
of new ways to use genetic tools, faster computers, new
imaging techniques, and electronic medical records in the
drug evaluation process. This ongoing
project, while still in its infancy, holds the potential
to break down barriers between regulators and industry and
to expedite the often complicated journeys of lifesaving
innovations from researchers to regulators to patients.
In support of the Critical Path Initiative, the Center for
Medical Progress at the Manhattan Institute convened 25 experts
from industry, government, and the scientific community in
a task force on the 21st century FDA. In spirited and wide-ranging
discussions, participants considered how advances in genomics
and other disciplines might be used to optimize the drug approval
process. This working paper distills the problems, principles,
and proposals that surfaced during that dialogue.
In our discussions, a general consensus emerged that FDA,
scientific researchers, and pharmaceutical companies can collaborate
- Integrate biomarker validation into every stage of the regulatory
review for drug, diagnostic, and biologic applications.
- Improve clinical trials by creating one standard for collecting
and using data from electronic medical records.
- Utilize validated biomarker-based studies to assess the
safety and effectiveness of specific drugs for specific subpopulations
at specific doses.
- Articulate the importance of congressional appropriations
sufficient to implement the FDAs Critical Path activities,
as well as providing the $5.9 million earmarked for Critical
Path purposes in the FDAs 2007 budget proposed by President
George W. Bush.
This working paper is intended not as the final word on
the Critical Path but as a springboard to continuing discussion
and debate. Although this paper focuses on drug development,
participants in the task force's discussions recognize that
the FDA has vital responsibilities in other areas, such
as bioterrorism and food safety. Nevertheless, the authors
of this report believe that the Critical Path Initiative
should be a priority within the FDA and within government.
By steering us toward a drug approval process that is driven
more by science and restricted less by regulationby
unleashing the powers of American enterprise and pathbreaking
sciencethe Critical Path Initiative can improve health
and save lives.
The Critical Path to Personalized
The biopharmaceutical industry has the means
to bring revolutionary new medicines to market faster, more
safely, and less expensively than current industry or government
policy allows. Yet the FDA, drug developers, and scientific
researchers have only begun to evaluate the new technologies
that may optimize testing-and-approval of new drugs.
We are living in a period of enormous innovation
in the biological sciences. New fields like genomics, proteomics,
and other "-omics" sciences are being linked to
powerful new computers and statistical modeling that allow
researchers an unprecedented view into the inner workings
of human biology.
Biopharmaceutical companies have embraced this
knowledge and are rushing to translate new discoveries into
powerful, safer, and more effective therapies that can treat
diseases based on their underlying genetic roots, hastening
the day when patients will, as a matter of course, receive
individualized or targeted drug therapies.
The scientific challenges facing the development
of personalized medicine are considerable. The current system
of drug development and approval is outdated, inefficient,
and expensive. It often uses technological standards developed
in the 1960s to evaluate drug candidates identified using
the latest advances in basic science. Under the FDAs
current framework, the development of personalized therapies
remains more expensive and less efficient than it could be,
slowing the translation of new knowledge into new treatments.
The current regulatory and industry approach
is focused on ensuring that every product is safe and effective
for the general population. At a presentation on drug safety
convened at the Institute of Medicine, senior FDA staffers
remarked that current drug development technologies are
"largely empirical in nature" and that "this
tradition focuses on population means and observations of
outliers" that result in "trial and error"
The FDA and its stakeholders recognize that
this approach makes failure likely for products that otherwise
might be safe and effective for specific subpopulations
or individuals. The challenge facing industry and regulators
is to develop valid standards for identifying these "high
responders" at earlier stages in the drug development
process. The goal is to speed development of important new
drugs to market and ensure that people receive the medicines
that are best for them.
To its great credit, the FDA is taking a dramatic
step to catalyze the use of new technologies to focus and
streamline drug development. The FDAs Critical Path
Initiative, announced in March 2004, aims to use genetic
tools, faster computers, new imaging techniques, and electronic
medical records in the drug evaluation process.
This report breaks new ground by bringing together the analyses
and ideas of experts and stakeholders to identify serious
challenges to the Critical Path and to suggest solutions
for those challenges.
Many task-force participants believe that the
media creates a false dichotomy between "bad"
drugs (which reach the market, but shouldn't have) and "good"
drugs (which are, at least in the publics mind, "safe").
In truth, "bad" drugs with unacceptable safety
profiles are usually weeded out by the development process
before they reach market. It is true, however, that the
development process focuses on a new drugs broad-based
safety and efficacy profile. Consequently, drugs that have
bad effects on small subsets of patients may slip through
In fact, physicians need better knowledge about
how to use medicines safely and effectively in individuals
and subpopulations. Some drugs, such as Vioxx, may cause
problems for a small subset of people. Others drugs, such
as thalidomide, may beintolerable for broad populations
but useful in subpopulations; thalidomide has been widely
used to treat certain cancers. Instead of taking "'bad"
drugs off the market, or plastering them with interminable
warning labels, regulators and industry should work together
to develop personalized medicines that can better ensure
that people who can safely benefit from these drugs get
them and that those who are at risk avoid them.
The hope is that scientific advances will eventually
enable pharmaceutical companies to use genomic screening
techniques linked to biomarkers early in the drug development
process to identify drugs likely to cause serious side effects
in a substantial number of people.
FDA and industry can facilitate this goal by working together
to create standards for biomarker validation that can be
used in clinical testing to screen for rare, but unavoidable,
side effects such as liver, kidney, and heart damage. This
technology is not currently available, although these side
effects are often the reason that medicines do not reach
the market or are withdrawn from the market. A more targeted
approach to drug development and evaluation would make medicines
more effective and safer. The first step is to find and
validate likely biomarkers.
Biomarkers are measures of disease progression,
pharmacology, or safety that can identify unique disease
mechanisms or responses to medicines. FDA guidelines can
specify how biomarker-based tools and alternative drug evaluation
techniques can be used for drugs, biologics, and diagnostics
as well as a combination of medicines and tests.
Because of advances in our understanding of how
genetic variations shape response to medicines and disease,
researchers have reason to hope that biomarkers can become
an important new tool for the personalization of medicine.
But validating these biomarkers will require an unprecedented
degree of collaboration and cooperation among many stakeholders
in the biopharmaceutical community.
Figure 1. Innovation Gap
© 2004 Burrill & Company. Confidential and Proprietary
How can the FDA, working with industry and other
partners, best promote and advance personalized medicine?
That is the vital question confronting policymakers, industry
leaders, the scientific community, and the FDA itself.
Public Health and the Critical
The drive to streamline the drug development
and approval process through biomarkers, better animal models,
improved surrogate end points, and innovative clinical
trial designs has important public health implications,
especially for the development of new vaccines and an antibiotics.
The markets for these drugs are often smaller and much less
reliable sources of industry revenue than for chronic
ailments like heart disease and cancer. As a result, lengthy
and expensive regulatory requirements can act as powerful
disincentives for companies to invest in research
for these vital public health tools.
Public health officials actively discourage
physicians from prescribing new antibiotics, hoping to delay
the evolution of drug-resistant pathogens. This practice
may be prudent medically, but it has the side effect of
reducing industry revenue for investing in the next generation
of antibiotics. Vaccines, on the other hand, can be subject
to government price con controls trols
and are administered to targeted populations at infrequent
intervals. As a result, the annual U.S. sales for a single
statin drug, Lipitor, are greater than those of the entire
global vaccine industry. Since companies view these products
as finan financially unattractive, the pipeline for new
vaccines vaccines and antibiotics
antibiotics to combat resistant
pathogens, emerging diseases, and potential bioterror attacks
has grown worryingly thin. Streamlining the drug development
process and lowering de de-velopment costs should spur additional
research into these product areas.
Shortcomings of the Current Drug Development
and Approval Process
In the last decade, U.S. pharmaceutical research
and development expenditures have risen 250 percent, and
from 1999 to 2003 the National Institutes of Health (NIH)
budget for biomedical research doubled, from approximately
$13 billion to over $27 billion.
Although these expenditures have led to many advances in
basic biomedical science, the number of new drug and biologic
applications per year submitted for FDA approval over this
period has declined.
Figure 2. 10-Year Trends in Major Drug and
Biological Product Submissions to FDA
The Tufts Center for the Study of Drug Development
estimates that the industry must spend $800 million to $1.7
billion and 12 to 15 years of research and development on
average to bring a product to market.
These costs must be recouped predominantly during a limited
period of patent protection or marketing exclusivity if
drug development is to remain financially viable. With such
great amounts to recoup in such a limited time, the drug
industry is forced to charge increasingly higher prices.
Because of rapidly escalating prices for branded prescription
medications, national drug policies and price controls are
being considered, which would threaten the future of the
research-based pharmaceutical industry in the United States.
Higher development costs also limit accessibility of medications
and discourage development of medications for orphan diseases
and diseases that affect primarily low-income populations.
Individual companies have devoted enormous resources
to identifying potential biomarkers that would help streamline
the development and approval process. To date, however,
critical improvements have proved elusive. Many biomarkers
have been discovered, but the task of validating them is
laborious, and many do not prove reliable in the validation
process. For instance, there is still no biomarker to predict
hepatic injury (liver damage), nor is there a good animal
model available. Achieving even one of these goals would
represent a major breakthrough.
The FDA has concluded an analysis of the causes
for the delays in drug development and has called for collaborative
research to develop and validate new tools and methods for
testing new medicines. The FDA is confident that this research
and the resulting new tools will enable more rapid and informative
drug development, such as occurred for AIDS drugs in the
1980s and 1990s. In response to the AIDS crisis, the FDA
worked closely with the pharmaceutical industry to develop
innovative methods for the rapid development of new drugs
for AIDS and HIV, resulting in
development times as short as two years for these drugs.
During the same period, the average development time for
all drugs slowed to less than twelve years. This experience
clearly demonstrates that it is feasible to accelerate drug
development without taking unnecessary and dangerous shortcuts.
It is clear that FDA leadership is committed to
the Critical Path Initiative. It is increasing the number
of training sessions for reviewers on new statistical and
drug study methods. The Interdisciplinary Pharmacogenomics
Review Group (IPRG) advises and educates reviewers on how
drug evaluation can utilize pharmacogenomics (the study of
how variations in the human genome affect the response to
But the agency needs new organizational mechanisms
and additional resources to implement the Critical Path
Initiative fully and to ensure that drug advisory committees
utilize its tools. While the amount of this additional funding
should be determined through consultation with the FDA,
Congress, and FDA stakeholders, there is a glaring need
for additional funding. The FDA is unable to routinely send
staff to important collaborative scientific activities in
such areas as bioinformatics, biomarker development, nanotechnology,
clinical trial design, and imaging. For example, a recent
meeting on storing, collecting, and analyzing tissue samples
drew scientists and managers from the National Cancer Institute,
the army, CDC, private companies, and academia, but
no one from the FDA attended, despite the fact that the agency
will be one of the single biggest repositories of genetic
samples in the world. Similarly, there is no FDA office responsible
for ensuring that companies adopt Critical Path Initiatives.
Bioinformatics is a branch of clinical research
that analyzes biological information using computers
and statistical techniques. An ever-growing discipline,
it includes analyzing data from drug studies, evalu
evaluating and mining clinical data of patients
in the real world, collecting and storing genetic
material, and combing patient health records to
develop predictive models of health care.
The FDA's senior management has collaborated
with industry, other government agencies, community-based
research, and academia to develop new ways to evaluate drugs
during and after the development process. The Critical Path
process will be most successful, however, when collaboration
expands beyond senior management and FDA reviewers are comfortable
using validated Critical Path tools.Medical reviewers, for
example, within drug divisions could actually begin to use
non-frequentist trial designs (such as Bayesian models)
or virtual clinical trials for diseases where small treatment
populations make traditional clinical trials extremely time-consuming
or expensive. Although the agency is developing guidances
that implement new sciencebased standards, industry can
do much more to share the clinical data necessary to validate
the standards. Safety biomarkers, for example, have the
potential to expedite the creation of new guidances.
Through conferences, consortia, and other means,
the Critical Path Initiative has encouraged collaborative
efforts in the following areas:
- Testing and development of molecular and imaging biomarkers
for regulatory approval and use
- Specific directions for use of biomarkers in clinical
trials during drug development (a clear regulatory framework
- Collection and evaluation of genomic and molecular information
to develop assays for predicting the toxicity of drugs at
given doses and identifying who benefits most from which
- Evaluation of the impact of drugs in the real world through
the use of electronic patient records
Bayesian analysis is an important statistical tool for confirming
that smaller groups of patients are benefiting from new drugs
and devices and identifying the connection between how a product
works and clinical outcome.
When comparing two hypotheses using the same information,
traditional statistical methods would typically result in
the rejection or non-rejection of the original hypothesis
with a particu particular degree of confidence, while Bayesian
methods would yield statements that one hypothesis was more
probable than the other. Rather than assuming that we know
nothing prior to con conducting an experiment and then conducting
an experiment to see if a cause and an effect (drug and clinical
outcome) happen so frequently that it is most likely not a
matter of chance, Bayesian analysis presumes that we have
knowledge about other causes and effects and uses that knowledge
to shape the experiment and come up with an estimate of whether
the cause and effect are the result of chance. Such estimates
and experiments are continually updated in light of new knowledge.
The FDA recognizes that Bayesian computations can be used
in combination with these two forms of data. The FDA has
used Bayesian statistics to accelerate the approval and
improve the safety of coronary stents. Harvard statisticians
used Bayesian statistics to analyze seven ran randomized
trials of FDA-approved stents involving 5,806 patients stored
at the Harvard Clinical Research Institute (HCRI) to develop
an objective performance criterion for medical
device clinical trial. 
With input from stakeholders, the FDA is working to clarify
standards and guidelines for the application of these new
tools in the regulatory process. In that spirit, the FDA
and stakeholders can continue to employ statistical measures
that identify smaller groups of patients more likely to
benefit from a product compared with those less likely to
Such approaches incorporate the sort of confirmatory
evidence encouraged and allowed under the FDA Modernization
Act of 1997.14 The act allows drugs to be approved with
data from one adequate and well-controlled clinical trial
investigation and with confirmatory evidence to establish
effectiveness for risk/benefit assessment. Under this model,
validated biomarkers will be combined, as reliable data
emerge, with existing studies to speed up drug development
and narrow the group of patients for whom a medicine works
In the past, when the FDA granted "accelerated
approval" of a drug, it was based on the results of
one or more adequate and well-controlled studies establishing
that the drug has an effect on a surrogate end point that
is reasonably likely to predict clinical benefit. Thereafter,
the FDA requires studies, once the drugs are available,
to re-establish clinical benefit. The Critical Path Initiative
accelerates approval and, in theory, makes it available
to drugs and diagnostics or a combination of the two. Combining
a genetic test that identifies who responds best to a drug
could become more widespread as collaborative efforts identify
benchmarks that can accelerate the development of products
targeted to particular populations.
Recommendations for Promoting Collaboration
- A new cross-centers products task force can, in collaboration
with relevant review communities within the Center for Devices
and Radiological Health (CDRH), the Center for Biologics
Evaluation and Research (CBER), and the Center for Drug
Evaluation and Research (CDER), review the agenda of advisory
committees for biomarkerbased diagnostics, biologics, and
- The Interdisciplinary Pharmacogenomics Review Group can
develop guidelines for the use of biomarkers in combination
with small and adaptive trial designs. It should develop
specific training, recruitment, and reorganization goals
to be funded through the Prescription Drug User Fee Act
(PDUFA) budget increases.
- Standards can be set for small and adaptive trial design
and promoted throughout the divisions to replace, where
appropriate, Phase 3 pivotal trials.
- The agency can separately build upon the exploratory
IND (investigational new drug) guidelines and work with
consortia, such as the Critical Path Institutes biomarker
safety consortium, to improve and increase computerized
simulations of drugs to complement Phase 1 human testing.
Consistent with the rationale of the exploratory IND, the
can collaborate to develop guidelines for the use of these
computerized models. An organization such as the Critical
Path Institute could sponsor meetings to help the FDA develop
methods for doing so. Such a program would allow reviewers
to work more closely with their scientific peers outside
- To encourage familiarity with Critical Path tools throughout
the FDA's rank and file, FDA senior leadership can make
knowledge of the Critical Path an integral part of performance
and incentive reviews. Without this performance review,
be wide variance in the use of Critical Path tools among
different product reviewers, even within the same center.
Genomics and Postmarket Drug Safety
The fast moving field of genomics can also impact drugs
on the market now, which may not have benefited from the
use of validated biomarkers during their development and
regulatory approval. Novel DNA markers may provide an important
contribution to postmarket drug safety by helping to quantify
an individuals risk of suffering an adverse events
from the use of currently approved medications. Yet research
into this field also raises questions on how to apply genomic
phar pharmacosurveillance data to drug labeling. It also
raises questions about how to best coordinate the roles
of the pharmaceuti pharmaceutical industry, the research
community, and insurance providers in dealing with novel
diagnostic tests that may be developed post approval for
Eploratory INDS and Validating Biomarkers
The FDA and industry share responsibility for
developing better tools for clinical evaluation. To that
end, early in 2006, the FDA announced a new method for early
stage pharmacokinetics (drug metabolism over time) and pharmacodynamics
(the effect that the drug is having over time) clinical
testing. This new approach, called the exploratory IND,
was developed by the Interagency Oncology Task Force (IOTF).
An exploratory IND study, sometimes called a "Phase
0 trial," involves "very limited" human exposure
to a compound and has no therapeutic or diagnostic intent.
The exploratory IND process increases the number of potential
drugs that can be tested in micro-doses in small numbers
of patients instead of testing pill-size quantities in large
clinical trials. While unexpected and serious adverse reactions
may still arise, the IND process may allow companies to
identify promising drugsor reject drugs with poor
safety or efficacy profilesbefore entering into a
Phase 1 clinical trial.
As FDA Deputy Commissioner Dr. Woodcock has noted,
"The purpose of an exploratory IND [study] is to learn
about new discoveries before embarking on extensive human
trials... Thus, we think eventually not only will this lead
to new knowledge about many new discoveries, it will save
people from being exposed to higher doses of compounds that
ultimately turn out not to be useful."
Many companies, in association with the FDA,
are seeking to identify and validate potential biomarkers.
For example, the FDA and BG Medicine, a Massachusetts-based
biotechnology research company, are seeking to validate
biomarkers to discern signs of human liver toxicity in the
beginning of the drug development process.
The NIH, CDC, FDA, National Institute of Technology
Standards, Department of Energy, companies, and Department
of Defense are working independently of one another and are
not sharing information with nonprofit centers and companies.
The FDA currently lacks the resources to be a
full partner in these important activities. There is not enough
staff to be part of all the relevant committees or the meetings
and scientific programs within government or the scientific
community as a whole.
Absent FDA leadership, the various federal agencies
involved in genomic research sometimes find it difficult
to work together. For example, the NIH, CDC, FDA, and companies
have been meeting about the development of a biomarker for
a rare drug side effect called QT prolongation that causes
But the FDA currently lacks the resources to
send the relevant medical reviewers to these meetings. In
this case, Duke University would like EKG data to identify
genetic variations linked to the heart problem. Companies
are ready to share EKG data, but the absence of the FDA
is stalling progress. Similarly,
the FDA's Clinical Pharmacology and Biopharmaceutics office,
which is responsible for receiving genomic data, and scientists
who review drug and diagnostic applications are unable to
fully participate in collaborative efforts to create genomic
analysis platforms and to evaluate pharmacogenomic data.
Recommendation for Validating Biomarkers
- The FDA should be given additional funds sufficient to
sustain its Critical Path activities, particularly for maintaining
a leadership role in biomarker development and use.
The Critical Path Institute is a nonprofit organization
created in 2005 to support the FDA in its effort to implement
the Critical Path Initiative. Based in Tucson, Arizona,
the C-Path Institute has been given $10 million in public
and private seed funding for five years. The institute is
working with the FDA, drug companies, and other scientific
stakeholders to collab collaborate on a variety of biomarker
- The Cardiovascular Safety Biomarkers Initiative will develop
tools for assess assess
ing and preventing idiosyncratic adverse cardiac events.
- A QT biomarker initiative aims to assist the FDA in accelerating
approval by in increasing the likelihood of effective safety
screens and risk management programs.
- The Toxicogenomic Biomarkers Initia Initiative will
explore ways to incorporate new technology into methodologies
for evaluating general toxicity related to the drug development
Post-Market Drug Evaluation
The FDA and the entire biomedical community recognize that
prior to marketing, with current tools and technology, there
is no way to detect rare, unexpected side effects short
of performing studies with sample sizes that exceed tens
of thousands of patients. Such massive clinical trials are
not practical or sustainable and would bring drug development
to a halt. Nor does it make sense to rely on doctors and
patients to submit reports of possible problems when information
technology permits real time and continuous reporting of
Rather, as the Critical Path report notes, "[S]afety
issues should be detected as early as possible, and ways
to distinguish potential from actual safety problems should
be available. Unfortunately, in part because of limitations
of current methods, safety problems are often uncovered
only during clinical trials or, occasionally, after marketing."
The goal of post-market drug evaluation need not be limited
to safety but can also include the ability of doctors and
patients to choose medicines and treatments that are best.
Further, computerized analysis of clinical data can help
pinpoint which patient subgroups will be more likely to
benefit from one medicine or avoid side effects from another.
In short, post-market data can be the source of information
to develop faster studies that more accurately predict and
measure safety and benefits.
To achieve this goal, the FDA, consumers, health plans, and
companies must use disease registries, biobanks, and electronic
patient records to coordinate medical information that can
be used to further personalize medicine (see sidebars). Several
agencies are already banking DNA samples using various approaches
and standards. The FDA is cooperating with the National Cancer
Institute to adopt NCI standards for DNA submissions. Such
collaboration is critical to create a common platform for
the evaluation of genetic materials and for the establishment
of best practices in the future. At present, the FDA's participation
in the creation of this important source of post-market information
is limited by resource constraints. Congressional approval
of the FDA's requested $4.7 million for drug safety evaluation
would facilitate the agency's participation in these collaborative
Biobanks are "actual repositories of collected human
tissueblood, bone, serum, or sometimes just individuals'
DNA. But they become
valuable because of the clinical
information captured about the patient and the molecular
data generated from the sample. When this data is integrated
in a robust, secure fashionor a clinical genomics
environmentresearchers can use biobanks for many different
purposes, such as hunting for reasons for the underlying
genetic processes that cause different diseases or identifying
molecular markers that may provide early warning signs."
Larger health plans and hospitals, as well as the Medicare
program, are switching to electronic patient records that
contain information on many individuals' characteristics,
their medical diagnoses, the medicines they took, and how
they fared. The health systems of the Mayo Clinic, Kaiser
Permanente of Northern California, and several disease-specific
patient registries together comprise millions of patients'
records with information that can be useful for proactive
post-market drug evaluation.
The Mayo Clinic used electronic patient records earlier than
many large health systems did. It has a base of 4.4 million
electronic records that can generate outcome data using standardized
entry criteria. Mayo has a full-scale program for the development
of biomarkers in adjusting drug dosing and safety profiling.
With the help of IBM, Mayo will be able to link its outcomes
data to very large external sources of genomic and proteomic
data such as the National Cancer Institute. Mayo is also generating
its own genomic data by collecting genetic samples and integrating
with outcome data. These efforts provide a promising model
for the future use of medical records
in post-market surveillance.
Use of these data sourceswith appropriate safeguards
to protect patient privacy and prevent the abuse of medical
informationholds great promise for improving drug
safety, health outcomes, and the reliability of drug development
studies. If a safety problem emerges,
it will be more precisely identified in terms of patient
characteristics, dosing, and genetic variations. Such information
can be used to update and further refine medical treatments
to avoid safety problems as well as to maximize benefit.
The next step is to use electronic medical record systems
to mine patient data in new ways and to compare outcomes
among patients with similar disease characteristics and
genomic makeup. Because researchers can look at dozens of
patient characteristics and hundreds of treatment steps,
observational studies designed to detect individual differences
in response to medicine or other treatments can be fairly
small but still yield powerful conclusions. Studies have
found that carefully designed post-market trials have the
same explanatory power as randomized, controlled trials.
DISEASE REGISTRIES, BIOBANKS, AND ELECTRONIC PATIENT RECORDS
Disease registries have become a powerful tool to identify
populations of patients most ap appropriate for a given clinical
trial. Disease registries are computerized systems that capture
and track key patient information. They are longitudinal,
ongoing databases that collect and maintain information on
patients with specific diseases. Registries keep track of
patients' signs and symptoms, what medications they may be
using, various alternative therapies that they may have tried,
and such issues as psychosocial aspects of the disease and
functional status. All data are collected from physician visits
or encounters with the health system.
Each patient's privacy is maintained by coding his or her
history. Many registries are simply lists of patients with
the disease or disorder, and some provide enough clinical
data to provide a "snapshot" of the characteristics
of the clinical expression of the disease. Few registries
actually provide an up-to-date assessment of the natural
history of the disease needed for the design of prospective
clinical trials, most likely because of the difficulty in
obtaining the required clinical data on an ongoing basis.
One example of a comprehensive approach to patient registries
is the C-Path Institutes Orphan Drug Registry, which
will create an electronic medical record (EMR) with additional
specific modules for each disease. The EMR will automatically
be updated from the medical-care pro providers and include
a portal for patients or their families to submit their
own data on the course of the disease and how it is being
managed. It will identify the standards for accurate diagnosis
and characterization of these and other rare diseases. It
will also identify a population of patients who are readily
available for participation in clinical trials, and it will
provide a basis from which to conduct post-market safety
The FDA, however, must weigh the need for a complete overhaul
of its bioinformatics operation against other pressing agency
priorities. For now, it is seeking to create standards that
make FDA data easily available to researchers outside the
agency and to establish common formats that allow the sharing
and pooling of data from registries and electronical medical
In addition to better data, researchers and
the agency must work with other organizations in an open
way to develop terminology standards and interoperability
standards for use in animal and human studies. The FDA is
part of the Clinical Data Interchange Standards Consortium
(CDISC) HL7 (standards for electronic interchange of clinical,
financial, and administrative information among health-care
computer systems) to ensure that FDA bioinformatic activities
are consistent with those in the private sector. The National
Cancer Bioinformatics Grid, Centers for Medicare and Medicaid
Services, FDA, and Centers for Education and Research on
Therapeutics (CERTS, a program of the Agency for Health
Quality Research in the Department of Health and Human Services)
have taken important strides toward data sharing but do
not work together on a regular basis.
Recommendations for Preclinical and Post-Market
- Create a Center for Clinical Bioinformatics within the FDA
and a corresponding Bioinformatics Interagency Task Force.
This center could allow stakeholders to create a single standard
for collecting and using information from electronic patient
which would improve medicines and clinical trials. This could
include companies, Medicare, health plans, employers, NIH,
- The FDA should become a full participant along with NIH
and NCI as part of the NCIs Biospecimen Coordinating
Committee and NCI's Wide Repository Committee.
- Companies submitting genomic data should establish
specific protocols for the collection, storage, and sharing
of tissue samples and serums from which genomic, protein,
and metabolic profiles information is generated. One approach
that should be considered is the forthcoming NCI best-practice
standards. The FDA can require companies to contribute
all relevant clinical trial data and biospecimens in standard
format by a date set by the agency. The Center for Clinical
Bioinformatics can develop partnerships with large health
systems such as the Mayo Clinic and integrate its post-market
program with larger efforts to mine data for genetic and
clinical patterns. If this is implemented, there must
be changes in the legal and intellectual property infrastructure
to allow the FDA to share data among pharmaceutical companies,
and pharmaceutical companies must agree on which data
can be shared and which are proprietary.
A practical driver for personalized medicine and the study
of drug response variation is the realization that extremely
rare and catastrophic side effects that require drug withdrawals
from the market may be trac tractable for formal study.
The process would include monitoring the epidemiological
occurrence of adverse events and corre corre- lating the
cases with a higher frequency of certain genetic markers,
which result from genome screens. The utilization of genomic
data for surveillance constitutes a powerful application
of personalized medicine, which is now feasible with the
advent of array diagnostic technologies. By focusing on
both common and rare side effects, the practice of personalized
medicine should accept the challenge of drug safety and
in the process could re relieve some of that burden from
the clinical trial process.
It is important to let the public know what personalized
medicine would look like from the standpoint of the Critical
Path Initiative. These new tools could accelerate approval
for a wide range of drugs, diagnostics, and devices targeted
to specific subpopulations.
Critical Path activities and tools leading to
targeted approval of a drug could go through a process such
as the following:
- An exploratory and confirmatory phase (up to targeted
approval) to determine the safety and effectiveness of
a drug for a specific group of patients at a specific
dose. This would occur in Phase 2 testing and would include
biomarker-based studies to identify how specific groups
of people respond to medicines. Tools used to develop
targeted subgroups would include gene-expression profiling,
gene sequencing, proteomics, and molecular imaging.
- At this point, a medicine could be granted targeted approval.
Access could be limited to the specific subpopulation to control
early market access.
- Drug safety and effectiveness could be monitored in a registry-type
setting or in cooperation with the NIH, academic medical centers,
or health plans with acceptable EPR systems.
- Companies could replace direct-to-consumer advertising with
a communications plan designed to improve prescriber and individual
knowledge of the relative risks and benefits of the product
for that defined patient population while prospective and
confirmatory trials were conducted.
- Expanded approval could be given to other patients after
updated safety assessment and clinical outcome. Any uses for
broader patient groups could be applied through a streamlined
process similar to the drug's original targeted approval mechanism.
Recommendations for Accelerated Approval
- Companies that rely upon validated biomarkers in Phase
2 testing to identify drugs that work for subpopulations
with increased benefit and smaller risk or provide an
unmet medical need should be able to make their medicines
available to people who meet the pharmacogenomic criteria
of their clinical trials on the basis of one study with
convincing proof of efficacy in the relevant population.
- Companies can participate in registries or in postmarket
monitoring of their products within a national interoperable
electronic medical records program also used by the FDA.
The 21st CENTURY FDA
The key to making medicines safer and more effective is to
make them more personalized and targeted. Moreover, the way
to personalize medicine is to transform the FDA from an organization
of rulebased regulators to a public healthfocused agency
staffed with 21st century science-based standard setters.
By collaborating with academic institutions, private companies,
and other government agencies, the FDA can utilize genetic
information and better bioinformatics to create a template
that will allow us to move from trial and error or one size
fits all medicine to predictive and ersonalized care.
Much of 21st century health care might be shaped by policies
and actions that are outside the realm of science. If we
are not vigilant, third-party payers and the tort system
could force drug companies and the FDA to shift investment
away from personalized medicine. We recognizes the scientific
and regulatory challenges, as well as the impact, that personalized
medicine will have on the manufacturing and marketing of
medicines. Steering drug development to smaller, even orphan
markers will require a significant investment on the part
of companies, without a certain return for their efforts.
The marketing methods of the past, geared to broad populations
and after the fact detection of safety problems, will give
way to informing and educating small groups of patients
and physicians whose understanding of the mechanisms of
new medicines and participation in data collection will
be critical. To the extent that the FDA evolves into a science-based
standard setter for translating genetic knowledge into medicines,
great progress is possible.
The task force strongly commends the FDA's Critical Path
Initiative and the scientists in government, academic, and
private settings whose insights made it possible. We share
their commitment to personalized medicine as a template
for both drug development and public health in the twenty-first
century. These reforms will help promote a future where
treatment is predictive, rather than haphazard and empirical.
They will help usher in an era in which drugs are targeted
by biomarkers and diagnostics rather than marketed to large,
and perhaps inappropriate, populations.
The sequencing of the genome has made possible a revolution
in human health. Personalized medicine is a possibility that
depends ultimately on our ability to create the tools and
marshal the will to make its many benefits a reality. The
recommendations of the task force are intended to promote
the Critical Path with a positive discussion of the specific
resources and actions needed to achieve this goal. We look
forward to making them, and the vision they seek to sustain,
a reality in the years ahead.