The Promise of Personalized Medicine
Imagine being able to walk into your doctor's office and present a "smart card"
encoded either with the sequence of your genome itself or with an access code
granting permission to log on to a secure database containing your genomic information.
Armed with a complete and accurate understanding of your unique genome, your
physician would be able to prescribe the right drug in the right dosage at the
right time to effectively treat your condition, with little or no concern that
the therapy won't work or that you will suffer adverse side effects.
That day of truly personalized medicine is still just a gleam in the eyes
of the scientists engaged in pharmacogenomics, but they are unanimous in their
belief that it is achievable and that it will arrive.
Just as genomics is the study of the entire genome while genetics is the study
of individual genes, pharmacogenomics looks at inheritable response to drugs
over the entire genome while pharmacogenetics identifies interactions between
drugs and individual genes. Pharmacogenomics seeks to uncover significant associations
between genomic patterns and clinical outcomes--correlations that produce useful
predictive knowledge, allowing clinical treatment decision making to be based
upon more rational criteria than today's probabilistic approach, which is largely
based upon educated guesswork.
Drug treatment is fundamentally a well-controlled environmental challenge
to the individual, says Gualberto Ruaño, vice chairman and chief scientific
officer of Genaissance Pharmaceuticals in New Haven, Connecticut. "What we learn
from pharmacogenomics will also apply to envirogenomics," he says, "as relates
to exposures to other challenges relevant to environmental health, such as pollution,
toxins, radiation, heat and cold, and even food."
Although the enormous variability in people's responses to drugs cannot be
attributed solely to their genotype, s †cientists believe that by understanding
the genetic underpinnings of how people absorb and metabolize drugs, they will
eventually quantify a great deal of that variability and be able to tailor therapies
accordingly in order to optimize treatment and avoid adverse effects. The influential
physician Sir William Osler summarized the problem aptly in 1892: "If it were
not for the great variability among individuals, medicine might as well be a
science and not an art." By solving much of the riddle of variability, pharmacogenomics
may contribute to swinging the balance of medicine much further toward science.
Why Pharmacogenomics?
If pharmacogenomics can do nothing more than help reduce the frequency of
adverse drug reactions (ADRs), it will have a tremendously positive impact on
morbidity and mortality. According to a landmark meta-analysis appearing in
the 15 April 1998 issue of the Journal of the American Medical Association,
more than 2.2 million hospitalized patients in the United States had serious
ADRs in 1994, resulting in more than 100,000 deaths and making ADRs the fifth
leading cause of death in the nation. If a patient is genetically predisposed
to be a poor metabolizer of a particular drug or class of drugs, when that drug
is administered, even at normal dosages, amounts of the agent retained in the
system can quickly build to toxic concentrations, leading to an ADR.
Experts believe that ADRs are likely to be the first area in which pharmacogenomics
will benefit patients. "The first advancement
that's already taking place is that we will understand better what people
will have a high drug level if we give them a drug, versus a low drug level,"
says David Hein, Peter K. Knoefel Professor and Chairman of the Department of
Pharmacology and Toxicology at the University of Louisville.
Today's trial-and-error, one-drug-fits-all approach to prescribing also means
that all too often a medicine is ineffective, re †sulting in wasted treatment
time, high health care and drug costs, and, most importantly, therapeutic failures.
Pharmacogenomic analysis can help identify patients who are abnormally high
metabolizers of certain drugs. These people metabolize so much of the agent
that it passes through their system without its intended effect.
image credit: Alamy |
Like ADRs, ineffective drug therapy is a widespread problem in clinical practice,
and it can have serious consequences, particularly in the treatment of diseases
in which delays in determining effective therapy can be disastrous, such as
psychiatric disorders, hypertension, and cancer. "There's an element of time
in which patients are taking a drug, and it takes some period of time for that
drug to become efficacious," says Michael Murphy, president and chief executive
officer of Gentris, a Morrisville, North Carolina, company offering pharmacogenetic
laboratory services and diagnostic tests. "Depending on the illness, that could
be pretty devastating, to be taking a drug and waiting for something to happen,
only to find out that it doesn't."
Penelope Manasco, chief medical officer and executive vice president of First
Genetic Trust, a Deerfield, Illinois-based company involved with genetic
data handling and bioinformatics, agrees. "Now most drugs are between thirty
and fifty percent effective," she says. "That's a really big deal when it's
something like a drug for depression. It can take people six months to a year
to get on the right drug. And with pharmacogenomics, they'll be able to have
a test that will say instead of having a thirty or forty percent chance, they'll
have a seventy or eighty percent chance that this medicine will work."
Scientists in pursuit of personalized medicine believe that although ph †armacogenomics
has already begun reaping benefits, the field is still in its infancy, and change
will come gradually. "In the short term," says Hein, "we're going to have less
drug toxicity. In the long term, we're going to have much better drug effectiveness."
Roger Ulrich, senior scientific director of Rosetta Inpharmatics, a subsidiary
of Merck engaged in the application of pharmacogenomics to the drug discovery
process, feels that it's going to be a while until there's a practical, everyday
application of pharmacogenomics approaches. Clinical trial data already show
there's variation in response for almost any agent, he says. "I think over the
next several years, we'll understand why we see variation in response. . . .
And there will be a sort of gradual assimilation of that data into practice,"
he adds. "I don't think any of us are going to wake up one morning and go, 'Wow,
we've finally entered the era of individualized medicine.' However, within research,
the impact of pharmacogenomics has already been positively felt, from the way
we discover and validate therapeutic targets to the way we explore drug safety
and design clinical trials."
Rochelle Long, chief of the Pharmacological and Physiological Sciences Branch
of the National Institute of General Medical Sciences (NIGMS), foresees a similar
pattern of steady development in knowledge and implementation of pharmacogenomics:
"I think progress will be incremental. . . . Within the next one to five years,
we're simply going to understand enough to, in a systematic way, better use
some of the drugs that are already on the market." Within 5-10 years after
that, she says, there should be a little more progress in the area not of drug
metabolism, but of variants in target receptors themselves; within the period
after that, people are going to start better understanding the genetic basis
of complex diseases such as hypertension, and drugs will be designed based on
that understandin †g.
Pharmacogenetics Begat Pharmacogenomics
image credit: CDC |
Pharmacogenomics may be in its infancy, having only recently come into its
own on the heels of the advances in knowledge, method, and technologies generated
by the Human Genome Project. But the discipline has deeper roots in pharmacogenetics,
a field of study that has been formally recognized for more than 50 years, and
that existed in practice much earlier.
"Avoid fava beans." So Greek philosopher Pythagoras instructed his followers
in the sixth century b.c., supposedly because he noticed that consumption of
fava beans made some people sick. Pythagoras may or may not deserve to be called
the father of pharmacogenetics, but his observation was on the money--in the
twentieth century, scientists discovered that ingestion of uncooked fava beans
can cause acute hemolytic anemia, a serious red blood cell disorder, in certain
populations. In the 1950s, it emerged that an inherited deficiency of glucose-6-phosphate
dehydrogenase, a red blood cell enzyme, caused this reaction, known as favism.
Today, this enzyme deficiency is known to be a relatively common disorder among
certain populations, and has subsequently been linked to sensitivity to a variety
of drugs, particularly antimalarial agents and sulfa antibiotics.
The emergence of pharmacogenetics in the twentieth century followed a path
forged by advances in molecular biology and genetics. In the mid-nineteenth
century, scientists learned that ingested substances were excreted in different
forms, establishing the concept of metabolism. In 1902, the physician Sir Archibald
Garrod, investigating alkaptonuria, a rare inherited enzyme deficiency, suggested
that enzymes were important in the detoxification of f †oreign substances, and
that genetically determined differences in the operation of enzymes (characterized
by Garrod as "inborn errors of metabolism") could be responsible for ADRs. The
year 1931 saw what is regarded as the first pharmacogenetic findings, when chemist
Arthur L. Fox reported on "taste blindness," an inherited difference in subjects'
ability to taste phenylthiocarbamide.
Before technology allowed the study of individual genetic variation, the field
concentrated on identifying racial and ethnic variations in response to drugs.
Most notable among several landmark studies, perhaps, was University of Toronto
professor emeritus Werner Kalow's investigation in the 1950s of the occurrence
of prolonged paralysis and rare, unexplained deaths in surgical patients receiving
succinylcholine, a neuromuscular blocker tolerated well by most patients. Kalow
discovered that a genetically based deficiency in the metabolizing enzyme pseudocholinesterase
was responsible, and proceeded to describe the population incidence of the various
alleles responsible for the deficiency. Similar studies confirmed the genetic
basis of the variability, seen in response to a wide variety of drugs.
image credit: Photodisc |
As the wider fields of genetics and molecular biology progressed, so did pharmacogenetics.
Now essentially folded into the burgeoning science of pharmacogenomics, the
discoveries that have emerged from the progenitor field are today available
and in use in the diagnostic arena, helping to screen patients who fall into
broad populations that, due to their metabolic genotypes, should not receive
specific drugs. As Murphy states it, "The reality is that for a lot of genes
that we've known about for the last twenty or thirty years, the need is to have
a cli †nical test that defines two or three patient populations [i.e., normal,
high, and low metabolizers], and we can do that now."
SNPs and Haplotypes
The remarkable innovations that led to the sequencing of the human genome
spawned a great leap forward, as pharmacogenetics spawned pharmacogenomics.
Technological breakthroughs such as polymerase chain reaction, high-throughput
robotic sequencing, and DNA microarrays, as well as simultaneous advances in
bioinformatics--which brought the ability to mine the mountains of data produced
for nuggets of useful knowledge--have allowed the field to move forward quickly,
as the genome begins to reveal some of its age-old secrets. Perhaps most significant
to pharmacogenomics has been the relatively recent discovery of two related
genetic phenomena--single-nucleotide polymorphisms, or SNPs, and haplotypes.
They have transformed the notion of personalized medicine from fond fantasy
to realistic goal.
SNPs are single-letter variations in DNA sequence that happen in at least
1% of the population (lower-frequency variations are considered to be mutations).
They occur every 100-300 bases along the 3 billion base pairs making up
the human genome. By collecting and analyzing the DNA of a diverse group of
many individuals, researchers are working toward identifying SNPs that are relevant
markers of drug response and disease susceptibility, an endeavor they hope will
ultimately yield diagnostic tests and targeted drugs based on genotype.
The discovery that SNPs tend to occur in patterns or blocks called haplotypes
may help speed the process of squeezing clinically relevant information out
of the human genome. Haplotypes are inherited groups of SNPs that occur within
a defined region of the chromosome, and some of them may influence drug response
more than individual SNPs do. Some experts believe that identifying haplotypes
of interest will yield more useful biomarkers of response by accounting for †
genomic variation in the multiple genes often involved in drug response.
Genaissance Pharmaceuticals is one biotechnology company banking heavily on
the value of haplotypes. "The haplotype is composed of multiple SNPs, but it
has the advantage and the power that it has the SNPs grouped into an alignment
as to how they occur in the chromosome and code for different versions of the
gene," says Ruaño. "Because of that resolution and symmetry with the
physiology and the function, the haplotype is therefore a much higher-resolution
technique for looking at genetic associations."
Murphy agrees to a point: "Sometimes when we don't know how mutations segregate--that
is, the pattern of how they fall on the two copies of every gene that we get
from our parents--then we have to do haplotyping. Where we do know [how mutations
segregate] . . . haplotyping would just be overkill. So sometimes it's needed,
but sometimes it's not, and you just have to take it on a case-by-case basis.
Whether it be SNP analysis or haplotyping, the most important thing is that
we can predict phenotype, or clinical outcome."
image credit: Photodisc |
Pharmacogenomics Initiatives
The rapid development of pharmacogenomics has led to an encouraging amount
of scientific cooperation and collaboration among government, industry, and
academia. "There's always some competition, but actually I think people have
been very collaborative," says Manasco. "One of the key things that's going
to be needed is more money for this translational research . . . so that the
people who actually win are the patients."
Ulrich voices similar sentiments. "It's a whole different approach to science,
this large-scale international consortium approach, †" he says. "There are pockets
of opportunity for each individually, but because of the size of the challenge,
it's going to take a continued joint effort."
Large-scale collaborative initiatives are making significant contributions
to the effort to eventually bring the benefits of pharmacogenomics to the bedside.
One such effort, the Pharmacogenetics Research Network (PGRN), was established
in 2000 and currently funds 13 academic research groups conducting basic research
describing pharmacogenetic phenotypes and relating them to genetic and genomic
information. The PGRN is spearheaded by the NIGMS, with the participation of
the National Cancer Institute, the National Heart, Lung, and Blood Institute,
the National Human Genome Research Institute, the National Library of Medicine,
and the NIEHS, making it a true trans-NIH effort.
Long, the NIH program director for the PGRN project, says the network stresses
high-quality research. "We want interdisciplinary groups of researchers," she
says, "who are coming together and putting their brains and expertise together
to design the very best pharmacogenetic projects, and then execute them, collect
the data, and put it in the database."
image credit: Steve McCaw/Image Associates |
The database she refers to is the Pharmacogenetics and Pharmacogenomics Knowledge
Base, or PharmGKB (http://www.pharmgkb.org/),
developed by and based at the PGRN grantee group at Stanford University. Long
emphasizes that although the data from PGRN members are the core resource of
the PharmGKB, the database is open to access and contributions by one and all.
"These data are all made public: the tools and resources that [PGRN members
are] generating, primers, or †any sort of chips they're developing, or reagents,"
Long says. "The intention also is to make [these resources] available to the
community at large. . . . So we're impacting all pharmacogenetics researchers,
whether or not they're presently a member of this network." [For more information
on both of these efforts, see "The Pharmacogenetics Research Network and the
Pharmacogenetics and Pharmacogenomics Knowledge Base," p. A575
this issue.]
In 1999, the International Life Sciences Institute Committee on the Application
of Genomics in Mechanism Based Risk Assessment was formed. This international
consortium, with participants from industry, government, and academia, evaluates
experimental methodologies for measuring alterations in gene expression, and,
in collaboration with the European Bioinformatics Institute, is building an
extensive database of microarray assays and analyses, which is scheduled to
be made public in 2004. A March 2003 white paper reporting on the committee's
status and recent findings is publicly available at http://rsi.ilsi.org/file/ACF539D.pdf.
Also formed in 1999, The SNP Consortium (TSC) is a nonprofit collaborative
effort among several major pharmaceutical companies, technological companies,
and academic research centers, along with the Wellcome Trust, with the target
of identifying 300,000 SNPs of biomedical interest. The discovery phase of the
project, which is now essentially complete, in the end identified 1.8 million
SNPs. The group viewed this high-density SNP map, which is publicly available
online at http://snp.cshl.org/,
as an important resource for defining haplotype variation across the genome,
and a rich source of new genomic information about disease susceptibility, drug
response, and novel therapeutic target †s.
image credit: Chris Reuther/EHP |
The SNPs identified by TSC contributed a major source of data for another extensive
public library of variations, this one hosted by the National Center for Biotechnology
Information. This library, dbSNP, located online at http://www.ncbi.nlm.nih.gov/SNP/,
now contains more than 4.1 million human SNPs, a significant portion of the
estimated 10 million common SNPs in the human genome.
More recently, a $100 million public-private research consortium called
the International HapMap Project was launched in October 2002. Expected to take
three years to complete, the HapMap will map the haplotypes in the human genome,
obviating the need to study all 10 million SNPs. With the abundant information
embedded in haplotypes and their variation across populations, the HapMap is
expected to be a powerful new tool for researchers to conduct association studies.
This will allow them to precisely identify significant genetic variations in
disease susceptibility, drug response, and even infectious disease resistance
and longevity [also see "HapMap: Building a Database with Blocks." EHP
111(1T):A16 (2003)].
Pharmacogenomics and the Pharmaceutical Industry
The pharmaceutical industry has seen the future. In a 9 April 2003 presentation
to a U.S. Food and Drug Administration (FDA) Science Board Advisory Committee
meeting, Brian Spear, director of pharmacogenomics at Abbott Laboratories in
Abbott Park, Illinois, put it succinctly: "[Pharmacogenomics] is not something
that a company here and a company there have taken a chance on. This is now
a standard part of the drug discovery and development process in every one of
†the drug discovery and research companies."
The industry's avid pursuit of pharmacogenomics, as evidenced by a recent
spate of acquisitions of pharmacogenomically oriented biotechnology firms by
the major pharmaceutical companies, runs the gamut from drug discovery to enhancing
the safety and efficacy of drugs that have been on the market for many years.
Just as the days of the one-drug-fits-all treatment approach may be numbered,
so too may be those of the present pharmaceutical industry business model, which
relies heavily on the periodic introduction of blockbuster drugs (typically
defined as products with annual revenues in excess of $1 billion) to generate
profits and fund research and development. The hope is that the application
of pharmacogenomics and other genomics technologies will enable a new paradigm
to emerge, with lower development costs, fewer candidate drug failures, revitalized
existing products, the possible resuscitation of withdrawn drugs, and a "portfolio"
approach to the introduction of new agents, with drugs available in different
formulations to maximize safety and efficacy in specific phenotypic populations.
According to A Revolution in R&D: How Genomics and Genetics Are Transforming
the Biopharmaceutical Industry, a 2001 report by The Boston Consulting Group,
it currently takes on average $880 million and 15 years to bring a new drug
to market. Failed candidate compounds represent a large proportion of that development
cost--right now, only about 10% of compounds that enter clinical development
make it to the marketplace. The Boston Consulting Group estimates that the effective
application of genomics technologies could reduce that staggering investment
by as much as $300 million and two years.
Ulrich is optimistic that pharmacogenomics can play a major role in increasing
drug development productivity. "The real cost savings will be that we have fewer
failures going into development," he says, adding †, "Making better choices is
really what it's all about. In the end, you might develop just as many compounds,
but you're going to have more successes. You'll spend just as much money, but
you'll get a greater return on investment."
image credit: TSC |
Ruaño anticipates that pharmacogenomics will contribute to all pharmaceutical
products, from new candidates to old warhorses. "The bottom line is that if
you match the chemical and pharmacological properties of the drug to a target
population that benefits from that drug, you have a new product," he says. "I
believe that can be applied to old, new, and in-the-middle drugs, in-the-middle
being the ones that are in clinical design and trials, the new being early-discovery
ones from genomic targets, and the old, the ones that we have already on the
market."
Many pharmaceutical companies are already using pharmacogenomics to screen
participants in clinical trials. Murphy, whose company provides such screening
services to the major pharmaceutical companies, has seen this concept evolve
from novel idea to accepted necessity. "We have a number of clients now who
routinely screen every single volunteer at Phase I for the important drug metabolism
genes," he says. "That's a big paradigm shift. And then they consider those
same inclusion/exclusion criteria, or stratification as they call it, as they
continue to develop that same drug in Phases II through IV."
The FDA appears to be solidly on board the pharmacogenomics bandwagon as well.
The agency has reportedly been working hard recently to acquire the necessary
in-house expertise to facilitate the submission, interpretation, and implementation
of pharmacogenomics data. "These studies are being widely done, and they may
have tremendous progress †," said Janet Woodcock, director of the FDA Center for
Drug Evaluation and Research, at the agency's 9 April 2003 Science Board Advisory
Committee meeting. "We need to find a way to get the information in, develop
our policies, develop a regulatory framework, . . . and help to move this field
along." Later this year, the FDA is expected to issue guidelines designed to
enable the free exchange of information between the industry and its regulators,
with the goal of bringing the benefits of pharmacogenomics to the bedside as
soon as possible.
Applied Pharmacogenetics
Most of the promise of pharmacogenomics remains to be fulfilled. However,
the concept of using known genetic associations to prevent patients from taking
drugs that would likely be ineffective or harmful is already available and used
in clinical practice in certain specific arenas, thanks mainly to the steady
progress made in pharmacogenetics over the past several decades.
Cancer therapy today includes two shining examples of applied pharmacogenetics.
First, there is now a commercially available diagnostic test measuring a patient's
ability to produce the metabolic enzyme thiopurine S-methyltransferase
(TPMT), which is essential for the metabolism of thiopurine medications used
to treat acute lymphoblastic leukemia (ALL), the most common form of childhood
cancer. Genetic testing gives clinicians the ability to classify ALL patients
according to their TPMT genotype, which allows optimized dosing. Doses in patients
with alleles rendering them deficient in TPMT (who are thus less tolerant of
thiopurine medications) are reduced by as much as 95%. This means TPMT-deficient
patients can tolerate the drug, yet enough is still metabolized to retain efficacy.
Second, the breast cancer drug trastuzumab (trade name Herceptin), which is
marketed in tandem with a diagnostic test, is often cited as an early indicator
of the value of the concept. Trastuzumab is effect †ive only sin the 25-30%
of breast cancer patients whose tumors overexpress the human epidermal growth
factor receptor (HER2) protein. The drug was developed specifically to exploit
that characteristic; it binds to HER2, which slows tumor growth. The diagnostic
test measures HER2 expression in the tumor and is thus predictive of the potential
efficacy of the drug; patients who do not overexpress HER2 are not given the
drug, because it will not work.
image credit: PGRN |
This is a unique combination today. But such diagnostic-agent pairings
will become more commonplace as pharmacogenomics progresses and strides are
made in disease genetics, in which a variety of diseases (particularly cancer)
are being genetically subclassified, often significantly redefining treatment
strategies.
An intermediate step toward such pairings is illustrated by work being done
with the cytochrome P450 (CYP450) family of enzymes, which is responsible for
a large segment of human drug metabolism. It is the metabolic pathway of choice
for about 60% of the drugs on the market today. It has also been the focus of
a great deal of research attention through the years, and the numerous CYP450
subtypes are well characterized, as are the important phenotypes of variation
in response. Several companies now offer CYP450 genotyping tests to the pharmaceutical
industry for clinical trial subject inclusion/exclusion based upon metabolic
profile, and now such tests are making their way into the clinical diagnostic
marketplace. Gentris, for example, soon expects to market five kits to physicians
for pharmacogenetic testing of their patients. Genelex Corporation of Seattle,
Washington, has taken the concept one step further, marketing tests dir †ectly
to the public for three of the major CYP450 pathways--CYP2D6, CYP2C9, and CYP2C19.
Once a consumer has placed an order for the test, Genelex sends them a blood
collection kit, and the consumer either sees their own doctor or Genelex will
refer them to a phlebotomist in their area.
image credit: Photodisc |
Pharmacogenomic tests appear to be just over the horizon. In March 2003, at
the 52nd Annual Scientific Session of the American College of Cardiology, Genaissance
presented results from its STRENGTH (Statin Response Examined by Genetic Haplotype
Markers) prospective clinical study, which showed that haplotype variations
are associated with response to treatment with the statin class of cholesterol-lowering
drugs. The associations discovered in the study were strongly predictive of
efficacy. Genaissance plans to eventually develop the information into a point-of-care
diagnostic test that will help physicians choose the safest and most effective
drug for individual patients, maximizing the prevention of cardiovascular disease
afforded by the statins. The company has performed similar studies of response
to asthma drugs, and other researchers in industry, academia, and government
are making substantial progress in establishing variability associations for
drugs used to treat hypertension, depression, HIV, cancer, and several other
conditions whose patients stand to benefit from optimized prescribing.
Ethical Concerns
On one level, the ethical issues involved with pharmacogenomics are similar
to those raised by genomics in general--broad concerns about research integrity,
privacy, confidentiality, informed consent, the specter of genetic discrimination
or stigmatization, and access to information or to specialized care. "There
are cer †tainly inherent problems that will not go away with pharmacogenomics
research," says Patrick Terry, president of PXE International, a patient advocacy
and research support group for victims of pseudoxanthoma elasticum, a rare genetic
disorder that affects connective tissues. "They're certainly not new and different
for pharmacogenomics research. Privacy, confidentiality, and misuse or misappropriation
of data were the same twenty years ago as they are today."