杏吧原创

AFTER much suffering, a deeply troubled young woman drags herself to her
doctor for help. The doctor diagnoses severe depression and says reassuringly,
鈥淣ow we鈥檒l just test your DNA to see what meds will work best for you,鈥� as he
reaches over and plucks a hair from her head. 鈥淲hen we get the results this
afternoon, I鈥檒l call in a prescription and you鈥檒l soon be feeling better.鈥� For
the first time in a long time, she smiles hopefully.

Such a scenario may not be far off. According to many pharmaceuticals and
biotech companies, patients could soon be taking drugs tailored to their genetic
makeup, saving time and bundles of money now wasted on ineffective treatments,
as well as minimising debilitating side effects. Today doctors would tell this
young woman to try a medication鈥攕ay, Prozac鈥攁nd then wait four to
six weeks to find out whether she is one of the lucky 40 per cent the drug
helps. It might take another six months and four different drugs before she
finds the best treatment for her depression. A quick genetic test to replace
this torturous trial-and-error would mean 鈥渁 big improvement in the way patients
get treated鈥� and lower healthcare costs for society at large, says Daniel Cohen,
a geneticist at a Paris-based genomics company called Genset.

That鈥檚 the promise of pharmacogenomics, an emerging science that aims to
describe at the genetic level precisely why some people respond well to certain
drugs and others don鈥檛. Such information will be used to create diagnostic tests
to help select the right drugs for each patient. In some cases, this gene-based
approach may even save lives by identifying people who are likely to develop a
fatal reaction to a drug.

Pharmacogenomics will be part of 鈥渢he next revolution in medicine,鈥� says
Francis Collins, director of the National Human Genome Research Institute near
Washington DC. Doctors will soon routinely test patients to ensure that the
drugs they take really are the best for them, he predicts.

But consumers aren鈥檛 the only ones who stand to benefit from 鈥減ersonalised
medicine鈥�. Pharmacogenomics should also boost corporate profits: biotech
companies expect to sell genetic test kits, and drugs companies hope to increase
the number of medications that survive clinical trials and make it to market.
鈥淚f we could identify who will strongly benefit [from a drug], we could promote
it to a defined segment of the population; that should also make it easier to
show it鈥檚 safe and effective,鈥� says Brian Spear, director of pharmacogenetics at
Abbott Laboratories in North Chicago. In fact, drugs companies have already
begun collecting DNA samples from people who take part in their clinical trials.
One day they hope to find a genetic pattern that will distinguish people who
might be helped by a drug from those the drug might harm.

The prospect of developing customised medicines drew corporate researchers to
New York for a conference on pharmacogenomics last month and the topic was
recently the subject of a special issue of Nature Biotechnology (vol
16, supplement). But people like William Haseltine, head of Human Genome
Sciences in Rockville, Maryland, wonder whether such enthusiasm may send drugs
companies heading off down the wrong path, spurred on in part by exuberance over
new technologies for rapidly cataloguing and comparing the minute genetic
variations of large numbers of people.

Haseltine fears that the focus may shift from finding the right drugs to
treat individuals to pinpointing the people who are 鈥済enetically right鈥� for the
drugs pharmaceuticals companies want to sell. Such screening might leave a
significant portion of the population medically out of luck. 鈥淭hat鈥檚 not where
we want to go,鈥� he says. 鈥淲e still want new drugs that treat as many people as
辫辞蝉蝉颈产濒别.鈥�

Genset鈥檚 Cohen disagrees. 鈥淗umans are polymorphic and diseases are also,鈥� he
says. 鈥淧harmaceuticals companies dream of a drug that treats everyone, but life
is such that this is not 辫辞蝉蝉颈产濒别.鈥�

Skilled physicians have always understood that individual patients respond to
drugs differently. And many factors contribute to this broad variability. Age,
gender, health status and whether the person is taking other drugs all dictate
whether a drug will work and what its side effects might be. But it鈥檚 now clear
that genes have a considerable influence over how someone reacts to a drug.
Individuals inherit specific versions of enzymes that affect how they
metabolise, absorb and excrete drugs. So far, researchers have identified
several dozen enzymes that vary in their activity throughout the population and
that probably dictate people鈥檚 response to drugs鈥攚hich may be good, bad or
sometimes deadly.

Adverse drug reactions caused more than 100 000 deaths in the US in 1994,
according to a recent article in The Journal of the American
Medical Association (vol 279, p 1200). Perhaps more frightening is that
these reactions often occur in patients receiving a standard dose of a
particular drug. As an example, doctors in the 1950s would administer a drug
called succinylcholine to induce muscle relaxation in patients before surgery.
More than a few patients, however, never woke up from anaesthesia鈥攖he
compound paralysed their breathing muscles and they simply suffocated. Doctors
discovered that these unfortunate patients had inherited a mutant form of the
enzyme that clears succinylcholine from their system.

The drugs don鈥檛 work

And that鈥檚 not the only time genes have been implicated in a toxic drug
response. As early as the 1940s doctors noticed that a good number of
tuberculosis patients treated with the antibacterial drug isoniazid would feel
pain, tingling and weakness in their limbs. These patients were unusually slow
to clear the drug from their bodies鈥攊soniazid must be rapidly converted to
a nontoxic form by an enzyme called N-acetyltransferase. The 鈥渟low
acetylator鈥� phenotype isn鈥檛 exactly rare鈥�40 to 60 per cent of Caucasians
have a less active form of the enzyme than 鈥渞apid acetylators.鈥� Again, this
difference in drug response, it was later discovered, is due to differences in
the gene encoding the enzyme.

In the past few years, researchers have also found that variations in certain
genes can determine whether a drug treats a disease effectively. For example, a
cholesterol-lowering drug called pravastatin won鈥檛 help people with high blood
cholesterol if they have a common gene variant for an enzyme called cholesteryl
transfer protein (CETP). And several studies suggest that the version of the
ApoE gene that is associated with a high risk of developing Alzheimer鈥檚
disease in old age goes hand in hand with a poor response to an Alzheimer鈥檚 drug
called tacrine.

This is where pharmacogenomics might be able to help. Right now, doctors can
run tests to determine whether a patient is likely to react badly to a handful
of drugs. However, these tests usually measure a phenotype鈥攕uch as the
amount of enzyme activity in a person鈥檚 blood鈥攔ather than which form of a
gene a person has. For instance, doctors can use an enzyme test developed by
researchers at the Mayo Medical School in Minnesota to determine which dose of
6-mercaptopurine to give children with leukaemia. About 1 in 300 Caucasian
children have severe鈥攐ften fatal鈥攔eactions to this chemotherapy
because the enzymes they need to metabolise and detoxify the drug are defective.
So physicians around the world send blood samples to a lab鈥攖he Mayo Clinic
analyses nearly 1000 a year鈥攖o determine whether it鈥檚 best to give a child
a full dose or as little as one-fifteenth of the normal prescription.

Such tests for phenotype are fine for identifying people who may have bad
reaction to a drug because they lack certain enzymes. But identifying people
whose cholesterol can鈥檛 be lowered by pravastatin treatment, for example, is not
so simple. Although toxic reactions often involve defects in single genes, good
or bad treatment responses are likely to be determined by multiple genes, says
Allen Roses, head of worldwide genetics research for the British-based drugs
giant Glaxo Wellcome. And the more genes involved, the more possible
combinations will need to be tested before the best treatment can be prescribed.
That鈥檚 why researchers are turning to the power of pharmacogenomics to look
directly at the variations in the genes themselves. By comparing the genetic
profiles of people who respond well with those who have a bad reaction to a
drug, scientists should be able to identify the genes involved, and which
versions lead to the best outcome.

Recent advances in rapid gene sequencing and the statistical comparison of
data collected from large populations are now making such studies possible. With
laboratories all over the world churning out massive amounts of human DNA
sequence, researchers can now focus on finding the hot spots in human
chromosomes that vary in sequence from person to person and then look to see
whether these are related to differences in drug response. The US Human Genome
Project鈥攑art of the international effort to sequence all 3 billion base
pairs of DNA in the human genome鈥攁ims to discover and map 50 000 to 100
000 of these hot spots, called SNPs (single nucleotide polymorphisms, pronounced
鈥渟苍颈辫蝉鈥�).

To chart these tiny differences in an individual鈥檚 DNA, many researchers are
turning to a new tool called a SNP chip, a tiny microarray studded with genetic
variations commonly found in human chromosomes
(see 鈥淪peed freaks鈥�, p 46). A
technician prepares a solution containing a patient鈥檚 DNA. When this sample is
washed over the chip, gene fragments that match the SNP sequences bind to the
chip and fluoresce. A computer then analyses the resulting pattern, determining
which variations are present in the patient鈥檚 genes. Researchers then tally the
variations present in two different groups of people鈥攕ay, people who have
a good or a poor response to a drug. With the help of sophisticated new
statistical techniques that can simultaneously compare thousands of SNPs from
thousands of patients, scientists look for associations between particular SNPs
and different responses to drugs. By mapping out the location of these SNPs,
researchers can generate a trail that will lead them to the genes involved in
drug response.

Abbott Laboratories and Genset have teamed up to do just that. Genset has put
together a rough SNP map of the human genome and Abbott has collected DNA
samples from people in clinical trials for zileuton, its asthma drug. Genset is
now analysing these DNA samples and looking for differences that may reveal
which genes control a person鈥檚 response to zileuton, a drug that can damage the
liver in 3 per cent of patients. Their study may eventually lead to a simple
test to filter out patients who might not respond well to the drug.

Tailored therapies

Soon SNP chips may be available for use in clinics, an advance that would
mark the real beginning of 鈥減ersonalised medicine鈥� based on gene screening.
Affymetrix, a biotech company in Santa Clara, California, has developed a chip
to detect 12 different variants of two genes that encode the cytochrome P450
enzymes. This family of enzymes is involved in the metabolism of at least 20 per
cent of all commonly prescribed drugs, including the antidepressant Prozac, the
painkiller codeine, and high-blood-pressure medications such as captopril. In
the near future, clinics may use the chip to determine the correct drug dose to
give a patient, says Robert Lipshutz, vice-president of corporate development at
Affymetrix.

Before they can even think about matching the right drugs with the right
patients, though, drugs companies will have to find out during clinical trials
who will be compatible with a product and who will not. Many companies, such as
Glaxo Wellcome and Abbott, have begun collecting DNA samples from people taking
part in their clinical trials. But Glaxo, for one, is holding onto the DNA until
a SNP map of the entire human genome becomes available. Then they will look for
a recognisable overall pattern in the people who respond well or badly to each
of their products. Theoretically, doctors would not even have to understand how
or why patients respond to certain drugs, says Glaxo鈥檚 Roses. They鈥檇 just
identify the good responders and give them the drugs that DNA tests indicate
should work best.

Haseltine, for one, is sceptical of Roses鈥檚 suggestion that physicians might
prescribe drugs, based on gene testing, without understanding why a patient
might respond well or poorly to the treatment. These new DNA diagnostic tests
are not yet fully understood or even 100 per cent accurate, he notes. A patient
whose treatment doesn鈥檛 work鈥攐r worse, makes him sick鈥攎ight sue. 鈥淚
don鈥檛 think any pharmaceuticals company would do it because of liability.鈥�

And in the end, many pharmaceuticals companies may be more concerned with
profit margins than with public health. Indeed, if the goal were truly to reduce
adverse drug reactions, many simple tests could be developed now, says Daniel
Nebert, a human geneticist at the University of Cincinnati. For example, a
genotype test could be developed to detect slow acetylators because the genes
and their variations are known. Says Nebert, 鈥淵ou could pick up 95 per cent of
slow acetylators with just a couple of DNA tests.鈥� And it would truly save
lives, he says. If a slow acetylator receives procainamide, a drug commonly used
after a heart attack, the patient has a 60 per cent chance of developing a liver
disease which could kill him.

At the same time, finding out such information about people鈥檚 genotypes opens
a Pandora鈥檚 box of other ethical issues. Epidemiological studies suggest that
slow acetylators are more susceptible to some environmental insults. So a person
known to be a slow acetylator may have trouble getting health insurance. One
study found that among post-menopausal women who smoked cigarettes, slow
acetylators were four times as likely to develop breast cancer as rapid
acetylators.

There鈥檚 always a danger that genetic information will be misused. But 鈥渢he
fight should be against the misuse, not the science鈥�, argues Cohen. Whether
society deals with the potential ethical consequences or avoids them,
researchers will continue to plug away at identifying and sequencing drug
response genes, and developing clinical tests. Pharmacogenomics may not meet all
the promises being made, says Richard Weinshilboum of the Mayo Clinic. 鈥淏ut I
see this as the most exciting time in the history of medicine as far as helping
to explain disease and treat it.鈥�

Drugs can be harmful. And sometimes one ethnic group is affected more than
others. During the Second World War, for example, African-American soldiers
given the antimalarial drug primaquine developed a severe form of anaemia. The
soldiers who became ill had a deficiency in an enzyme called glucose-6-phosphate
dehydrogenase (G6PD) due to a genetic variation that occurs in about 10 per cent
of Africans, and very rarely in Caucasians.

Such ethnic variation in reaction to drugs is not uncommon. Of the 45 or so
genes that are involved in drug metabolism, 37 show ethnic differences, says
Werner Kalow, a pharmacologist at the University of Toronto in Canada.

How do such differences arise? The answer may lie in the environment, say
Daniel Nebert of the University of Cincinnati and Frank Gonzalez of the National
Institutes of Health near Washington DC. Take the cytochrome P450 family of
enzymes, for example. These enzymes evolved some 400 million years ago to
protect plant-eating animals from being poisoned. As plants evolved more
powerful toxins to defend themselves, animals evolved better enzymes for
detoxifying them.

After dozens of generations, certain gene variants for these metabolic
enzymes become prevalent in some human populations because of their diet. A
population that鈥檚 dined on mostly goat meat and milk for 6000 years is likely to
evolve different forms of drug-metabolising enzymes than one that has subsisted
on tropical fruits and plants, says Nebert. Thanks to genetic differences in
P450, for example, 6 to 10 per cent of Whites, 5 per cent of Blacks, and less
than 1 per cent of Asians are poor drug metabolisers.

Other environmental factors can also play a role. G6PD deficiency probably
became more common in Africans because it confers some protection against
malaria.