杏吧原创

OUR genes hold the keys to curing or even preventing cancers, heart disease,
strokes and countless other serious illnesses. But the first step is linking
genes to the right diseases. To speed the process, British researchers hope next
year to start analysing DNA from half a million people in the Population
Biomedical Collection.

Armed with such information, doctors will eventually be able to spot people
at risk from certain cancers and intervene early. Others at risk from heart
disease, say, will be able to make pre-emptive adjustments to their lifestyle
and begin treatment before their arteries start to clog. The Medical Research
Council alone has pledged 拢20 million for the giant project. The Wellcome
Trust says it will stump up further millions.

These huge sums will be money well spent, say the leading lights of British
medical research. The project is twice the size of the world鈥檚 largest existing
effort to chart the genetic roots of illness in 270,000 Icelanders. So far,
criticisms of these mega gene projects have focused on ethical concerns. Some
observers have asked whether participants鈥� gene secrets will be kept truly confidential
(New 杏吧原创, 2 December, p 7).

Too little time

However, as the finishing touches are made to the British plans, a more
pressing logistical concern is emerging: some experts fear Britain鈥檚
hard-pressed family doctors will not be able to diagnose illness, record
symptoms, and take medical and family histories with anything like the degree of
accuracy needed. They warn that failure to make the right links between
patients鈥� symptoms and their genes鈥� phenotype and genotype鈥攃ould
even send researchers on wild goose chases after worthless drugs.

One critic is Ronald Worton, president of the American Society of Human
Genetics. 鈥淕etting the diagnosis right is crucial,鈥� he says. 鈥淚f you don鈥檛, the
genetics gives you the wrong answers.鈥� But it鈥檚 not only onlookers who have
concerns. Even Mike Dexter, head of the mighty Wellcome Trust, which will pour
some of its millions into the research, expressed fears at a genetics meeting at
the Royal Society in London in September that general practice might not be
equipped for the task.

In theory, family doctors will take personal and family medical histories of
the 500,000 DNA donors. They will then report fresh signs and symptoms of
disease to teams of geneticists. Armed with each donor鈥檚 DNA, or 鈥済enotype鈥�, the
geneticists would then look for common gene patterns in volunteers who develop
specific diseases.

But incorrect information, apart from misleading medical researchers, could
make some people mistakenly think they are at risk of developing a serious
disease. Conversely, doctors might miss people who are genuinely at risk.

No one is blaming the family doctors themselves. In an ideal world, say
critics, all the diagnoses and assessments would be done by trained clinical
geneticists who know exactly the signs and symptoms of the diseases in which
they are interested.

This already happens in many studies linking genes, environment and disease.
One example is the Avon Longitudinal Study of Parents and Children (ALSPAC) in
Bristol that has been running since 1991. ALSPAC鈥檚 clinical geneticists give the
14,000 children a painstaking annual examination lasting a whole afternoon, and
have a meticulously drawn-up list of tests and measurements for each condition
they are interested in.

People recruited for gene studies by Gemini Genomics, a company in Cambridge,
undergo 8-hour overhauls during which the investigators make between 500 and
1500 clinical measurements.

Such thoroughness will be beyond the resources of the MRC/Wellcome study. The
only affordable alternative, it seems, is to rely on family doctors. The
trade-off is the added statistical power of the study gained by including so
many subjects.

So where are mistakes likely to appear? Ian Purves, head of the Sowerby Unit
for Primary Care Informatics at the University of Newcastle, says that the most
reliable data come from 鈥渉ard鈥� phenotypes such as cancer where you can get
physical evidence of disease from biopsies of tissue.

Something like asthma is more difficult to diagnose because symptoms and
their severity vary greatly. 鈥淵ou can never be certain [of the diagnosis],鈥� says
Purves.

Behavioural and psychiatric disorders are the most notoriously 鈥渟oft鈥�
phenotypes, and pose diagnostic problems even for psychiatrists. David
Weatherall, an eminent geneticist at the University of Oxford, has similar
misgivings. He notes that, worryingly, even genetically simple diseases, where
the illness is linked to a single faulty gene, can manifest themselves in a
variety of ways.

His own research on the blood disease thalassaemia, for example, has shown
that patients who have exactly the same defect in the gene may be extremely
sick, mildly ill or completely healthy. 鈥淭his is the message, that even in
simple monogenic diseases, the relationship between genotype and phenotype is
complex,鈥� he says.

Weatherall argues that the capacity for misjudging signs and symptoms will be
massively amplified when the study looks at common multifactorial diseases such
as cancer, heart disease and schizophrenia.

Critics agree it would be unfair to pin the blame for such failings on family
doctors. Most surgeries in Britain are stretched to the limit, with little
chance of consultations longer than 10 minutes per patient.

鈥淚n culling information about genetic risks, GPs face some significant
problems,鈥� says Mike Pringle, chairman of the Royal College of General
Practitioners. First, he says, it鈥檚 difficult for GPs to establish family
histories of disease. Second, patients themselves don鈥檛 always know what their
relatives died of. Third, there鈥檚 a high chance in some parts of the country
that the man a patient calls 鈥渇ather鈥� is not the biological father. 鈥淵ou must
accept there鈥檚 no such thing as a complete and accurate medical record,鈥� says
Pringle.

So is there a solution, or will the survey simply have to live with the
expected errors? 鈥淚f you think the phenotype is not clear enough, you must do a
more specific project,鈥� says Purves. 鈥淵ou鈥檇 take a more detailed sample, and
look at patients in more depth, but it鈥檚 not affordable to do that across the
产辞补谤诲.鈥�

But some observers, such as Andrew Wilkie, a clinical geneticist at the
University of Oxford, wonder why, at this stage of the project, precise
guidelines have not been drawn up for all the GPs involved. He thinks each GP
could be trained and equipped to take specific measurements which would at least
provide unequivocal baseline data. 鈥淭here should be some kind of uniform
protocol,鈥� he says.

Tom Meade, who chairs the MRC expert working group that is refining the
study, is confident of finding a solution. 鈥淲e accept the importance of
phenotype determination,鈥� he says. 鈥淏ut it can be tackled.鈥�

One idea gaining favour is for highly trained reviewers at centres in each
geographical region to go through each GP鈥檚 data thoroughly, rating its accuracy
and validity against a checklist of clinical signs. Any dubious reports would be
discarded or investigated further.

Another is for highly trained research nurses in surgeries to take detailed
baseline data for each patient at the beginning of the study. 鈥淚t鈥檚 pretty
labour intensive, and you need these reviewers, but reliable phenotypic data can
be obtained provided you鈥檙e willing to spend time and effort on it,鈥� says Meade.
鈥淥f course it costs money, and that will influence what we can do,鈥� he adds.
鈥淏ut I don鈥檛 accept that we鈥檒l get results which are any less accurate than any
other large prospective study.鈥�

The value of the whole enterprise could in the end hinge not on the DNA
itself, but on the accuracy with which each illness is dissected. Paul Kelly,
the chief executive of Gemini Genomics, puts it more bluntly. 鈥淕arbage in,
garbage out,鈥� he says.