A Genomic Treasure Hunt May Be Striking Gold
June 18, 2002
By NICHOLAS WADE
REYKJAVIK, Iceland - In the year 874, Viking crews from
western Norway started to drop in on Ireland, capture an
allotment of young Celtic women and sail off northwest to a
remote island beyond the reach of retribution.
Eleven centuries later, a direct descendant of those
Icelandic pirates and their slave wives, Dr. Kari Stefansson, says he
is starting to extract a tremendous prize, made possible by Iceland's
tiny, isolated population and its obsessive interest in genealogy: a catalog
of the deviant genes that cause the most common human diseases.
He hopes to snatch this prize away from competitors who are
hunting for the same errant genes by a different method,
including those relying on information from the Human
Genome Project, a $3 billion program financed largely by
the United States government.
Dr. Stefansson's six-year-old company, Decode Genetics,
says it has already mapped the general location of the
errant genes for 20 of the 50 common diseases on its list,
including Alzheimer's, anxiety, asthma, hypertension, obesity, Parkinson's
disease, macular degeneration and
rheumatoid arthritis, and has also found a region holding a
Within the 20 mapped locations, the Decode scientists say,
they have identified specific disease-causing genes in 3,
those for schizophrenia and two forms of stroke.
Discovery of the variant genes that promote a disease is a
first step to developing drugs that treat its root cause. It is also a
way of quantifying the genetic components of disease.
Icelanders who possess the variant stroke gene, Decode
researchers have found, have a risk of stroke five times as
great as normal, making it a larger risk factor than smoking, high blood
pressure, cholesterol and age.
The claims of the company are hard for others to assess
because so far it has published articles on only two of the
disease locations it has found. But Dr. Jonathan Knowles,
head of global research at Hoffmann-La Roche, said that
Decode's progress had been extraordinary and that 13 of the
disease gene locations, found under contract to Roche, had
been verified for Roche by independent experts.
Beyond mapping, finding the actual genes is an even more
impressive claim. "If they have really identified three
genes for three complex disorders, that would be outstanding," said
Dr. Nancy Cox, a statistical geneticist
at the University of Chicago. Only a minuscule number of
complex disease genes have been found so far, she said.
Dr. Stephen Warren, editor of The American Journal of Human
Genetics, said: "On face value, their approach is very compelling.
It's logical, and I'm not surprised at their success."
A spokesman for the National Institutes of Health office that finances
the Human Genome Project said Decode's progress could not be judged until
it published its claims.
Discovering the genetic roots of common disease was one of
the promised fruits of the Human Genome Project. But taking
this next step has proved enormously difficult. The sequence of DNA letters
in the human genome, obtained in draft form in June 2000, says nothing
of the changes in the DNA that cause disease.
Although geneticists have pinpointed the genetic errors
behind certain single-gene diseases, like Huntington's and
cystic fibrosis, these diseases are comparatively rare. Most common illnesses,
including cancer, stroke, arthritis and schizophrenia, are thought to
be caused by several errant genes acting together.
Because each errant gene makes only a small contribution to
the disease, it has no obvious pattern of inheritance and its presence
is hard to spot among the natural variation in DNA sequence, making progress
But by focusing on Icelanders, a population of just 278,000
whose careful genealogic record keeping allows disease genes to be traced
back more than 10 generations, Dr.
Stefansson believes he has found a general method for analyzing diseases
involving several genes and is well ahead in the race. "The genome
isn't doing that much at the moment," he said.
Dr. Stefansson, 53, is a former Harvard neuropathologist
who decided that his native Iceland, because of its small
population and rich genealogy, was the ideal place to seek
the causes of multigenic diseases. Rejected for a National
Institutes of Health grant on the ground that his idea would not work,
he was able to raise private money instead.
But his proposal caused a furor. Critics accused him of risking Icelanders'
medical privacy to profit from their genetic patrimony. After a long debate
in Iceland's Parliament, the Althing, Decode was allowed to proceed, though
with stringent privacy safeguards, including the encryption of patients'
Luring back Icelandic scientists from abroad, as well as
many Americans and other experts, Dr. Stefansson has built
up a 600-person biotechnology company in Iceland, a beautiful but semi-barren
expanse of volcanic rock with few
industries apart from fishing and aluminum smelting. "It's
a world-class operation, and extraordinary to find it in
Reykjavik, previously best known for research on fish," Dr.
The key to Decode's approach is a genealogical database
that extends back 1,100 years. It has been reconstructed
from a mixture of sources, including calfskins that hold the first 300
years of records, church archives and three complete censuses starting
in 1703. The database contains the names of about half the Icelanders
who ever lived, including 85 percent of those born in the 19th century
and almost everyone born in the 20th, said Dr. Thordur Kristjansson, a
To look for variant genes that dispose a person to asthma,
say, Decode collects the names of asthmatic patients from
doctors and runs them through the genealogy database. The
computer spits out family trees that are often surprisingly
deep, linking patients who had no idea they were related.
One cluster of 104 asthma patients turned out to be descended from a single
common ancestor, born 11 generations back in 1710, said Dr. Hakon Hakonarson,
Decode physician who is looking for asthma genes.
In a room-size freezer in the basement of Decode's
Reykjavik headquarters, a robot named Goliath is swaddled
with insulating clothing. Designed for life in an auto
assembly plant, Goliath sits surrounded by racks of blood
samples from some 65,000 Icelandic residents, about a third
of the adult population. The robot can pull out requested
samples and send them upstairs to a fleet of 56 DNA
sequencing machines, part of what Decode says is the
largest genotyping facility in the world.
Genotyping is a way of analyzing a person's genome to see
which sections of DNA a cluster of asthma patients, say,
may have in common. In Decode's method, DNA from a
patient's blood cells is analyzed at 1,000 different sites
along the genome. These marker sites define the intervening
sections, enabling computers to pick out which particular
section a group of patients has inherited more often than
Outside Iceland, several teams hunting the genes in
multigenic diseases have succeeded in implicating a section
of DNA but failed to find the causative gene because the
flagged section contained so many candidates. Decode's
flagged sections are much smaller, Dr. Stefansson said,
thanks to the depth of the Icelandic genealogy.
Because of the mingling of genetic material at each
generation, which progressively whittles down any block of
genes from a given ancestor, the disease-linked sections
identified by Decode's computers tend to hold just a
handful of candidate genes.
Another feature of the Icelandic population helpful to gene
hunters is its relative homogeneity. The founders were some
10,000 to 15,000 men, mostly Vikings from Norway, and about five times
as many women as men, mostly from Ireland. The population remained small
and isolated until recent times, and its genetic diversity was probably
reduced by three major disasters: the death of almost half the population
from an outbreak of pneumonic plague in 1402, the loss of another third
from smallpox in 1708 and a 20 percent reduction in the famine that followed
a volcanic eruption in 1784.
The loss of diversity means that many Icelanders with a
disease have inherited exactly the same causative mutation,
which is therefore easier to identify. In larger populations, a particular
gene may have different DNA changes, or mutations, in different patients,
making each mutation harder to spot.
Even Icelanders have three different versions of the disease-causing stroke
gene detected by Decode.
It is not yet clear whether drugs based on errant genes
found in Icelanders will work elsewhere. Icelanders may not
possess the full spectrum of human genetic weaknesses, or
they may have them with different frequency. "Even if
Decode finds the biggest gene for schizophrenia in Iceland,
the biggest in England may be different," said Dr. Cox of
the University of Chicago.
But Icelanders have much the same overall pattern of common diseases as
do Europeans and Americans, prompting Decode scientists to hope that whatever
drugs or diagnostics they may develop will work for others too.
Identifying the major human disease genes has been an
important goal of the Human Genome Project. Having decoded the sequence
of most of the three billion DNA units in the human genome, the project's
leaders are now compiling a catalog of common variations in the genome
They hope that these variations, single unit changes in DNA
known as SNP's or "snips," can be associated with patients
who have particular diseases. (Celera, the private company
that is the public project's chief rival in the race to sequence the genome,
is not hunting for disease-causing genes.)
Dr. Stefansson expressed confidence that his genealogy-based approach
would prove more efficient. One reason is that he first analyzes each
patient's genome at only 1,000 sites, compared with the 100,000 sites
that may be needed with the SNP approach.
Since each analysis costs about 50 cents, the difference is
significant. The Decode gene hunters do use SNP's, but only
after they have pinpointed the section of chromosome where
a suspect gene lies.
Though Dr. Stefansson compares his gene-hunting operation
to that of the Human Genome Project, American geneticists
note that everyone, Decode included, depends on the
information the project has obtained and placed in the
public domain. They also say there is so much to discover
that Decode is unlikely to snatch the whole Pandora's box
of human disease genes away from rival investigators.
"I don't think they will make a clean sweep, but they will
make significant contributions," Dr. Warren said.