Kári Stefánsson remembers exactly where he was when he formulated the idea for his company, deCODE Genetics. It was 1995, and he had recently moved from the University of Chicago to Harvard University. “I was sitting in Starbucks at the Beth Israel Hospital, and I put together this narrative for identifying genetic variants of a large number of diseases using data from a large number of people and also figuring out the structure of a population,” he says. “I proposed to do this in Iceland because the Icelandic population has the advantage of having started from a small number of colonizing individuals,” leading to what’s called a “founder effect.” He reasoned that because of this founder effect, the number of disease-associated genetic variants would be relatively small for each individual disease, so they’d be easier to identify among people in Iceland.
Stefánsson was a seasoned molecular biology and protein biochemistry researcher with a focus on neuroscience, but he had never done human population genetics studies. Still, he realized that such genetic analyses were the only way to probe the nature of human diseases in a model-independent way. “You can look for [genetic] variants without prior ideas of what causes the disease, so you free yourself from the necessity of beginning with a hypothesis, which I thought was very liberating.”
After that day, there was no turning back, Stefánsson says. He was determined to form this company. “I realized that to make a significant contribution to the field of genetics, the scope was too large to fit into academic research and to obtain the proper grant funding.” So Stefánsson went to venture capitalists in New York City, San Francisco, and Los Angeles to convince them to fund his company.
While his goal was to find genetic variants at the root of common diseases, he sold his idea to investors by proposing that deCODE would turn a profit by finding novel drug targets for pharmaceutical companies and by using human genetics data to identify subpopulations that would be more likely to respond to certain treatments—what is today called precision medicine.
Pitching this idea, he convinced seven companies to give him a total of $12 million in the span of six weeks. “I didn’t know much about how to raise funding,” he says. “I was just this eccentric man from Iceland who was telling potential investors his story.” Stefánsson moved back to Iceland in 1996 and began to build deCODE from scratch. “I thought that would take four years,” he says, “but instead it took almost 20.”
Starting with books
Stefánsson was born and grew up in Reykjavik. He was the second youngest of five children. “I was in the worst position in the order of children that you can think of, the child that tends to get the least attention,” he says. He doesn’t recall being starved of his parents’ attention, although that is how his eldest sister remembers it, he says. Stefánsson’s father was a radio journalist, writer, and then a member of Iceland’s parliament, while his mother stayed home with the kids.
At a young age, Stefánsson was more interested in books and writing than in science. His father wrote biographies and published an autobiography of his early life, growing up in an Icelandic fishing village, that “straddled fiction and nonfiction,” Stefánsson says. Stefánsson was a prolific reader and aspired to become a writer too. “When I was growing up, popular writers in Iceland had the status pop stars have elsewhere,” he says.
You free yourself from the necessity of beginning with a hypothesis, which I thought was very liberating.
Iceland was a relatively poor country then. “Our family was economically not at all privileged, but we had cultural privileges. . . . We were encouraged to read good literature, to write, and to be creative.” Stefánsson also enjoyed the outdoors, fishing with his father in the summer and riding horses.
Stefánsson excelled academically. He entered the College of Reykjavik in 1966 and majored in math. After graduating, Stefánsson went to medical school—following the lead of his best friend because he couldn’t decide what to do next. At the University of Iceland Medical School in Reykjavik, Stefánsson initially wanted to pursue psychiatry but then switched to neurology. “I was interested in the brain for the same reason as I am interested in it still today—the brain is the last frontier of biology, it is the organ we don’t understand, the organ of consciousness and emotion, which define us as a species,” he says. “I think it requires an extraordinarily dull mind not to be fascinated by the brain.”
After completing medical school in 1976, Stefánsson was accepted into the neurology residency program at the University of Chicago. He had missed the deadline to apply for that academic year, but the neurology department chair, Barry Arnason, was a Canadian of Icelandic origin and was impressed that an Icelander had applied. Arnason offered Stefánsson a position in his lab for a year before Stefánsson could join the clinical program. In Arnason’s lab, Stefánsson initially conducted laboratory research, isolating and culturing glial cells called oligodendrocytes from sheep. Oligodendrocytes produce myelin—a substance composed of fat and protein that wraps around the axons of nerve cells. The goal of the research was to better understand the pathology of multiple sclerosis, a myelin-related disease. This was Stefánsson’s first lab experience, and he became fascinated with research, publishing on the work in 1980. “Looking back, I am pleased to have done hands-on experiments where 95 percent of the time you are generating data and 5 percent of the time you are analyzing it,” he says. “In comparison, at deCODE, we spend 95 percent of our time on the data analysis.”
Stefánsson completed his clinical residency in neurology, trained as a neuropathologist, and then spent another year in a lab at the University of Chicago, where he found that Müller cells, the glial cells in the retina, had similar structural properties to oligodendrocytes, including the presence of myelin-associated glycoprotein.
In 1983, Stefánsson became an assistant professor in neurology at the university, running a lab to study the role of myelin in multiple sclerosis and other autoimmune diseases and specializing in multiple sclerosis as a clinician. In 1991, he became a full professor, and in 1993, he moved his laboratory to Harvard University to become the chief of the neuropathology division at what was then called Beth Israel Hospital, now known as Beth Israel Deaconess Medical Center.
Having secured funding for deCODE in 1996, Stefánsson moved back to Iceland, originally turning an old building that had housed a photography store into a laboratory. He hired a few colleagues from the US, and within a year, the company grew to about 100 employees.
The deCODE team started constructing a database of the existing genealogy of almost the entire Icelandic population, going back hundreds of years—and making that information accessible to the people of Iceland. The researchers initially asked, with just birth and death data, whether individuals who lived until at least the age of 90 were likely to have relatives who also lived past 90. In fact, those who did live to 90 were more likely to be related to each other, indicating that there is a genetic component to long life. Analyzing the pattern of longevity among families, the researchers also showed that the people who live longer are more likely to inherit a positive factor—one or several genes—that can stave off multiple diseases. Stefánsson’s team then used a similar approach, adding medical record data, to show that there is a genetic component to the most common late-onset form of Parkinson’s disease.
DeCODE also explored how the human genome has evolved. In 2004, the team analyzed how reproductive success is affected by the way that the maternal and paternal genomes are scrambled during the formation of egg and sperm. They used genome-wide microsatellite data on 23,066 individuals and came to the conclusion that a small portion of the genome, just 10 percent, has a higher rate of recombination, suggesting that characteristics driven by genes in these regions evolved faster than phenotypes coded by genes in other parts of the genome.
I think it requires an extraordinarily dull mind not to be fascinated by the brain.
It’s still not clear what these characteristics are, according to Stefánsson, but the team is investigating this further. And while new mutations across the genome are more likely to come from the father (because spermatogenesis is a continuous process throughout a man’s life while oocytes do not divide postnatally), in this 10 percent of the genetic code both the father and mother equally contribute to the de novo mutation rate. The analysis also showed that a high recombination rate was linked to increased viability of a fetus, particularly in older mothers. In addition, mothers who had a high oocyte recombination rate were more likely to have more children. Again, the results raised more questions than they answered, so Stefánsson and his colleagues are continuing to dig deeper.
As Stefánsson had envisioned on that fateful day in Starbucks, human disease risk is another focus of the company. In 2009, the deCODE team identified common genetic variants that result in an increased risk of developing schizophrenia. “We showed that those who carry the variant but don’t develop schizophrenia have a departure from the norm in terms of cognition, in a similar way as schizophrenics but not as severe,” says Stefánsson. “This suggests that what comes first is the abnormality in thought and then the schizophrenia, and not that the schizophrenia is a prerequisite for thinking differently.”
For Stefánsson, this is among the first clues to figuring out how thoughts develop in the brain. Another clue is that there’s an overrepresentation of individuals with bipolar disorder and of healthy siblings of people with schizophrenia and bipolar disorder among creative professions, but, according to Stefánsson, it hasn’t been clear whether creativity and these mental illnesses had shared biology. “We showed that Icelandic writers, painters, and others in creative professions have a higher predisposition for schizophrenia, suggesting that at least in part, creativity and schizophrenia share biology.”
In another study, deCODE researchers uncovered a genetic link underlying the phenomenon that individuals who obtain a higher education typically have fewer children, showing that genetic predisposition to obtain education may also predispose an individual to have fewer children. Analyzing about 130,000 people in Iceland, the team identified genes associated with completing higher education (educational attainment) and found that these genes decreased in frequency in the population between 1910 and 1990. In other words, the genes associated with educational attainment were under negative genetic selection. If an individual harbored these “education” genetic variants, they were also more likely to have later and fewer children. It was not simply a matter of investing more time in education and thus less in childrearing: those who had these gene variants but did not attain higher education also had fewer children than individuals who did not harbor the “education genes,” suggesting a link between these education-associated variants and fertility.
A genomic frontier
In 2015, Amgen bought deCODE, but Stefánsson says the biotechnology company has left him and his team alone to continue to make new discoveries and publish their basic human genetics research. DeCODE now has genealogy data on all Icelanders, blood samples from about 160,000 of those individuals, and whole genomes on about 60,000 volunteers who have given deCODE informed consent to use their data for research. “We have sequence information on most of the nation, which puts us in a position to do fairly powerful genetics but also places a high responsibility on our shoulders on privacy and data protection,” Stefánsson says.
Reading is as important for me as eating. I cannot survive without it.
Iceland’s governmental bioethics committee oversees the data. The deCODE database includes individuals’ disease-causing genes, such as a mutation in BRCA2 that’s common among Icelanders and confers an 86 percent probability of developing cancer on the women who carry it, as well as increasing the risk of prostate cancer in men. The Icelandic government, however, does not allow itself or any private institution to provide people with their genetic information without explicit consent. “It’s a complicated issue,” says Stefánsson. “It seems self-evident that anyone should have access to their data.” But, because deCODE is sequencing data for bulk population studies, and not in a clinical-grade laboratory, the company cannot guarantee that each individual’s sequence is completely accurate. Therefore, because the national law prohibits both deCODE and the healthcare system from contacting those with disease-related genes, the company’s solution is a website where the team has deposited better-quality sequencing information on BRCA2 and other genes, allowing individuals to find their data if they wish.
After 23 years at deCODE, Stefánsson says he has achieved more than what he expected when he hatched his plan for the company in Starbucks, and he has no plans of slowing down now. “I am extraordinarily pleased with the contributions we’ve made to human genetics and evolution.”While staying engaged with his research, Stefánsson still manages to read about 40 novels a year. “Language is the equipment with which you think, and the best way to train yourself on the use of language is to read good literature,” he says. “Reading is as important for me as eating. I cannot survive without it.”