by Stephanie Overby

Iceland’s Dilemma: Privacy Versus Progress

Jul 15, 2001 14 mins

A small Icelandic startup has been granted a 12-year license to create and manage adatabase of the entire nation's medical and genetic records. Can it make medical history without violating patient privacy?n

The first thing you see on the Icelandair video screen after takeoff on the flight from the United States to Keflavik International Airport isn’t a sweeping tour of Iceland’s famous glaciers, geysers, healing hot springs or vast lava fields. It isn’t even an introduction to the country’s lucrative fishing industry or its legendary Norse sagas. It’s a three-minute presentation on Iceland’s newest national treasure, DeCode Genetics.

It’s not your average airplane fare, but DeCode is not your average company. Since its founding in 1996, the U.S. genomics company headquartered in Reykjavik has become as much a part of the landscape there as Iceland’s natural wonders, history and culture. DeCode gained celebrity status last year when the Icelandic government granted it an exclusive 12-year license to create and manage an electronic database of the country’s medical records, previously scattered in clinics around the island. More important, the license also permits DeCode to cross-reference that data with genetic and genealogical information it is collecting on individual Icelanders. As a result, a private company now has unprecedented access to almost every Icelander’s most intimate personal information.

The embarrassment of data riches puts DeCode at an advantage over any other company doing similar genetic population studies. And DeCode has capitalized by developing sophisticated software tools to locate genetic variations that may be found in a number of diseases. The company’s goal is to discover then market new drugs and diagnostic tests for those ailments.

Last October, for example, the company announced the discovery of a gene it believes is linked to schizophrenia. DeCode scientists studied DNA samples from 400 Icelandic schizophrenia patients and their unaffected family members to locate genetic variations believed to be involved in the disorder. DeCode made that discovery while working with the Swiss drug giant F. Hoffman-La Roche in what observers say is the largest deal ever between a pharmaceutical company and a genomics company.

Still, DeCode’s success is far from certain. Indeed, the fact that the company has been granted state-sanctioned access to personal information has sparked an intense controversy at home and abroad. Mannvernd, an Icelandic grassroots organization, plans to take the company to court for violating Icelanders’ right to privacy. One-third of the doctors belonging to the Icelandic Medical Association are refusing to hand over medical records to the database without first obtaining their patients’ written consent. So while DeCode is forging ahead in true Viking fashion, its path is strewn with roadblocks as large as the boulders that pockmark this remote island. Companies worldwide are watching closely. After all, if DeCode can overcome the political and ethical barriers to using sensitive medical data, its success will provide priceless lessons to every organization seeking to mine personal data for profit.

The Perfect Population Base

Dr. Kari Stefansson strides into the conference room, portable phone in hand. The 51-year-old CEO of DeCode fits the stereotype of an Icelander — tall and lean with a shock of spiky white hair and beard. He sounds like Sean Connery, but without the charm. He answers his high-tech phone in the middle of our interview and hangs up on the caller without a word as soon as he deems him insufficiently important.

A native Icelander, Stefansson left a cushy academic environment to develop plans for DeCode in 1996. As chief of neuropathology at Boston’s Beth Israel Deaconess Hospital, Stefansson had grown frustrated with the pace of academic research. At the time, he was studying multiple sclerosis and had become convinced there were inherited factors involved. And he was beginning to see the disease and its genetic roots as an information technology issue. “The fact of the matter is that the fundamental unit of life is a bit of information,” Stefansson says. He felt strongly that the most efficient way to study those bits of information would be in the private sector. And he knew the perfect place to locate his startup.

Iceland, an island the size of Kentucky just below the Arctic Circle, has seen little immigration in the last 1,100 years. The population of 280,000, kept small throughout the centuries by plague and natural disaster, is almost entirely descended from a small group of ninth century Norse and Celtic settlers. Thanks to a national obsession with ancestry, there are records tracing those family ties to 75 percent of all the Icelanders who have ever lived. And, owing to a nationalized public health-care system, some medical records date back to 1915. Well-documented medical and genealogical records on this relatively homogeneous group of people mean that genetic mutations should be as conspicuous as geysers in the middle of the Icelandic countryside.

Even before DeCode was given the go-ahead to create the national medical records database, it had begun to work on the genetics of specific diseases, using the enormous amount of data available on the local population along with blood samples from willing participants. Now that it can plumb Icelanders’ medical records and match that information with genetic data culled from volunteer blood samples, DeCode hopes to identify and describe the function of genes or genetic variations that contribute to the onset of 35 common illnesses, including Alzheimer’s, lung and prostate cancers, Parkinson’s and osteoporosis.

One disease DeCode has focused on is asthma. Asthma patients and their relatives give blood to DeCode through a network of collaborating physicians. DeCode then processes the DNA in its laboratories, obtaining genotypes (variations in the genome) for each asthma sufferer. Those are then matched with the patient’s medical records, linking genotype to phenotype data — physical details, including the person’s age and weight. DeCode fits each genotype into a web of family inheritance patterns to study the genetic differences between affected and unaffected relatives. The company then uses a suite of statistical analysis tools and sequencing equipment to locate common genes that may be involved in the development of asthma. Once those genes and the proteins they code for are revealed, DeCode and other companies can experiment with drugs that can directly target key proteins and thereby halt the disease process.

Thus far, DeCode has announced the discovery of about 24 genes it believes are related to more than 15 diseases. And Roche has begun programs to develop new drugs for three diseases: schizophrenia, stroke and peripheral arterial occlusive disease, a narrowing of the arteries in the arms and legs.

The Science Begins with Software

When DeCode first began the process of creating and mining its enormous genealogical and clinical databases, Stefansson quickly realized his company would have to become a software development business as well. There simply wasn’t a lot of existing software that could handle the mammoth databases that DeCode was exploring. So DeCode began creating a suite of tools, ranging from statistical analysis programs to security software, to aid its disease-gene discovery process, protect its data and provide a potential stream of revenue for the fledgling company. In fact, when you enter DeCode’s headquarters on the outskirts of Reykjavik, you won’t see scientists in lab coats or any of the 50 ABI Prism 3700 DNA analyzers the company recently bought. Instead, the nondescript office building, which looks like an insurance company inside and out, is mostly populated with employees at workstations. There is not a test tube in sight.

Hakon Gudbjartsson, who is in charge of DeCode’s software development efforts as vice president of infomatics, is the soft-spoken, even-tempered converse of Stefansson. Gudbjartsson returned to Reykjavik from postdoctorate work at Boston’s Brigham and Women’s Hospital to join the company in 1996. To create effective software for this emerging field, he works hand in hand with many of the 500 DeCode employees who use IT — from statisticians and disease project managers to anthropologists and genealogical historians.

The first piece of software DeCode developed was Project Manager, so named because it tracks and remembers all the blood samples and medical and genealogical data for each disease project. Project Manager also allows DeCode to analyze the data using statistical algorithms created in-house and elsewhere. Among the unique tools wrapped into the Project Manager is DeCode’s HaploTool, a statistical analysis program used to determine what versions of a gene are shared among families being studied for particular diseases. HaploTool builds on Allegro, a multipoint linkage analysis tool developed by DeCode that is used to examine the genotypes of families by placing genes or genetic markers on chromosomes and determining their proximity to each other. The closer the markers are, the more likely they’ll be inherited together — important information because many diseases, such as Alzheimer’s and MS, are the result of many genes working together.

DeCode uses these IT tools to determine what unique genetic traits or sequences people with a certain disease share. Its scientists use Project Manager, HaploTool and Allegro to find a location or tiny piece of the chromosome that is shared by disease sufferers but not by unaffected participants.

“You’re just narrowing, narrowing, narrowing, narrowing the chromosome down to a very small region so you can work with it,” explains Dr. Hakun Hakonarson, DeCode’s director of pharmacogenetics.

The researchers then use GeneMiner, a software package created by DeCode, to sequence the targeted region on the chromosome to determine if there in fact is a candidate gene there and if so, what it does. For example, when studying schizophrenia, researchers found more than one candidate gene in the same location, but they determined that only one of the genes codes for a protein is involved in regulating the central nervous system and thus is more likely to be involved in schizophrenia.

DeCode’s IT employees have developed a tool for cross-referencing the three data sets the company works with — the medical records database, the genealogical database and the genotypic data from its disease projects. An analytical tool called the DeCode Clinical Genome Mineris will link the three. The company hopes to make money by selling subscriptions to the Clinical Genome Miner and licensing its software tools, including its system for encrypting medical identifiers. And some analysts think that is a viable strategy for the young company. Ian Smith, a London-based biotechnology analyst for Lehman Brothers, says the company’s early accomplishments illustrate the strength of DeCode’s software. “Some of the announcements they’ve made recently and the milestone payments they’ve received for drug targets that Roche is now working on offer proof that they can do what they set out to do,” Smith says.

Besides its high-profile partnership with Hoffman-La Roche, which could be worth more than $200 million, the startup is collaborating with Partners HealthCare of Boston to compare the results of population genomics in a relatively homogenous population (Iceland) with the results in a more heterogeneous one (Boston). Partners is conducting its own genomics studies on osteoarthritis, schizophrenia, asthma and narrowing of the arteries.

The Price of Privacy

Not everyone is thrilled about DeCode’s work, however. Opposition has been growing ever since the Icelandic Parliament, the oldest functioning parliament in the world, passed the Health Sector Database Act in 1998. The Act enabled the Ministry of Health to grant a license to create and operate a national database of medical records, and in January 2000 it awarded that license to DeCode.

Petur Hauksson is the psychiatrist who founded Mannvernd (an Icelandic word that means human protection), an organization formed to overturn the Health Sector Database Act. Hauksson says his countrymen haven’t been exploited to this degree since they broke free from Danish rule in 1944. To show one example, he points to what he sees as an unusual degree of collusion between DeCode and the Icelandic government. After all, it was Stefansson, DeCode’s CEO, who originally came up with the idea for the centralized database and supported legislation that established the database license.

“Many of my psychiatric patients and their relatives were very concerned about this from the beginning. They’re used to discrimination, and their health information is very sensitive,” Hauksson says. “But I think all medical information is sensitive.” Hauksson says people who become seriously ill or whose DNA may predict the likelihood of contracting some condition might also be discriminated against.

One of Mannvernd’s main complaints is that the Health Sector Database Act is based on the presumed consent of Icelandic citizens. Furthermore, Icelanders who consent to give blood for one of DeCode’s disease studies must agree to have their DNA used for other studies without knowing what those might be. Thanks to Mannvernd’s complaints, Icelanders can now opt out of having information entered into the database by filling out a form, and so far 20,000 of Iceland’s population of 280,000 have. Of course, if their information is already in the database, they don’t have the right to opt out.

But now Mannvernd has a new dragon to slay. Until March 2001, the medical data, blood samples and genealogical information DeCode has access to was sent first to the Icelandic government’s Data Protection Commission, where the personal identification numbers on them were encrypted and then sent on to DeCode. One or two government workers did the encryption, an iffy system that created some lag time. “It’s a definite security risk if a government worker has an encryption key in his wallet that he might lose,” Gudbjartsson says.

To address such risks, DeCode developed an encryption tool known as the Identity Protection System. After months of testing the system, the government replaced those individuals carrying encryption keys in their pockets with the automated encryption software. The encryption takes place at the Noatun Research Services Center, a blood collection facility financed by DeCode, where doctors send those patients who agree to participate in DeCode studies. The Data Protection Commission of Iceland members can log in remotely and find out what data has been sent back and forth and halt the process if they so choose.

But Hauksson and others say that doesn’t go far enough. In a small country like Iceland, they argue, it’s easy to figure out individual identities based on clinical data, even if personal identification numbers are encrypted. Mannvernd has filed a case in court, suing the government and DeCode for violating Icelanders’ right to privacy.

DeCode says a majority of Icelanders support its work, at least according to a recent Gallup Poll it commissioned. But Hauksson attributes that result to a DeCode marketing campaign that appealed to Icelanders to contribute to the global fight against diseases. Though Hauksson thinks the Icelandic population does offer a wealth of data for genetic disease research, he would prefer that public institutions use the data in the interest of medical progress rather than profit. Neither DeCode nor its partners, Hauksson points out, have revealed any data on their disease-gene discoveries for others in the genomics community to verify or build on.

Opposition to DeCode’s information monopoly isn’t limited to Icelandic naysayers. Dr. George Annas, a medical ethicist and professor at Boston University’s Schools of Law, Medicine and Public Health, says DeCode should obtain informed consent from individual Icelanders before including their medical records in the database and should allow them to withdraw from studies they don’t approve of. The Data Protection Commissioner of the European Union and the World Medical Association has echoed those beliefs.

DeCode’s Stefansson shrugs off the criticism. “The opposition isn’t large; it’s just been somewhat noisy,” Stefansson insists. Pointing to DeCode’s third-party encryption technology, he says it would be a lot easier to get information on Icelanders by simply walking into a doctor’s office and grabbing their records.

Annas says that the controversy over DeCode is mild compared with what would erupt if other companies and national governments attempted similar collaborations. And while no one expects a U.S. company to gain the kind of monopolistic access to national medical, genealogical and genetic information that DeCode has, similar databases are expected to proliferate on a smaller scale, generating the same ethical concerns.

“[DeCode] took advantage of a very different climate in Iceland about these issues,” says Dr. Russ B. Altman, president of the International Society of Computational Biology. “I don’t think a U.S. company could do the same thing, but many companies may form alliances with smaller medical entities, like Kaiser Permanente or another large health-care organization, to assimilate some of this data.

“Ethical issues,” Altman concludes, “are paramount in this kind of research.”