MIT biologist and Nobel laureate Phil Sharp recently warned that while knowledge generated by the genome-mapping effort is sure to lead to rapid advances in the diagnosis and treatment of disease, there's a void in the areas of securing patient consent and assuring genetic privacy. Sometimes, one man's void is another man's opportunity.
Arthur Holden, of SNP Consortium fame, created a company last October to address just those issues. Holden founded First Genetic Trust with Andrea Califano, who previously led the computational biology group at IBM, and David Wang, former head of genomics and bioinformatics at Motorola.
Through their involvement in the SNP Consortium, all three had ample opportunity to talk about what was needed to translate insights from genomics into medicine. And they realized that as the pharmaceutical industry moves closer to individualized medicine, one question would inevitably come to the fore: How will individual information be protected as the scale of genetic research expands?
"For years, people threw their arms up over that question, saying, 'Let the politicians deal with it,'" Holden says. "But I can't think of a singularly less qualified group to tackle this problem."
Instead, Holden and team have taken it upon themselves. And their solution is a bank.
Relying on Califano's technological prowess, First Genetic Trust is constructing a database for safeguarding genetic information and a network infrastructure to transfer it in a secure way — touted as comparable to the highest-level security systems in the commercial world. The company also plans to facilitate clinical studies by managing and analyzing data for universities, hospitals, and drug companies, and by using an Internet-based system to obtain patient consent.
Study sponsors would have ready access to data, but information that could reveal the identity of trial participants would be kept confidential. In fact, participants would hold private accounts at the bank that give them exclusive access to the most sensitive information, the company says.
It's an intriguing idea. But the questions it raises are numerous. Among them: How will First Genetic Trust serve the competing needs of potential clients — pharmaceutical companies, academic researchers, hospitals, patients, and HMOs — while still making a buck? Holden says he's got it all figured out. Initially, First Genetic Trust will generate revenue by supporting genetic research for drug companies or biomedical centers. It will also make money through genetic data analysis — an activity that Holden says could eventually become bigger than financial banking — and by setting up computer systems to handle data at hospitals. The company ultimately hopes to bank not only genetic information, but medical files and samples, so that when people change doctors or hospitals, they'll always know where to find their records.
To date, the company has raised about $15 million — most from venture capital firms Venrock Associates and Arch Venture Partners. And Holden was heartened by a recent meeting with 500 clinical lab directors, "many of whom are doing genetic diagnostics without getting informed consent," he says. "They're just not equipped to handle that step, and they let me know they need this stuff now. So I think our timing is great. If anything, we're a little bit behind the curve. But we're still ahead of everyone else."
For now, First Genetic Trust has 20 employees split between the business group in Deerfield, Ill., where CEO Holden and Executive VP David Wang are based, and the R&D group in Lyndhurst, NJ, headed by Chief Technology Officer Califano.
For Holden, the formation of this company represents a career trajectory that has "basically gone from big to small." His strategy has been to obtain management experience — as an executive at Baxter International, as CEO of Celsis International, and as head of the SNP Consortium — while waiting for an idea to launch him into an "entrepreneurial phase."
Wang, who trained as a physician in Beijing, left Motorola "to work more closely at the interface between technology and medicine."
Califano had been seeking "a bridge to the clinical world." The energetic 40-year-old, who gave up physics for computational biology, was on their wave-length even before hooking up with Holden and Wang, "but everything clicked when Arthur, David, and I got together," Califano says. "I also want to be involved with the people defining the bioethical standard."
"[Califano's] an amazing guy with a rare combination of technical virtuosity and a highly developed understanding of ethical and social issues," says Alan Buchanan, a University of Arizona professor who serves on the Human Genome Project advisory council. "He has a much greater awareness of ethical issues than most scientists."
FGT's essence is tucked away in a side room: Three racks of computers worth millions of dollars constitute the bank in its earliest stages. One houses the genetic database. The second, a bioinformatics cluster, does data mining. A third rack is devoted to software development, and that's where most of Califano's attention is now aimed.
His top priority at the moment is to install the computer architecture that will manage the data of patients enrolled by outside researchers in clinical trials. Components of the system he's assembling include: a genotypic information database; a phenotypic database; a user database for storing the identity of patients separately from their genotypic information; software packages for managing patient consent and collecting genetic and clinical information; and a Web-based application for managing interactions with patients, physicians, and laboratory researchers.
Operating requirements are stringent; confidentiality is crucial. But, as opposed to typical systems that anonymize patient data, FGT will avail the most sensitive, identifying information only to the patients themselves. Through a process called dynamic informed consent, patients specify how their genetic information can be used in clinical trials and subsequent research. Then, with a novel method for protecting privacy while keeping track of identities, patients can be contacted if new treatments become available or new studies are planned. Clinical researchers can also question patients after an experiment, which is not possible with de-identified samples. The approach is original enough, says Califano, "that we've filed patents around the entire process."
These patents cover both the consent protocol and security apparatus — concerns that lie at the heart of First Genetic Trust's mission. "In the science world, most computer systems used in genetic research are designed to deal with the mouse genome or C. elegans, but they're not secure enough for dealing with humans," explains Wang. The company is building highly secure systems comparable to any in use. Its computers will be protected by up to a half-dozen independent barriers "that would take hackers a very long time to get through," Califano says.
These firewalls provide perimeter security to filter Internet traffic between First Genetic Trust's databases and the outside world. "You're basically dividing the network into distinct zones and then drawing up a set of rules that governs what kind of traffic can travel from one zone to another," explains senior scientist Peter Young, a programmer recruited early on from the Whitehead Institute.
The company will also employ proprietary encryption methods for concealing all identifying information, as well as genotypic information. "There's also a link connecting sensitive information with non-encrypted information, and that link — a coded number that allows the system to keep track of identities — will be encrypted as well," says bioinformatics director Aris Floratos, who worked with Califano at IBM.
As a further precaution, every entry is individually encrypted. "If you manage to decrypt one message, hacking your way in by brute force, that's all you'll have," explains Califano. "But the system as a whole will not suffer."
This entire infrastructure has to come together on a tight schedule, with a limited system test set for mid-June and a full-system test in September. The next milestone is to have the system ready for clinical trials before the end of this year. Califano is confident that his team can meet the challenge. "We've had a specific timetable in place since November for getting our technology ready, and so far we've hit every point on the dot," he says.
Their efforts are specifically geared toward research that probes links between genetics and disease susceptibility, drug efficacy, and drug side effects. The last is a huge issue, according to Holden. Adverse drug reactions are the fourth leading cause of death in the US, costing $75 billion to $100 billion in health care expenses.
FGT is close to signing deals for clinical studies with several pharmaceutical companies. And breast-cancer trials with Memorial Sloan-Kettering, its first confirmed client, are due to begin late this year. The company will secure informed consent during testing and treatment, and provide genetic counseling and updated information over the Internet. "We'll manage [and analyze] the data for Sloan-Kettering," Califano says. They'll also look for associations between genetic markers, the onset of certain cancers, and patient outcomes.
If Sloan-Kettering researchers want to do additional studies, First Genetic Trust will take care of contacting and securing additional patient consent via the Internet. "If there's a hint of a genetic component in drug efficacy or side effects, we'll help them devise a new study that yields more definitive answers," Califano adds.
Conventional approaches to data mining work reasonably well for single-gene diseases like cystic fibrosis, says Califano, "but most diseases and drug interactions are not regulated by a single gene. Suppose you have a million SNPs and you go through them one at a time, asking, 'Is this SNP related to this disease?'" For a single-gene disease, there'd be a million possibilities — a tractable number. If two genes are involved, there are a trillion possibilities — a much more daunting computational task. With six genes, the number of possibilities expands to 1036, says Califano, "and no computer in the world could handle it."
He and Floratos employ mathematical tricks to narrow down the search, so they can "explore the full space without looking at all possible combinations. For example, there are  possible combinations of six genes and we just need to look at one." Similarly, if it's been learned through experiment that two genes do not interact, every combination of six genes that includes those two genes can be tossed out. "The idea is to use heuristics to make the search much more efficient," Califano says.
Mark Sobel, a genetic privacy expert with the American Society for Investigative Pathology, sees benefits for both parties in this arrangement. "Getting informed consent can be a paperwork nightmare," he says. "It's costly, time-consuming, and can delay research. Any group that can smooth the way will help everyone." But, he points out, First Genetic Trust will still have to secure approval from Institutional Review Boards, local ethics panels that oversee these studies.
Holden isn't worried about the IRBs, "since we're just developing an infrastructure to execute what they say." In fact, in a recent meeting, an IRB head who reviewed a First Genetic Trust draft research protocol told Holden, "I've been waiting 20 years for this type of system."
The company is working with the Food and Drug Administration and its European counterparts to make its system compliant for clinical trials, Holden says. Satisfying those requirements is not always easy, Young says, "because FDA doesn't tell you what to do, they just tell you what is needed for approval." European agencies are also a consideration "as drug studies become more and more globalized," Young says. "In the future, it might be hard to do much with a system that is only FDA compliant."
In the Balance
In addition to having questions about how First Genetic Trust will fare in an unpredictable regulatory landscape, observers such as Mark Sobel also wonder how secure FGT's technology will really be. "There's no such thing as an absolute, 100 percent guarantee," he says. "Hackers can even get into the FBI."
But Sobel doesn't see that as a reason to ignore the concept. "We can have something like 99.9 percent confidence, and it's worth that small level of risk, given the potential benefits from the research," he says.
Consumers' trust is crucial for the whole enterprise. "In order for genetics to advance, it's essential to get patients to participate in these studies," Holden says. To investigate multi-gene diseases and drug safety or efficacy, large populations are needed. But people tend to fear that their information could fall into the wrong hands, so they don't volunteer. First Genetic Trust is trying to address this concern head-on with what Holden calls a "Swiss bank-grade" approach to security.
Still, the question remains: Will people entrust their most sensitive personal information with a for-profit company? "Every bank makes a profit, but we still put our money there," Holden notes. "Because managing sensitive information is our main mission, our accountability is far greater than any other entity. If we mishandle private data, we're out of business."
Bearing High Interest
As trust in the company grows over time, and more people participate in clinical trials, says Califano, "we'll build a unique database through all these research studies that can be mined for scientific discoveries." This database will be used by First Genetic Trust scientists, as well as by outside researchers who might pay some kind of access fee.
"The value of our system will grow with the numbers in our database," says Floratos. "These algorithms are not simple, so places like Sloan-Kettering are not likely to have them," he adds. "That's why they might outsource the job to someone like us."
Meanwhile, Califano's already looking ahead to the next phase in life sciences, which he calls predictive or theoretical biology. "Right now there's almost nothing we can do predictively, because we don't have equations like a Faraday's law or Ohm's law of the cell," he says. "But there's a huge potential for representing complex biological systems by equations and modeling them. That's an interest of mine that could become an important activity for this company in the future."
Sidebar: The Gene Brokers
First Genetic Trust is trying to carve out a unique niche, but it overlaps with other companies in areas such as procuring and storing genetic information, analyzing data, and communicating with patients over the Internet.
DeCode Genetics The Iceland-based population genomics company made waves in December 1998 when the country's parliament passed a bill granting the company a 12-year license to create and manage a health database for its population. Some critics have questioned whether a single company should profit from the personal records of an entire nation, and others have challenged DeCode's consent policy — the company uses presumed consent, rather than informed consent, which enables it to use personal information unless told otherwise.
DNA Sciences Based in Freemont, Calif., the company focuses on the genetics of diseases such as asthma, diabetes, breast cancer, and colon cancer. Through the Gene Trust project, the company has recruited more than 5,000 people through the Internet to donate blood samples for genetic research. DNA Sciences promises not to share personally identifiable information with any outsiders, but researchers have access to de-identified information. The company provides research updates to Gene Trust volunteers but, unlike First Genetic Trust, does not make data available to participants.
Genomics Collaborative The Cambridge, Mass.-based firm has recruited more than 90,000 patients for studies — done in-house or with collaborators — aimed at finding the genes associated with common diseases and identifying pharmaceutical targets. Patients are given choices about how their samples will be used. "Unlike First Genetic Trust, we don't make data accessible to patients," says CEO Michael Pellini. "I don't see how they can utilize this information, given that the vast majority of physicians still can't interpret it."