Several years ago, Dick Mallery learned that his wife had developed a rare form of gynecological cancer. The diagnosis shocked the couple, especially because she had consciously avoided carcinogens in her diet and lifestyle. They sought the proper course of treatment and a biopsy was taken. To Mallery's amazement, the medical center discarded the sample.
"I'm not a scientist," says Mallery, who has been easing toward retirement from his Phoenix law practice in the past two years. "Yet the disease was so rare, I understood right away how [the sample] might have been able to help someone in the future."
Inquisitive by nature, Mallery began a search to understand why tumor samples such as his wife's aren't preserved for study. He discovered that no two private tissue banks operate alike, each using its own standards and naming conventions to analyze similar tumor types. And he learned that cancer centers all have their own methods for handling samples.
Along the way, he met oncologist Daniel Von Hoff of the University of Arizona Cancer Center in Tucson who had a brainchild for addressing exactly the problem Mallery was investigating. Von Hoff's idea was simple, but compelling: If US medical centers would standardize their methods for collecting, storing, and analyzing tumor samples, as well as their protocols for obtaining the consent of patients to have their tissue studied, then the data could be stored in a central repository. Von Hoff envisioned a global system by which all sample analyses, annotated in a uniform manner and submitted to a database, could be accessed by both private and public researchers.
Von Hoff, who has tested new cancer drugs with the pharmaceutical industry for years, first presented his concept, which he dubbed the International Genomics Consortium, at the American Association of Cancer Research meeting in April 2000.
Jeffrey Trent, then a prominent genome researcher with NIH's National Human Genome Research Institute, had known Von Hoff for nearly two decades. "Every time I ever met with Dan, he would be after me about how we could incorporate genetics and expression profiling into the treatment of cancer," says Trent, an Arizona native.
Trent was a career scientist eager to pursue a more applied path after years of basic genome research at NIH. But it wasn't until he and Von Hoff hooked up with Mallery in the summer of 2000 that the vision began to become reality. Mallery, a mergers-and-acquisitions attorney, was their rainmaker. He established the nonprofit structure of the IGC and began building consensus among research institutions and industry. He dove into the project obsessively after his wife passed away, seeking to build support for the IGC and biomedical research in Arizona.
For the Greater Genomics Good
With the nearly finished map of the human genome, researchers now have access to an abundance of genotype data. Yet phenotype and gene expression data exist in a jumble of incompatible languages and disparate databases. Moreover, medical centers regard phenotype data as proprietary and biopharma companies can only access such data selectively through clinical trials.
A consortium uniting medical centers and pharmaceutical companies, Von Hoff and Trent reasoned, could maximize the research strengths of both groups. Adamant that public sector researchers should be availed of equal access to the resource, they determined from the start that funding partners would not be given preferential access to the database.
Arthur Holden, chairman of the nonprofit SNP Consortium who advised Von Hoff and Trent as their idea took shape, says the International Genomics Consortium is a logical successor to the Human Genome Project and the SNP Consortium. Those two undertakings "created environments of [genotype] data that can be used and reused in research," says Holden, who is also CEO of First Genetic Trust, a company concerned with the technological challenges of storing, distributing, and protecting genetic data.
"SNP data is just the letters, or the alphabet, of genomics. It's not the words or paragraphs," he says. "That's why the IGC is so important. It establishes a platform, like the SNP Consortium, where phenotypic and gene expression data [in cancer initially] can be reused." Holden adds, "The IGC's objective is for the greater public good, and a public-private consortium is the only cost effective way to achieve it."
The first conceptual meeting for the IGC was held in December 2000 in Phoenix. A month later, Trent contacted Stephen Fodor, chairman and CEO of Affymetrix, to appeal for funding assistance. Opportunities for the use of the chipmaker's technology in the project were abundant, and Affymetrix readily agreed to help out with an undisclosed sum of cash. At the same time, Trent and Nic Dracopoli, executive director of clinical discovery technologies at Bristol-Myers Squibb, which has yet to decide if it will back the project, asked Holden to bring his experience with the SNP Consortium to the IGC at the same time.
More planning meetings followed at Johns Hopkins University and during the 2001 cancer research conference. Affymetrix executive Grace Colón attended a number of those before signing on as IGC's interim chief operating officer in October 2001. Colón laughs when she recalls that her tenure was intended to be three months but stretched to nine. Even after she returned to her full-time post at Affy in January 2002, Colón continued to work weekends for the IGC.
Of course, Affymetrix saw the consortium as an opportunity for the use of its microarrays in treating cancer, she says. But the real draw was the chance to contribute to the standardization of all phases of tissue collection and analysis, which Affy believes will be essential to advancing cancer research.
"Everyone understands it's very confusing to [compare] results of gene expression when different storage and collection techniques are used," Colón says. Researchers already have been able to draw some conclusions in spite of the confusion, she notes, but to advance cancer research, hundreds of samples have to be collected and fully validated with robust signatures and interpretations. The IGC is the logical public-private forum to establish standard protocols that use similar algorithms of analysis.
"Moving to standardized [research] protocols is key for the future of cancer therapy," says Colón.
Affymetrix, Agilent, and Motorola donated chips and equipment to launch a pilot project to analyze 100 brain tissue samples in October 2001. An Arizona charity, the Flinn Foundation, donated funds for consulting services, and the pilot moved ahead while Von Hoff and Trent promoted their idea to medical centers across the country. They succeeded in recruiting, among others, the Mayo Clinic, Duke University, Emory University, the M.D. Anderson Cancer Center, and the Moffitt Cancer Center.
Three of the medical centers' home bases, North Carolina, Atlanta, and Houston, lobbied to host the IGC, as did Phoenix and Baltimore. But with Trent leading a separate effort in his home state to form a biomedical collaboration among Arizona's three major universities, and with Mallery boosting medical research there too, Phoenix was the strong favorite.
In June this year, state and local governments in Arizona agreed to supply funding that would house the IGC in Phoenix, as well as an additional $92 million to fund The Translational Genomics Research Institute, a complementary outfit also to be directed by Trent (see "Tribal Translational Genomics," this page). Before yearend, IGC organizers expect to secure commitments for the $35 million they need to fund their first three-year project, which will be called expO — the Expression Project for Oncology.
Once they've secured funding, the organizers hope to install a permanent CEO with the same type of industry and public sector consensus-building experience that Holden brought to the SNP Consortium. Holden, who has been watching IGC's development closely, says, "The bread is about half-baked and soon will come out of the oven."
A Daunting Scale
The expO project aims to collect and analyze 10,000 tumor samples during a span of three years. Each analyzed tumor sample will produce a microarray of 10,000 spots, and each of those will be linked to the donor's basic clinical information and treatment regimen data. Each patient sample record could contain as much as a gigabyte of data, organizers estimate. Collecting the tumor samples and distributing them to participating institutions with the efficiency they desire will require a computing cluster capable of performing 7.5 trillion calculations per second.
The scale of the project would be too massive for any single research organization to undertake. The consortium approach simply makes good sense, IGC's organizers argue.
In addition, a collaborative research arrangement could be even more appealing to big pharmas during the current economic climate. In the '90s, gene-sequencing startups hustled to pull together data they could sell to drug companies. These days, pharmas aren't exactly looking for new ways to invest research dollars, BMS's Dracopoli says. They are more inclined to invest in late-stage drug candidates or research, such as tumor expression data, that can be plugged into existing oncology pipelines and proteomics programs.
The IGC would give pharmas economies of scale, and, with its extensive network of medical centers, access to a wider range of patients than any of them could possibly realize on their own. Not surprisingly, Bristol-Myers Squibb, Pfizer, and eight other biopharmaceutical companies have indicated they will work with the IGC and the 19 medical centers that have signed letters of intent to participate in expO.
"Pharma companies are looking for non-proprietary sources of reliable [gene expression] data that offers good standardization," notes Alan Williamson, a retired pharma exec, formerly of Merck and Glaxo. "No single company would wish to collect the extensive data that IGC can offer but all would like to have access thereto. Because owning the data is not important to pharma companies, sharing makes sense."
Gene Logic, Genomics Collaborative, and CuraGen are among a handful of companies that have made a business out of selling biocontent and tumor expression databases. The IGC might at first glance, appear to be a competitive threat to these firms. But observers say that the sorts of projects that IGC will undertake, such as expO, are beyond the scope and means of such small businesses.
In fact, Mike Pellini, CEO of Genomics Collaborative, is a member of the IGC pathology committee. He notes that whereas gene sequence data could be generated from a few anonymous DNA samples, obtaining cancer tumor samples requires major legwork. Cancer manifests itself in a multitude of forms, and each form is seen in several variations. For research, scientists need access to multiple samples of the same form.
"Recruiting patients and collecting [tumor] specimens is a lot more difficult than people think," Pellini says. Navigating patient consent and confidentiality code is too cumbersome for most genomics companies to take on. Medical centers tend to admit and process patients according to their own procedures, which means that companies such as Genomics Collaborative are faced with non-standard records and incompatible data. In addition, when patients are asked to consent to the use of their tissue in a particular study, the use of the sample and data is restricted to that single study.
One of the IGC's goals is to figure out a way to identify and collect the minimum yet most appropriate data that can be reused in future studies. "You want the [patient] consent and re-consent to use the data in comparison population studies," says Holden. By applying the data to future population studies, researchers can verify and validate the conclusions drawn from smaller studies now completed.
While the IGC's goals to unify researchers and standardize protocols are certainly ambitious, the clinical reach of expO remains modest. While some 3 million tumor biopsies are taken each year in the US, Gene Logic has gathered just 10,000 tissue samples in the past six years. And in four years, Genomics Collaborative has collected DNA, tissue, and sera samples from only 115,000 patients. Those low numbers attest to a shortage of quality, uniformly annotated samples with consistent RNA integrity.
For certain types of cancer where only small studies with inconsistent RNA integrity exist, it's very difficult for the drug industry to develop new therapies, says Dracopoli. The IGC could help advance treatments in cancer where there is little economic incentive for biopharma or medical centers to invest.
Indeed, by establishing standards for data collection and annotation, the IGC could also be a catalyst for the development of new informatics and bioinformatics. Some day, the organizers say, clinical algorithms could be sent back to the treating physician.
The first step, however, is to standardize protocols for obtaining patient consent, protecting privacy, handling tissue, conducting microarray hybridization, and storing RNA expression data, which would allow for much larger sample sizes and higher levels of accuracy and detail.
"We have good bioinformatics tools available today," says Pellini. "But the tools need to be refined and evolve." That can't happen until researchers have larger sample sizes and data sets. Pellini predicts that next-generation bioinformatics tools will emerge as the IGC puts together a database of uniform samples.
Examples could include genomic profiles that differentiate between tumor types and stages of tumor development, or tools to allow scientists to mine expO to define which biomarkers will induce what type of tumor, or to explain how far a tumor has progressed.
The IGC also says the standard protocols and language it establishes first in expO could lead to new informatics standards that could be applied in other diseases.
Intellectual property rights stood out as one of the thorniest issues in the private-public effort to sequence the human genome. To avert legal battles over rights to its content, the IGC has secured commitments from its partners that the samples and expression analyses will be treated as pre-competitive information. Gene expression profiles stored in expO will be deemed baseline information, regardless of patent positions a member might hold in a particular type of cancer.
Individual institutions and companies can examine, and subsequently attempt to patent knowledge of, how a given gene expression profile responds to a certain type of therapy. Or an institution may draw conclusions from mRNA by comparing it to its own proprietary data. But the expression profile itself, collected prospectively once the IGC begins to operate, would be available to all participating researchers.
The only filings the IGC expects to make with the Patent and Trade Office, says Trent, would be so-called "blocking patents" that establish dates when data became publicly available. Blocking patents would deter users from filing broad utility patents of novel mRNA activity in the profile. The SNP Consortium has filed similar blocking patents.
While the consortium participants have agreed to the process, Trent concedes that some medical centers might refrain from making all their goods available to the IGC. "We've had some [partners] say they will send head-and-neck tumors, for example, but not prostate or breast," says Trent. "But the same partners recognize the IGC is an important resource for everyone, so they want to participate, even on a limited basis."
Dracopoli notes that Bristol-Myers is primarily interested in understanding how its drugs respond in clinical trials conducted at cancer centers. Whether a cancer center sends the patient's tissue sample and core clinical history to the IGC or retains it for its own analysis doesn't matter to Bristol-Myers, as long as data regarding the drug's activity remain under wraps until a regulatory decision is finalized. Releasing such information prior to a regulatory decision would violate trial protocol.
For example, patient age, estimated progression of cancer, and family history could be made publicly available once a biopsy is conducted. But Bristol-Myers would not want public disclosure of how the same patient responded to an investigational drug during a clinical trial. The therapeutic data would remain blinded and proprietary, but the gene expression data would be regarded as pre-competitive, even if Bristol-Myers was funding the trial.
The Buy-in Promise
While several biopharmas already subscribe to Gene Logic's database of tissue samples, IGC organizers see plenty of opportunity for expO to add value. Gene Logic has strength in numbers in the three most widely researched cancers — breast, prostate, and colon cancer — notes Mike Bittner, a senior science advisor to the IGC and a colleague of Trent's at NHGRI. However, the phenotypic and gene expression data generated by the IGC could enhance current research in the big three cancers, as well as in other forms.
The IGC says that its projected costs and contributions are in line with other recent consortia initiatives, and given the potential long-term benefits of membership, the fee seems reasonable. Organizers are requesting $4.2 million per member, spread across three years. SNP Consortium partners contributed an average of $4.7 million, while other recent consortia initiatives have received average contributions of about $5.8 million, according to estimates by a consultant. Biopharma customers such as GlaxoSmithKline, Pfizer, and Vertex Pharmaceuticals also pay between $2 million and $16 million annually to gain various degrees of access to and analysis from Gene Logic databases.
Gene Logic currently provides more detailed and custom analyses for its price, but IGC founders expect big pharma to be lured by perks such as access to Trent's work at TGRI and data from cancer trials conducted at participating medical centers. Partners also will get data analysis of blood and sera from the tissue samples collected, a recent modification to the IGC platform made to aid the development of biopharmas' proteomics programs.
"The drug industry cannot afford not to get behind this," argues Holden. "For each [pharma] to do this internally makes no sense. Drug company researchers want access to this data — they recognize how their work has advanced though the SNP Consortium — and the medical centers don't have the scale to exploit [their] patient samples."
The next hurdle will be getting financial and infrastructure commitments. Affy's Colón says that even though the IGC effort is progressing a bit more slowly than anticipated — she blames the economy — a series of recent verbal commitments signify that the first domino is likely to tip soon, and the rest will soon follow. IGC leaders are counting on most pharmas to recognize that they have much to gain and little to lose by anteing up.
"The key for IGC and our company is to get scientists to buy into [the idea that] what you are trying to accomplish is gathering the data," says Pellini. "Unless the clinicians support the scientific basis of the idea, nothing will happen."
Tribal Translational Genomics
Large biomedical research projects in the US typically gravitate to four metro areas — Boston, Washington-Baltimore, San Francisco, and San Diego. Arizona was never on the radar screen. That is, until Jeff Trent came along.
Trent, an Arizona native and most recently the scientific director of the NHGRI, worked closely with Governor Jane Dee Hull, Senator John McCain, and the presidents of the state's three universities to bring the Translational Genomics Research Center to Phoenix.
At the new institute, Trent will continue his NHGRI work in translational genomics with an initial focus on melanoma and pancreatic cancer. Later programs will be directed at diabetes, obesity, and cardiovascular disease. All tissue and cancer data will be sent to the International Genomics Consortium, where Trent is a senior science advisor.
The disease choices were not accidental. To fund the TGRI, Senator McCain persuaded 17 Native American tribes in the state to participate. For years, researchers have studied the Native American Salt River Pima tribe to understand the genetic basis of obesity. With TGRI now based in Arizona, the institute will be in a position to conduct long-term studies and trace the genetics of the Salt River Pima through family histories, just as DeCode Genetics and Roche reportedly have mapped genes associated with obesity, rheumatoid arthritis, and anxiety in Iceland.
Recognizing the potential for improved treatment, the Salt River Pima contributed about five percent of the $92 million raised to establish TGRI's infrastructure. State and local governments in Arizona, along with the universities and private foundations, contributed the lion's share.
Trent says the institute will aim to move genomic targets to treatments as rapidly as possible. About 30 researchers should be at work by the end of the year and another 70 by next summer. The number should rise to 250 in three years, when a new building is to be completed. Most of the new researchers will receive faculty appointments at the state's three universities.
"We're not going to become a major center like San Francisco or San Diego. The staff will compete for grants like everyone else, but very few will go after basic science," says Trent. "The research of the TGRI has to be targeted and focused to make a difference."