Single Nucleotide Polymorphisms: Big Pharma Hedges its Bets

SNP CENTRAL: A genetics researcher takes to the bench at the Wellcome Trust's Sanger Centre in Cambridge, England. The sequencing center and its London sponsor provided key leadership in the SNP Consortium, a public-private venture to find and map 300,000 single nucleotide polymorphisms. The Wellcome Trust helped entice 10 pharmaceutical firms to join the consortium by putting up $14 million of the project's estimated $45 million price tag. The Sanger Centre will provide much of the radiation hybrid mapping, which will help define the role of these common genetic variations. This past April, 10 rival companies entered an agreement that stands to garner little direct or immediate profit for any of them. Why the apparent altruism? They had a choice: Either team up with their competitors and contribute a few million dollars for research that could benefit all involved, or go it alone and spend tens of millions slowly accumulating potentially redundant results. The companies not only embraced their rivals; they invested millions in a research project whose potentially valuable results will immediately be made public rather than kept private and patented as intellectual property. Even with the alliance, there's still no guarantee that the data gained will translate into a significant return on their investment. Several of the world's pharmaceutical giants--including AstraZeneca, Bayer, Pfizer, SmithKline Beecham, and Novartis--along with their five academic partners (see list, page 9), are united on a common goal: To, within the next two years, scour the human genome to uncover 300,000 new heritable DNA sequence variations called single nucleotide polymorphisms (SNPs). As the single-base differences that genetically separate individuals within a population, SNPs could act as markers in the search for the origins of diseases caused by the interaction of multiple genes. A large SNP library could help researchers pinpoint the genetic origins of diseases, and, potentially, the most appropriate drugs to treat them. SNP value hinges on the assumption that a finite number of common DNA variants in the human population make a significant contribution to genetic risk for common diseases, a hypothesis called common disease-common variant (CD-CV). Although unproven, the CD-CV hypothesis has held true in several cases, perhaps most notably in the case of the ApoE4 allele and its association with Alzheimer's disease. By using SNPs as random markers, researchers would like to, in essence, create a highly detailed map of genetic variation to help them find disease-linked genes with more precision. A researcher might, for example, compare a group of patients that suffer from cardiovascular disease with a group of patients that do not. He'd then look to see which ones share which markers. If one group has a higher frequency of some common markers, he'd reason that a modified gene in that region might help cause cardiovascular disease, and only then would try to hunt down the gene. But the SNP consortium's mapping effort and others like it will not uncover SNPs targeted to specific candidate regions of the genome, nor will they uncover gene function. Each company will do that work on its own using the SNP mapping--indeed, some companies have already started to identify "functional SNPs," the real sources of profit. Although there are likely several million SNPs in the entire genome, only a few percent will have high commercial value from a diagnostic standpoint. Non-consortium member Incyte Pharmaceuticals of Palo Alto, Calif., for example, has about 60,000 SNPs in its database. "But we really know very little about those SNPs. Some of them will turn out to be sequencing errors, and some of them will turn out to be rare SNPs that are not terribly interesting," explains Randy Scott, president and chief scientific officer at Incyte. Only about 20,000 of Incyte's SNPs have a high probability of being SNPs that are observable in a patient, and far fewer have actually been screened against a number of different individuals to find out their frequency within that population. "The map itself has no real value," says Daniel Cohen, chief genomics officer at Paris-based Genset, also not a consortium member. "The issue is how to use this information ... to find genes." According to Cohen, a gene discovery company such as Genset will largely benefit from, not compete with, the SNP mappers. While Scott welcomes the consortium effort and, like Morgan and Cohen, expects that it will be a valuable tool for everyone, his company prefers to, as he puts it, "turn the paradigm upside down." Rather than finding genetic markers, examining patients, and then trying to locate the gene, they're starting from the genes they think are involved in the given disease and looking at variation within patients. In the case of rheumatoid arthritis, for example, researchers would look at the SNP level to find genetic variation within the target enzyme--well-known arthritis culprit cyclooxygenase-2 (COX-2) in this case. To find diagnostics for patients' drug response, they'd then see if different people with different variants of COX-2 responded differently to the given COX-2 inhibitor. "It's a pharmacogenetics approach versus a sort of classical genetic inheritance approach," says Scott. "We think the primary value of patents in SNPs is around coding SNPs that actually change the amino acid content of a protein. They have a physical biological effect as opposed to SNPs that are just markers along the genome." Interest in both targeted SNPs and SNP markers has blossomed in the last few years. Following a 1997 National Human Genome Research Institute (NHGRI) workshop at which leaders in the field of genetics of common disease voiced the need for a public SNP database, the National Institutes of Health, in January of 1998, issued a request for applications (RFA) for the discovery and scoring of SNPs. Because the sequencing component of the publicly funded Human Genome Project (HGP) will base its master human DNA sequence on only one individual at most positions, information on human variation will only appear in areas of overlap between adjacent clones, and only then if the clones derive from different chromosomal homologs. Although this will produce a cluster of SNPs every 250 kilobases on the average, a higher density is needed for many applications. As a result, the sequencing itself is not a source of publicly available SNP information. As part of the RFA, the NIH made $30 million available over three years for the compilation of 100,000 SNPs. Francis Collins Prior to the official funding of the NIH's efforts, pharmaceutical companies Genset and Abbott Laboratories, Abbott Park, Ill., had already teamed up to identify a planned 60,000 SNPs. Genset's deal with Abbott "certainly made the point that [SNPs] were valuable," says NHGRI Director Francis Collins. Subsequent to the NIH's RFA, Celera Genomics of Rockville, Md., started its own database, to which interested parties will have to pay to subscribe. Because their "shotgun" sequencing approach, which aims to blast genomes into millions of fragments then reassemble them like a giant jigsaw puzzle, naturally generates SNPs, Celera does not need a separate SNP effort. About half of the NIH-funded proposals also center around the discovery of random SNP markers; the other half focus on finding SNPs in coding regions of genes that might be candidates for involvement in particular diseases. Celera's database consists of random markers; Genset's SNPs are mostly targeted. "I think those pharmaceutical companies that decided to participate [in the consortium] were concerned about a scenario where the number of SNPs being produced by the NIH might not be sufficient for the applications they wanted to see happen," Collins remarks. "If 100,000 weren't enough and they needed 500,000, then what would they do?" Collins expects that companies probably would have ended up having to buy access to SNP databases from companies such as Celera at a considerably greater cost than their consortium share, $3 million. The Wellcome Trust, a nonprofit institution and consortium partner, kept membership costs low and enticed companies to join by putting up the bulk of the money--$14 million. "This looked like a very good opportunity to join together with a group of pharmaceutical companies ... in order that we got the [desired] objective as cheaply as possible and without competition," says Wellcome Trust Director Michael Morgan. "We decided that, as a pharma company, we were not really interested in selling SNP maps but we were interested in using them," explains Paul Herrling, the head of research at Novartis Pharmaceuticals. "And rather than everyone building them independently, why not ... put them in the public domain, and share them with everybody else." According to Morgan, the pharmaceutical companies in the consortium have no advantage over other nonparticipating organizations in terms of reaping the benefits of the SNP data gained. "There's no prior look," he says. "That's very important." Revealing SNPs to researchers in academia will, in theory, seed new discoveries on disease-linked genetic modifications. This will not only make university scientists happy, but will give companies more potential targets to work with in developing new disease therapies and will, as a result, eventually present the opportunity to profit. The consortium will also cut costs and maintain quality while getting results faster and more efficiently. And if SNPs do indeed turn out to be important tools for fine-tuning drug design and minimizing drug side effects, the consortium could offer a crucial "gold standard" SNP map that would serve as a valuable reference not only for academia and industry, but for the FDA and other regulatory agencies around the world. So everyone's happy--that is, unless the coveted SNPs fail to live up to the rather high expectations. "The value [of SNPs] is all potential value," admits Morgan. "We'll have to wait and see." While Morgan is certain that SNPs will be extremely useful as research tools, he's less sure of their value as guides to treatments for multifactorial diseases. And few will benefit from newly discovered SNPs, says Morgan, without the development of high-throughput genetic screening. Some argue that the project is more trouble than it's worth. "What you're really after ... is knowledge about the determinants of drug response, not the tool that you use to find those," claims William A. Haseltine, chairman and CEO of Human Genome Sciences Inc., Rockville, Md. "Most single nucleotide polymorphisms are only being used as landmarks to speed mapping of drug response genes, not in and of themselves as the determinant." Human Genome Sciences, says Haseltine, is more interested in genes with definite medical uses. "We know where they're expressed in the body, we know where they map, we know a good deal about their biological function, which allows us to predict medical utilities." Regardless of how useful the consortium's and others' SNP markers turn out to be, having a sizable amount in the public domain will at least, in principle, create a more level playing field for industry and academia. Indeed, that may well be the consortium's unstated intent. Even so, Celera and those with enough money to subscribe to its catalog will likely have a considerably larger database at their disposal. But, says Collins, Celera's strategy is more likely than NIH's or the consortium's to produce candidate SNPs that turn out to be due to gene families, not to gene variants. And unless combined with the public resources of the HGP, Celera may end up with a significant fraction of the candidate SNPs unmapped. According to Collins, it will likely not be the size of the catalog that matters, but how you use it. "It's not at all clear that anyone would want to use 10 million SNPs if they were available," he comments, adding that most analyses indicate that 500,000 SNPs (the projected total between the consortium's and the NIH's databases), or possibly fewer, are sufficient for case-control whole genome association studies. In fact, the consortium may actually reduce the number of private database subscribers if companies reason that the half million SNPs soon to be freely available prove sufficient. Celera's director of policy planning, J. Paul Gilman, isn't worried about the potential competition. "Some of the members of the consortium say that the consortium's product is sufficient for their needs," he acknowledges. But he adds that potential customers have nevertheless been "knocking on our door." And, says Gilman, the consortium could even inadvertently help Celera--a point some members of the consortium confirm. "We'll be able to incorporate all the information they come up with into our own database," he says.

Single Nucleotide Polymorphisms: Big Pharma Hedges its Bets

Interested in reading more?

Become a Member of