SAN FRANCISCO—Genomics research has unearthed a wide range of approaches that drug companies can use to treat diseases. Now, the big question for these companies is how to choose the right genes to go after in drug development.
Finding ways to optimize this discovery process was the subject of a Tuesday panel at BIO, a biotechnology industry conference here.
Since the advent more than a decade ago of high-throughput screening, pharmaceutical and biotechnology companies have spent vast sums to obtain information on the genetic variations associated with the emergence of particular diseases.
And in that task, they have been largely successful, due largely to the basic information available from the complete mapping of the human genome and the Hap Map project by the public-private SNP Consortium, which identifies single nucleotide polymorphisms, DNA sequence variations among individuals.
But given this wealth of targets, many of them novel, the challenge in recent years has shifted to validating which targets actually have the potential to be effectively treated.
Advances in IT have flooded scientists with possible research targets. Now, the increasing linkages between genetics, clinical pathway and disease expression will help them extract the most valuable ones.
This transition was focus of the panel. Partaking in the discussion were the research vice presidents of two major pharmaceutical companies, two leading genomics research tool companies, and the director of the National Human Genome Research Institute.
Only about one in 10 drug molecules that start out in the research pipeline actually make it to market, according to data presented by Ismail Kola, vice president of basic research at Merck & Co. And that rate varies greatly by indication.
According to Kola, central nervous system and oncology molecules succeed only 7 percent and 6 percent of the time, respectively. But cardiovascular and infectious disease molecules make it to market 19 percent and 16 percent of the time, respectively.
More Tools Needed
Overall, biologics go all the way through the pipeline almost one-quarter of the time, a comparatively high success rate.
Safety and toxicity concerns eliminate about 60 percent of drug candidates, but the remaining 40 percent fall prey to poor efficacy. Genomics research tools alone are ineffective in sorting which targets are useful to pursue, Kola said.
Understanding the disease through an examination of the disease pathway and phenotypical expression is crucial to choosing which targets are involved in the establishment of disease.
Also helping to refine the mountains of targets produced are the method of hypothesis testing and the examination of extreme and opposite cases.
Nicholas Dracopoli, vice president of clinical discovery technologies at Bristol-Myers Squibb Co. of New York, came to a similar conclusion.
Even representatives at research tool companies Perlegen Sciences Inc. and San Diego, Calif.-based Sequenom Inc. stressed that putting genomics data into context greatly enhances its usefulness.
Even though the understanding of the genes related to particular diseases is rapidly evolving, which genes are key and what role they play in creating an outcome is often still unknown.
In complex diseases, dozens of genes may be implicated, making it difficult to determine which are the most important to target, the executives said. And some targets are unable to be treated with drugs.
The costs of genomic analysis have dropped substantially in recent years. In 1989, it cost $200 million to discover the genetic basis of cystic fybrosis. Sequenom estimates that the cost of genomic analysis of a particular disease today is $500,000.
The next step in linking knowledge of the disease to discoveries of genomics may be to enable systematic understanding of disease pathways and phenotypes.
In an article published last week in the journal Nature and again at the BIO conference, Francis Collins, director of the National Human Genome Research Institute, called for the establishment of a biobank in the United States.
This would create a publicly accessible, longitudinal database containing the biological material of at least half a million people. It would allow scientists to track diseased and nondiseased populations, as well as to access data on an individual preceding the onset of disease.
The panelists seemed hopeful, yet cautious, that a biobank could be established. A similar effort in the United Kingdom is stymied so far.
The biggest challenge to such an effort, Perlegen CEO Brad Margus said, is the need to design the study impeccably upfront, since the usability of any subsequent data would hinge entirely on the original design.