By Malorye Branca
July 11, 2002 | PRINCETON, N.J. -- In an intriguing approach to pharmacogenomics, New Jersey businesses Orchid GeneShield and Merck-Medco are teaming up to find genetic variations that predict response to asthma drugs.
Under the terms of the deal, the two companies will collaborate on a study combining genotypes (tests used to find genetic variations), health claims, and pharmacy data from 2,000 patients with asthma. The data, except the genotypes, will come from Merck-Medco’s patients claims database, making the study retrospective -- that is, based on data that has already been collected.
“We can see from this data which [asthma] treatment protocol is best for people with a particular genotype,” says Scott Rakestraw, executive director of business development strategy at Orchid GeneShield.
Orchid GeneShield, one of Orchid BioSciences Inc.’s four business units, leverages Orchid’s SNP technology to develop pharmacogenomics tools and services. Merck-Medco is the country’s leading pharmacy benefits manager. Its clients include employers, HMOs, unions, and other groups that provide health insurance coverage.
Maintaining patient confidentiality is critical in such a trial. The names of patients who agree to be genotyped and participate in the trial will not be linked to their data. “We will be destroying the genetic samples we get and the electronic data at the conclusion of this study,” says Rakestraw. “Security has been a major consideration in the design of our data center.”
The Orchid GeneShield data center is based in Arlington, Va. It has 2.6 terabytes of storage capacity and is built around a 24-processor, massively parallel Beowulf computer cluster. The center’s network is redundantly secured.
As part of the deal, Orchid also acquires rights to proprietary data-mining algorithms from Merck-Medco. These algorithms are specialized for analyzing data on how patients use health care resources (i.e., doctors, drugs, hospitals), and can be applied to large data warehouses. Orchid also has some of its own pharmacoepidemiologic and pharmacogenomic tools for health risk analysis. “Our algorithms let us predict accurately what the odds are that a particular pattern indicates a patient has an elevated risk for an adverse event,” says Rakestraw.
Using this type of data is an innovative approach. “Pharmacy data is the most reliable type of health claim data available,” says Craig Fitzgerald, a partner at health-care marketing firm HealthCarta LLC and former vice president of managed care at GlaxoWellcome (now GlaxoSmithKline). “And it’s easy to analyze.”
Scheduled to be spun off from parent Merck & Co. Inc. this year, Merck-Medco has data on more than 65 million patients, which should provide plenty of prospects for the study. “One of the problems with pharmacogenetics is that you can’t draw good conclusions from a study with 100 or so patients,” says Rakestraw. Orchid has calculated that it will require a pool of about 2 million prospective participants to end up with 2,000 qualified patients, he says.