By Mark D. Uehling

April 15, 2003 | Stephen Friend, vice president of molecular profiling at Merck Research Laboratories, has a lot to live up to. While Merck shareholders might still be wondering whether the company got value for money when it acquired Rosetta Inpharmatics, the company Friend co-founded in 1996, for $620 million, Friend’s keynote presentation at the Bio-IT World Conference & Expo here last month hardly resolved the issue.

But Friend did spotlight key areas, notably in cancer diagnostics and complex trait mapping, where Rosetta’s technology may literally pay dividends for Merck.

For years, as an oncologist, molecular biologist (he isolated the first tumor suppressor gene, for retinoblastoma, in 1986), and bioinformatic visionary, Friend has been one of the pioneers in the effort to harness gene expression data by the thousands to help diagnose cancers. Rosetta’s microarray software package is one of the most highly regarded tools for manipulating gene chip data.

In results published last December in the New England Journal of Medicine, Friend and colleagues at the Netherlands Cancer Institute presented microarray data showing that a 70-gene signature associated with the fatal metastatic stage of breast cancer is evident early on in the disease. “It may be possible to intervene very early in patients’ tumors and start treating them before the tumors have gotten aggressive characteristics,” Friend said.

Nevertheless, it’s clear that the Rosetta gene expression data cannot reliably predict every patient’s risk for cancer. Friend acknowledged the potential of proteomics in such diagnostic settings, but he cautioned: “The importance of profiling proteins and use of blood serum and cerebrospinal fluid markers is basically five years away.”

Friend also described an exciting new application of microarrays to identify eQTL, or expression quantitative trait loci. Using microarray gene expression arrays and conventional genetic mapping techniques, Friend and colleagues have identified a host of eQTL across the mouse genome. Technically, these are genetic variations that profoundly alter the activity of one or more genes, providing important clues to genes that control other genes. A subset of these eQTL have been implicated in obesity (results published in the March 20 issue of Nature).

What Comes with the Pill?
Friend fully embraced the field of pharmacogenomics based on microarray profiling but argued that the term “personalized medicine” should refer to the identification of subdivisions of patients suffering from a common disease, rather than tailoring drugs to individual patients.

While such a transformation would mean the number of appropriate recipients for each new drug will fall in the genomic era, Friend does not believe drug company revenues will suffer necessarily—h e says he is confident that drugs carefully matched to patients will be more likely to work. He said he also believes that doctors and insurers will be willing to pay for results: “If you tie in the success of a treatment so there is a more successful outcome, that helps the economics significantly.”

Another research challenge: the advent of extras bundled with genomically targeted drugs. Both diagnostic tests (to determine a patient’s genetic makeup and suitability for a drug) and monitoring services (to check how a patient responds to a drug) will have to be developed hand-in-glove with the medicines themselves. “The diagnostic cannot come as an aftereffect,” Friend said, hinting that Merck might either develop such lab tests in-house or form alliances to do so.