Biomarkers: What are they good for? It's complicated

The New York Times is running a news analysis on biomarkers in the wake of scandal at Duke University over allegedly shoddy research, and a recent article in JAMA that calls into question whether the entire science of biomarkers is good for anything. The article highlights the extreme difficulty of finding patterns in genes, molecules or proteins that have significant meaning and then the problem with confirming those meanings independently and using them in any significant way.

A good summary comes from MD Anderson Cancer Center statistician Donald Berry, who tells The New York Times that gene or protein patterns "are very difficult to get right," and finding them "is like looking for a needle in a haystack when you can't tell the needle from the hay." Reading this, you'd get the feeling that everybody working on biomarkers should give up and find something more productive to work on. But Lajos Pusztai, a breast cancer researcher at MD Anderson, tells the NYT that, yes, it is difficult work, but that is no reason to abandon the quest for cancer signatures.

It is frustrating work, too, the article points out, because it involves a little intuition and a whole lot of gazing at numbers to find statistical patterns in cancer, which is turning out to be a much more complicated disease than once thought. And there is little incentive to go through the expensive, time-consuming process of evaluating biomarkers because the FDA does not require it, NYT points out, and companies are not reimbursed as much for tests based on biomarkers as they are for new cancer drugs.

One solution: researchers who think they've found a new genomic signature for cancer should publish enough of their data for others to verify their work. Yet few have done so. And the only reason the Duke research was found to be flawed was that it relied on data that was already publicly available. Still, it took about 2,000 hours to find all the mistakes the Duke researchers had made.

- read the NYT analysis