Biotech Values

[DewDiligence]

Re: Predicting who benefits from HCV treatment

This would be very big news if the algorithm really worked, but
my guess is that it’s yet another case of finding an association
that has no prognostic value.

The general principle (which is highly relevant to biotech
investing): If you look at a data set determined to find some
kind of association, it is almost a certainty that you will.

http://www.reuters.com/article/marketsNews/idINN2251608420081222

›Study May Predict If Hepatitis C Drugs Will Work

Mon Dec 22, 2008 6:44pm EST
By Julie Steenhuysen

CHICAGO, Dec 22 (Reuters) - Doctors hope to be able to better predict which patients will respond to traditional treatment for the hepatitis C virus using a new method for identifying slight variances in the virus' genetic makeup.

U.S. researchers said on Monday that the technique may prove useful for other viruses such as HIV as well. The finding could be used to develop a test that would analyze a patient's specific virus strain before treatment was started.

A team at Saint Louis University in Missouri analyzed genetic patterns of the virus in patients infected with Hepatitis C to see if they could tell why many patients fail to respond to standard treatment with pegylated-interferon and ribavirin.

The year-long therapy activates the body's natural defenses against viruses, but patients often feel as though they have a bad case of influenza. Only about half of the people who suffer through the treatment actually respond.

"This is a very difficult therapy to take. It's really hard on the patient," said John Tavis, a professor of molecular and microbiology at Saint Louis University, whose study appears in the Journal of Clinical Investigation.

"If you can identify those patients who aren't going to respond anyways because they've got a strain that is highly resistant to the drug, then you just don't treat those patients and you save them $20,000 to $30,000 in medical bills just from drugs alone -- not to mention the side effects," Tavis said in a telephone interview.

He and colleagues studied the ribonucleic acid or RNA chains of the hepatitis C virus, looking for patterns that would explain why some people responded to the treatment while others did not.

Using a math formula, they zeroed in on a specific pattern of changes called "covariance networks" that differed depending on whether the drug worked. And these patterns proved to be a strong indicator of whether the virus was especially resistant to therapy.

"What we found will allow a doctor to predict whether or not a medication will work in a patient," Tavis said in a statement.

The finding also may have implications for other types of RNA viruses, such as human immunodeficiency virus or HIV or the influenza virus.

"It's a pretty easy process. The algorithm can be applied fairly quickly," he said. Whether or not it turns up a pattern that will be useful is less clear, he said.‹