Saturday, June 23, 2007

Stroke does not strike where lab predicts

If you want to know your future health, you may go to a medical checkup where your blood and urine is taken and analyzed in a lab. You may just as well read the tea leaves if you are a woman after menopause under hormone therapy and want to know more about your stroke risk. This is the surprising outcome of the Women's Health Initiative trials where more than twenty-seven thousand women have been examined with a follow-up of more than five or six years on average. The women have taken either hormone or placebo (double-blind), their blood has been analyzed, even for stroke-promoting genes, and all stroke events during follow-up have been recorded.

The intriguing outcome: Some lab values (so-called biomarkers) that were thought to predict a higher stroke risk did exactly the opposite and lowered the stroke risk in combination with hormones. Hormones in turn are thought to increase the stroke risk in women by about forty percent. In other words: A risk factor combined with another risk factor lowers the risk. The authors of the study are so puzzled that they conclude this «may be due to chance». You better read the tea leaves then.

What are biomarkers good for?

The idea of biomarkers is early detection of a disease before it has evolved. The concept is somewhat similar to risk factors but is confined to factors that can be detected by physiological measures (such as blood pressure) or by analysis of body liquids (cholesterol, sugar and the like).

There is an ongoing debate about the usefulness of biomarkers. Big business such as lipid-lowering drugs is based upon biomarkers. This business may be at stake if it turns out that improving the biomarkers does not safe lives.

Another biomarker case is calcified plaques in the arteries that increase the risk of a heart attack. Women after menopause have less plaques when they take hormones. But other studies have shown that estrogen supplements increase the risk of a heart attack. This brings the study authors to the conclusion that «estrogen has complex biologic effects and may influence the risk of cardiovascular events and other outcomes through multiple pathways». In other words, they strongly warn against relying on the plaque biomarker.

The case of population studies

In epidemiology, biomarkers are often used as ersatz outcomes if researchers cannot wait until their subjects fall sick or die. The flaw of this method is a dilution problem. If for instance a certain diet has been shown to lower cholesterol by x percent and other studies have shown that in the lowest quartile of cholesterol the risk of a heart attack is y percent lower than in the highest, it cannot be concluded that the diet will lower the risk of a heart attack.

If this sounds too complicated, just imagine billiard balls that are made of sand-filled rubber. If the first ball hits another, the second ball will move, but not very far and if it hits a third ball, this latter may not move at all. Conclusion: Be sceptic when biomarkers are used in population studies.

Photo credit:

No comments: