That is, unfortunately, only a slight exaggeration when it comes to the PSA test for prostate cancer. First came studies finding that early detection of prostate cancer does not improve survival (there are a number of such studies, but one in the Archives of Internal Medicine, published by the American Medical Association last year, is particularly devastating). That led biomedical researchers to tweak the test and claim that, okay, even if absolute levels of prostate specific antigen (PSA) does not indicate whether prostate cancer threatens your life or can be left alone ("watchful waiting"), then maybe taking several readings and calculating the rate of increase in PSA levels might do the trick.

No such luck. Neither PSA levels at the time of a man's prostate cancer diagnosis nor its rate of change indicates whether the cancer is likely to be lethal, scientists report in the April 4 Journal of the National Cancer Institute. That is bad news for several reasons. First, men with localized prostate cancer have excellent long-term survival rates even if their cancer goes untreated, so doctors want to be able to tell men when they can safely choose this option. Changes in PSA were supposed to provide such guidance "little to no change, watchful waiting should be okay; significant change, take action. But although the rate of increase of PSA before treatment has been associated with prognosis, the study (of 267 men whose PSA levels were measured for two years after diagnosis) finds that initial PSA levels and the rate of change did not accurately predict lethal cancer.

The scientists say PSA measurement is still "an important monitoring tool," but does "poorly in distinguishing those who develop a lethal prostate cancer from those at low or no risk of disease progression."

The point is not to dump on the PSA test. But somehow people have gotten the impression "partly through testimonials from men who swear their PSA test saved their life "that if they're dutiful patients and have all the tests their doctors advise, they'll be just fine or, at worst, determining the best treatment or course of action will be obvious. Would that it were so.