Hopkins scientists were the first to show that simply checking a man's blood regularly, and watching for a rise in his levels of PSA can predict cancer years before it can be diagnosed by any other means (see below). Any recurrence of PSA after treatment for prostate cancer -- even more specifically, how soon it comes back, and how fast it rises -- can give important clues about the nature of the cancer (whether it's aggressive or mild-mannered, for instance, and how best to attack it). Furthermore, Hopkins scientists including H. Ballentine Carter, M.D., have even helped delineate and characterize subtypes of PSA -- "bound" and "free" -- within the blood, and decoded still more precise early warning signals of prostate cancer.

Despite all these advances, there are still some basic questions about PSA -- which, by the way, still rattles or just plain stumps many physicians, who discourage use of the test because they are unsure how to interpret it, or because they ding to the persistent but obsolete belief that prostate cancer screening will never work. Among the most obvious questions is the starting point: When should a man start getting his blood tested for PSA levels? When he's 50? That's what the American Cancer Society recommends unless the man is at higher risk, if he's African American or has a family history of prostate cancer. In that case, testing should start at age 40. In any event, testing should be done on a yearly basis afterward.

But is this the best way? "All of the recommendations that we make for individuals in terms of when to start PSA testing, how often to do it, and when to stop -- there's really not a lot of scientific basis to support any of that," says Carter, professor of urology. "We've been very conservative." From a medical standpoint, conservative is good -- doctors want to catch as many cases of prostate cancer as they can, as early as possible. But the conservative approach is not always the most cost effective, and it also may not be the most efficient. However, says Carter, such guidelines are impossible to determine with large screening studies, "because it just costs too much to vary all of these factors."

So Carter, with Kevin Ross, a graduate student in epidemiology at the University of North Carolina, and UNC epidemiologist Harry Guess, decided to try something that has worked well in the field of cervical cancer research: A highly sophisticated computer model, called a Markov model, which mathematically simulates the progression of a disease in a group. "Basically, it makes a hypothetical population of individuals and walks them through life," using established statistics -- how many American men die at what ages of heart attacks, accidents, and a host of other specifics, including cancer. "We know what happens," says Carter. "We know what the chance is of a 40-year-old not making it to age 45. So we put that into the model. And we know his odds of dying of a heart attack before 50, so we put that into the model. And we know the chance of dying of prostate cancer, developing prostate cancer, developing curable prostate cancer, developing noncurable prostate cancer -- we put all of that information into the model, and we walk a population of men from age 40 all the way to age 80." Setting up the model is the hard part. "But then, you simply ask, how would you affect the death rate of prostate cancer if you started screening at age 40, or if you started screening at age 50. You can vary anything you want: So you can say, 'I'm going to start screening at age 40, and I'm going to test every year, and see what happens." The big questions included: How did various screening plans affect overall deaths from prostate cancer, and how many PSA tests and biopsies were needed to detect each cancer?

The result was unexpected. For men who are not at higher risk of developing prostate cancer: "It turned out that the most cost-effective way to detect prostate cancer was beginning PSA testing at age 40, repeating the test at age 45, and then at age 50 testing every other year, instead of every year," says Carter. "That was the only strategy that did three things: It reduced the death rate of prostate cancer, reduced the overall number of PSA tests, and reduced the overall number of prostate biopsies for each cancer detected " Although Carter emphasizes that this is preliminary, it does make clear that more work is needed to find the most cost-effective way to detect the most prostate cancers, "and that starting at age 50 and testing every year may not be the best way."

Several years ago, in a landmark study, Johns Hopkins scientists used a large data base called the Baltimore Longitudinal Study of Aging to find out just how good the PSA test was at predicting cancer.

The BLSA is a remarkable tool for scientists astute enough to ask the right questions: Since it was begun in 1958, about 1,500 men have participated in this study, returning every other year for physical examinations and a battery of medical tests. Their blood samples from every check-up are stored for future research. In this particular study, the investigators studied 20 years' worth of blood samples from men with BPH, men with prostate cancer, and a control group of men with no prostate disease. Their results were startling: By tracking changes in PSA levels, they were able to detect prostate cancer up to 10 years before it could be diagnosed by other means. PSA, it turned out, was a veritable crystal ball at predicting prostate cancer-for those who knew how to read it properly.

Recently, Carter wondered whether PSA could be just as accurate at showing men whose destiny does not include prostate cancer. In this study, he and Merck Epidemiologist Jay Pearson, Ph.D., tracked PSA levels in men who ultimately did, and men who ultimately did not, develop prostate cancer, in two groupsone of men from age 60 to 75, and the other of men from age 65 to 75. "We asked a simple question," says Carter: "Are there men with very, very low PSA levels at an older age, who are at very low risk of ever developing prostate cancer-and who therefore don't need frequent screening?"

The answer was yes-and one less worry for hundreds of thousands of men. "We found that a man who has a PSA less than 1 at age 65 is extremely unlikely there's over a 90- to 95 percent chance that he will not-to develop prostate cancer over the next decade." Furthermore, although the scientists cannot yet make a strong recommendation for what these men should do, "what we can say is that these men probably do not need to be tested yearly, and the interval for testing could be two to five years. I think it's good news for men age 65," concludes Carter. "That's a large percent of the population. These are people who should be opening a bottle of champagne."