Dr. DeAntoni provides a timely, critical review of the concept of age-specific prostate-specific antigen (PSA) ranges, as well as other frequently used attempts to improve the accuracy of serum PSA testing in the diagnosis of unsuspected prostate cancer. His review is complete, and his assessments of each of the modalities reflect not only the majority view but also realistic appraisals of the limitations of this less-than-perfect test.
Specificity and Sensitivity of PSA
Foremost to be remembered in any discussion of PSA as a diagnostic test for prostate cancer is that PSA is not a cancer-specific marker. Only in the absence of the prostate after surgical removal or in the face of advanced disease does PSA truly function as a cancer-specific marker. The positive predictive value of an elevated PSA of more than 4 ng/mL is only 25% to 35%.
The very fact that so many manipulations exist of a single blood test (eg, PSA density, age-specific PSA, and the newest variation, the free-to-total PSA ratio) or two blood tests (in the case of PSA velocity) provides ample evidence of the struggle of clinical researchers to make this imperfect test better. For the vast majority of males subjected to this test once or on a regular basis, most elevations of PSA above the "normal" value of 4.0 ng/mL are due to the many noncancerous causes discussed by Dr. DeAntoni.
Prostate-specific antigen, as a single value or with many of the artificial perturbations introduced by clinicians, is simply not a highly sensitive or highly specific cancer marker.
An article by Babaian and colleagues (the author's reference 29) points out nicely that attempts to increase the sensitivity and specificity of PSA simultaneously are impossible. In that report, Babaian et al showed that the use of age-specific reference ranges causes more biopsies to be performed in younger males but substantially fewer biopsies to be done in older males. Overall, fewer biopsies are performed and fewer cancers diagnosed.
Although this approach may allow more aggressive cancers to be diagnosed earlier in younger males and avoid the diagnosis of many cancers in older males, its validity remains to be proven. Such an approach is most appealing to third-party carriers or at-risk contracted parties who must bear the cost of treatment when cancer is found and the decision to treat is made.
Other Prognostic Markers Needed, Not More Studies of PSA
DeAntoni calls for "large national studies...to establish appropriate and cost-effective methods of interpreting a man's PSA test results over time," but I would suggest a different approach. Why spend millions of dollars to continue to massage an imperfect cancer marker such as PSA? Instead, spend that money to investigate other clinical prognostic markers of prostate cancer to find that one test or panel of tests that will help diagnose and prognosticate more accurately the clinical virulence of an individual's prostate cancer. The likelihood of finding such a marker or markers is greater than the chances of improving what is clearly a relatively insensitive, nonspecific serum test for prostate cancer.
Fortunately, prostate cancer, overall, is a relatively slow-growing cancer, but unfortunately, as a result, any beneficial impact of early diagnosis and intervention requires many years to become evident. This provides ammunition for the naysayers, who rightfully point out that any benefit of early detection and intervention remains to be proven. However, the slow natural biology of the disease presently limits our ability to assess the more rapid technologic improvements in the diagnosis and treatment of prostate cancer.
This should not cause us to shirk from our responsibility to intervene in men in whom the physician and patient have together agreed that intervention would be appropriate if the disease were to be diagnosed. Serial screening of healthy asymptomatic males has already been shown to cause a marked shift to earlier-stage disease at the time of diagnosis, as compared with the stage distribution of cancers found on digital rectal examination without PSA testing (the author's reference 15).
The clinician does not need to adhere to a strict set of guidelines when interpreting an individual patient's PSA. Much of medicine remains an art, and factoring in the multitude of characteristics to decide who to test and how to interpret the test is a fine example. Should the exceptionally vigorous, 68-year-old with a PSA of 4.4 ng/mL undergo evaluation with possible biopsy? What about the 68-year-old with multiple medical problems, no voiding symptoms, and a PSA of 6.0 ng/mL?
Age-specific ranges for PSA provide an additional way to interpret the test, based on economics and yield. Until a better test or tests are developed, it is reasonable for the clinician to use these as well as other modifications of PSA that are useful in individual cases. However, spending large amounts of additional funds to prove that this twist is better may not be as smart as taking a bigger step forward to attempt to discover a new, more meaningful, and more sensitive and specific test or prognostic marker.