This review of prostate-specific antigen (PSA) by Pannek and Partin, two experts in the prostate marker field, comes at a very good timea point at which great changes are occurring after a relatively long period of stability. I expect that this trend will continue. Moreover, given the rapid developments occurring in this area, some of the statements made in both the review and my commentary will probably need to be modified within the next 12 months, with further revisions necessary thereafter.
When reading this article, it is important to recall where we were not so very long ago. From 1972 to 1979, the National Prostatic Cancer Project (NPCP) was focusing its efforts primarily on trying to improve the marker prostatic acid phosphatase by various assay methods (eg, radioimmunoassay, solid-phase chromatography, and liquid chromatography, among others). In prospective, randomized trials, less than 20% of individuals newly diagnosed or followed with metastatic prostate cancer had an elevation in prostatic acid phosphatase at a given point in time. Unfortunately, very few of these patients had any measurable disease or soft-tissue lesion that could be assessed. One was left relying on a bone scan, which, at that time, was in need of further quantitation.
Thus, it is no small surprise that the work from Roswell Park and the NPCP came at a particularly good point in 1979. As the Pannek and Partin paper states, we should not dwell on the fact that, well before that date, some substance of interest was identified in the seminal fluid, but the information sat around on various dusty laboratory benches for a period of time. (Even now, there may be other markers languishing in laboratories around the world that have yet to be demonstrated to be associated to a significant degree with the prostate.) Rather, we should focus on the significance of the findingthat there was something in the prostate that was associated with malignancy (hence, the name "prostate-associated antigen," which later became known as prostate-specific antigen).
In 1980, PSA was shown to be in the serum, and by 1981, a prospective, randomized trial conducted in multiple centers with central laboratory assays demonstrated its usefulness in diagnosis and follow-up. That is the important history.
Accuracy of PSA Assays
From 1979 to 1985, events occurred that led to the development of a commercial assay. Then the tweaking began. The authors correctly point out that PSA velocity, PSA density, and other variations have been devised in an effort to increase the diagnostic accuracy and clinical correlation of PSA. This article provides a fairly detailed description of those endeavors.
What it does not discuss, however, is the accuracy of the assay itself. At the low range of diagnostic areas, from 1 to 10 ng/ml, the significance of equimolarity of the assaythe ability of the assay to equally detect both free PSA and complex or bound PSA in serumis most important. BPH tends to produce a higher proportion of free to total PSA than prostate cancer, and prostate cancer tends to produce more complex PSA, that is, PSA bound to the protease inhibitor alpha-1 anti-chymotrypsin, than free PSA. If the molar ratio in the assay is not equal, that is, if the assay at these lower ranges is not equally sensitive in detecting free PSA and bound PSA, the readings may be skewed.
The second Stanford Conference devoted some initial attention to this, principally through the work of Dr. Howard C. B. Graves. Graves pointed out that when you are measuring more than one substance in the serum, as we do with PSA, the molar ratio of the marker must be equal for both substances. If they are not, there will be skewing and overrating or underrating. Although the Hybritech-Beckman PSA assay is equimolar, not all commercial assays are. As a result, certain tests may miss the diagnosis due to overreading or, perhaps, underreading.
At this point, there is no benefit in trying to place blame, other than to say that this whole issue needs to be resolved. The American Cancer Society has asked the FDA to ensure that all future markers be equimolar, and that this be made one of the requirements for approval of any new ones. Such steps will resolve this issue.
Another point that we have to recognize is that prostate cancer is not a young man's cancer. Although the number of cases in men under age 70 has increased nearly 44% since 1986, the mean age of those diagnosed continues to be over age 62.
The 1997 figures, when published, will show a continued rise in incidence with a larger relative rise in morbidity. It has been documented, however, that the number of advanced or regional cases of prostate cancer has declined since 1986, from over 44% to roughly 30% in 1997, and that this is associated with an increase in the number of localized cases reported. These are valid objective measurements that indicate that the introduction of PSA into the diagnostic arena has been useful. They do not suggest that PSA alone is responsible for these favorable trends, but I believe any reasonable observer would conclude that there is an association, since, until PSA testing was introduced there was no decline in the incidence of advanced disease at diagnosis. This happened only after PSA came on the scene and helped identify more early cases of prostate cancer in asymptomatic men.
This review also discusses the desirability of having age-related guidelines, higher for older and lower for younger men. This, unfortunately, is not a good recommendation and has not been proven by prospective studies that have addressed the issue.
The NPCP of the American Cancer Society has now followed a cohort of 3,000 men of various ages for nearly 10 years. The data accumulated by the NPCP show that raising the PSA level as an indicator for diagnosis in elderly men would result in more diagnoses being missed. Lowering the PSA level in young men would only result in an increased number of negative biopsies. No change in morbidity and mortality are achieved, but one sacrifices both specificity and sensitivity. In other words, the old 0- to 4-ng/mL range is still effective and, after all test models and analyses were applied, was found to be the most cost-effective.
New Markers and Assays
Of course, we are currently assessing new markers, particularly free PSA. To date, studies in multiple centers have not shown any improvement with free PSA over PSA density or total PSA, at least in screening populations.[5,6] It has shown some benefit, however, in troublesome diagnostic cases. In a high-risk population, depending on the cut-off level, free PSA assays can help the clinician decide whether a repeat biopsy is needed. In a recently completed study, we showed that this was the case.
I am sure that the cut-off percentage will also be debated further. Whether it is between 15% and 24%, the precise number will not be so important as some area of agreement on it, as indicated by Pannek and Partin.
The authors do not mention the need for new assays, particularly an assay that has prognostic value. Fortunately, such an assaya serum test for prostate-specific membrane antigen (PSMA)is being developed by our laboratory in collaboration with Hybritech-Beckman after a few years of preliminary work.[8-10] We have shown that PSMA in the serum, measured by Western blot, is very definitely prognostic and is associated with evidence of clonal resistant states that can be detected pretreatment, post-treatment, and, in particular, in the presence of hormone-refractory cancer.
As suggested by our earlier work,[11-13] the FDA has approved PSMA radiolabeled and chelated for prostate SPEC scanning. Now that antibodies to the extramembrane portion of PSMA have been developed, a convenient radioimmunoassay should be achievable in the next few months. Multicenter national trials of this assay will be conducted and are expected to confirm its value. Thus, we will have not only a diagnostic PSA test but also a prognostic PSMA test.
Human kallikrein 2 (hK2) is also currently under development and may be shown to be better than PSA. That will require confirmation, of course.