WHO first international standards for prostate-specific antigen: the beginning of the end for assay discrepancies?

Despite the success of prostate-specific antigen (PSA) as a tumor marker for the early detection and monitoring of disease in prostate cancer, both clinical and analytical limitations of PSA remain (1 ). Since the introduction of assays for total PSA in the mid-1980s, assay results have been found to differ among manufacturers (2, 3). Much has been learned in the last two decades about the biology of PSA and the molecular forms of PSA present in serum. Differences among assays still persist, however, even among assays from the same manufacturer that have the same antibodies but different assay formats. With the availability of the First International Standards from WHO, is this the beginning of the end for assay discrepancies? The lack of standardization of PSA was recognized early on as contributing to disparities among assays. Similar to some other tumor markers, a lack of defined antigens, differences in calibrator composition and specific molecular forms, and differences in calibrator assignments as well as the lack of a reference method are some of the issues hampering assay standardization. The lack of standardization of PSA prompted several organizations as well as individual researchers to convene meetings and conferences to address the issue. The Second Stanford Conference on International Standardization of Prostate-specific Antigen (4 ) was held in 1994. At this conference, the use of a standard, prepared at Stanford by Tom Stamey, consisting of 90% purified PSA-a1-antichymotrypsin (ACT) and 10% free PSA (90:10) on a molar basis was proposed with the rationale that these proportions of PSA represented the proportions most similar to those observed in patients with prostate cancer. Subsequently, the NCCLS issued an approved document (5 ) recommending a set of three distinct materials for calibrating immunoassay procedures. The materials would contain 100% free PSA, 100% PSA-ACT, or 90% PSA-ACT:10% free PSA. In this issue of Clinical Chemistry, Rafferty et al. (6 ) evaluate the Stanford 100% free PSA and 90:10 PSA preparations as potential WHO international standards in an international collaborative study involving 10 laboratories in six countries. The authors should be congratulated on their efforts to identify a material that will allow manufacturers to set their calibrations to common values to achieve the goal of overall PSA standardization. The WHO Expert Committee on Biological Standardization has established the 1 mg free PSA and 1 mg total PSA in the 90:10 PSA preparation as the First International Standards, 1999. In the study by Rafferty et al. (6 ), three human serum samples were analyzed by the participating laboratories using in-house reference materials and the 90:10 PSA standard as calibrators. Using the common 90:10 standard, the between-laboratory CVs decreased by 22–59% at concentrations of 1–15 mg/L. Although standardization among assays can reduce interassay variability, it is only a first step, as other factors contribute to differences among assays. A major factor relates to the ability of the assay to measure equally the free forms of PSA in serum and the forms bound to protease inhibitors, primarily ACT. Assays that measure free and bound forms of PSA equally are termed “equimolar” response assays, whereas those that do not are termed “non-equimolar” or “skewed” assays (7, 8). Non-equimolarity may result from the choice of antibodies and the PSA forms used to generate the antibodies. Polyclonal antibodies tend to be heterogeneous and recognize various epitopes on PSA. Some of these epitopes may be blocked by ACT. More than one polyclonal antibody may also bind to a PSA molecule, producing a larger signal and hence higher results. These circumstances may cause a non-equimolar response. Epitopes recognized by 83 antibodies provided by various researchers and manufacturers of PSA assays were studied in detail and characterized with respect to linear or conformational-type epitope and cross-reactivity with PSA-related molecules such as human kallikrein 2 (9 ). An epitope map for PSA, generated from these studies, made clear the potential variation from choice of antibodies in PSA assays. Non-equimolarity may also reflect assay kinetics where free PSA, with a low molecular weight, may bind preferentially in assays with shorter incubation times. A nonequimolar assay with an incubation time of only 7.5 min appeared to be an outlier in the study by Rafferty et al. (6 ) when the 100% free PSA preparation was analyzed, giving results higher than the other laboratories. The effect of non-equimolar assays on PSA values is most evident in specimens containing a high percentage of free PSA. We recently showed this phenomenon in proficiency testing specimens (10 ), such as those from the College of American Pathologists (CAP), which were prepared by adding semen, which is composed almost entirely of free PSA, to specimens. Others have also reported this phenomenon (11–13). We found CVs of 54–56% for three different Ligand Survey specimens among 10 individual total PSA assays (10 ). Assays known to be non-equimolar had results 119% and 72% higher than the mean of 6.07 mg/L (median, 5.36 mg/L), respectively, in one of these specimens. Variability among assays was minimized when, as in the study by Rafferty et al. (6 ), pooled patient specimens were analyzed. In contrast to the Ligand Survey (which used specimens containing .90% free PSA), when the three CAP Reference Materials, with 7–15% free PSA, were analyzed by these same 10 assays, the among-method CVs were only 6–13%. Despite the reduction in variability among assays shown by Rafferty et al. (6 ) with the 90:10 PSA standard, use of this standard will not correct all problems resulting from non-equimolar assays. In contrast to equimolar Editorial