Dr. Roach initiates his discussion with the relevant statement that how we detect, stage, and treat carcinoma of the prostate, as well as subsequently evaluate treatment efficacy, has forever been dramatically altered by the availability of prostate specific antigen (PSA), which has been labeled "the most useful tumor marker available" . However, as Dr. Roach also notes, new information and insights generate new questions and uncertainties about the best applications of this valuable tumor marker. Some of the uncertainties are discussed in his review and include, but are not limited to:
1) different monoclonal/polyclonal assay methodologies (in the United States there are 5 different PSA assays available to the physician and in Europe there are now 36 different assays);
2) no universally accepted reference standard;
3) variable physician-assigned intermediate PSA end points (as an example, the "undetectable" PSA level after surgery has ranged from equal to or less than 0.1 ng/mL to equal to or less than 0.6 ng/mL and after definitive irradiation, from 0.5 ng/mL to 4.0 ng/mL); and
4) biologic variations (eg, an individual's serum PSA level will vacillate within short periods and is determined by varying contributions from benign prostatic hypertrophy and such other prostatic perturbations as infection, infarct, and basement membrane permeability associated with aging, in addition to the parameter of primary interest, cancer volume).
The Cancer- and Organ-Specificity of PSA
Major efforts have been directed toward increasing the cancer specificity of PSA and thereby reducing the number of false-positive results. These include age-specific reference ranges, PSA density, PSA velocity, and, most recently, free-to-total PSA ratios. While urologists await the validated free-to-total PSA ratio cut-off points to hopefully increase the specificity of early cancer detection, radiation oncologists will analyze post-radiation free-to-total ratios to identify patients whose PSA falls to levels between undetectable and normal, who may experience a favorable prognosis. Will there be a subgroup of patients with a detectable PSA after external-beam radiation therapy who, by virtue of their high ratio of free to total PSA, will fare more favorably than a subgroup with the same total PSA level but a low ratio?
Although increasing the specificity of PSA for cancer detection is a major objective, ample information is simultaneously appearing to challenge the validity of the prostate organ-specific label we have attached to this 30-kd protease. Prostate-specific antigen has been found in breast tissue, both normal and cancerous, as well as in the ovary, endometrium, colon, liver, and kidney . It appears that any organ with hormone-receptor activity can express PSA with appropriate ligand stimulation .
Dr. Roach appropriately emphasizes the importance of PSA as a pretreatment stratification/ prognostic factor. Several series have demonstrated the prognostic value of PSA and, with multivariate analysis [3,4], have identified it as the most important prognostic pretreatment variable. Our analysis of pretreatment PSA distribution in selected contemporary surgical and radiation series treating stage T1/T2 prostate cancer clearly demonstrated a statistically more favorable distribution in the surgical series . Based on this less favorable PSA tumor mix within the same tumor stage, it is anticipated that the outcome of the radiation series also will be less favorable.
The high failure rate by PSA criteria after radiation , the identification of PSA failure in a relatively short time interval of 12 to 24 months after radiation, and the recognition that pretreatment PSA levels more than 10 ng/mL, and certainly more than 20 ng/mL, predict for a significant subsequent PSA failure rate have prompted implementation of treatment-intensification strategies. These have included neoadjuvant or adjuvant androgen deprivation or a combination of both with radiation therapy, conformal radiation therapy with dose escalation, and wide-field radiation therapy to include the pelvic nodes. Some of these modifications are addressed in the author's Table 2, which describes the multiarm RTOG protocol 9413. Other strategies combine brachytherapy, radiation sensitizers, hyperthermia, neutrons, or protons with traditional photon therapy.
All of these strategies key initially on achieving a low PSA nadir as an intermediate end point. However, this focus should not detract from the practical reality that many patients remain clinically well not only during their post-treatment PSA nadir but also during its subsequent rise. The lead time between a detectable PSA and clinical failure for patients with clinical stage B2 in our series was 54 months . Clearly, for patients in the eighth decade of life, irradiation can often provide freedom from clinical progression for their anticipated life span.
Defining the Appropriate Post-Radiation PSA
Nevertheless, a major effort has been directed toward defining the appropriate PSA level to expect after external-beam radiation therapy. Indeed, this is an important effort, not only as one yardstick for assessing the relative merits of new therapies, but also to standardize reporting of results from different institutions. A consensus conference to address this important issue will be held this fall under the auspices of the American Society for Therapeutic and Radiation Oncology (ASTRO).
Over the past several years, the nadir PSA level acceptable to radiation oncologists after external-beam radiation therapy of localized prostate cancer has decreased from a PSA of equal to or less than 4.0 ng/mL to equal to or less than 2.0 ng/mL  to equal to or less than 1.5 ng/mL  to equal to or less than 1.0 ng/mL , and, most recently, to equal to or less than 0.5 ng/mL . The intermediate- and long-term biomarker (PSA) progression-free outcomes of any series will understandably vary according to which level has been used as an acceptable post-radiation nadir. The cohort with the best prognosis are patients who have a PSA of equal to or less than 0.5 ng/mL . These patients have been described as meeting the criteria of a bloodless prostatectomy. The group most likely to reach the nadir of equal to or less than 0.5 ng/mL are those with a pretreatment PSA of equal to or less than 4.0 ng/mL. (In our Eastern Virginia Medical School series, only 13% of patients had a pretreatment PSA at or below this level).
This finding raises an interesting paradox; namely, the most favorable treatment outcomes are seen among patients in whom our best tumor marker has not indicated the presence of tumor. Presumably, these patients were diagnosed based on an abnormal digital rectal examination, generally considered to be very insensitive to the presence of tumor or the extent of tumor volume. This serves as a reminder that, while biomarkers are in the process of evolution, testing, and application, physicians involved in the detection, treatment, and care of patients with prostate cancer need to practice and maintain their clinical skills.