As a tumor marker, prostate specific antigen (PSA) has revolutionized the detection and management of adenocarcinoma of the prostate. From its discovery in the early 1970s to its application in the 1980s and finally widespread use in the 1990s, PSA has profoundly affected the way in which we treat prostate cancer. Many researchers in basic science and clinical practice have helped to create the PSA story, and the authors of this manuscript have made major contributions to our understanding of PSA as a tumor marker.
Brief History of PSA
Prior to the 1930s, there were no biochemical markers for adenocarcinoma of the prostate; the diagnosis was made by clinical history and physical examination. In 1938, Gutman and Gutman discovered that serum acid phosphatase was frequently elevated in patients with metastatic adenocarcinoma of the prostate. Over the next 30 years, acid phosphatase was used as a serum marker for prostate cancer. However, acid phosphatase was not a sensitive marker for detection of the disease, and new markers were needed.
Several researchers claim to have first discovered PSA. Working independently in the late 1960s and early 1970s, they each identified prostate antigens, which were later found to be what is now known as PSA. Ablin and associates reported on antigens from prostatic tissue in 1970, but whether their finding was PSA cannot be confirmed.[2,3] In Japan, Hara and associates described the protein gamma seminoprotein, which they isolated from seminal plasma in 1971. Li and Beling further isolated and purified this same protein and published their findings in 1973. They reported that the molecule had a weight of 31,000 daltons and called it E1 antigen due to its mobility in conventional electrophoresis.
Sensabaugh used immunoelectrophoresis to show that this protein was semen-specific and could thus be used as a marker for semen identification. He reported its molecular weight to be 30,000 daltons and thus called the protein P30. Graves, Sensabaugh, and Blake proposed that P30 could be used to identify semen from rape suspects. Finally, Wang, Valenzuela, and Murphy reported on the purification of the protein PSA from prostate tissue.
Other PSA Milestones
Working with Wang, Valenzuela, and Murphy, Papsidero used immunoelectrophoresis to show that the protein PSA purified from prostatic tissue was, in fact, identical to PSA in human serum. Advances in amino acid sequencing have confirmed that the proteins described by all these investigators were probably also PSA.[10-12] Riegman and associates described and sequenced the PSA gene in 1989. Lilja and his group suggested in 1985 that the biological function of PSA is to liquefy the seminal coagulum,[14,15] and subsequently reported that PSA existed in serum bound mainly to alpha-2 macroglobulin and alpha-1 antichymotrypsin.
The US Food and Drug Administration (FDA) approved the first commercial immunoassay for PSA in 1986. In 1987, Stamey et al reported that the half-life of PSA is 2.2 ± 0.8 days, but in 1988, Oesterling et al measured it to be 3.2 ± 0.1 days.[17,18] Nadji and colleagues demonstrated the utility of PSA as a tumor marker in prostate cancer in 1981.
Refining the Role of PSA
In light of the overlapping causes of a rising serum PSA level, investigators have sought to refine the use of this parameter to detect, stage, and monitor the progression of prostate cancer. In the early 1990s, Carter and associates introduced the concept of PSA velocity using sera collected as part of the Baltimore Longitudinal Aging Study.[20,21] Benson and associates proposed the use of PSA density to increase the sensitivity and specificity of prostate cancer detection by PSA.[22,23] The correlation of PSA with increasing prostate gland volume was first shown by Stamey and associates in 1987 and confirmed by Babaian and associates in 1990.[17,24]
Age-specific PSA reference ranges were originally proposed by Oesterling and associates and Dalkin and his group.[25,26] The use of either free or bound PSA to improve the sensitivity and specificity of prostate cancer detection was introduced in the early 1990s by Stenman et al and Lilja et al.[27-30] Partin and associates demonstrated the clinical utility of looking at total PSA in conjunction with clinical stage and Gleason score to improve the preoperative prediction of final pathologic stage.
PSA as a Surrogate End Point
The authors review the role of PSA as a surrogate end point for both hormonally naive and refractory prostate cancer patients. They cite our Southwest Oncology Group (SWOG) study of orchiectomy, with or without flutamide(Drug information on flutamide) (Eulexin), as an example. To date, however, no study has clearly shown a significant relationship between nadir PSA level and survival. Many trials use a PSA value of < 4 ng/mL as normal. In my experience, patients with hormonally naive prostate cancer whose PSA drops to undetectable levels live significantly longer than those whose PSA does not reach nadir. Although serum PSA may not be a good surrogate for survival, other parameters such as bone scan response, change in measurable disease on computed tomography (CT) scan, and improvement in performance status are also less-than-perfect tests.
Unlike hormone-dependent disease, hormone-refractory disease seems to be better suited to the use of PSA as a valuable surrogate. In a number of clinical trials in hormone-refractory prostate cancer patients, a PSA decline > 50% is associated with improved survival. This response should be validated by a second PSA value taken 4 or more weeks later. Used as a surrogate in this fashion, PSA response represents a potentially useful marker for screening the activity of chemotherapeutic agents in hormone-refractory prostate cancer patients. That said, differentiation agents may not lower the PSA level but may still improve the patient survival rate.
TNM Staging in Prostate Cancer
The shortcomings of the tumor-node-metastasis (TNM) staging system are also discussed. This system does not always conform well with prostate cancer because of the long natural history of the disease, the age of the patients, and the fact that many patients die of other causes. Moreover, TNM staging does not assist the clinician in determining survival rates for many stages and grades of prostate cancer.
In an attempt to deal with the most common presentation of advanced prostate cancer, we published a modification of the staging system that addresses both biochemical failure and early hormone-refractory disease. For lack of a better term, we labeled a rising PSA after biochemical failure D1.5, and hormone-refractory disease (defined as a rising PSA only), D2.5. These are the most common presentations of advanced and hormone-refractory cancer in 2002.
Partin et al describe the dynamic staging system proposed by Scher and Heller. Another approach to staging that we have used employs artificial neural networks that process data sets from large clinical trials. These neural networks incorporate large amounts of data and develop an outcome. We have a website (www.prostatecalculator.org) that utilizes numerous neural networks to help predict organ-confined disease, biochemical failure, and survival in patients with advanced disease.
New Directions in Phase II and III Trials
The authors review conventional phase II and III clinical trial designs in prostate cancer. It should be noted that many phase III trials require the enrollment of more than 100 to 300 patients if the equivalence of treatment results is being evaluated.
I agree with the authors’ statements regarding new directions in phase II and III clinical trials in hormone-refractory prostate cancer. During my tenure as chairman of the Genitourinary Committee of SWOG for the past 20 years, a number of chemotherapeutic agents have been evaluated in hormone-refractory prostate cancer patients. Early in our evaluation, only patients with measurable disease (a small subset of those with hormone-refractory disease) were included. Although recent trials have incorporated PSA as a surrogate for response, improvement in survival continues to be the desired end point. Unfortunately, no agent or combination of agents has been shown to improve survival.
Our current trial involves the evaluation of mitoxantrone(Drug information on mitoxantrone) (Novantrone) and prednisone vs docetaxel (Taxotere) and estramustine(Drug information on estramustine) (Emcyt). PSA is being considered as a surrogate for survival.
Challenges in Chemoprevention Trials
The authors review the role of PSA in the diagnosis and management of prostate cancer. Their discussion includes chemoprevention of prostate cancer and the possible role of PSA in identifying and following high-risk cohorts of patients. Chemoprevention trials in this disease require large samples and lengthy follow-ups. One strategy for dealing with this challenge is to include high-risk cohorts.
The inclusion of high-risk patients reduces the number of subjects required to evaluate chemopreventive strategies. Age, race, family history, high-grade prostatic intraepithelial neoplasia (PIN), and high PSA levels are well-defined risk factors. For example, if a study includes men with familial prostate cancer, whose risk approaches nearly 50%, several hundred men would be required to demonstrate a 20% to 30% reduction in the incidence of the disease. If the general population were studied, the number of patients required would be much larger. To detect a 25% reduction in prevalence in subjects taking the chemopreventive agent vs placebo would require the enrollment of 18,000 men. If the estimated difference is smaller, a study population of 50,000 men or more would be necessary.
Ideally, the incidence of cancer should be the end point in trials of chemopreventive agents for prostate cancer. However, such trials may require 10 or more years to reach this end point. To gain time, intermediate end points might be chosen, but no reliable biomarker has been validated in such a way that it can be used as a replacement for definitive end points.
Nevertheless, PSA appears to be a useful surrogate for following patients in chemoprevention trials. A rising PSA level is associated with an increased risk of the development and/or progression of prostate cancer. The major advantage of using PSA is that it represents an inexpensive noninvasive method of selection.
Defining a High-Risk Cohort
Given that the risk of prostate cancer parallels PSA level, it is not surprising that men with an initial PSA > 10 ng/mL have nearly a 50% risk of disease. The use of serum PSA levels to define a high-risk cohort was investigated in a study of men who participated in annual screening for 3 years. A total of 6,804 men with an initial PSA level < 4 ng/mL (IMx assay) and a normal digital rectal examination were extracted from a database of 55,000 men. Those with a PSA level < 2 ng/mL had a very low risk of an abnormal PSA within 3 years (< 3%), whereas those with a level of 2.1 to 4.0 ng/mL had a 34% chance of developing an abnormal PSA during follow-up. The latter group seems to represent an ideal cohort for chemoprevention trials.
Agents that have a suppressive effect on PSA represent another challenge in such trials. Finasteride(Drug information on finasteride) (Propecia, Proscar), antiandrogens, differentiation agents, and other hormonal agents can lower the PSA level but may not have a salutary effect on the disease.
We recently reported the results of annual prostate cancer screening in a cohort of men in the Prostate, Lung, Colorectal, Ovarian (PLCO) Screening Trial, with the risk of developing an abnormal PSA predicated on initial blood levels. For men with a PSA < 1 ng/mL, the risk of developing an abnormal PSA (ie, > 4 ng/mL) at 5 years was less than 1.6%. However, men with a PSA ranging from 3.1 to 4 ng/mL had an 83% chance of developing an abnormal PSA within 5 years. This high-risk cohort would be the appropriate focus of chemoprevention trials.
In summary, the authors have provided a scholarly overview of PSA as a marker of disease activity in prostate cancer. PSA is perhaps the most important tumor marker in oncology and has increased the detection of organ-confined disease. It offers promise in assessing response to treatment for hormone-sensitive and refractory disease and as a surrogate to identify men for chemoprevention trials.