Clínica de Urologia Dr. Charles Rosenblatt

A New Marker for Prostate Cancer

Herbert A. Fritsche, PhD
Chief of Clinical Chemistry, Biochemist, and
Associate Professor
The University of Texas M.D. Anderson Cancer Center
Houston, Texas, USA

Prostate cancer continues to be a serious health threat worldwide. In the United States, for example, estimates suggest that, by the time figures for 1998 are in, there will have been 180,000 new cases diagnosed during the year and approximately 40,000 more men will have died from prostate cancer.1 Table 1 lists the age-adjusted prostate cancer death rates for selected countries.2
As with breast cancer in women, for which prevalence rates in the United States are comparable to those for prostate cancer in men, early detection of prostate cancer permits effective surgical treatment and the opportunity for cure. The serum prostate-specific antigen (PSA) test has contributed greatly to the early detection of prostate cancer, and the assay is now included as a routine component of diagnostic algorithms along with the digital rectal examination (DRE) and transrectal ultrasonography. The American Cancer Society recommends that a serum PSA determination be performed with the annual DRE for men 50 years and older-and starting at age 40 if there is an increased risk of prostate cancer.3
Early detection of prostate cancer and the concept of cancer "screening" itself, however, have not gone without criticism. Controversies have revolved around reliability of the evidence that early detection of prostate cancer leads to improved patient outcome and proof of the cost effectiveness of cancer screening. Questions over improved outcome center on the types of cancers that are being detected by PSA testing. Are those cancers that are detected early actually indolent tumors of low grade, which will never present a health threat to the patient and therefore do not need to be detected? Or, are they extracapsular tumors for which there is no effective treatment and for which early detection will not improve the patient's survival time?
Such questions can be answered only after prospective clinical trial data are obtained. However, it should be noted that early detection programs have contributed to the increasing number of early-stage tumors being diagnosed and that most of these cancers meet the current criteria defining what is clinically significant-and thus require early detection and immediate treatment.
A chief concern over the early-detection programs for prostate cancer is the high false-positive rate associated with the serum PSA test when using the recommended cut-off value of 4.0 ng/mL. The number of positive tests associated with disease, known as the positive predictive value (PPV), is approximately 25% when serum PSA is in the range of 4 to 10 ng/mL. In other words, 3 out of 4 men who do not have cancer will be erroneously identified for additional diagnostic workup and biopsy. When the serum PSA is >10.0 ng/mL, the PPV is approximately 50%. Thus, the major diagnostic problem occurs when serum PSA is in the 4 to 10 ng/mL range, commonly referred to as the diagnostic "gray zone."
For some urologists, the gray zone is limited to the 4 to 6 ng/mL range, as patients who have serum PSA values >6.0 ng/mL will automatically be designated for additional workup or biopsy. As listed in Table 2 and discussed below, various attempts have been made to improve the specificity of the PSA test and to reduce the false-positive rate that occurs in the diagnostic gray zone.

Age-Referenced PSA Values
With the age-referenced PSA value, clinicians vary the PSA cut-off value depending on the subject's age. For example, the PSA cut-off value of 3.5 ng/mL would be used for men in their 50s, 4.5 ng/mL for men in their 60s, and 6.5 ng/mL for men in their 70s.4 Since the size of the prostate gland typically increases as men get older, age has been proposed as a surrogate marker for prostate gland volume. In reality, the serum PSA value is a reflection of prostate gland volume, particularly the volume of the transition zone of the prostate gland.
Age-referenced cut-off values improve the clinical specificity; but they do so at the expense of clinical sensitivity, allowing some tumors to be missed in older men. This clinical trade-off of increased specificity for decreased sensitivity has not been widely accepted.

PSA Density
The measurement known as PSA density is the ratio of the serum PSA to the prostate gland volume, as calculated from ultrasound measurements. Recently, the transition- zone volume has been shown to be a better index than total gland volume, since benign prostatic hypertrophy (BPH) occurs almost exclusively in the transition zone.5
Most investigators believe that the measurement of PSA density is helpful in the differential diagnosis of BPH and prostate cancer when it is performed by a well-trained individual using modern ultrasound equipment. But the test is too expensive and too slow for general use in early detection programs.

PSA Velocity
Serum PSA velocity is the rate of increase of serum PSA over a designated time interval. The measurement of velocity is based on the observation that rapidly rising serum PSA values occur as a result of prostate cancer, as compared to slower increases, which are due to BPH.
This approach has several major limitations. One is the wide variation of serum PSA demonstrated by some men. This biologic variation may be as high as 30% for serial measurements and could be responsible for overestimation of the velocity.6 Second, technical limitations of current PSA methods and some laboratories may not permit attainment of the long-term test precision necessary to insure the reliability of the velocity calculation. Finally, it has not yet been established what number of measurements should be made or over what period of time they should be made to calculate the velocity.

Free-to-Total PSA Ratio
Total serum PSA is a measure of the noncomplexed or "free" forms of PSA and the PSA complexed to several inhibitor substances present in serum. The predominant measurable form of complexed PSA is bound to alpha-1 antichymotrypsin (ACT); PSA complexed to alpha-2 macroglobulin is not detectable by conventional immuno-assays. Free PSA exists either in a nicked form, in which one or more peptide bonds in the molecule are broken, or in a proenzyme form. In both of these cases, the conformation of the PSA molecule prevents its binding to ACT.7,8
Christensson9 and Stenman10 were the first to demonstrate the role of the free-to-total (F/T) serum PSA ratio as an aid in the differential diagnosis of prostate cancer and BPH. Recently, the F/T PSA test has been proposed for use in ruling out biopsies for men who have total serum PSA values in the 4 to 10 ng/mL range. A low F/T ratio (<10%) is associated with cancer while a high F/T PSA ratio (>25%) is indicative of BPH. The test is not proposed for use in identifying cancer patients, but rather to identify BPH patients and thus avoid unnecessary biopsies.11
While the F/T PSA test has been subjected to intense study and has received approval from the US Food and Drug Administration, there are concerns related to its use in routine practice.12 These concerns involve the cut-off value selected and the nature of the cancers that are missed13; the stability of free PSA14; the necessity for the use of two PSA test procedures (free and total); and the imprecision of the assays and the effect of this measurement on the confidence interval of the calculated value.15 Clearly, there are limitations to the application of the F/T PSA ratio to routine clinical practice.

Complexed PSA
The complexed PSA (cPSA) assay* (Bayer Diagnostics, Tarrytown, New York), available on the Bayer Immuno 1™ system, uses a monoclonal antibody (anti-epitope E) to block the free PSA from binding to magnetic particles coated with antibodies to fluorescein, which are used to capture the fluorescein-labeled MM1 antibody as depicted in the accompanying Figure.16 The MM1 antibody is able to bind only the PSA-ACT complex. A PSA polyclonal antibody labeled with alkaline phosphate is used as the indicator antibody. The cPSA assay has excellent performance characteristics, with a detection limit of 0.016 ng/mL and excellent precision (CV = 2.3% at mean = 3.34 ng/mL and CV = 2.0% at mean = 14.88 ng/mL).
In a multi-site evaluation, sera from 202 cancer patients who had a total PSA measurement in the range of 4 to 10 ng/mL were tested for cPSA. A cut-off value for cPSA of 3.75 ng/mL had been previously defined as the value that provided a 95% sensitivity. Using this cut-off value, investigators found that the clinical specificity for the cPSA assay in a group of 237 noncancer subjects was 21%.
Comparison of the sensitivity and specificity for cPSA at the 3.75 ng/mL cut-off value with the F/T PSA using a 25% cut-off shows the cPSA assay to be more specific than the F/T ratio (21% vs 13%), with similar sensitivities (93% vs 96%) for samples in the 4 to 10 ng/mL range. When the PSA range is reduced to 4 to 6 ng/mL, the specificity of the cPSA assay is improved to 37%, while the specificity of the F/T ratio remains at 13% (Table 3 ). There is an apparent decrease in the sensitivity for the cPSA test, but this is due to the fewer number of cases studied (74 vs 202). Overall, the cPSA assay demonstrates much greater specificity than the F/T PSA ratio in the 4 to 6 ng/mL range. A recent report from Brawer and colleagues also demonstrates a two-fold improvement in specificity of the cPSA assay when compared to the F/T ratio for samples in the 4 to 10 ng/mL range.17

Conclusions
The multi-site evaluation of the cPSA assay suggests that the direct measurement of complexed PSA is equivalent to the F/T PSA ratio for sensitivity and may offer improved specificity for samples with total PSA values in the 4 to 10 ng/mL range. In addition, cPSA may have its most useful application in the 4 to 6 ng/mL range, where it appears to have much greater specificity than does the F/T ratio.
Additional studies are underway to confirm these initial observations. In addition to the possibility of offering greater specificity for cancer, the cPSA test offers a less complicated test alternative to the F/T ratio, since it eliminates the need to reflex to a second test and it measures a more stable analyte.

Prostate-Specific Antigen as a Prognostic Predictor for Prostate Cancer

Martin E. Gleave and S. Larry Goldenberg
British Columbia Cancer Agency and University of British Columbia, Vancouver, British Columbia, Canada
Nicholas Bruchovsky
British Columbia Cancer Agency, Vancouver, British Columbia, Canada

I. INTRODUCTION
The ideal tumor marker is not available and likely does not exist. All tests used clinically in medicine are imperfect, with false positives and false negatives, and require judgment and experience in their interpretation. If one were to exist, the ideal tumor marker for prostate cancer would be accurately and reliably measured, with minimal diurnal variation would be tumor specific, predictive of presence of early, low-volume, or organ-confined cancer that is likely to progress; and would predict clinical stage by identifying patients at high risk of having extracapsular extension, microscopic nodal, or osseous metastases. It would also help predict response to either surgical, radiation, or endocrine therapy, and identify preclinical recurrences. Over the past decade, increased confidence and clinical experience has identified prostate-specific antigen (PSA) as the best tumor marker for prostate cancer and the best overall tumor marker in oncology [1-3]. PSA is a tissue-specific tumor marker used by urologists and oncologists to monitor treatment responses, prognosis, and progression in patients with prostatic cancer (Table 1). The role of PSA in staging and screening of prostate cancer remains limited because of the wide and overlapping ranges in serum PSA levels that exist in many patients with either localized disease or advanced metastatic disease. These wide ranges in PSA levels may be related to genetic (tumor heterogeneity) or epigenetic (e.g., changes in tumor cell microenvironment or hormonal, growth factor, or extracellular matrix milieu) factors. PSA testing has revolutionized how clinicians approach men with prostatic disease. It is not perfect, however, and requires experienced clinical interpretation and judgment in its application to individual patients, like all tests in medicine.

1. Monitoring systemic therapy of metastatic disease
2. Early detection of recurrence after radical prostatectomy or radiation therapy
3. Immunohistochemical diagnosis of poorly differentiated carcinoma
4. Diagnostic adjunct in patients with suspicious DRE
5. Staging adjunct of diagnosed cases
6. Early detection for men aged 50 to 70 years with normal DRE

Prostate-specific antigen was initially identified in human seminal plasma by Hara et al. [4] and termed gamma seminoprotein; it was later used as a semen marker in rape victims [5]. In 1979, Wang et al. [6] isolated and purified a protein produced only in prostate epithelial cells (i.e., prostate-specific) that was distinct from prostatic acid phosphatase (PAP) and identical to that found in semen, and termed it PSA. Development of enzyme-linked immuno-absorbant assays permitted detection of serum PSA to levels as low as 0.1 µg/L. The use of PSA to diagnosis, to predict tumor volume and stage, and to predict prognosis in patients with prostate cancer is improving as factors regulating its production and clearance become better defined.

II. BIOLOGY OF PROSTATE SPECIFIC ANTIGEN

A. Biomolecular Characteristics
Prostate-specific antigen is a serine protease produced exclusively by human prostatic epithelium [7]. This 34 KDa glycoprotein consists of a single polypeptide chain of 240 amino acids and closely resembles other human tissue kallikreins [6,8]. The PSA gene is a member of a small family of genes encoding kallikrein-like serine proteases, which include PSA, tissue kallikrein, and human glandular kallikrein (hGK-1) [9]. The three genes are highly homologous and clustered in the kallikrein locus at chromosome 19q13.2-13.4 [10]. The PSA gene is approximately 6 kilobases in size and is composed of 4 introns and 5 exons. PSA and mRNA or protein is detectable only in epithelial cells of normal prostate, benign prostatic hyperplasia (BPH) tissue, and both primary and metastatic prostate cancer cells [11].

B. Prostate-Specific Antigen Gene Regulation
Transcriptional activation of the PSA gene increases PSA mRNA levels, which, following translation, leads to increased PSA protein levels. Initiation of transcription occurs through binding of transcription factors to specific DNA-binding domains in the 5' promotor region of the PSA gene. Regulation is likely complex and involves one or more activator and inhibitor binding regions, as well as protein-protein interactions between transcriptional factors.

Androgen-Dependent Prostate Tissue
The regulatory action of androgens is mediated through the androgen receptor, which is a member of the superfamily of ligand-responsive transcription factors [12]. Androgen receptors contain three major domains: an N-terminal region involved principally in transcriptional activation; a DNA-binding domain, which is required for interaction with specific gene sequences as well as transactivation; and a steroid-binding domain in the C-terminal end of the molecule.
The expression of the PSA gene in normal and BPH tissue, and in androgen-dependent prostate cancer, is directly regulated by androgens through their binding to the androgen receptor [9,13,14] (Fig. 1) Transcriptional regulation by androgen receptors is mediated via direct binding to specific enhancer-like DNA sequences in the promotor regions of target genes termed androgen responsive elements (ARE). The promotor region of the PSA gene contains a functional ARE that is closely related to the hormone response elements for the progesterone and glucocortoid receptors [15]. Additionally, the androgen receptor interacts with general and specific transcription factors to form a stable transcription preinitiation complex that permits efficient transcription of target genes.

Figure 1 Prostate-specific antigen gene expression is androgen-regulated. Testosterone is converted to its more active metabolite dihydrotestosterone, by 5 -reductase, and then binds to the steroid-binding region of the androgen receptor. Once activated, the androgen receptor binds to specific DNA-binding region (androgen response elements, ARE) of the promoter region of the PSA gene, which activates gene transcription (copyright Dr. N Bruchovsky).

Although PSA transcription is regulated by androgens, little is known about the actual mediation of this process, which seems to be influenced by tissue-specific cofactors [16,17]. PSA synthesis is stimulated by vitamin D [18], and phorbol esters down-regulate the androgen induction of PSA [19). Most growth factors have minimal or no effect on PSA mRNA levels in LNCaP cells over a broad range of concentrations [20]. Epidermal growth factor (EGF) and fibroblast growth factor (FGF) may decrease the secretion of PSA, while both EGF and transforming growth factor-alpha can interfere with the androgen regulation of PSA [21]. Recent investigations suggest that insulin-like growth factor-I (IGF-1) may also stimulate PSA gene expression, possibly through the AR, because its stimulatory effect can be inhibited by the antiandrogen Casodex [22]. Furthermore, PSA mRNA expression is down-regulated in a dose-dependent fashion by the differentiation agents dimethylsulfoxide (DMSO), dimethylformamide, and fenretinide, but up-regulated by the differentiation agent sodium butyrate (M. Gleave, unpublished data). Taken together, these observations suggest that PSA gene expression in androgen-dependent tissues is regulated primarily by androgens, but can be modified by other transcriptional factors through mechanisms that remain poorly defined.

Androgen-lndependent Prostate Cancer
Following castration, serum PSA levels decrease due to cessation of androgen-regulated PSA gene expression and to decreased tumor volume because of castration-induced apoptosis. Increases in serum PSA after castration are the earliest sign of progression to androgen independence [23] and reflect nonandrogen regulation of a previously androgen-regulated gene [24,25]. The molecular basis for the increased production of PSA in the androgen-independent state has received little attention. Although factors that down-regulate PSA production have been identified, nonandrogenic factors that increase PSA production during progression to androgen independence have not been characterized. How prostate cancer cells regain their ability to synthesize PSA in quantities similar to precastration levels in the absence of androgens is unknown, yet it can be safely assumed that greater understanding of this process is necessary before strategies for delaying or averting androgen independence can be developed.
The LNCaP prostate tumor model permits rapid assessment of the molecular events that mediate progression to androgen independence [26,27]. Of the current human prostate cancer cell lines available for study, only the LNCaP system is androgen responsive, PSA secreting, and immortalized in vitro. Because of its unique ability to produce PSA both in vivo and in vitro, this model system serves as a novel and powerful tool to define aspects of the regulation of PSA. LNCaP tumors give rise to androgen-reduced serum PSA levels in athymic mice that correlate with tumor volume [20]. Immediately after castration, the synthesis of PSA by the tumor decreases by 80%, but remains androgen sensitive because PSA synthesis increases sharply after testosterone replacement. However, PSA production gradually returns to precastrate levels in the absence of testicular androgens beginning 4 weeks after castration (Fig. 2).

Figure 2 Changes in serum PSA in the LNCaP tumor model mimics the course of human prostate cancer. Serum PSA levels increase in intact male mice proportional to increases in tumor volume (a), and decrease rapidly by 80% after castration (b). Prostate-specific antigen levels begin to rise 4 weeks after castration, heralding the onset of androgen-independent PSA gene expression.

Animal model data suggest that escape from androgen-regulated PSA gene expression may result from LNCaP cells adapting to an androgen-deprived environment through the up-regulation of alternative nonandrogenic pathways of signal transduction [28,29]. To permit more detailed mechanistic studies, LNCaP tumors grown in castrate mice that had progressed to androgen independence were cultured to obtain cell lines [24,25]. A cell line was obtained from a tumor-bearing animal 4 weeks after castration and termed the C-4 LNCaP cell subline. The spontaneous recovery of PSA synthesis in Al LNCaP cells appears to be an acquired stable phenotype because the C-4 LNCaP cell line continues to overexpress PSA in the absence of androgen stimulation. Furthermore, when the conditioned medium of the C-4 LNCaP cell line is incubated with parental LNCaP cells, PSA production in the parental LNCaP cell line is stimulated fourfold (Fig. 3). These observations suggest that Al C-4 LNCaP cells produce a soluble nonandrogenic factor that stimulates PSA production in an autocrine fashion, and is likely partly responsible for increases in PSA with progression to androgen independence; its production may represent an adaptive response to changes in the tumor cells' hormonal milieu.

Figure 3 (a) Androgen-independent C-4 LNCaP cells produce a soluble protein factor that up-regulates PSA gene expression in a dose-dependent fashion in parental LNCaP cells (M line) as well as C-4 cells, and is illustrated schematically in (b).

C. Determinants of Serum PSA Levels
Circulating measurable serum PSA levels represent a steady state between the total production of PSA, the fraction that backleaks into the serum, the percentage that is protein bound and not immunodetectable, and the rate of excretion.

Total Production
Total PSA production is determined by the level of PSA gene expression and rate of PSA protein secretion per cell, and the total number of PSA-producing cells. Factors regulating PSA gene expression are reviewed above; post-translational modifications and rate of transport through the endoplasmic reticulum and Golgi apparatus with subsequent secretion remain largely undefined. Aside from medications that decrease or block testosterone, most medications do not affect circulating PSA levels. An exception may be high intake of carotenes or other vitamin A derivatives, which may decrease the production of PSA [30].
Benign and epithelial cell volume is a major determinant of serum PSA levels. Using the Yang assay, Stamey et al. [31] calculated that BPH raises the serum PSA level at a rate of 0.3 (0.5 using Hybritech Tandem E assay) µg/L/g of BPH tissue, while cancer raises the serum PSA level 10-fold higher on a gram-for-gram basis. Cancer cells do not produce more PSA on a cell-for-cell basis than BPH cells do; rather, the increased contribution to serum PSA levels by cancerous tissue is the result of the greater degree of "backleak" because of blind-ending malignant acini or stromal infiltration by malignant cells, and to a denser cell population per cubic millimeter of tissue in cancer, which is not characterized by large glands with intervening stroma as in BPH.

"Backleak"
Under normal conditions, PSA is secreted into acini by luminal epithelial cells and then excreted through prostatic ducts into the seminal fluid in concentrations one million times higher than serum concentrations. PSA concentrations in seminal fluid range from 0.24 to 5.50 g/L, averaging 1.92 g/L [32]. The resorption of PSA into interstitial fluids and serum is minimized in the presence of patent ducts, and intact epithelial cell junctions and basement membrane. Conditions that result into ductal obstruction or disruption of tight junctions or basement membranes will lead to ''backleak'' of PSA into the interstitium of the prostate gland with subsequent entrance into the circulation. Increased serum PSA levels may result from inspissated secretions or distorted architecture in BPH; disruption of basement membrane integrity may result from prostatic infarction, prostatitis, or prostatic instrumentation.
Serum PSA levels do not appear to fluctuate significantly throughout a 24-hour period, with serial repeat measurements within individuals differing by less than 10% [33]. Serum PSA levels do decrease following hospitalization, which may be the result of lack of physical activity or immobility [31]; however, levels are not significantly increased by sexual activity or digital rectal examination [34,35]. Cystoscopy can increase serum PSA levels four-fold, and needle biopsies and transurethral resections can temporarily increase PSA levels up to 50-fold [1], all due to an increased "backleak" of PSA into the serum.

Protein Binding
Enzymatically active PSA that backleaks into the extracellular fluids and serum is usually inactivated by irreversible binding to -1-antichymotrypsin (ACT) or -2-macroglobulin [36,37]. Only a small fraction of PSA in sera is actually free, and it remains unclear whether the free form in serum manifests any enzymatic activity. Both ACT and -2-macroglobulin are protease inhibitors produced by the liver and are present in 104-fold molar excess to PSA in serum. PSA complexed with -2-macroglobulin is not accessible for immunodetection because of steric shielding, while that bound to ACT has a sufficient number of antigenic epitopes exposed to interact with anti-PSA antibodies. The PSA-ACT complex is the major molecular form of PSA in serum; a small portion exists in a free, noncomplexed form, despite the several thousand-fold molar excess of serum ACT. Recent investigations report that the proportion of serum PSA complexed to ACT is significantly higher in cancer than in BPH [36-38]. The higher proportion of PSA complexed to ACT in cancer compared to BPH may be due to structural changes in the PSA molecule produced by cancer cells. The difference in complexed versus free PSA in cancer and BPH may provide a method for improving the ability of PSA to differentiate between these two common clinical conditions, as discussed below.

Prostate-Specific Antigen Clearance
The half-life of serum PSA was determined by Stamey et al. [31] using the Yang assay to be 2.2 0.8 days, and by Oesterling et al. [39] using the Hybritech assay to be 3.2 0.1 days. These determinations of PSA clearance were made from data obtained from patients in the postoperative period after radical prostatectomy. Clearance of PSA from the serum follows an exponential decay pattern characteristic of first-order elimination kinetics, as indicated by the straight line obtained when the natural logs of serum PSA values are plotted as a function of time. Because of the relatively long half-life of PSA, coupled with slow resolution of inflammation following biopsy or prostatitis, it may take several months for serum PSA to reach its baseline after transurethral prostatic resection (TURP), biopsy, or infection. This is particularly important in assessing elevated PSA levels in patients with resolving prostatitis or after surgery, where reevaluation of serum PSA levels is generally performed after 3 months.
The site and mechanisms of PSA clearance from the serum remain unknown. Animal and human evidence, however, suggests that PSA is most likely cleared through the liver. Although PSA is detectable in the urine of patients after radical prostatectomy, this is due to PSA production from the periurethral urethral glands in the bulbar and membranous urethra [40]. PSA does not appear to be excreted by the kidneys. Using the PSA-producing LNCaP tumor model, Gleave et al. [20] found that bilateral nephrectomy or ureteral ligation in mice bearing LNCaP tumors did not change PSA levels or PSA half-life (Figure 4a). Similarly, PSA half-life in an anephric man on dialysis who underwent cystoprostatectomy for clinically localized prostate cancer was 3 days, demonstrating that PSA clearance is not affected by end-stage renal failure (Fig. 4b). After hepatic injury with either carbon tetrachloride or common bile duct ligation, PSA half-life is slightly prolonged in mice bearing LNCaP tumors. However, a lethal hepatic injury is required before PSA clearance is affected, which suggests that hepatic dysfunction in most men will not delay PSA clearance rates. Furthermore, hypogonadism is common in men with advanced liver disease, which will decrease PSA production and actually lower serum PSA levels. In summary, PSA complexed with ACT is likely cleared by the liver, but PSA clearance rates do not appear to be an important determinant of circulating PSA levels under most conditions.

Figure 4 (A) Prostate-specific antigen clearance in the LNCaP tumor model follows first-order elimination kinetics similar to that seen in humans. PSA half-life is prolonged by common bile duct (CBD) ligation, but not by bilateral ureteral ligation, which suggests a hepatic route of elimination for PSA. (B) PSA half-life in an anephric man after radical prostatectomy is normal at 3 days, which illustrates that PSA clearance is not affected by renal insufficiency.

D. Biological Activity of PSA
Normally, PSA is secreted into seminal fluid in high concentrations, where it is enzymatically active and directly involved in liquefaction of the seminal coagulum. Lilja et al. [8] (1985) determined that PSA cleaves a high molecular weight seminal vesicle protein into several basic, low molecular weight proteins, which then results in liquefaction of the coagulum. Subsequently, PSA was shown to possess chymotrypsin-like and trypsin-like enzymatic activity [41].
The high degree of preservation of PSA production with tumor progression, coupled with its proteolytic activity, suggests the possibility that PSA is not just a passive marker of tumor cell activity but that it may play a functional role in progression itself, by either direct or indirect stimulation of cancer cell growth or invasion. The proteolytic activity of PSA reduces the affinity of insulinlike growth factor (IGF) for IGF binding protein-3 and increases the availability of mitogenic IGF [42]. Also, PSA itself may increase the growth of primary tissue cultures of prostatic epithelial cells. However, other in vitro experiments have failed to demonstrate a correlation between PSA mRNA expression and tumor cell growth rates [14]. In addition, the non-PSA-producing human prostate cancer cell lines PC-3 and DU-145 are more tumorigenic and invasive than the PSA-producing LNCaP cell line. The presence of such conflicting evidence makes it difficult to determine whether PSA has an active biological role in promoting the growth or progression of prostate cancer.

III. PROSTATE-SPECIFIC ANTIGEN AS A DIAGNOSTIC PREDICTOR OF PROSTATE CANCER
The role of PSA in the early detection of prostate cancer is currently the subject of much debate and research. Recent and accumulating data indicate that when used as a screening tool in the appropriate population, serum PSA is the single best test for the early detection of prostate cancer, and compares favorably with screening tests for breast and cervical cancers. Because PSA is objective, reproducible, and relatively inexpensive in itself, the debate over its role in screening primarily focuses on its cost-effectiveness and two related issues: (1) Is PSA testing (along with digital rectal examination (DRE)) specific and sensitive enough to detect organ-confined curable cancer and (2) will it detect an excessive number of clinically insignificant small-volume cancers and lead to overtreatment.
Because BPH tissue variably contributes to serum PSA concentrations and BPH has a high prevalence in men older than 50 years of age, careful interpretation of an elevated PSA level is required. On an individual basis, serum PSA by itself is not capable of distinguishing between patients with early organ-confined prostate cancer in an ocean of BPH. Although PSA lacks adequate specificity to be diagnostic of prostate cancer, it remains the single best test for stratifying men into groups with a high risk of having prostate cancer-who should undergo definitive testing with prostatic biopsy-and those with a low risk of having prostate cancer-who can be reassured and followed without additional testing [43-46]. Attempts to improve PSA sensitivity with lowering of the upper limit of normal leads to an increased number of false positives and reduction in specificity, which is most important for a screening tool. Approximately 25% of men with BPH may have elevated serum PSA levels, usually in the intermediate range between 4 and 10 µg/L. Although elevated serum PSA levels can occur in men with BPH, elevations occur more frequently and are higher in men with prostate cancer.

A. Detection of Cancer on Initial PSA Determination
Cross-sectional studies involving tens of thousands of men from several countries have been published and show very similar results regarding the ability of PSA to predict the presence or absence of cancer following transrectal ultrasound (TRUS)-guided biopsy of the prostate. Cooner et al. [44] reported on a group of 1807 men referred to a urologic practice (i.e., not primarily screened) who were evaluated with PSA, DRE, and TRUS. Biopsies were obtained only when DRE or TRUS results were abnormal, irrespective of serum PSA levels. The results of this historically significant study demonstrated that the positive predictive value of a PSA between 4 and 10 µg/L was 20% when DRE was normal and 45% when DRE was abnormal, which increased for PSA levels greater than 10 µg/L to 31% when DRE was normal and 77% when DRE was abnormal. The overall detection rate in this urological practice was 14.6%, which illustrates that the detection of prostate cancer can be increased by combining DRE, PSA, and TRUS.
Subsequent studies involving community-based populations similarly suggest that use of serum PSA increases the sensitivity, specificity, and positive predictive value of DRE in the diagnosis of prostate cancer. Andriole and Catalona [47] recently published the Washington University experience results from a screening population of 20,000 men. Overall, about 10% of screened men older than 50 years of age will have a PSA greater than 4 µg/L, and one-third of these were found to have cancer on subsequent biopsy, for a cancer detection of 3%. The probability of having cancer varied with the degree of PSA elevation. PSA levels were less than 4 µg/L in 92% of men, 4 to 10 µg/L in 6.5%, and greater than 10 µg/L in 2%. In patients with serum PSA levels between 4 and 10 µg/L, 22% had cancer, and in those with PSA levels greater than 10 µg/L, 67% had cancer. When PSA was the only abnormal parameter, cancer was diagnosed in 20%. However, when both PSA and DRE were abnormal, 31% of men had cancer; and when PSA, DRE, and TRUS were abnormal, 56% had cancer (Table 2).

Brawer et al. [46] reported similar results and concluded that use of serum PSA resulted in the best performance characteristic for the diagnosis of prostate cancer. They evaluated 1249 men and noted a positive predictive value of 31% for PSA and a 2.6% detection rate. Labrie et al. [48] randomly screened 1002 men between 45 and 80 years of age in Quebec City and, using a cutoff of 3.0 µg/L, reported the sensitivity and specificity for PSA to be 80.7% and 89.6%, respectively. Based on their findings, Labrie et al. calculated the positive predictive value of PSA to be 25% at levels above 3.0 µg/ L, 33% at levels above 4.0 µg/L, 51% at levels above 10 µg/L, and 90% at levels above 30 µg/L.
Each of these cross-sectional community-based studies have their minor flaws, yet despite significant variability in the populations studied and indications for biopsy, they are remarkably consistent and indicate that serum PSA levels above 4 µg/L are associated with a 30% to 40% chance of finding cancer on TRUS-directed biopsies (Table 3). The positive predictive value of a serum PSA above 10 µg/L ranges between 50% and 77%, and is influenced by DRE findings. Overall, 8% to 12% of screened men over the age of 50 will have an elevated serum PSA greater than 4 µg/L. The detection rate of prostate cancer using PSA in community-based populations is 2.2% to 3%, which is approximately twice that when DRE alone is used 145,46,49] and approximately two to three times the detection rate of breast cancer using mammography-based screening programs [50,51].

B. Detection of Cancer in Longitudinal Studies
The evaluation of PSA sensitivity and specificity from cross-sectional studies is subject to criticisms because the true negative disease status of each study participant can never be known. For example, what proportion of men with normal PSA and DRE, or with elevated PSA and negative biopsy, at initial evaluation will subsequently develop clinically significant disease in the future? Longitudinal studies involve measuring PSA levels at the "start" of follow-up in stored blood from a cohort of healthy men who later are or are not diagnosed with prostate cancer. The incident cases of prostate cancer allow a less biased estimation of test validity. Based on a retrospective, longitudinal study from the Baltimore Longitudinal Study of Aging including 54 men followed from 7 to 25 years, Carter et al. [52] demonstrated that consistent increases in PSA of more than 0.75 µg/L identified men with prostate cancer an average of 5 years before the usual diagnosis with a 72% sensitivity and 90% specificity. Differences in PSA velocity between patients with BPH and cancer were apparent as early as 9 years before diagnosis, and even longer in patients diagnosed with metastatic cancer.
A prospective longitudinal study evaluating the ability of PSA to predict for prostate cancer has recently been published. Gann et al. [3] prospectively evaluated PSA testing by measuring levels in stored blood at the start of follow-up in a cohort of 366 healthy men who were subsequently diagnosed with prostate cancer (usually without the aid of PSA) and 1098 health controls. They concluded that PSA had impressively high sensitivity and specificity. Their results, which are in general agreement with smaller prospective [53] and retrospective [52] longitudinal studies, noted that a single PSA level would have detected 80% of all aggressive cancers diagnosed within 5 years and about 50% of aggressive cancers as early as 9 to 10 years before diagnosis. Only 96 of 1098 men who remained free of prostate cancer diagnosis over a 10-year period had an elevated PSA, which means that false positive test remained very low over this period of time. The mean lead time provided by PSA testing is estimated to be 5.4 years. On the basis of their data, the authors conclude "that PSA has the highest validity of any circulating cancer screening marker discovered thus far."
PSA should not be used alone to exclude the possibility of prostate cancer. PSA is a useful adjunct to DRE in the early diagnosis of prostate cancer, and permits a rational guide to use of TRUS and biopsy. When used in conjunction with DRE, the positive predictive value of PSA is improved. Use of PSA and DRE detects 27% more cancers than would have been detected by PSA alone, and 34% more than by DRE alone [45,46]. For example, in a screening study of 6630 men using PSA and DRE, 18.2% of cancers detected were in men with normal PSA levels [54,55]. The positive predictive value of an abnormal DRE when PSA is normal is 10%; conversely, the positive predictive value of an abnormal PSA when DRE is normal is 20% to 30%.

C. Does PSA Predict for Clinically Significant Cancers?
The study reported by Gann et al. [3] (1995) suggests that PSA screening will not lead to an increase in the diagnosis of clinically insignificant tumors. PSA testing may be sensitive enough to detect clinically aggressive cancers whose natural history may be altered by early detection and therapy, but not sensitive enough to identify the highly prevalent small-volume indolent cancers. Use of PSA increases the proportion of cancers detected and treated at an organ-confined and potentially curable stage. More than 95% of the cancers detected in the University of Washington screening study were localized to the prostate, and about two-third have pathologically organ-confined disease [47].
However, it is possible that screening with serum PSA will increase the detection of clinically insignificant cancers and lead to overtreatment of some men whose quality of life or life expectancy would otherwise not have been affected by their prostate cancer. The clinical significance of a cancer is determined not only by the volume of the cancer and its grade, but also by the life expectancy of the patient, which determines duration of time he is at risk for progression. Evidence thus far suggests that less than 15% of PSA-detected cancers are smaller than 0.5 cc and probably clinically insignificant [56-58]. Pathologic evaluation of stage T1c cancers detected through elevations in serum PSA reveal that these nonpalpable tumors are similar to clinical stage T2 tumors in terms of their tumor volume, grade, and pathologic stage. However, as reviewed by Stamey [59], the risk of diagnosing and treating potentially clinically insignificant disease increases to 30% when sextant biopsies contain a single positive clinically insignificant disease increases to 30% when sextant biopsies contain a single positive core containing well or moderately differentiated cancer less than 3 mm in length. To reduce the risk of overtreatment, Stamey recommends repeat biopsies in an effort to validate the need for aggressive treatment.

D. Enhancing the Positive Predictive Value of PSA
Because of the substantial overlap in serum PSA between men with early prostate cancer and those with BPH, PSA lacks sufficient specificity to be considered an ideal screening test. Consequently, research efforts are focusing on the development of methods that improve the ability of PSA to predict for the presence of clinically important early prostate cancers (Table 4).

Table 4 Methods to Improve PSA Specificity in the Early Detection of Prostate Cancer

1. Serial PSA determinations for PSA velocity
2. Age-specific PSA reference ranges
3. Assays that measure complexed vs. free serum PSA
4. PSA density

Prostate-Specific Antigen Velocity
The rate of change of PSA over time, termed PSA velocity, is one such method. Based on the retrospective, longitudinal study by Carter et al. [52], a PSA velocity of greater than 0.75 µg/L is highly predictive of the presence of prostate cancer. The specificity for diagnosing prostate cancer increases to 90% when PSA velocity is greater than 0.75 µg/L in one year, compared to 60% for a cutoff of 4.0 µg/L for a single cross-sectional serum PSA determination. Differences in PSA velocity between BPH and prostate cancer patients were apparent as early as 9 years before the diagnosis of cancer was made. Catalona et al. [56] also suggested that a cutoff for PSA velocity of 0.8 µg/L per year helped distinguish between BPH and cancer patients. The mean velocity for men with cancer was 2.18 µg/L, compared to 0.48 µg/L for men without cancer. In their screening cohort of 693 men, Brawer et al. [55] suggested that a PSA velocity of greater than 20% per year was highly suggestive of cancer and warranted further investigation. The positive predictive value of men with PSA velocity higher than 20% per year was 17.1% in the second year [55] and 18.6% in the third year [60].
Serial PSA measurements over time in individual men permits longitudinal measurements of PSA, which provides a more accurate reflection of pathological changes within the prostate than does a single determination and will increase both the positive predictive value of PSA and the likelihood of diagnosing cancers while they are organ-confined. However, one problem that has become apparent with PSA velocity results from the individual variation in serum PSA levels from one determination to another. This variability can be as high as 55% to 70% [2]. The optimal utility of PSA velocity is not yet clearly defined and many questions regarding number of determinations, intervals between testing, and the influence of aging on interpretation of PSA velocity remain unknown. Because of the significant variation between measurements over time, it is likely that at least three annual PSA determinations will be necessary to establish an accurate PSA velocity for any given individual.

Prostate-Specific Antigen Density
Another method to improve PSA specificity is to correlate serum PSA with prostatic volume, using a parameter called PSA density, which is the quotient of serum PSA divided by prostate gland volume as determined by TRUS. PSA density is based on the concept that cancer, on a gram for gram basis, will increase serum PSA levels to a greater extent than will BPH or normal prostate tissue. Initial studies suggested that PSA density helped differentiate between BPH and early nonpalpable cancer [61,62]. Benson et al. [61] reported a mean PSA density of 0.04 in 41 patients with BPH compared to a PSA density of 0.58 in 20 cases of cancer, and they proposed a cutoff level for PSA density of 0.15. However, no biopsies were obtained from men with PSA densities below 0.15 in this preliminary study and therefore the true negative rate could not be calculated. Subsequent studies reported by Seamen et al. [63] suggested that PSA density is most helpful in determining which patients with mild elevations of PSA (4 to 10 µg/L) and normal DRE and TRUS should have a biopsy or just be followed with serial evaluations.
Although additional investigators also found PSA density helpful in distinguishing between BPH and cancer in men with intermediate elevations of PSA [64,65], others reported that PSA density offered no improvement over PSA alone in men undergoing prostatic biopsy [55,60,66]. Inconsistencies in the ability of PSA density to provide more information than PSA alone is due to difficulty in obtaining accurate and reproducible volume determinations of the prostate gland using TRUS, as well as the heterogeneous stromal and epithelial composition between prostate glands, which leads to marked variation in the amount of PSA produced per gram of prostate tissue. Use of PSA density as a strict guide to biopsy will miss some early cancers and may not be appropriate in younger men with intermediate elevations in PSA.

Age-Specific PSA Reference Ranges
Increases in serum PSA levels with increasing age is a well-documented observation. Oesterling et al. [67] reported that the correlation of serum PSA and age is significant, and suggested that the performance status of PSA would be improved by selecting different age-specific cutoff levels, rather than using a single arbitrary value. Based on this data, the upper limit of normal for serum PSA in each decade of life is 2.5 µg/L during the fifth decade, 3.5 µg/L during the sixth decade, 4.5 µg/L during the seventh decade, and 6.5 µg/L during the eight decade. The use of age-specific ranges as a guideline to biopsy will theoretically increase the detection rate of early cancers in younger men (i.e., cancers most likely to be a health threat over time) and decrease the biopsy rate and detection rate of early cancers in men over 70 years of age (i.e., cancers that may not be a health risk). Therefore, consideration of the age of the patient in interpretation of a given PSA level will help increase the sensitivity of PSA as a tumor marker in men under 60 years of age, and increase its specificity in men older than 60.

"Complexed" versus "Free" PSA
An area of considerable investigation is now focusing on measurement of bound and free PSA in the serum. Backleak of enzymatically active PSA into the extracellular fluids and serum is usually inactivated by binding to -1-antichymotrypsin or -2-macroglobulin [36,37]. PSA complexed with -2-macroglobulin is not accessible for immunodetection, whereas that bound to ACT has a sufficient number of antigenic epitopes exposed to interact with anti-PSA antibodies. The PSA-ACT complex is the major molecular form of PSA in serum; a small portion exists in a free, noncomplexed form. Lilja and his colleagues have developed assays capable of differentiating between free and PSA-ACT complexed forms to determine whether differences existed in the ratio of free to complexed PSA in BPH and cancer. These investigators suggest that measurement of the proportion of free PSA in serum may be a more accurate way of differentiating between men with BPH and those with cancer. Preliminary data suggest that BPH is more efficiently differentiated from cancer by the free-to-total serum PSA ratio, which is higher in men with BPH than in patients with prostate cancer [38,68,69]. The ability of free-to-total PSA ratios to distinguish BPH from cancer was examined in samples with PSA levels ranging from 1 to 2 µg/L from 47 BPH and 39 cancer subjects. By using an abnormally low free-to-total PSA ratio (<0.2) to detect cancer, specificity was increased by up to 25% compared to serum PSA levels alone. Although the molecular basis for the differences in ACT binding between cancer and BPH remain undefined, these observations suggest that use of free-to-total PSA ratios may increase the specificity and positive predictive value of PSA in the early detection of prostate cancer. Assays are being developed and more research is required, but this approach is offering some hope of improvement in the ability of PSA to predict for the presence of an early cancer.

IV. PROSTATE-SPECIFIC ANTIGEN AS A STAGING PREDICTOR OF PROSTATE CANCER
Use of PSA is changing the profile of clinical stage at diagnosis, with migration toward earlier stage disease. Prescreening in men over 50 years of age with PSA should reduce the number of men diagnosed with advanced stages of prostate cancer. Use of PSA increases the proportion of cancers detected and treated at an organ-confined and potentially curable stage. More than 95% of the cancers detected in the University of Washington screening study were localized to the prostate, and about two-thirds of patients have pathologically organ-confined disease [47]. Several studies have demonstrated that serum PSA correlates well with tumor volume and advancing clinical and pathological stage [31,45,46,70,71]. Do preoperative serum PSA levels help identify patients who are more likely to have organ-confined versus extraprostatic disease?

A. T-Stage
Although serum PSA correlates well with advancing T-stage and tumor volume (Fig. 5a, b), there is excessive overlap between stages to allow clinicians to use PSA preoperatively to accurately stage individual patients. However, preoperative PSA levels can stratify men into groups at low and high risk of having extraprostatic tumor extension. When preoperative PSA levels are below 4 µg/L, up to 80% of men have organ-confined tumors; approximately 60% of tumors are organ-confined when serum PSA is between 4 and 10 µg/L. A consistent finding in most series is that preoperative serum PSA levels above 10 µg/L predict for a higher risk of pathological upstaging. Over 50% of men with PSA levels above 10 µg/L have extraprostatic extension [39,45,46,70,71]. Despite its ability to stratify men into low and high risk groups, when used in isolation preoperative PSA levels cannot be relied on to distinguish between organ-confined and extracapsular disease. Oesterling et al. [39] reported a false-positive rate of over 65% for PSA cutoff levels of either 4 or 10 µg/L in predicting extracapsular disease.
Although PSA is not accurate enough as a staging tool when used alone, the staging accuracy of PSA can be improved by combining it with DRE findings and tumor grade on needle biopsy [71].

Figure 5 (a) Serum PSA levels at the time of diagnosis are roughly proportional to clinical stage, but significant overlap exists within any given stage. (b) Mean and median serum PSA levels increase gradually from stage A to stage Dl, and increase most dramatically in stage D2.

B. N-Stage
Although serum PSA levels tend to be higher in men with lymph node metastases, and most men with PSA levels above 50 µg/L have positive pelvic lymph nodes, serum PSA levels between 4 and 50 µg/L are not able to accurately predict for the presence of nodal metastases. However, a low serum PSA is a good negative predictor of the presence of pelvic lymph node disease. In patients with PSA levels below 10 µg/L and no high-grade disease on needle biopsy, the risk of finding positive nodal metastasis is negligible; it may be possible to avoid the need for pelvic lymphadenectomy in this group of patients [39,72].

C. M-Stage
Staging evaluation for newly diagnosed prostatic carcinoma has traditionally included DRE, serum PSA, and nuclear bone scan. Historically, one-third of patients presented with stage D2 disease, but with increasing emphasis on early detection fewer patients are presenting with osseous metastasis. Serum PSA levels are directly proportional to clinical stage and may identify patients with a low probability of having osseous metastases. Chybowski et al. [73] correlated tumor grade, DRE findings, and PSA with bone scan results of 521 patients and found PSA to be the overall best predictor of bone scan findings. Only 1 of 306 patients with a PSA level less than 20 µg/L had a positive bone scan. They calculated the negative value of a serum PSA less than 20 µg/L to be 99.7%. These investigators conclude that the routine staging bone scan is not necessary in patients with newly diagnosed, untreated prostate cancer who have a low serum PSA. Unfortunately, extrapolation from this important study is somewhat compromised because only seven men in their cohort had a positive bone scan.
To determine whether the expense of a bone scan can be safely avoided, we retrospectively reviewed 490 evaluable patients with prostate cancer at Vancouver Hospital since 1990, and identified only 5.7% (28/490) with a positive bone scan on initial evaluation [74]. Patients were stratified according to clinical stage, grade, and PSA levels to identify factors that may predict for presence or absence of osseous metastases. Although the risk of positive bone scans increased with increasing PSA levels, PSA is a poor positive predictor of positive bone scans. However, serum PSA levels below 10 µg/L are strong negative predictors of positive bone scans; 0 to 290 patients with PSA levels below 10 µg/L had positive bone scans. Four (4%) of 88 patients with PSA 10-20 µg/L, and 24 (21%) of 112 patients with PSA greater than 20 µg/L had positive bone scans at presentation (Fig. 6). Although the risk of positive bone scans increased with increasing stage and grade, tumor stage and grade were poor negative predictors of positive bone scans: 2.2% of patients with well-differentiated tumors, 3.7% with moderately differentiated, and 12.2% (all with PSA > 20) with poorly differentiated disease had positive bone scans. Our data suggest that 50% of newly diagnosed prostate cancer cases present with PSA levels below 10 µg/L; these patients can be spared the expense of a bone scan because none were positive. Based on these data, bone scans are indicated in patients with PSA levels above 10 µg/L, or with stage T3 or poorly differentiated tumors.

Figure 6 Pretreatment PSA levels below 10 µg/L are the best negative predictor of positive bone scans, with none of 290 patients having a positive scan. Because about 50% of newly diagnosed cases of prostate cancer present with PSA levels below 10 µg/L, the expense of bone scans can be avoided in this group because none is positive.

V. SERUM PSA AS A MARKER OF RESPONSE AFTER TREATMENT
PSA determination appears to be a reliable marker for disease response and progression with minimal random fluctuations between serial measurements in patients with localized or metastatic tumor [75-77]. This feature allows it to be a valuable marker of response to therapy and of the early detection of residual or recurrent carcinoma.

A. Prostate-Specific Antigen After Radical Prostatectomy
It seems intuitive that serum PSA should become "undetectable" following radical surgery because all prostatic tissue has been removed. Frazier et al [78] recently reported on the difference between clinical failure based on standard criteria (increased acid phosphatase, positive bone scan, or positive biopsy) and biochemical failure based on postoperative serum PSA levels. In 266 patients who had undergone radical perineal prostatectomy for clinically localized prostate cancer they observed a PSA failure rate ranging from 10% of the group with organ-confined cancer to 66% of the group with positive surgical margins at a median interval of less than 2 years postoperatively. It is noteworthy that they have 11 patients with increased postoperative PSA but no clinical evidence of failure at greater than 36 months of follow-up. In an expanded series of 895 men, Paulson [79] reported similar probabilities of biochemical and clinical recurrence rates (Table 5). Several other studies have also investigated the significance of a detectable serum PSA within the first year following surgery. Stein et al. [80] reported that in 50% of men who had detectable serum PSA levels within 1 year following surgery, either local recurrence or distant metastases subsequently developed. At 10 years the clinical evidence of freedom from disease was 72%; however, this number falls to 41% if a delectable PSA is considered as biochemical evidence of failure. Lange et al. [81] (1990) reported 100% progression for a group of patients with postoperative PSA levels exceeding 0.4 µg/L. They noted that needle biopsy of the anastomosis was positive in 42% of patients with PSA >0.4 µg/L and in none with a PSA <0.2 µg/L. A similar 45% positive biopsy rate was found by Foster et al. [82]. Goldenberg et al. [83] suggested that routine application of a transrectal ultrasound guided biopsy could increase this number significantly and decrease sampling errors. Earlier detection of locally recurrent cancer would lead to earlier adjuvant therapy with potentially increased control of both local recurrence and subsequent metastases.

aModified from Paulson [79]; probability of recurrence at 13 years, follow-up.
bPSA >0.4 ng/ml.

Is it possible to predict whether a detectable PSA level after surgery indicates a local recurrence or distant metastases? This may be relevant in selecting between potentially beneficial radiation therapy and androgen withdrawal. Partin et al. [84] followed 51 patients with isolated elevation of PSA until local recurrence or distant metastases developed. Using linear mixed effects regression analysis they were able to show that a combination of PSA velocity, pathologic stage, and Gleason score distinguished between the types of relapse. They have published a nomogram that may be useful in the clinical situation.
The use of ultrasensitive PSA assays has been studied as a means of detecting residual cancer [76,85]. Theoretically, at least, this may allow for the study of the effects of earlier adjuvant intervention. Stamey et al. [76], in a retrospective review of men with residual cancer, found that the ultrasensitive assay detected recurrence much earlier (median 202 days) than the standard assay. The optimal time between PSA measurements has not yet been established and awaits further study.

B. Prostate-Specific Antigen After Radiation Therapy
The utility of PSA in assessing the disease status and prognosis of men treated with radiation therapy is becoming better defined. However, unlike after radical prostatectomy, the significance of a detectable serum PSA level after radiation therapy is not as clear. Because the prostate remains in situ, serum PSA will not fall to undetectable levels as frequently as is seen after total resection. Zagars et al. [86-88] reporting the data from MD Anderson Cancer Center on patients treated either by brachytherapy or external beam therapy, suggested that greater than 95% of men show a significant decrease in serum PSA after radiation therapy and that approximately 80% of patients demonstrate a decrease to the normal range within 6 months following therapy. However, when a rising serum PSA is used as an intermediate end point for evaluation of treatment efficacy, it becomes apparent that total and permanent eradication of localized prostate cancer with radiation therapy is not achieved as often as previously believed. Few studies demonstrate significant continued decrease in post-therapy serum PSA levels beyond 12 months. Link et al. [89] reported that only 8% of men continued to show a decline in serum PSA after the first year of therapy and that only 25% had low PSA levels at 5 years after therapy. In a multivariate analysis of post-treatment values of 427 men treated at MD Anderson Cancer Center, Kavadi et al. [90] reported that PSA nadir after radiotherapy was an important independent predictor of subsequent relapse, with recurrence rates increasing as PSA nadir increased. In 40% of patients the PSA nadir decreased below 1 µg/L, and only patients with a PSA nadir lower than 1 µg/L did well, with a 5-year recurrence rate of 17%. If the PSA level was below 2.0 µg/L at 3 years, 95% of patients were free of disease [91]. The nadir serum PSA levels after radiation therapy appears to be one of the best post-treatment predictors of outcome, with low PSA nadir levels below 1 µg/L identifying a group of patients with lower risks of disease recurrence.
The prognostic significance of pretreatment serum PSA levels has recently been shown to be the single most significant pretreatment predictor of disease outcome after radiation therapy. Using a rising PSA as an intermediate end point of disease relapse, Zagars and von Eschenbach [88] identified four prognostic groups based on their pretreatment grade and PSA level (Fig. 7). For pretreatment PSA below 4 µg/L (any grade), patients did well, with a 3-year relapse rate below 10%. When serum PSA is between 4 and 10 µg/L, 3-year recurrence rates are 20% for well-differentiated tumors and 55% for moderate to poorly differentiated tumors. When pretreatment PSA levels are between 10 and 30 µg/L, recurrence rates are about 50% for well to moderately differentiated tumors but 90% for moderate to poorly differentiated disease. Finally, any serum PSA level above 30 µg/L, regardless of grade or local T-stage, predicted a 90% relapse rate at 3 years. Similarly Russell et al. [92] noted that patients with pretreatment PSA levels over 4.0 ng/mL had a 30% CR rate compared to a rate of 82% in men with normal pretreatment PSA. These pretreatment prognostic variables can therefore identify patients at high risk of disease recurrence who may benefit from multimodality therapy such as neoadjuvant androgen withdrawal therapy.

Pre-Treatment
PSA Level (µg/L) Figure 7 Pretreatment serum PSA levels help predict for treatment failure following radiation therapy for clinically localized prostate cancer. Patients with low PSA levels below 4 µg/L did well; those with pretreatment levels above 30 µg/L did poorly, regardless of grade. The presence of high-grade tumors increased risk of treatment failure for PSA values between 4 and 30 µg/L.

The rate and magnitude of change in serum PSA following radiation will vary depending on several known prognostic factors. Zagars and Pollack [87] followed 154 patients with at least four serial PSA determinations and concluded that rate of change of PSA did not predict outcome. Schellhammer et al. [93] described a high correlation between clinical stage and 10-year progression-free rates. However, only 10% of men in their series had undetectable PSA levels 10 years after therapy. Other factors that influence the PSA in irradiated patients include the volume of normal prostate, BPH and malignant tissue present, the degree of tumor differentiation, the dose and delivery of therapy, and the effect of radiation on testicular androgen production.
Serum PSA levels may increase after achieving a normal nadir. Sources of rising PSA after radiation include residual radio-resistant tumor within the prostate gland, de novo growth of new tumor within the prostate, BPH or normal prostate remnants, metastatic disease at the time of radiation, or a combination of these factors. Regardless of the source of the detectable PSA after radiation therapy, increases above baseline portend disease recurrence. Stamey et al. [94] reports persistingly low PSA values in only 20% in patients with stages A to Dl prostate cancer. Kaplan et al. [95] found that 38% of men had increasing serum PSA levels 3 years after radiation therapy and 68% of these had subsequent clinical evidence of relapse or progression. Crook et al. [96] has evaluated the ability of DRE, PSA, and TRUS to identify patients who are at high risk of having positive biopsy results after radiation therapy. They found that serum PSA was the best predictor of the presence of a biologically active tumor. Only patients with a rising serum PSA had a persistently positive biopsy; those with a normal (<4 µg/L) or falling PSA had either stable clinical disease, negative biopsies, or conversion of a positive biopsy to a negative one on subsequent follow-up biopsies.
In summary, studies regarding PSA and radiation therapy reviewed above have identified pretreatment serum PSA levels and post-treatment PSA nadir levels as very powerful prognostic tools that help stratify patients at high and low risk for disease recurrence, and can help guide the clinician in applying multimodality therapy to those patients less likely to do well using radiation therapy alone.

C. Prostate-Specific Antigen After Androgen Withdrawal Therapy
Medical or surgical castration will result in the down-regulation of PSA gene expression, resulting in rapid and dramatic lowering of serum levels. In 1981, Kuriyama et al. [97] were the first to measure serum PSA levels in men with advanced prostate cancer. Stamey et al. [98] noted that virtually all men had a significant initial response to therapy, with serum PSA levels decreasing rapidly within the first 6 months. In this series 31% of the patients had a normalization of PSA, including 9% into the undetectable range.
Serum PSA levels are also very useful in reassuring men that they have stable disease when their PSA levels remain stable while on androgen suppression therapy, and conversely, in predicting eventual disease progression to androgen independence when PSA levels begin to rise. In more than 90% of patients whose disease progresses, serum PSA levels increase before evidence of tumor progression becomes apparent, with a mean lead time between PSA elevation and tumor recurrence of 6 to 12 months [23,99,100].
Serum PSA nadir following androgen therapy can stratify patients into good and poor prognostic groups. Bruchovsky et al. [100] observed that at least 32 weeks of treatment are necessary in order to bring the serum PSA into the normal range in about 70% of men with advanced prostate cancer. In the remaining 30%, serum PSA will decrease temporarily and then increase; if a plateau is reached it will be short-lived or stabilize outside the normal range. This is usually a sign of early progression to androgen independence and is associated with a poor prognosis. In this series, if the serum PSA remains above 4 µg/L between 24 and 32 weeks of treatment, the median survival time is only 18 months (Fig. 8). On the other hand, if the serum PSA is below 4 µg/L between 24 and 32 weeks of therapy, the median survival time is more than twice as long at 40 months. It has been observed that the rate of normalization is faster between time zero and 10 weeks than in the succeeding interval between 10 and 32 weeks of initial treatment. From this point onward, the serum PSA result will remain abnormal in about 30% of patients (as described above), with little chance of treatment producing a normal result beyond 32 weeks. Similarly, Miller et al. [23] have noted longer survival times in patients with a nadir in serum PSA into the normal range (less than 4.0 µg/L) compared with those who did not achieve such a nadir (42 months vs. 10 months)

Figure 8 In approximately 70% of patients treated with androgen withdrawal therapy, serum PSA decreases into the "normal" range below 4 µg/L. Survival is roughly [text missing in original paper]

The LNCaP tumor model has also been used to determine if changes in serum PSA after castration can predict duration of time to androgen-independent progression [29]. Although time to progression is variable (range 14 to >100 days, mean = 28 days), androgen-independent progression invariably occurs. Precastrate PSA level, precastrate tumor volume, or earlier versus later castration did not predict time to androgen-independent progression. PSA nadir did not predict time to progression, because low precastrate levels reach low nadirs but time to progression was not prolonged. The PSA nadir fraction, defined as the percent decrease in PSA after castration, was the only factor identified that directly correlated with time to progression (Fig. 9). Time to androgen independence ranged from 14 days with a PSA nadir fraction of 1.2 to 90 days with a PSA nadir fraction of 19.8. LNCaP tumor volume stabilizes but does not decrease after castration; hence, higher PSA nadir fractions may reflect more diffuse decreases in PSA production in more intrinsically androgen-sensitive tumors. These observations suggest that PSA nadir fraction can be useful in identifying good and poor responders to androgen ablation therapy.

Figure 9 The PSA nadir fraction, defined as the percent decrease in PSA after castration, is the best predictor of time to androgen independence in the LNCaP tumor model.

Stamey et al. [98] reported that within 2 years after androgen deprivation therapy, only 9% of men still had undetectable serum PSA levels; 22% had serum PSA levels within the reference range; and 72% of those who had good initial response within the first 6 months experienced increases in serum PSA within the second 6 months after therapy. They also observed that a decline to below normal within the first 6 months after therapy predicted a favorable response and increased survival.
Thus it is clinically useful to monitor serial PSA levels in patients treated with androgen ablation. Androgen suppression is the only effective systemic therapy for metastatic prostate cancer, where survival is inversely proportional to serum PSA levels prior to treatment. The pattern of response can distinguish patients with favorable from those with unfavorable outcomes. Progression-free survival is longer if serum PSA decreases to a stable nadir in the normal range, identifying patients who may be candidates for intermittent androgen suppression [101]; the others may be eligible for chemotherapy or various experimental protocols before their overall performance status declines

Facts and Statistics- Prostate Cancer
Prostate cancer is diagnosed every 2 ½ minutes, approximately 200,000 new cases each year. It is the most commonly diagnosed cancer in America among men.Nearly 40,000 American men lose their lives to prostate cancer each year, one death every fifteen minutes.Prostate cancer incidence rates increased 192% between 1973 and 1992.One in six American men is at lifetime risk of prostate cancer. If a close relative has prostate cancer, a man's risk of the disease more than doubles. With two relatives, his risk increases fivefold. With three close relatives, his risk is about 97%.In 1999, more than 1.2 million new cases of cancer were diagnosed in the United States, and more than 560,000 lives were lost. That's more lives lost than by the U.S. military on all battlefields this century.African American men have the highest prostate cancer incidence and mortality rates in the world. The incidence rate is about 35% - 50% higher than - and mortality rate double - that of Caucasian males, who have the second highest rate.In the next 24 hours, prostate cancer will claim the lives of over 100 American men.Prostate cancer represents 29% of all new cancer cases in American men.This year, more cases of prostate cancer in men under the age of 65 are expected than the combined number of men of ages who are victims of leukemia, Hodgkin's disease, melanoma, and brain tumors.

Facts About Prostate Cancer Research & Funding
Prostate cancer accounts for approximately 15% of all cancer cases in the United States and 15% of male cancer deaths. Yet, on average, only about 5% of federal cancer research dollars have been devoted to beat the disease.AIDS research receives approximately $1.7 billion in federal dollars. Breast cancer research will receive nearly $700 million next year. Compare that to $335 million for prostate cancer research.The United States invests approximately $4,000 to find a cure for each life lost to prostate cancer; more than $14,000 for each life lost to breast cancer, and about $100,000 for each life lost to AIDS. It's not that research for other diseases receives too much funding. Prostate cancer receives too little.Overall, the total cost of treating prostate cancer in the U.S. amounts to several billion dollars per year. Since most men diagnosed with the disease are over 65 years of age, most of the cost is paid for through Medicare.Since its inception in 1996, the National Prostate Cancer Coalition (NPCC) has, through its activist participation, infused more than $500 million in new federal dollars into prostate cancer research laboratories and clinics around the country.

Facts on the High Risk of Prostate Cancer for African-American Men
African American men have the highest prostate cancer incidence and mortality rates in the world. The incidence rate is about 35% - 50% higher than - and mortality rate double - that of Caucasian males, who have the second highest rate.African-American men have the highest risk of developing prostate cancer and are twice as likely to die from it as other men with the cancer.During this year alone, 18,500 African-American men will be diagnosed with prostate cancer.6,100 African-American men will die from prostate cancer this year.One of every four deaths in this country is from cancer.Ten percent of the world's cancer cases are recorded in the United States. However, ninety percent of the world's investment in cancer research originates here.Prostate cancer death and occurrence rates among African-Americans are higher than other racial or ethnic populations in the United States.In 1999, the most commonly diagnosed cancer in African-American men was prostate cancer (29%).Prostate cancer is the second-leading cause of cancer death among African-American men.Although prostate cancer incidence rates are high in whites, the rate for African-Americans is even higher -- 50% higher than the incidence in white men.

General Cancer Facts
One tenth of one percent of the U.S. budget is invested in cancer research.The cost of cancer to the economy is more than $100 billion annually. Within a decade, it is likely to exceed $200 billion. We invest just over $3 billion annually to cure cancer.If cancer were cured today, the economic value to the United States would exceed $46 trillion, more than the entire financial assets of the country.When adjusted for inflation, federal funding for cancer research has increased only slightly better than one percent during the past decade. Funding for all medical research has increased 15%.American men have a one in two lifetime risk of developing cancer. For women, the risk is one in three.Cancer risk, including prostate cancer risk, increases after the age of 50. Baby boomers turn 50 at the rate of one every seven seconds.In 1971, President Richard Nixon declared America's war on cancer, promising to end its toll within a decade. Each subsequent administration has reaffirmed this commitment, yet the number of cancer cases and deaths continue to grow.In 1938, President Franklin Roosevelt declared war on polio, and within two decades, polio had been successfully controlled in this country. In 2000, the World Health Organization anticipates the complete eradication of polio on the planet.At its worst, polio claimed about 17,000 American lives and caused about 52,000 new cases in the United States in a single year. Cancer takes 33 times as many lives and causes 23 times as many cases in a single year.Cancer will surpass heart disease as the number one killer of Americans shortly after 2000.

CAN COMPLEXED PROSTATE SPECIFIC ANTIGEN AND PROSTATIC VOLUME ENHANCE PROSTATE CANCER DETECTION IN MEN WITH TOTAL PROSTATE SPECIFIC ANTIGEN BETWEEN 2.5 AND 4.0 NG./ML.

KOJI OKIHARA; HERBERT A. FRITSCHE~; ALBERTO AYALA; DENNIS A. JOHNSTON; W. JEFFREY ALLARD; R. JOSEPH BABAIAN*±
From the Departments of Urology, Research Laboratory Medicine and Biomathematics, University of Texas M. D. Anderson Cancer Center, Houston, Texas, and Bayer Corporation, Business Group Diagnostics, Tarrytown, New York
THE JOURNAL OF UROLOGY 2001;165:1930-1936

ABSTRACT
Purpose: We assessed whether complexed prostate specific antigen (PSA) and complexed PSA referenced variables would enhance prostate cancer detection in men with serum total PSA between 2.5 and 4.0 ng./ml.
Materials and Methods: Transition zone and total prostate gland volumes were determined in 151 men who underwent prostate biopsy using an 11 core biopsy strategy. In addition to measuring the Bayer§ complexed PSA assay, we also calculated 2 computed complexed PSA values (Hybritech|| total PSA - Hybritech free PSA and Bayer total PSA - Hybritech free PSA). We calculated 8 volume referenced variables using total and complexed PSA, and 2 computed complexed PSA values by dividing each value by the total prostate and transition zone volumes.
Results: Of the 151 patients 37 (24.5%) had cancer. In 10 of the 37 men with cancer (27%) a positive core was present in only 1 or more of the 5 alternate regions not sampled by conventional sextant biopsies. At 92% sensitivity a cutoff value of 2.3 ng./ml. for complexed and 31% for free-to-total PSA provided 42% and 11% specificity, respectively (p <0.001). In the 116 men with a total prostate volume of 30 cc or greater at 92% sensitivity the specificity of complexed PSA density (55%) and complexed PSA adjusted for transition zone volume (52%) were better than that of complexed (40%) and free-to-total (11%) PSA. In the 35 men with a total prostate volume of less than 30 cc at 92% sensitivity the specificity of complexed PSA (50%), complexed PSA density (55%) and complexed PSA adjusted for transition zone volume (55%) were significantly better than that of free-to-total PSA (8%, p <0.001). The area under the curve of complexed PSA was almost identical to that of the 2 computed complexed PSA calculations.
Conclusions: A substantial proportion of men with total PSA values between 2.5 and 4.0 ng./ml. had prostate cancer. Complexed and computed complexed PSA were more specific than the free-to-total PSA ratio when total PSA was between 2.5 and 4.0 ng./ml. A 2.3 ng./ml. threshold for complexed and computed complexed PSA appears to stratify prostate biopsy results in men with total PSA between 2.5 and 4.0 ng./ml. The computed complexed PSA calculation appears to be equivalent to the complexed PSA serum assay for detecting cancer. Volume referenced complexed PSA performed better than complexed PSA in men with a total prostate volume of 30 cc or greater compared to men with a total prostate volume of less than 30 cc.
Key Words: prostate; prostatic neoplasms; prostate-specific antigens; biopsy

J Urol 2001 June;165(6):1930-1936
Copyright © 2001 American Urological Association, Inc. ®. All rights reserved
Published by Lippincott Williams & Wilkins

RE: COMPLEXED PROSTATE SPECIFIC ANTIGEN PROVIDES SIGNIFICANT ENHANCEMENT OF SPECIFICITY COMPARED WITH TOTAL PROSTATE SPECIFIC ANTIGEN FOR DETECTING PROSTATE CANCER

M. K. Brawer, C. D. Cheli, I. E. Neaman, J. Goldblatt, C. Smith, M. K. Schwartz, D. J. Bruzek, D. L. Morris, L. J. Sokoll, D. W. Chan, K. K. Yeung, A. W. Partin and W. J. Allard
J Urol, 163: 1476-1480, 2000
To the Editor. We agree that a single test would reduce the costs of free and total prostate specific antigen (PSA), and avoid possible quotient bias. Due to similar improvement in specificity for the total PSA range of 4 to 10 ng./ml. the authors suggested that complexed PSA may be an alternative to percent free PSA. However, there are important limitations to using complexed PSA alone compared to the conventional determination of free and total PSA and the calculation of percent free PSA. Contrary to the results of the authors we could find neither a substantial advantage of complexed compared to total PSA nor comparable results of complexed PSA alone to percent free PSA.1 These findings were also confirmed by others.2,3
Complexed PSA strongly correlates with total PSA and can only provide better specificity than total PSA within the small total PSA range of 4 to 5 ng./ml. or 4 to 6 ng./ml. using the 3.75 ng./ml. complexed PSA cutoff. Assuming an average complexed PSA of approximately 79% for patients with benign prostatic hyperplasia (BPH), only those with total PSA less than 5 ng./ml. would have a complexed PSA of less than 3.75 ng./ml.1 Assuming a 90% average of complexed PSA in patients with prostate cancer, all of those with total PSA greater than 4 ng./ml. will be identified correctly, which means that with only between 4 and 5 ng./ml. total PSA the complexed PSA alone can enhance specificity compared to total PSA. The authors conceded that their population had no normal distribution of PSA values. In their cohort with an overrepresentation of patients with BPH in the 4 to 5 ng./ml. total PSA range this lack of distribution could lead to specificity improvement of complexed PSA for the entire 4 to 10 ng./ml. total PSA range. Already 30% of all patients (71 of 237) with no evidence of malignancy and total PSA between 4 and 10 ng./ml. had total PSA between 4 and 5.06 ng./ml. However, only 3 of 56 patients with total PSA at 5.06 to 6 ng./ml. had a complexed PSA less than 3.75 ng./ml. and were subsequently identified correctly. This result indicates the limitation of complexed PSA for patients with BPH and total PSA greater than 5 ng./ml., and inability to distinguish it from prostate cancer.
Moreover, patients with prostate cancer and total PSA less than 4 ng./ml. also could not be correctly identified with a single complexed PSA determination because they always had complexed PSA less than 3.75 ng./ml. This restriction seems to be most important, since about 20% of all cases with prostate cancer are expected to be in this total PSA range.4 The current results of the European Randomized Study of Screening for Prostate Cancer indicated that nearly 50% of all detectable cancers could be diagnosed with PSA between 0 and 4 ng./ml.5
We conclude that with the widespread use of the total PSA range of 4 to 10 ng./ml. and especially for the 2 to 4 ng./ml. total PSA area percent free PSA cannot be replaced by complexed PSA alone, which may only be helpful in a small total PSA range between 4 and 5 ng./ml. Whether the complexed-to-total PSA ratio may be an alternative to the free-to-total PSA ratio remains to be studied further.1,2,6

1. Respectfully,

*Bayer Diagnostics, Tarrytown, New York.
Reply by Authors. Stephan et al stated, "Contrary to the results of the authors we could find neither a substantial advantage of complexed PSA compared to total PSA nor comparable results of complexed PSA alone to percent free PSA." They support this statement by citing their published work (reference 1 in letter), and that of Filella et al (reference 2 in letter) and Wang et al (reference 3 in letter). While the data of Jung et al did not support the use of complexed PSA as a single test, their data cannot be compared to other published results on the clinical usefulness of complexed PSA for a number of reasons. In addition, we believe that the authors misinterpreted our results as well as those of others concerning the clinical usefulness of the complexed PSA method.
The study design used by Jung et al is significantly different from others on complexed PSA in a number of respects. All previously published studies on complexed PSA used sera from men with a followup prostate biopsy. In contrast, the BPH population in the study of Jung et al included men with biopsy confirmed BPH and those with clinical symptoms of BPH without prostate biopsy. The population with clinical BPH is known to have occult prostate cancer, which may have been detected on biopsy. The population of men in the study of Jung et al also included 1.6-fold more men with prostate cancer (144) than with BPH (89). This population is not representative of a typical screening population or a urology referral population. Previously published studies of complexed PSA used sera from populations of men with the expected proportion of prostate cancer of 20% to 30%.1-4
To obtain a population of men more representative of a typical male population Jung et al hand selected a small subset of patients (80) with equal numbers of those with cancer and BPH, and with equivalent concentrations of total PSA in the range of 2 to 10 ng./ml. Again, this population is not representative of screening or referral populations. There is no reason to expect that such populations should include men with cancer whose PSA concentrations are equal to those of men with BPH. This result would obviate the value of the PSA test. Additionally, the population of men studied by Jung et al demonstrated a high proportion with PSA concentrations in the low total PSA range. Lastly, the sensitivity reported in the study of Jung et al for total PSA using a cutoff of 4.0 ng./ml. was 45%, which is a particularly low sensitivity and again reflects the differences in the population selected for this study. For all of the aforementioned reasons the population of men studied by Jung et al is not representative of a screening or urology referral population. Therefore, results from this cohort cannot be extrapolated to other studies using more representative populations of men.
The report of Jung et al actually indicated that complexed PSA enhanced specificity at almost 10% compared with total PSA at 90% sensitivity. They further demonstrated that the area under the curve for complexed PSA was 0.632 compared to 0.568 for total PSA. Filella et al confirmed these findings and concluded that the area under the curve was greater for complexed than total PSA, and complexed PSA gave enhanced specificity compared to total PSA in the truncated total PSA range of 2 to 20 ng./ml. Others have demonstrated superior diagnostic performance of complexed PSA compared to total PSA.1-6
Finally, it is inappropriate for Stephan et al to compare our results to those using the assay of Wang et al due to the reported excessive cross-reactivity of their antibodies with cathepsin-G complexes.7 Additionally, Wang et al used an assay that specifically measures PSA complexed -1-antichymotrypsin, which is different from the Bayer* assay that measures all immunodetectable complexes of PSA, including minor forms of complexed PSA,8 and not just PSA bound to area under the curve. Studies have indicated that several PSA- 1-antichymotrypsin methods may not be accurate due to technical problems in the development of these assays,9-11 which is in contrast to that reported for complexed PSA.1,6 Therefore, it may be inaccurate to state that because a PSA- 1-antichymotrypsin method does not have enhanced diagnostic performance that the same will be true for the Bayer complexed PSA method or for other PSA- 1-antichymotrypsin methods.
Considered together, all studies reported to date have indicated that complexed PSA enhances diagnostic performance over total PSA measurement in the detection of prostate cancer. Some have demonstrated equivalent or greater enhancement compared to the use of the free-to-total PSA ratio. We believe that the different results may be related to differences in patient populations tested and/or volume of the prostate gland, and a large multicenter prospective trial is currently in progress to address these issues. In conclusion, we suggest that complexed PSA be considered as a first line test to replace total PSA in the detection of prostate cancer.

J Urol 2000 November;164(5):1666-1672
Copyright © 2000 American Urological Association, Inc. ®. All rights reserved
Published by Lippincott Williams & Wilkins

COMPLEXED PROSTATE SPECIFIC ANTIGEN PROVIDES SIGNIFICANT ENHANCEMENT OF SPECIFICITY COMPARED WITH TOTAL PROSTATE SPECIFIC ANTIGEN FOR DETECTING PROSTATE CANCER

MICHAEL K. BRAWER; CAROL D. CHELI; IRENE E. NEAMAN; JOAN GOLDBLATT; CAROL SMITH; MORTON K. SCHWARTZ; DEBRA J. BRUZEK; DEBORAH L. MORRIS; LORI J. SOKOLL; DANIEL W. CHAN; KWOK K. YEUNG; ALAN W. PARTIN; W. JEFFREY ALLARD
From the Northwest Prostate Institute, Seattle, Washington, Bayer Diagnostics, Tarrytown and Memorial Sloan-Kettering Cancer Center, New York, New York, and Johns Hopkins Medical Institutions, Baltimore, Maryland
THE JOURNAL OF UROLOGY 2000;163:1476-1480

ABSTRACT
Purpose: Determining serum total prostate specific antigen (PSA) has proved to be a valuable diagnostic aid for detecting prostatic carcinoma, although the lack of specificity has limited its usefulness. Studies indicate that the use of percent free PSA would improve specificity while maintaining sensitivity. Since complexed PSA represents the major proportion of measurable PSA in serum, we determined whether it represents a single test alternative to the use of percent free PSA for the early detection of prostate cancer.
Materials and Methods: Archival serum was obtained from 385 men with no evidence of malignancy on biopsy and 272 with biopsy confirmed prostate cancer. We determined the concentration and proportion of total, complexed and free PSA.
Results: Receiver operating characteristics analysis using total PSA results from all samples (range 0.32 to 117 ng./ml.) indicated that the areas under the curve for complexed PSA alone as well as the free-to-total and complexed-to-total PSA ratios were similar and significantly greater than those for total PSA alone. Within the range of 85% to 95% sensitivity receiver operating characteristics analysis revealed that the specificity of complexed PSA was higher than that of total PSA and equivalent to that of the free-to-total PSA ratio. We noted a similar improvement in specificity in the 4 to 10 ng./ml. total PSA range. Using published cutoff values for complexed, total and percent free PSA when total PSA was in the 4 to 10 ng./ml. range the sensitivity and specificity of complexed and percent free PSA were similar. Within the 4 to 10 ng./ml. total PSA range the population of patients with no evidence of malignancy and complexed PSA below the upper limit was different with respect to total PSA from that with no evidence of malignancy and free PSA greater than 25%.
Conclusions: The measurement of complexed PSA represents an alternative to the use of percent free PSA, although the patient populations identified by the 2 tests are different.
Key Words: prostate; prostate-specific antigen; prostatic neoplasms; ROC curve

J Urol 2000 May;163(5):1476-1480
Copyright © 2000 American Urological Association, Inc. ®. All rights reserved
Published by Lippincott Williams & Wilkins

A PROSPECTIVE STUDY TO EVALUATE THE ROLE OF COMPLEXED PROSTATE SPECIFIC ANTIGEN AND FREE/TOTAL PROSTATE SPECIFIC ANTIGEN RATIO FOR THE DIAGNOSIS OF PROSTATE CANCER

I. D. C. MITCHELL; B. L. CROAL; A. DICKIE; N. P. COHEN; I. ROSS
From the Departments of Urology and Biochemistry, Grampian University Hospitals Trust and Health Services Research Unit, University of Aberdeen, Aberdeen, United Kingdom
THE JOURNAL OF UROLOGY 2001;165:1549-1553

ABSTRACT
Purpose: Patients are increasingly undergoing prostatic biopsy to identify localized prostate cancer. The decision to perform a biopsy is often made on the basis of total prostate specific antigen (PSA). However, this value lacks adequate specificity for this task. We evaluate the role that a number of these tests, including the Bayer complexed PSA (Bayer Diagnostics, Tarrytown, New York) and free/total PSA ratio, may have in our clinical practice.
Materials and Methods: A total of 160 consecutive patients attending a prostate assessment clinic were enrolled during an 18-month period in our study. All patients had a previously recorded total PSA (range 2.6 to 20.0 ng./ml.). Before transrectal ultrasound biopsy of the prostate gland, a blood sample was taken with patient consent. The findings on ultrasound were then recorded, including prostate volume. Serum samples were immediately sent for subsequent storage and analysis.
Results: Of the patients enrolled 109 had benign histology while 51 had prostatic carcinoma. The 2 patient groups were well matched for age. In our series patients with prostate cancer had significantly smaller prostates and higher mean total PSA. At a high sensitivity, such as 95%, it appeared that Bayer complexed PSA performed better than the other tests and ratios, with an estimated specificity of 24.8% compared with 17.4% for Bayer total PSA and 15.6% for Abbott free/total PSA (Abbott Laboratories, Abbott Park, Illinois). Receiver operator characteristics curves were drawn, and when the areas under them were calculated, we demonstrated that the area under the curve for Bayer complexed PSA (0.706) was between the values for total PSA (0.671) and free/total PSA ratio (0.731). However, the only statistically significant improvement in performance was in Bayer complexed PSA over the total PSA assays.
Conclusions: Our study revealed that the overall diagnostic performance of Bayer complexed PSA appears to be better than the other PSA tests and ratios studied. The use of Bayer complexed PSA may lead to a reduction in the number of men undergoing unnecessary prostatic biopsy.
Key Words: prostate; prostate-specific antigen; prostatic neoplasms

J Urol 2001 May;165(5):1549-1553
Copyright © 2001 American Urological Association, Inc. ®. All rights reserved
Published by Lippincott Williams & Wilkins

SERIAL PROSTATE SPECIFIC ANTIGEN, FREE-TO-TOTAL PROSTATE SPECIFIC ANTIGEN RATIO AND COMPLEXED PROSTATE SPECIFIC ANTIGEN FOR THE DIAGNOSIS OF PROSTATE CANCER

WILLIAM J. ELLIS; RUTH ETZIONI; ROBERT L. VESSELLA; CHENGCHENG HU; GARY E. GOODMAN*
From the Division of Public Health Services, Fred Hutchinson Cancer Research Center and Department of Urology, University of Washington, Seattle, Washington
THE JOURNAL OF UROLOGY 2001;166:93-99

ABSTRACT
Purpose: The free-to-total prostate specific antigen (PSA) ratio and complexed PSA have been introduced as adjuncts to total PSA for prostate cancer screening. Little data exist on the use of these tests for serial PSA screening. We compared serial total PSA, the free-to-total PSA ratio and calculated complexed PSA in men diagnosed with prostate cancer and matched controls in a population based study.
Materials and Methods: We identified 90 men diagnosed with prostate cancer between 1988 and 1996 with at least 3 serial serum samples obtained at 2-year intervals who were participants in the -Carotene and Retinol Efficacy Trial for the prevention of lung cancer. Samples were available up to 10 years before diagnosis. A total of 90 age matched men from the same cohort without prostate carcinoma were identified as controls. Free and total PSA was measured by the Abbott AxSYM~ system.
Results: Baseline demographics of cases and controls were similar. At baseline and diagnosis the men with prostate cancer had higher total and complexed PSA, and a lower free-to-total PSA ratio than controls. Mean followup was 5.2 years in cases and 5.5 in controls. The yearly change in PSA parameters in cases versus controls was 20.7% versus 3.5% for total, -3.4% versus 0.2% for free-to-total and 21.5% versus 3.4% for complexed PSA (p <0.0001). At diagnosis PSA alone was estimated to perform with more than 90% specificity in our model.
Conclusions: In this population based study total PSA was superior to the free-to-total PSA ratio for predicting the development of prostate cancer. While serial changes in free-to-total PSA ratios with time were statistically significantly different in men diagnosed with prostate cancer and controls, the magnitude of these serial changes were slight enough to render them clinically insignificant.
Key Words prostate; prostatic neoplasms; prostate-specific antigen; mass screening; diagnosis

J Urol 2001 July;166(1):93-99
Copyright © 2001 American Urological Association, Inc. ®. All rights reserved
Published by Lippincott Williams & Wilkins