Prostate-specific antigen: A review of the validation of the most commonly used cancer biomarker

Authors

  • Javier Hernández M.D.,

    1. Division of Urology, The University of Texas Health Science Center at San Antonio, San Antonio, Texas
    Search for more papers by this author
  • Ian M. Thompson M.D.

    Corresponding author
    1. Division of Urology, The University of Texas Health Science Center at San Antonio, San Antonio, Texas
    • Division of Urology, University of Texas Health Science Center at San Antonio, 7703 Floyd Curl Drive, San Antonio, TX 78229
    Search for more papers by this author
    • Fax (210) 567-6868


Abstract

BACKGROUND

The widespread use of prostate-specific antigen (PSA) screening has had a tremendous impact on all aspects of the management of prostate carcinoma. Although PSA-based screening has resulted in a stage migration to more organ-confined tumors at the time of diagnosis, and has been temporally associated with a decrease in prostate carcinoma mortality, PSA screening is imperfect. A recent analysis of results from the Prostate Cancer Prevention Trial (PCPT) has provided insight into the positive predictive value of PSA in the so-called “normal” range.

METHODS

The history of the discovery, initial studies, and subsequent widespread application of PSA screening is reviewed.

RESULTS

The application of PSA for screening preceded the development of current prostate biopsy techniques and an upper limit of normal was established without complete disease ascertainment. More recent modifications of PSA-based screening have been adopted clinically without sufficient validation. With current methods, overdiagnosis of clinically unimportant disease almost certainly occurs and high-grade, aggressive disease may not be detected sufficiently early to allow successful treatment.

CONCLUSIONS

To the authors' knowledge, the optimal upper limit of normal for PSA for prostate carcinoma screening is unknown. New biomarkers of disease are needed; these must be linked with disease prognosis and must be validated in rigorously designed clinical trials. Cancer 2004. © 2004 American Cancer Society.

Measurement of serum prostate-specific antigen (PSA) has become the most common event leading to the diagnosis of prostate carcinoma and may be the most commonly used cancer clinical test. The introduction of routine PSA-based screening over the past 20 years has led to a dramatic increase in the rate of disease detection and a subsequent stage shift at the time of diagnosis. For the past decade or more, a PSA value of 4.0 ng/mL has been considered to be the upper limit of normal (ULN). Challenging this long-held notion was a recent report from the Prostate Cancer Prevention Trial (PCPT).1 In an analysis of 2950 PCPT participants randomized to the study's placebo group and who never had a PSA level > 4.0 ng/mL or an abnormal digital rectal examination (DRE), we found that PSA levels of 0–4.0 ng/mL were associated with a positive predictive value of between 6.6–26.9%. Overall, 14.9% of men with prostate carcinoma were found to have high-grade disease, with this rate reaching 25% among men with a PSA level of 3.1–4.0 ng/mL.

In this review, we examined the history of PSA, from its initial description to its widespread current use. We examined the design of initial screening studies as well as strategies used to enhance its performance. Finally, we examined future opportunities for more reliable screening methods that incorporate disease prognosis in an entity known to have highly variable outcomes.

Discovery, Isolation, and Physiologic Function of PSA

To our knowledge, the earliest investigations of tissue-specific antigens in the human prostate were conducted by Ablin et al. in 1970.2, 3 Other investigators independently discovered prostatic antigens in seminal plasma.4, 5 Subsequently, Sensabaugh and Crim isolated and characterized PSA from human seminal plasma while searching for a potential marker that could be used in the investigation of rape crimes.6 In 1979, Wang et al. isolated and purified an antigen from prostate tissue that was determined to be prostate-specific in nature.7 A member of the human kallikrein gene family, PSA is a 33-kilodalton serine protease secreted by the prostatic epithelium and the epithelial lining of the periurethral glands. It is functionally significant in its role in the liquefaction of the seminal coagulum to allow the release of spermatozoa.8

PSA as a Tumor Marker

Prior to the introduction of PSA, prostatic acid phosphatase (PAP) measurements were used as a marker for prostate carcinoma. Although somewhat useful in monitoring patients with advanced disease, its low sensitivity in localized disease prevented any consideration for its use in the screening of prostate carcinoma.9 The initial concept of PSA as a screening tool arose because of its improved sensitivity over PAP.10, 11

The widespread use of PSA in the clinical setting did not begin until the mid-1980s. The earliest observations had important clinical applications and included: 1) a decrease in the PSA level after hormonal therapy appeared to be correlated with response to treatment11–13; 2) an increase in the PSA level after treatment appeared to precede and herald disease recurrence11, 14, 15; and 3) after radical prostatectomy, PSA should be undetectable; if not, disease recurrence is the rule.11, 14, 15

Several of these early investigators rejected the possibility of using PSA for screening because of a substantial overlap in PSA values between patients with and those without carcinoma and the resulting poor test specificity.11, 16, 17 Recognizing the critical nature of specificity in a screening tool for prostate carcinoma, Guinan et al. established an ULN for PSA of 24 ng/mL in a case–control study.16 This value was selected because it was the mean plus two standard deviations (SDs) of a group of control men who had 1) a histologic diagnosis of benign prostatic hyperplasia (BPH) or 2) had no genitourinary symptoms. Of note, none of the patients had undergone a prostate biopsy. It is interesting to note that, among asymptomatic men in this study, the ULN was 3.59 ng/mL. In a similar case–control study, a serum PSA level of 10 ng/mL was established as the ULN despite the finding that 90% of apparently healthy study subjects had PSA levels < 2 ng/mL.17

In a subsequent study by Stamey et al., investigators evaluated PSA in 2200 serum samples from 699 men, 378 of whom had prostate carcinoma.11 Controls in this study included 157 men ages 21–76 years who had a mean (± SD) PSA concentration of 1.1 ± 0.7 ng/mL, normal DRE findings, no prior history of any prostatic disease, no urinary symptoms, and no abnormal urinalysis results. By comparison, patients with prostate carcinoma had disease stages ranging from A1 (T1a) to D2 (M1) (TNM and Whitmore-Jewett staging systems). Using these definitions, the authors concluded that the ULN for PSA should be 2.5 ng/mL. They also expressed concern regarding the low test specificity as well as the degree of PSA variation within individuals over time.18–20

PSA as a Screening Tool

The results of what to our knowledge were the first large-scale studies of PSA screening among healthy men appeared in the late 1980s and early 1990s.21–25 These studies examined both men volunteering for screening21–23, 25 as well as populations seen within a general urologic practice.24 It is important to understand the unique attributes of these studies because design considerations affected the results and therefore PSA screening over the past decade. Even more important was the fact that the populations being screened had never been examined with a sensitive screening test before. Prior to the use of PSA for screening, studies of DRE for prostate carcinoma had found rates of prostate carcinoma detection of 0.8%26 and 1.4%.27 When disease was detected by DRE, a test with low sensitivity, approximately two-thirds of those patients who were staged pathologically were found to have extraprostatic disease.26 Thus, in this effectively unscreened population, if a screening test with only slightly improved sensitivity were implemented and if the prevalence of the disease was relatively high, the test would be expected to have a high rate of disease detection. This is precisely what occurred in the period between 1989–1991 when PSA screening dramatically increased across the U.S.28–30

Five initial series of PSA screening set the stage for this widespread test adoption: those of Brawer et al., Terris et al., Cooner et al., Catalona et al., and Wang et al..21–25, 31 Against the backdrop of current practice and our current knowledge of the likelihood of a positive prostate biopsy in men with normal PSA levels, it is important to examine several aspects of theses studies. Table 1 summarizes these characteristics.

Table 1. Early Experience with PSA-Based Screening for Prostate Carcinoma
 Series
Catalona et al.25Brawer et al.21, 22Terris and Stamey23Cooner et al.24Wang and Kawaguchi31
  1. PSA: prostatic-specific antigen; DRE: digital rectal examination; TRUS: transrectal ultrasonography; ULN: upper limit of normal.

No. of men165312494781807720
% WhiteUnknown89.4%UnknownUnknownUnknown
Testing (e.g., PSA, DRE, TRUS)PSA - allPSA - allPSA - allPSA - allPSA - 156
DRE - if abnormal PSADRE - if abnormal PSADRE - 474 of 478 patientsDRE - allDRE - unknown
TRUS - if abnormal PSATRUS - if abnormal PSA, done at time of biopsyTRUS - see belowTRUS - allTRUS - unknown
Frequency of testingEvery 6 monthsYearlySingle visitSingle visitSingle visit
Compliance with serial screening1 year - 66% of those with PSA < 4.0 ng/ml
ULN of PSA4.0 ng/mLYear 1 - 4.0 ng/mL2.5 ng/mL4.0 ng/mL5.0 ng/mL
Year 2 - 1.5 ng/mL
What prompted a biopsy?PSA > 4.0 ng/mL AND abnormal DRE ± abnormal TRUSYear 1 - abnormal PSAAbnormal DRE and/or PSA > 2.5 ng/mLAbnormal DRE, suspicious TRUS findingsUnknown
PSA > 10.0 ng/mL (regardless of DRE and TRUS findings)Year 2 - abnormal PSA or DRE
What fraction of men were recommended to undergo a biopsy?UnknownYear 1 - 187 (15%)180 (37.7%)958 (53%)Unknown
Year 2 - 260 (37.1%)
Of these, what fraction underwent the biopsy?112 men among 137 men with PSA > 4.0 ng/mLYear 1 - 105 (56.2%) 16 patients elsewhere69 (38.3%) TRUS only performed in 13 of 180 patientsAllUnknown
Year 2 - 82 (32%)
Biopsy technique?Directed only (random sextant if PSA ≥ 10 ng/mL normal DRE and normal TRUS)Random sextantUnknownDirected onlyUnknown
Was there a difference in the men who underwent the biopsy compared to those who did not?UnknownUnknownUnknownUnknownUnknown
Any biopsies performed in men with a normal PSA/DRE?NoSome men with PSA < 4.0 ng/mL underwent biopsy on year 2 (given ULN set at 1.5 ng/mL) - unknown number of men with normal DRE in this groupNo923 men with normal DRE/PSA (22%)204 biopsies performed19 (2.1%) positiveOne patient with PSA of 4.2 ng/mL found to have carcinoma prompting recommendation to change ULN

In retrospect, it is evident how the design for these various studies could have interfered significantly with the assessment of PSA for disease screening. In the face of findings that may not be applicable to all ethnic groups, major limitations in the ascertainment of disease among study participants, the lack of application of modern biopsy techniques (e.g., 6–12 cores obtained at needle biopsy), multiple questions regarding the validation of PSA-based screening for prostate carcinoma remained unanswered. To our knowledge, few of these authors evaluated patients with low PSA levels with prostate biopsy. Some authors did not include DRE as part of the evaluation in men with a “normal” PSA. It is interesting to note that some of these same investigators subsequently described relatively high carcinoma detection rates among men with abnormal DRE findings and PSA levels < 4.0 ng/mL.32

One interesting series by Wang et al. used a PSA of 5 ng/mL as the ULN, as recommended by the manufacturer of the Pros-Check PSA assay (Yang Laboratories, Bellevue, WA).31 Among the 720 patients who had a PSA level < 5 ng/mL, 46 were “studied.” Of these 46 men, 1 was found to have a PSA level of 4.2 ng/mL and subsequently was found to have prostate carcinoma, prompting the authors to advocate lowering the ULN to 4.0 ng/mL. The authors reported 5 additional patients within this group who had PSA levels between 4.0–5.0 ng/ml as having BPH based on clinical impression only (i.e., no biopsy was performed).

Subsequent Developments in the Application of PSA Screening

Although the test initially was approved for disease monitoring, with the dissemination of these data, widespread screening began.33 With experience, it became recognized that 1) patients diagnosed by screening most often had clinically organ-confined disease, 2) approximately 1 in 4 men with a PSA level > 4.0 ng/mL was found to have prostate carcinoma on biopsy, and 3) PSA levels were associated with risk of the disease. In accordance with the last observation, Catalona et al. evaluated 10,251 men undergoing PSA screening as well as 266 men without a PSA determination who had undergone a prostate biopsy for an abnormal DRE.34 Disease detection rates among screened men with PSA levels of 4.1–9.9 ng/mL and ≥ 10 ng/mL were 27% and 59%, respectively, during initial screening and 42% and 41%, respectively, with serial screening. A critical element of this study was that none of the 8727 men with a PSA level of ≤ 4.0 ng/mL underwent a prostate biopsy. Another unique aspect of this study was that patients with a PSA level > 4.0 ng/mL with normal DRE and transrectal ultrasonography (TRUS) findings did not undergo biopsy. The limitations of TRUS in the detection of prostate carcinoma have since been recognized.35, 36

Modifications to Improve PSA as a Screening Tool

With the general acceptance of PSA screening and the nearly universal acceptance of 4.0 ng/mL as the ULN, investigators sought to address several problems with the test, namely that 1) only 1 man in 4 with a PSA level > 4.0 ng/mL would be found to have prostate carcinoma, whereas in the other 3 patients, the biopsy was “unnecessary,” and 2) in men with persistently elevated PSA levels, a repeat biopsy frequently detected prostate carcinoma. Using similar study designs as those initially conducted (annual examinations with PSA and DRE, biopsy if either are abnormal, no biopsy if both are normal), a series of observations were made. These included the following.

Age-adjusted PSA

PSA increases in men with age.37, 38 Based on these findings, a lower ULN should be used for younger men whereas a higher ULN may be used for older men. This modification was meant to improve sensitivity in younger men and specificity in older men.

Effect of ethnicity

PSA performance appears to vary by ethnicity. Most investigators have found higher PSA values among African-American men,39–41 although some have not.42 Presumably, by decreasing the ULN in African-American men, test sensitivity would be increased.

PSA velocity

An increasing PSA level is often observed in men who are later diagnosed with prostate carcinoma.43, 44 Some investigators have suggested that an annual increase of 0.75 ng/mL per year should prompt a biopsy, regardless of the PSA level.

PSA density

Because higher levels of PSA are noted in men with larger prostates, some investigators have suggested correcting PSA for prostate size to improve the test specificity in men with large glands and the sensitivity in men with smaller glands.45 Others have not confirmed the benefit of PSA density.46, 47

PSA isoforms

A substantial fraction of circulating PSA is bound to plasma proteins. The proportion that is unbound (ratio of free-to-total PSA) has been found to be associated inversely with the risk of prostate carcinoma.48–50 Other PSA isoforms have been reported to be related to the risk of prostate carcinoma detection.51, 52

These observations have modulated recommendations for prostate biopsy in some settings. Most often, these factors have affected recommendations for a repeat prostate biopsy after an initial negative biopsy.

Prostate Carcinoma Screening in Men with Low PSA Levels

A number of series have reported on rates of prostate carcinoma detection at low PSA levels. In 332 men with a normal DRE and a PSA level of 2.6–4.0 ng/mL, Catalona et al. found prostate carcinoma in 22%.48 One Japanese study found no difference with regard to disease detection between men with PSA levels of 2.0–4.0 ng/mL and those with PSA levels of 4.1–10.0 ng/mL.53 Prostate carcinoma was diagnosed in 23.6% of patients in both groups.

Although a preliminary analysis of the European Randomized Study of Screening for Prostate Cancer (ERSPC) suggests that PSA velocity in men with PSA levels < 4.0 ng/mL does not improve the detection rate of prostate carcinoma,54 data from the PCPT suggest otherwise.1 Studies assessing the value of free-to-total PSA and complexed PSA in men with low PSA levels report detection rates higher than previously suggested.55–57 To our knowledge, the value of these “PSA modifications” compared with biopsy alone in this range of PSA is unknown.

One important question is whether tumors detected in the “normal” PSA range are biologically consequential. If not, detection would be unnecessary and potentially harmful. At odds with common belief, using commonly accepted definitions of a “significant” tumor, a substantial number of the tumors in this population may indeed be consequential. Among 129 men with PSA values < 2.0 ng/mL who were undergoing radical cystoprostatectomy for invasive bladder carcinoma, 30 of whom had an incidental diagnosis of prostate carcinoma, Ward et al. found that 60% of the prostate carcinomas were clinically significant when defined as volume > 0.5 cc, the presence of Gleason score 4 or 5 disease, pT3 disease, positive tumor margins, > 3 tumor foci, or an adverse ploidy status or proliferation index.58 Experience with screening both in Chicago and Tyrol, Austria have reported similarly high rates of clinically significant tumors among men with a low PSA level.59, 60 Notably, in 1 series of 82 patients treated for metastatic disease, 4 patients had PSA levels ≤ 2.0 ng/mL.61 Finally, in our recent report from Southwest Oncology Group Study 8894, we noted that 153 of 1238 men with metastatic disease enrolled in this Phase III clinical trial had a PSA level < 20 ng/mL at the time of diagnosis.62

Changes in Biopsy Technique

Further confounding the understanding of PSA as a clinical test were changes in prostate biopsy. Prior to the mid-1980s, prostate biopsies were performed using a wide range of techniques, using biopsy core needles that were large-bore and often unreliable, and there was substantial procedure-related morbidity.63, 64 As a result, often only two or four biopsy cores were obtained and on occasion, these cores had insufficient sample for diagnosis. In the mid-1980s, automated spring-loaded biopsy “guns” were developed using 18-gauge needles and with this, more reliable tissue sampling was achieved.65 This less painful and more rapid technique, along with the development of TRUS guidance, allowed safer and better sampling of the prostatic parenchyma.66

In 1989, Hodge et al. introduced the technique of ultrasound-guided, random sextant biopsy.67 This technique remained the clinical standard for nearly a decade until a number of reports of false-negative rates as high as 20–25%.68–70 Additional changes to improve biopsy sensitivity included directing the needles more laterally to sample more of the peripheral zone as well as increasing the number of biopsy cores. Currently, most physicians obtain ≥ 10 cores.71

It is important to recognize that these changes in prostate biopsy were occurring simultaneously with the rapid acceptance of PSA screening. As a result, the performance of the test (e.g., sensitivity, specificity) was changing, not due to the test itself but because of the method of disease ascertainment.

Clinical Significance of Tumors Detected by PSA-Based Screening

Since Whitmore first conceived of the prostate cancer paradox of whether cure was possible when it was necessary and whether cure was necessary when it was possible,72 the urologic oncology community has struggled with the definition of a “clinically significant” prostate carcinoma. From multiple series of patients who were managed with observation, there is abundant evidence that the disease is not uniformly progressive nor fatal.72–75 Unfortunately, previous experiences with the natural history of prostate carcinoma are seriously compromised by selection bias72 and limited follow-up,76 and, in perhaps the most widely cited article, very few individuals were reported to be at risk at 10 years and 15 years and all participants had their disease detected using DRE because PSA screening had not yet begun to be used.75

With the advent of PSA screening and the dramatic increase noted in the detection of organ-confined disease, the idea of “clinically-significant” disease became more important. Because of the concerns of possible overtreatment, two large-scale treatment-versus-observation clinical trials were initiated. In the early results from a Scandinavian prospective randomized trial comparing radical prostatectomy and watchful waiting, an improved disease-specific survival rate and a lower risk of distant metastasis were found among men assigned to undergo radical prostatectomy.77 However, it is important to note that no difference was found with regard to overall survival between the two groups. The U.S. Prostate Cancer Intervention Versus Observation Trial (PIVOT) is a randomized trial comparing radical prostatectomy versus expectant management for the treatment of patients with clinically localized prostate carcinoma. The primary endpoint of the study is all-cause mortality.78 This trial finished the accrual of 727 men in November 2001.

With the goal of predicting the behavior of an individual tumor in an individual patient, a number of criteria have been developed. In one approach, Eastman et al. and D'Amico et al. developed the concept of “good,” “intermediate,” and “poor” risk groups.79–81 Most systems combine PSA, DRE staging information, and tumor grade. Unfortunately, in the context of population-based, annual PSA screening, these systems break down because the majority of patients are categorized as “good risk.” Table 2 lists a number of definitions of clinically insignificant tumors.

Table 2. Definition of Clinically Indolent Prostate Carcinoma
SeriesDefinition
Smith and Catalona122Impalpable, focal, no Gleason Grade of 3, 4, or 5
Epstein et al.123Tumor volume < 0.2 cm3, confined, no Gleason Grade of 4 or 5
Stamey et al.124Tumor volume < 0.5 cm3
Dugan et al.125Tumor volume ≤ 20 cm3 at the time of expected death, Gleason's sum < tenths place of patient's age

To illustrate the troublesome nature of these definitions, it is instructive to examine the implications of the findings from the recent Prostate Cancer Prevention Trial article.1 There are approximately 50 million men ages 40–74 years in the U.S.82 The experience of the Prostate Cancer Awareness Week would suggest that approximately 92% of these men have a PSA level < 4.0 ng/mL.83–85 Among the 8% of men with a PSA level ≥ 4.0 ng/mL, 25% or 1 million men have prostate carcinoma. Based on data from the Surveillance, Epidemiology, and End Results (SEER) program, approximately 19% or 190,000 men would be anticipated to have high-grade disease.86 Using the PCPT findings, if the remaining 46 million men with a PSA level < 4.0 ng/mL underwent prostate biopsy, 15% or 6.9 million men would be diagnosed with prostate carcinoma and approximately 1 million of these men would be diagnosed with high-grade disease.

These estimates raise three concerns. First, because the lifetime risk of prostate carcinoma death in the U.S. is approximately 3%, the application of prostate carcinoma screening to the general population (in which prostate biopsy can indeed detect many more tumors) is nearly certain to result in disease overdetection. A second, very serious concern, is that the use of a PSA level < 4.0 ng/mL to identify men at no risk of prostate carcinoma may result in a very serious underdetection of biologically consequential disease. The finding that as many as 25% of men who undergo radical prostatectomy require secondary therapies within 5 years (suggesting failure of the initial therapy) confirms that waiting for a PSA level to exceed 4.0 ng/mL may not be appropriate in some men.87 Finally, the sheer number of Gleason Grade 7-10 tumors that are present in the general population and are not detected with current screening practices, contrasted with the relatively low lifetime risk of death from prostate carcinoma, suggests that our current methods to establish the prognosis of an individual tumor may perform quite poorly in many men.

New Directions in the Early Detection of Prostate Carcinoma

Given the highly variable natural history of prostate carcinoma and the concerns we have described, it is our opinion that we must now move beyond using PSA for the detection of prostate carcinoma to a detection strategy that incorporates disease prognosis. Ideally, a public health strategy would have three steps.

Step 1. Apply a test to identify the man at high risk of prostate carcinoma. In this man, prevent the disease.

This first step is extremely attractive because we now have evidence that it is possible to reduce the risk of prostate carcinoma.88 For the man found to have a low risk of disease, prevention may not be desirable because of the possible side effects of the prevention strategy. Ideally, this risk assessment could be applied with some frequency to ensure the man remains at low risk. Conversely, a man with a substantial risk of disease would be more likely to accept possible side effects for his greater likelihood of benefit. With finasteride, there is a small but measurable risk of sexual dysfunction and a possible risk of the development of high-grade disease.88 One example of this approach, The Breast Cancer Prevention Trial, used the Gail model to predict the risk of breast carcinoma.89 Almost certainly, a risk assessment for an individual man would incorporate a host of factors including PSA, age, family history of prostate carcinoma, ethnicity/race, genetic factors, diet and use of micronutrients, and behavioral factors (such as exercise and obesity).

Ideally, different preventive strategies may be used in different men. The preclinical evidence of substantial differences in the activity of 5-α reductase as well as the degree of inhibition by finasteride in variants of this enzyme coded for in different genetic variations of SRD5A2 is one potential application of this concept. Another possibility could be selenium supplementation for those individuals with low levels of natural selenium exposure. These strategies must be examined in the context of appropriately designed clinical trials.

Step 2. Among men at high risk of prostate carcinoma, apply a prognostic/diagnostic test that identifies the man who has disease that puts him at risk for morbidity or mortality. In this man, diagnose the disease.

The correlate to this is that in the man in whom the prognostic/diagnostic test is negative (identifying a tumor of low risk), it may be desirable for the diagnostic test to be negative. The advantages of such an approach are obvious: elimination of the cost, morbidity, and anxiety of the overtreatment of tumors of no clinical consequence while identifying and curing those that pose a risk to an individual man.

Step 3. In the man who is subsequently found to have prostate carcinoma that is of clinical consequence using this stepwise approach, treat the disease.

The end-product of this approach therefore is the man whose prostate carcinoma could not be prevented and puts him at a meaningful risk for complications over his lifetime. In this man, clinicians could have greater confidence that treatment was necessary. This would also allow for the intense focus of clinical trials to identify the optimal treatment for the disease.

Merging Prognosis and Diagnosis: through a Glass, Darkly

A number of indicators independent of the stage and grade of the tumor have been implicated as predictors of outcome among men with prostate carcinoma. Obesity has received considerable attention as a predictor of poor outcome after radical prostatectomy.90–92 Dietary fat93, 94 as well as levels of some micronutrients95 have been associated with a risk of worse outcomes as well. A number of genetic polymorphisms,96–98 ethnicity,99–102 age,103–105 and pretreatment PSA level106 also are reported to be correlated with outcomes. Some of these prognostic markers also may be useful for disease diagnosis.107–114 Although several series have failed to demonstrate any independent contribution of family history to the overall prognosis from prostate carcinoma,115–117 it is well recognized that individuals with a family history of prostate carcinoma are at a higher risk for developing the disease than the general population.107, 118, 119 The challenge for clinical research is the incorporation of appropriate risk markers into the most reliable and cost-effective models of disease outcome.

Contribution of New Technologies

The application of new technologies also may play a role in identifying high-risk men through further prostate carcinoma risk stratification strategies. The explosion in the field of proteomics may provide a key opportunity for rapid advances in prognosis. One proteomic tool, surface-enhanced laser desorption and ionization–time-of-flight (SELDI-TOF) mass spectrometry, has been reported to have very high sensitivity and specificity for the detection of prostate carcinoma.120, 121 This technique of “protein-profiling” allows for the rapid identification of mass/charge peaks in spectroscopy output that can be compared between cases and controls.

This technique is currently being evaluated by the Early Detection Research Network (EDRN) of the National Cancer Institute. In this rigorously designed validation trial, three stages are incorporated: 1) portability and reproducibility, 2) validation of protein peak classifier algorithms, and 3) validation of the methodology in a prospective clinical trial. Unique to this validation study are the definitions of cases and controls. Several groups of controls have been incorporated including men with a range of PSA values and a negative biopsy and men with persistently low PSA levels, as well as men with other tumors or inflammatory conditions and no evidence of prostate carcinoma. (This range of controls offers the greatest opportunity to examine the risk of false-positive results.) Most important, to our knowledge this study will be the first to examine the potential of a diagnostic test to incorporate prognosis. To accomplish this, two groups of men with clinical T1-2 prostate carcinoma will be included as cases: lower-risk tumors (Gleason Grade ≤ 6) and higher-risk tumors (Gleason Grade ≥ 7). This concept of diagnosis and prognosis is a key feature of validation studies of the EDRN.

Conclusions

The recent publication of data regarding the detection of prostate carcinoma in men with a “normal” PSA level makes this a challenging time for clinicians and patients. Which patients should receive treatment with finasteride to prevent prostate carcinoma? What is a normal PSA level? When should a prostate biopsy be recommended? There are no clearcut answers to these questions and it will require additional effort from all physicians to ensure that men are well informed. Nevertheless, these data provide an opportunity for the cancer prevention, treatment, and research communities to set the agenda to answer critically important questions that will remain so for decades to come.

Ancillary