Epdemiology

In 2010, it was estimated that 217,730 American men would be diagnosed with prostate cancer, accounting for 25% of all newly diagnosed cancers in men.¹ It appears that one in six American men will develop a clinically recognized invasive prostate cancer during his lifetime. Ninety-one percent of cases were expected to be diagnosed of local or regional stage disease for which the 5-year relative survival rate approaches 100%. The age-adjusted annual incidence rate increased 6.4% per year between 1983 and 1989,² with a 66% overall increase between the periods of 1975 to 1979 and 1987 to 1991.³ This change largely reflected the increased detection of localized, rather than regionally advanced or metastatic, prostate cancer cases.^2,4 From 2001 to 2005 the incidence of newly diagnosed cases dropped 4.4% per year.^5,6 This decrease in the incidence of prostate cancer may reflect stabilization related to PSA screening. Prostate cancer remains the second leading cause of cancer death in men, with 32,050 deaths being estimated in 2010. As with the incidence of newly diagnosed cancers there has been a significant drop in prostate cancer death rates from 1990 to 2005. The death rate from prostate cancer was 38.56/100,000 men in 1990 compared with 24.65/100,000 men in 2005 (p <.05), representing a 36% decline.¹

The incidence of incidental (or latent) prostate cancer observed at autopsy examination and the incidence and prevalence of clinically manifest prostate cancer increase substantially with age.^6–9 Information from the Surveillance, Epidemiology, and End Results (SEER) program indicates that approximately 1.8% of American men between ages 40 and 59 years had a clinical diagnosis of prostate cancer whereas 15% of men 60 to 79 years old would have prostate cancer by the time they attained this age. Although the prevalence of incidental carcinoma does not exhibit marked geographic variation, the regional disparity in the antemortem clinical expression of the disease and the cancer mortality rate of prostatic carcinoma is noteworthy.^7,10 Men in North America, Australia/New Zealand, Western and Northern Europe, and the Caribbean have a much higher incidence of this disease than men in Asia and China.¹⁰

Etiology

Cohort and case-control studies are the two principal methods used to ascertain the etiology of cancer.¹¹ Several reports describe various host-related and environmental factors that appear associated with prostate cancer (Table 51-1), but it is not yet understood which factors directly initiate or promote prostate carcinogenesis. As discussed by several authors,^12,13 long-term androgen exposure, which is required for normal prostatic development and for cancer growth and maintenance, advanced age,¹ race/ethnicity,^14–16 and familial history of prostate cancer^17,18 are presently viewed as the most likely risk factors. Other associations such as diet, body mass index, energy balance, and physical activity are becoming more suspect.^19–23

TABLE 51-1 Potential Risk Factors for Prostate Cancer

Prevention

Prostate cancer is an androgen-dependent tumor with a prolonged latency between initial malignant transformation and clinical expression, which are features well-suited to disease prevention efforts.⁷ Progression from tumor inception to invasive carcinoma often takes decades, allowing sufficient time for intervention.⁹ Chemoprevention strategies that use high-risk target populations, particularly those with premalignant lesions (e.g., high-grade PIN), have the greatest potential to identify promising agents in a time-efficient manner.⁹⁸ The results of focused studies such as these can then be confirmed in large-scale trials applied to the general population. The ability to alter the hormonal environment of the host provides an excellent opportunity to interrupt the multistep process that results in clinical expression of the disease.⁷ Advances in our understanding of the process of carcinogenesis and the availability of promising new chemopreventive agents, including those producing reversible androgen deprivation, have the potential to favorably affect the morbidity and mortality of prostate cancer in the foreseeable future.

Luteinizing hormone-releasing hormone analogues (e.g., goserelin, leuprolide) reduce luteinizing hormone and (secondarily) testosterone levels. The long-term use of these agents may cause anemia, atrophy of reproductive organs, diminished muscle mass, loss of libido, and vasomotor instability, which limits the utility of these agents for chemoprevention in the general population. Nonsteroidal antiandrogens (e.g., flutamide, bicalutamide) competitively bind to androgen receptors in target tissues. These agents are well tolerated in most patients, although adverse effects may include gastrointestinal disturbance, gynecomastia, and vasomotor instability.

Intracellular 5α-reductase converts testosterone to dihydrotestosterone (DHT), the hormone responsible for prostate epithelial proliferation. DHT has greater affinity for the androgen receptor and is the primary agonist leading to prostate maintenance and growth. Three isoforms of 5α-reductase have been identified with various levels in different tissues. Type 1 is expressed in benign prostate hyperplasia, and its expression is greatly increased in prostate cancer, especially high-grade tumors.⁹⁹ Type 2 5α-reductase expression is decreased in PIN and some early cancers but is increased in metastatic and recurrent prostate cancer.⁹⁹ The role of type 3 5α-reductase is less defined. Competitive inhibitors of 5α-reductase (e.g., finasteride and dutasteride) suppress intraprostatic dihydrotestosterone to castrate levels. The Prostate Cancer Prevention (PCP) trial was initiated to test the efficacy of finasteride as a chemoprevention agent in men at low risk of having prostate cancer.¹⁰⁰ This placebo-controlled phase III trial randomized 18,882 eligible men (age ≥55 years, normal digital rectal examination [DRE] and PSA levels <3 ng/mL) to finasteride (5 mg daily) or placebo for 7 years. There was a 25% reduction in the prevalence of prostate cancer over this 7-year period from 30.6% in the placebo group to 18.6% in the finasteride group.¹⁰⁰ Of note, however, is that more aggressive tumors, with Gleason score 7 to 10, were more common in patients who took finasteride: 37% of all tumors and 6.4% of all men on the finasteride arm versus 22% of all tumors and 5.1% of all men on the placebo arm. The increased incidence of high-grade cancers seen in the PCP trial has been a topic of great debate. The investigators have argued that the increase in high-grade cancers is due to a detection bias related to the reduced volume of prostate tissue and therefore a greater ratio of cancer to benign tissue. Furthermore, there was no dose effect from the finasteride with no significant increase in worse cancers with higher cumulative doses of the drug.¹⁰¹ The PCP trial was not designed or powered to detect differences in cancer-specific survival (CSS) or overall survival (OS). Finasteride did reduce urinary symptoms compared with placebo, but there were also significantly more adverse sexual side effects.¹⁰⁰ A reduced volume of ejaculate, erectile dysfunction, loss of libido, and gynecomastia were more common in the finasteride group (p <.001), but urinary urgency, frequency, retention, urinary tract infection, and prostatitis were less common with finasteride (p <.001).¹⁰⁰

The Reduction by Dutasteride of Prostate Cancer Events (REDUCE) trial, a phase III study testing inhibition of 5α-reductase with dutasteride, has completed its accrual of patients.¹⁰² Unlike finasteride, which blocks 5α-reductase type 2, dutasteride blocks both types 1 and 2, suggesting it may be more effective at preventing the development of prostate cancer.^99,103 Phase III trials evaluating dutasteride for the treatment of benign prostatic hyperplasia coincidentally showed a significant reduction in the incidence of prostate cancer.^102,104 The REDUCE trial is a randomized trial of placebo versus dutasteride, 0.5 mg, administered daily to evaluate it as a chemopreventive agent for prostate cancer. This trial completed its accrual of 8000 subjects in 2004. Unlike the PCP trial, which enrolled men with a low risk of prostate cancer, the REDUCE trial sought patients with a higher risk of developing or being diagnosed with prostate cancer (e.g., PSA >2.5).¹⁰²

The association of dietary consumption or serum levels of selenium and α-tocopherol and a low rate of prostate cancer have suggested that dietary supplements of these nutrients may protect men from the development of this disease.⁶¹ Despite this association, two prospective phase III clinical trials did not show a reduction in the incidence of prostate cancer with vitamin E or selenium supplementation. In the Selenium and Vitamin E Cancer Prevention Trial (SELECT) there was no significant reduction in prostate cancer incidence related to either selenium or vitamin E supplementation.¹⁰⁵ SELECT enrolled 32,400 participants to a phase III randomized, double-blind, placebo-controlled trial with a 2 × 2 factorial design. The study was designed to determine the efficacy of selenium, vitamin E, or a combination of the two to prevent prostate cancer. Unfortunately, neither supplement alone nor in combination lowered the risk of prostate cancer in healthy men.¹⁰⁵ The Physicians’ Health Study II enrolled 14,641 male physicians and randomized them to receive vitamin E and C supplements daily. Neither vitamin E nor vitamin C supplementation reduced the risk of prostate or other cancers.¹⁰⁶

Early Detection

Considerable controversy surrounds the use of early detection programs for prostate cancer. Some argue that early detection efforts are too costly and will lead to the recognition of an increased number of clinically insignificant tumors, because autopsy studies demonstrate a high prevalence of incidental tumors in older men.^9,10 Likewise, a study of prostate cancer discovered in organ donors found incidental prostate cancer in one third of men aged 60 to 69 and 46% in men older than age 70.¹⁰⁷

Contributing further to the arguments against prostate cancer screening are the limited sensitivity and specificity of serum PSA level, DRE, and ultrasonography in diagnosing cancer. Although DRE has high specificity for prostate cancer, it has a low sensitivity profile and is not considered an effective detection tool on its own.¹⁰⁸ In contemporary series,¹⁰⁹ PSA testing with a threshold of 4.0 ng/mL has a sensitivity of only about 20%. Although the sensitivity of PSA testing could be improved by lowering the threshold value for all men, this would compromise the specificity and increase the detection of clinically insignificant cancers. Early detection strategies have also been criticized for exaggerated improvements in cancer-specific survival related to an early detection bias. In an early report the use of PSA resulted in a diagnostic lead time of approximately 5 years^110,111 and longer in the detection of earlier-stage and lower-grade tumors.¹¹² Data from the European Randomized Study of Screening for Prostate Cancer (ERSPC) and the SEER registry suggest the lead-time bias could range from 5.9 to 7.9 years.¹¹³ Included in the debate over prostate cancer screening are the risks of overdetection and overtreatment. Overdetection occurs when men are found to have disease that would never have remained silent and contributed no morbidity in their lifetime. Overtreatment occurs when an intervention plays no role in extending a patient’s life or preventing morbidity from the illness. The real challenge for clinicians involved in the management of prostate cancer is the identification of clinically significant disease.

Arguments in support of prostate cancer screening include the 36% reduction in prostate cancer deaths seen between 1990 and 2005.¹ This trend began shortly after the introduction of PSA testing, and statistical models suggest that PSA testing contributed to this decline.^114,115 Furthermore, PSA testing is responsible for the migration of prostate cancer diagnoses to earlier and more curable stages.^1,116 These findings, as well as data from studies suggesting a survival benefit to treatment for early cancers, support a role for early detection and treatment. The Swedish randomized trial of surgery versus watchful waiting demonstrated an improvement in disease-specific and overall mortality in men undergoing radical prostatectomy (RP) for early-stage disease.¹¹⁷ In the United States, an observational cohort of 44,630 men from the SEER registry suggests a survival advantage with active treatment for low- and intermediate-risk prostate cancer in men aged 65 to 80 years.¹¹⁸

Early data from two large prospective randomized trials seeking to measure the benefit from prostate cancer screening contribute to this screening controversy.^119,120 The U.S. Prostate, Lung, Colon and Ovary (PLCO) screening trial registered 76,693 men at 10 study centers to determine the impact of annual PSA determination and DRE on the cause-specific mortality for cancers in each of these organ sites.¹²⁰ The primary exclusion criteria were a history of a PCLO cancer, current cancer therapy, and more than one PSA blood test in the 3 years prior to study enrollment. Subjects in the screening group were offered annual DRE for 4 years and annual PSA testing for 6 years. A PSA of more than 4.0 ng/mL or a DRE indicating nodularity or induration were considered suggestive of prostate cancer, and patients with these findings were advised to seek diagnostic evaluation. With a median follow-up of 11.5 years, although there were significantly more cancers diagnosed in the screened group there was no reduction in prostate cancer mortality.¹²⁰ The ERSPC recruited 182,000 men between the ages of 50 and 74 years from seven European countries. The men were randomized to receive PSA screening once every 4 years or to a control group that did not get PSA screening. There was country to country variability in enrollment criteria and screening regimens, age of eligibility, case selection criteria, thresholds for PSA levels, and the inclusion of DRE in the screening assessments. PSA values as low as 3.0 ng/mL were considered abnormal and would prompt referral for further testing that differed between participating countries. Of 162,387 eligible men enrolled, 5,990 prostate cancers were diagnosed in the study group compared with 4307 (20% fewer) in the control group. With a median follow-up of 8.8 and 9.0 years in the screening and control groups, respectively, the adjusted rate ratio of prostate cancer death in the screened arm was 0.80 (95% confidence interval [CI], 0.65 to 0.98, p = .04).¹¹⁹ In the ERSPC trial the number of men who would need to be screened to prevent one prostate cancer death is 1,410. More importantly, 48 men would need to be treated to prevent each prostate cancer death.¹¹⁹

There are important differences in these two screening trials that help explain the apparent differences in conclusions. The U.S. trial enrolled a group of men who in as many as 42% of cases underwent previous screening, effectively reducing the prevalence of cancer relative to that of a completely unscreened population. In addition, over the period of the trial 52% of nonscreened men had PSA testing, contaminating the control group. The follow-up of 11 years for prostate cancer mortality is relatively short for a group of men with screened cancers. The heterogeneity of eligibility and screening criteria in the ERSPC trial make it challenging to determine the exact population that would benefit from screening. Finally, the quality of life outcomes and cost-effectiveness analyses for both these trials have yet to be reported and these results will undoubtedly contribute significantly to our understanding of the merits of population-wide screening for prostate cancer.

One criticism of published trials of PSA screening is the use of singular thresholds to prompt further diagnostic evaluation. Because serum PSA levels directly correlate with age in men without prostate cancer, several investigators sought to improve the diagnostic accuracy of PSA by defining the normal test range as a function of age and race^121–127 (Table 51-2). In particular, DeAntoni and colleagues¹²⁷ found that the mean PSA level was significantly different for successive decades of age and there was increased variability in PSA values with advancing age. It was this age-related variability that largely accounted for the phenomenon of age-specific reference ranges. El-Galley and associates¹²⁶ studied the clinical impact of age-specific reference ranges in 2657 men who underwent prostatic biopsy. Analysis of the sensitivity, specificity, and positive predictive value profiles supported use of age-specific reference ranges because of increased detection sensitivity in younger men and increased specificity in older men.

Measuring PSA velocity (PSAV), defined as the change in serum PSA over time, is another method to account for prostatic changes that occur during the aging process. In men who develop prostate cancer, an exponential increase in PSA values begins approximately 5 years before the diagnosis is established,¹²⁸ and detecting a PSAV of 0.75 ng/mL/yr or more appears to be a sensitive means to distinguish these men from those without prostatic disease or those with benign prostatic hyperplasia.^128–130 However, owing to interassay variability, only a PSA change exceeding +7.5% may be considered significant according to Kadmon and colleagues.¹³¹

Additional attempts to improve the screening accuracy of PSA measurements are based on the observation that serum PSA level depends on cancer volume, tumor differentiation, and the amount of benign prostatic tissue.¹³² To account for coexistent benign prostatic hyperplasia, the concept of PSA density (PSAD) was introduced by Benson and colleagues.¹³³ PSAD is the total serum PSA value divided by the volume of the prostate gland, as determined by TRUS using the prolate ellipsoid formula (volume × length × width × height × 0.52). PSAD appears most useful for patients with a total serum PSA level in the range of 4 to 10 ng/mL, particularly when palpable prostatic abnormalities are absent. In this setting it is believed a PSAD of 0.15 ng/mL/cm³ or more best identifies men in whom prostatic biopsy should be considered.^134,135 In other investigations, however, diagnostic accuracy was not enhanced by the PSAD compared with using the upper normal PSA concentration, defined as 4.0 ng/mL, as a cutoff point for early detection.^136–138 Another PSA derivative, transition zone volume-adjusted PSA (PSAT) was introduced as an evolution in the PSAD concept for men with a serum PSA value in the indeterminate range (i.e., 4 to 10 ng/mL).¹³⁹ PSAT is calculated by dividing the serum PSA value by the TRUS-determined transition zone volume and is based on the rationale that benign prostatic hyperplasia results exclusively from transition zone hyperplasia.

Because the proportion of PSA complexed to α₁-antichymotrypsin is greater in patients with prostate cancer than in men with benign prostatic disease, the ratio of free-to-total (i.e., percent free) PSA will be lower in men with prostate cancer and may help discriminate between benign and malignant prostate conditions.^140–144 Although the free-to-total PSA ratio can be applied to any serum PSA level, performing a free PSA determination improves the specificity for prostate cancer detection when the total serum PSA range is 3 to 10 ng/mL.^142,145 To determine the optimal cutoff point that may warrant prostatic biopsy, various free-to-total PSA ratios were examined for their association with prostate cancer.^144,145 A multicenter clinical performance study demonstrated that a free-to-total PSA ratio of less than 7% was highly suggestive of cancer whereas a free-to-total PSA ratio of above 25% was rarely associated with malignancy. In association with other study results, a diagnostic algorithm for the detection of early-stage prostate cancer based on the percent free PSA has been suggested.¹⁴⁵ However, larger population-based trials must be conducted and the utility of prostate cancer screening must be ascertained before widespread application can be recommended.¹⁴⁶

Screening Recommendations

The controversy in prostate cancer screening is evident when the current recommendations from various professional societies are reviewed. The American Cancer Society (ACS) previously recommended annual DRE and PSA testing in men older than 50 years of age with a life expectancy of at least 10 years.¹⁴⁷ In 2009, after the publication of the interim results of the PLCO and ERSPC trials, the ACS no longer supports routine testing for prostate cancer. The ACS does support health care professionals discussing the potential benefits and limitations of early detection with an offer to test with annual PSA screening and DRE beginning at age 50 in men who are at average risk of prostate cancer with a life expectancy of more than 10 years.¹⁴⁸ The American Urological Association (AUA)¹⁴⁹ recommends PSA screening for well-informed men who wish to pursue early diagnosis beginning at age 50 and sooner for those men with a higher life-time risk (positive family history in a first-degree relative or African-American race). A baseline PSA value at age 40 above the median value (0.6 to 0.7 ng/mL) may identify a group of men with a significant risk of prostate cancer in the future.^150,151 Based on the findings of the PLCO and ERSPC studies, the U.S. Preventative Services Task Force (USPSTF) concluded that for men younger than age 75 years the benefits of screening for prostate cancer are uncertain and the balance of benefits and harms cannot be determined. For men 75 years or older there is moderate certainty that the harms of screening for prostate cancer outweigh the benefits.¹⁵²

Anatomy

Prostatic anatomy is readily understood by using the urethra as a reference point. The verumontanum is a major anatomic landmark and protrudes from the posterior prostatic wall at the midpoint of the urethra. Most prostatic ducts and the ejaculatory ducts empty into the urethra in this region, whereas the periurethral glands empty throughout the length of the urethra. Immediately proximal to the verumontanum is the müllerian remnant, the utricle, an approximate 0.5-cm length of epithelium-lined cul-de-sac. A circumferential sleeve of muscle surrounds the entire urethra and is composed of the proximal preprostatic smooth muscle sphincter, which prevents retrograde ejaculation, and a distal sphincter of striated and smooth muscle at the apex that is essential for urinary control.

The prostate is composed of three zones: peripheral, central, and transition. The peripheral zone includes approximately 70% of the prostatic volume and is the most common site of PIN and carcinoma. DRE often includes a description of the prostatic lobes based on palpation of the median furrow, which divides the peripheral zone into the right and left halves. The central zone is a cone-shaped area that includes the base of the prostate and encompasses the ejaculatory ducts; it represents approximately 25% of the volume of the prostate. The transition zone is the smallest component of the normal prostate, accounting for approximately 5% of the gland, but it usually enlarges as men age owing to benign prostatic hyperplasia and may grow to dwarf the remainder of the prostate.

The prostatic capsule consists of an inner smooth muscle layer and an outer covering of collagen, with marked variability in the relative amount of each in different areas. At the apex, acinar elements are sparse and the capsule is ill defined. As a result, the prostatic capsule cannot be regarded as a well-defined anatomic structure with constant features.

The nerve supply of the prostate is furnished by paired neurovascular bundles along the posterolateral edge of the prostate. Autonomic ganglia are clustered near the neurovascular bundles, and small nerve trunks arising at this site penetrate the capsule to form an extensive network within the gland.

The primary blood supply of the prostate is furnished by the internal iliac artery, and the venous drainage is directly into the prostatic plexus, which eventually empties into the internal iliac vein. This route may account for hematogenous metastases, most often in osseous sites.

The lymphatics of the prostate and the seminal vesicles were detailed through anatomic dissections described by Rouvière.¹⁵³ Lymphatic drainage of the prostate originates in an extensive intraprostatic network that coalesces to form a subcapsular system and henceforth coalesces into a periprostatic lymphatic network and the four pedicles of the collecting trunks: the external iliac, hypogastric (or internal iliac), and posterior and inferior pedicles that terminate in the external iliac (including the obturator), presacral, and hypogastric lymph nodes, respectively. The lymphatics of the seminal vesicle originate in the mucous and muscular networks and give rise to the superficial plexus, which culminates in the lymph nodes of the external iliac system. Modern surgical series of extended pelvic lymph node dissections report involved lymph nodes commonly in the obturator fossa and external and internal iliac regions.^154,155 These sites account for nearly 85% of lymph node metastases in prostate cancer.¹⁵⁵ Another 15% of involved nodes can be identified in the presacral, paravesicular, and pararectal sites.¹⁵⁵

Histology

The prostatic epithelium consists of three main histologic types: secretory, basal, and neuroendocrine cells. Despite having the lowest proliferative activity of the epithelial elements, secretory cells produce PSA, prostatic acid phosphatase (PAP), acid mucin, and other secretory products. The basal cells form a flattened layer of cells surmounting the basement membrane at the periphery of the glands. These cells possess the highest proliferative activity of the prostatic epithelium and are thought to act as stem or “reserve” cells that repopulate the secretory cell layer.¹⁵⁶ Basal cells, selectively labeled with antibodies to high-molecular-weight keratin (e.g., 34βE12), are used to differentiate benign acinar processes from adenocarcinoma, which lacks a basal cell layer. Neuroendocrine cells are the least common epithelial cell type and are often not identified in routinely stained sections.

The seminal vesicles have a variable anatomic distribution, portions of which may occasionally be found within the prostatic capsule where they may be mistaken for prostatic nodularity or induration on examination. The cells of the epithelial component have large, irregular hyperchromatic nuclei with coarse chromatin and prominent nucleoli that often show nuclear abnormalities. When encountered in needle biopsies, such “pseudomalignant” cytologic atypia may lead to a mistaken diagnosis of prostatic adenocarcinoma. Cowper’s glands consist of lobules of closely packed uniform acini and are noteworthy in that they may also be mistaken for prostatic carcinoma in biopsy specimens.

Pathology of Prostate Cancer

Grading

The Gleason system, based on a Veterans Administration study of more than 4000 patients, has become the global standard for grading prostate cancer (Fig. 51-1). The Gleason system is based on the degree of glandular differentiation and an overall pattern of tumor growth at relatively low microscopic magnification.¹⁶⁵ Five patterns of growth are recognized and numbered in order of increasing aggressiveness. Because tumors may display variable histology, two patterns are recorded for each case: a primary or predominant pattern (Gleason 1 to 5) and a secondary or lesser pattern (Gleason 1 to 5). The Gleason score is the sum of the primary and secondary patterns and ranges from 2 to 10. If only one pattern is present, the primary and secondary patterns receive the same designation. Gleason¹⁶⁶ noted that more than half of prostate cancers contain two or more patterns, and Aihara and colleagues¹⁶⁷ found an average of 2.7 different Gleason grades in a series of prostatectomy specimens. Interobserver and intraobserver variability has been reported with the Gleason and other grading systems, and Gleason¹⁶⁶ noted exact reproducibility of score in 50% of needle biopsies and a score of greater than 1 in 85% of cases.

Figure 51-1 Schematic drawing of modified Gleason grading system.

From Epstein JI, et al: The 2005 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma, Am J Surg Pathol 29:1228-1242, 2005.

In 2005, the International Society of Urological Pathologists held a consensus conference to update the guidelines for Gleason scoring. The modifications were believed to be necessary owing to the evolving roles of PSA screening and TRUS-guided biopsy techniques on the diagnosis and identification of cancer in needle core specimens. For example, the Gleason scoring system predated the use of immunohistochemical staining for basal cells. Many of the grade 1 + 1 = 2 adenocarcinomas diagnosed in the past would now be considered adenosis owing to an intact basal layer. Furthermore, it was made clear that Gleason scores 3 or 4 should rarely be applied to TRUS-guided biopsies because these are rarely present in the peripheral zone but more typically found by transurethral resection of the prostate (TURP) including the transition zone. In addition, the finding that a cribriform pattern is more likely included as a Gleason grade 4 pattern has shifted some Gleason score 6 cases to Gleason score 7. These changes have resulted in a grade migration over the past decade that has paralleled the stage migration witnessed as a result of PSA screening.¹⁶⁸

Although a correlation exists between the needle biopsy tumor grade and that found in the prostatectomy specimen, a higher grade (i.e., more poorly differentiated) is identified in approximately one third of cases.^165,169,170 The correlation is strongest for the primary Gleason pattern, but the secondary pattern also provides useful predictive information, particularly when used to create the Gleason score.¹⁶⁵ The incidences of undergrading and understaging are decreasing related to stage migration as well as the elimination of Gleason score 2 to 4 biopsy specimens.¹⁷¹ Whereas in the past downgrading of high-grade cancers was uncommon at RP, one third to half of high-grade cancers are now downgraded at surgery.^172–176

Molecular Biology

Increasing focus is being given to the role of the androgen receptor (AR) in the tumorigenesis and prognosis of prostate cancer.¹⁹⁷ Androgens, along with multiple co-regulatory factors, activate the AR, leading to transcription of AR target genes promoting the growth of normal and neoplastic prostate tissue. The ETS family of genes includes transcription factors that are involved in a variety of functions, including cellular differentiation and cell cycle control. Several ETS factors have been implicated in carcinogenesis through gene rearrangements. Tomlins and associates¹⁹⁸ have identified recurrent gene fusions of two ETS transcription factors, ERG and ETV1, to the TMPRSS2 gene. In their initial series the TMPRSS2-ERG gene rearrangement was identified in 47% of clinically localized prostate cancers. TMPRSS2 is expressed in normal luminal epithelial and neoplastic prostate tissue and is strongly induced by androgen in androgen-sensitive prostate cell lines. The TMPRSS2-ERG fusion gene can be detected in a proportion of high-grade PIN lesions, and this molecular rearrangement may be an early event that precedes chromosome-level alterations in prostate carcinogenesis.¹⁹⁹ The importance of the TMPRSS2-ERG fusion as an early event in the development of prostate cancer is supported by the homogeneous distribution of the fusion gene in a prostate cancer, the absence of the fusion in benign prostate tissue, and the association of TMPRSS2-ERG–positive cancers associated with the same fusion in high-grade PIN lesions.²⁰⁰ Mutation or changes in AR expression may account for the development of androgen-independent prostate cancer. ERG overexpression, mediated through TMPRSS2-ERG fusion, contributes to development of androgen independence through disruption of androgen receptor signaling. Recently, Yu and colleagues²⁰¹ found that ERG inhibits AR expression and activity and induces repressive epigenetic programs that contribute to androgen resistance and cancer progression.

Predictive Factors

Clinical Endpoints

Factors predictive of clinical endpoints may be grouped according to whether they are patient related, are intrinsic to the neoplasm (i.e., tumor related), or are treatment related. It is essential to recognize that a factor may have importance only in certain patient subsets (e.g., those with clinically localized disease) but not in other settings (e.g., metastatic disease). Furthermore, different endpoints may be considered (e.g., local tumor control, metastatic relapse, biochemical relapse, cause-specific or overall mortality), and a particular factor may have significance for some, but not all, endpoints.

Patient-Related Factors

Patient age had been identified as a prognostic factor in older series of patients managed with surgery, RT, or expectant management.^223–227 These reports have offered conflicting results, suggesting that younger patients did better with expectant management in some series^224,228 whereas older patients fared worse in others.²²³ Similarly, some series suggest a worse prognosis with surgery or RT for younger patients,^{225,227,229–231} whereas others do not.^232–238 A meta-analysis of 34 studies with more than 27,000 patients did not show an association between age and outcome in the PSA era.²³⁹ Parker and associates²³⁹ speculated that PSA screening may have eliminated the impact of age on prognosis owing to less delay in diagnosis, improved measures of outcome, a screening-related age-specific lead-time bias, and an interaction between age and tumor grade.

Racial and ethnic disparities for the outcome of patients after diagnosis and treatment exist for many malignancies, including prostate cancer. African Americans are more likely to be diagnosed with advanced stage and higher-grade prostate cancer than whites or non–African-American minorities.^1,138 The survival rate of African-American men is inferior to that of non-Hispanic white men with localized prostate cancer.^42,138 The reason for these disparities is unclear and may be related to variations in genetics, nutrition, socioeconomic status, clinical and surgical staging procedures applied to different races, access to care, or inherent differences in cancer biology. SEER data have shown a difference in the rate of RP and surgical staging being performed in African-American men with localized prostate cancer.²⁴⁰ In a secondary analysis of Radiation Therapy Oncology Group (RTOG) clinical trial data in which eligibility and treatment guidelines were prescribed according to study specifications, Roach and colleagues²⁴¹ reported that the observed inferior survival of black Americans was likely related to their presentation with more advanced disease and not due to inherent biologic differences in their cancer. In a review of Southwest Oncology Group (SWOG) trials,²⁴² the survival of African-American men with advanced-stage prostate cancer was worse than all other patients, even when corrected for differences in socioeconomic status. Two other studies^243,244 suggest that even when effects of age, presenting PSA value, Gleason score, and stage are controlled, a difference in progression-free survival persists. Moul and colleagues²⁴⁵ demonstrated differences in outcome in men treated with RP in an “equal access” medical system (i.e., the U.S. military). There is recent evidence for differences in presentation and outcome for Hispanic men with localized prostate cancer,²⁴⁶ but few to no reliable estimates exist for the Native American/Native Alaskan or Asian/Pacific Islander populations. Therefore it appears that some evidence suggestive of biologic differences in patient subgroups is emerging.

The presence of coexistent medical ailments (e.g., ischemic heart disease) in a patient with prostate cancer may have a significant impact on treatment decisions. These comorbidities may determine whether a patient is a candidate for an aggressive surgical intervention or whether the patient should be treated at all. These non–prostate cancer–related treatment selection criteria have a profound impact on outcome reporting that are unrelated to the effectiveness of any specific treatment strategy. Several instruments are available to quantify the severity of comorbidity. Albertsen and colleagues²⁴⁷ applied three comorbidity assessment instruments (i.e., the Kaplan-Feinstein and Charlson indices and the index of coexistent disease) to a cohort of men with localized prostate cancer to assess the impact of comorbidity on patient survival. Although the index of coexistent disease had a marginally better predictive value, each instrument was predictive of age-adjusted survival when corrected for the effect of Gleason score. The risk of death from causes other than prostate cancer (i.e., competing causes) strongly correlated with the severity of comorbidity, and a weak association with prostate cancer-related (i.e., cause-specific) mortality was also noted. The selection of a particular management approach based on overall health status has important implications in interpreting and comparing outcomes, because overall and, perhaps, cause-specific survival estimates will be influenced.^248–251 Additional research is essential to quantify the confounding effect of comorbidity in the setting of prostate cancer outcomes reporting. A nomogram has been described that may allow estimation of life expectancy and aid physicians and patients in choosing treatment strategies.²⁴⁹

Tumor Stage

The anatomic extent of prostate cancer is an important tumor-related factor predictive of patient outcome. Studies consistently demonstrate the association of primary tumor stage with several endpoints in patients with localized disease. Cause-specific and overall survival are directly related to the extent of the primary tumor in both RT^252–254 and expectant management^224,255,256 series. The probability of local tumor control²⁵⁷ and the risk of clinical,^253,258 metastatic,^257,259 and biochemical relapse^{233,235,236,254} are influenced in a similar manner. Increasingly, tumor stage is combined with other prognostic factors (e.g., initial PSA, tumor grade) to define prognostic groups and report outcomes. Despite this, the tumor stage remains an important factor in selecting optimal local therapy.

Surgical Margins

Modifications to surgical technique have resulted in an improvement in the rate of positive margins with an open prostatectomy technique²⁶⁰; however, even with laparoscopic and robot-assisted procedures the rates of positive margins remain as high as 50% and 28%, respectively.²⁶¹ In a systematic review of 37 comparative series, Ficarra and associates²⁶¹ concluded that laparoscopic and robot-assisted laparoscopic prostatectomy had comparable positive margin rates as open prostatectomy. Positive margins are typically located at the apex (48%), rectal and lateral surfaces (24%), bladder neck (15%), and posterior pedicles (10%).^180,262 The presence of a positive surgical margin increases the likelihood of biochemical recurrence, local recurrence, and the need for salvage therapies.^263–265 In a series of 11,729 men undergoing RP between 1990 and 2006, Boorjian and associates²⁶³ reported positive surgical margins in 31% of cases. Patients with positive margins had a biochemical relapse free survival (RFS) of only 56% compared with 77% (p <.0001) and a local RFS of 89% compared with 95% (p <.0001) in patients with negative margins. The patients with positive surgical margins also had a significantly worse systemic progression-free survival (PFS), cancer-specific survival (CSS) and overall survival (OS). In a pooled series of 7816 patients from eight academic urology centers, the presence of a positive surgical margin increased the risk of biochemical recurrence by 3.7-fold.²⁶⁵

It has been suggested that an apical positive margin may actually represent an artifact in the processing of prostatectomy specimens and it may not carry as great a prognostic significance as margins involved elsewhere.^266,267 Analyses of randomized trials evaluating the role of adjuvant RT failed to show any differences in outcome by site of surgical margin involvement.^268,269

The extent of surgical margin involvement has been reported to carry prognostic significance. Van Oort and colleagues²⁷⁰ reported a higher rate of biochemical relapse in patients when the length of a positive surgical margin was more than 10 mm compared with smaller margins. In a series from the Mayo Clinic, Kausik and colleagues²⁷¹ did not detect a difference in relapse rates in patients with EPE and one versus more than one positive margins. In a series of 2,468 patients, Lake and associates reported that even focally positive margins (≤3 mm) in patients with organ-confined disease conferred a significant decrease in disease-free survival (DFS).²⁷² The 10-year DFS for patients with organ-confined disease was 84%, 64%, and 38% with negative, focally positive, and extensively positive margins (>3 mm), respectively (p <.0001).²⁷² These data indicate strongly that the presence of a positive surgical margin has a profound impact on patient prognosis and that these patients should be considered for adjuvant therapy.

Nodal Involvement

Tumor spread to regional lymph nodes is associated with a particularly poor outcome after RT or surgical management alone.^273–278 The prognostic significance of the number of involved nodes was identified in some.^278–281 In a series of more than 10,000 patients treated with RP, Boorjian and colleagues²⁷⁸ reported that the risk of systemic cancer progression was threefold greater with one positive node and another twofold greater with two or more positive nodes. One positive node increased the risk of cancer-related death fourfold, and when two or more nodes were involved the risk was another twofold higher. Barzell and colleagues²⁸² quantified nodal cancer volume and evaluated its association with disease outcome. Schmid and associates²⁸⁰ found that patients with micrometastasis (i.e., ≤5 mm in largest dimension) had significantly better PFS and OS compared with patients with more extensive nodal involvement. In contrast, Srignoli and colleagues²⁸³ found no difference in the risk of subsequent distant metastasis. Cheng and colleagues²⁸⁴ reported that the risk of distant metastasis in patients with regional nodal involvement was directly proportional to nodal cancer volume and that it was the best predictor of 5-year metastasis-free survival. Extranodal extension of disease, as well as Gleason score of the primary tumor, may also be significant predictors of CSS.²⁸⁵ In a series of 102 patients with pathologically involved lymph nodes at RP, Fleischmann and colleagues²⁸⁶ reported inferior outcome if extranodal extension was present, but this was strongly correlated to tumor volume and it was not an independent prognostic factor.

Tumor Grade

Long-term case series of EBRT,^235,253,287 brachytherapy,^288–290 RP,^291–294 and expectant management^224,255 uniformly identify tumor grade as a strong predictor of disease relapse and mortality in clinically localized prostate cancer. Studies consistently demonstrate that patients with a poorly differentiated tumor (i.e., grade 4-5 or Gleason score ≥7) have an increased risk of metastatic disease progression and reduced OS and disease-specific survival (DSS).^255,257,295 For the patient with clinical T1 or T2 disease, tumor grade has greater predictive value than does the distinction between T1 and T2 categories.* The association of tumor grade with metastatic relapse^235,257,258 and OS²⁵⁷ is preserved when pretherapy serum PSA level is considered. Although tumor grade is also associated with DFS and freedom from clinically evident disease relapse,^{225,235,296,297} its impact on local tumor control after EBRT is less certain, because some reports noted an association^252,257 whereas others did not.²⁹⁶ PSA level (i.e., biochemical control) after local therapy is impacted by tumor grade. Multivariate analysis of pretherapy tumor-related factors in RT^235,236,257 and surgical^293,295,298 series confirms the significance of tumor grade as a predictor for biochemical relapse.

The relative amount of high-grade cancer in biopsy and surgical specimens predicts for clinical and biochemical outcomes.^299,300 In patients with Gleason score 7, a primary component of Gleason grade 4 (4 + 3) has been associated with a higher rate of metastatic, biochemical, and clinical progression than a primary Gleason grade 3 component (3 + 4) in both surgical^301–303 and RT series.³⁰⁴ However, whereas some series demonstrate that the presence of the higher primary Gleason grade is correlated with other adverse prognostic factors such as tumor volume, SVI, or positive surgical margins,³⁰³ it is not independently associated with a worse outcome in some surgical³⁰³ or RT series.^305,306 Increasingly, it is being recognized that tertiary patterns of high-grade adenocarcinoma (Gleason grades 4 or 5) negatively impact prognosis.^{300,307–309} Patients with cancers with a Gleason score of 7 with any higher-grade tertiary pattern have a prognosis after RP that is indistinguishable from patients with a Gleason score of 8.³⁰⁷

Prostate-Specific Antigen

An abundant literature attests to the importance of the pretherapy serum PSA level as an independent and highly significant predictor of therapeutic outcome.† In general, the pretreatment PSA level is a measure of tumor volume, although one must also consider the relative PSA contribution from the normal prostate tissue and the fact that some tumors, especially the poorly differentiated ones, may not produce PSA commensurate with the tumor burden. In assessment of post-RT outcome, Landmann and Hunig³¹² were the first to draw attention to this association, which was confirmed by the detailed statistical analysis of larger study populations.‡ Furthermore, the prognostic significance of PSA is not limited to patients treated with RT, because similar observations appear in surgical reports.^§

Pretreatment PSA is one of the strongest individual predictors of biochemical relapse.* The effect of pretreatment PSA on outcome increases in a continuous fashion.^236,257 Zagars and colleagues²⁵⁷ described a relationship between pretherapy PSA and clinical and/or biochemical relapse after RT. In this series there was a significant difference in the risk of relapse for pretherapy PSA categories of 4.0 or less, 4.1 to 10.0, 10.1 to 20.0, and more than 20 ng/mL, which was 16%, 34%, 51%, and 89% at 6-year follow-up, respectively. D’Amico and associates³¹⁸ described a risk grouping based on pretreatment PSA, Gleason score, and clinical stage. A slightly modified risk group category has been adopted by the National Comprehensive Center Network (NCCN) and is used for reporting outcomes and to assist in treatment decision making³¹⁹ (Table 51-5). Most institutional and other large series now report outcomes by risk stratification group.

TABLE 51-5 National Cancer Center Network Prostate Cancer Risk Grouping: 2010

Although numerous studies relate pretherapy PSA level to biochemical relapse, reports using clinical endpoints are not as abundant.^257,258,320 Nonetheless, Crook and colleagues³²¹ affirmed that local tumor recurrence correlated with pretreatment PSA in patients in whom serial post-RT prostatic biopsies were performed. Zagars and associates²⁵⁷ reported a similar finding in a study population in whom post-RT prostatic biopsies were generally obtained to identify an otherwise unknown source for biochemical relapse. Studies with long enough follow-ups to document clinical disease relapse (local and distant) are now being reported and correlate clinical endpoints with pretreatment PSA levels.^235,291,297

In an effort to improve the prognostic accuracy of pretherapy serum PSA, many investigators have examined derivatives of the serum PSA, including PSAD (a correction for benign prostate gland volume) or PSA kinetics (velocity of rise or doubling time). Zentner and associates³²⁰ evaluated PSAD, that is, the PSA value divided by prostatic volume, in patients treated with EBRT. Patients with a PSAD of 0.3 ng/mL/cm³ or less had a 100% DFS probability at 30 months, whereas the estimate for patients with a PSAD of more than 0.3 ng/cm³ was significantly worse (62%). Others have not shown this association, and PSAD does not appear to add significantly to the utility of serum PSA level as a predictor of outcome after EBRT, and it is not routinely used as a prognostic factor.^322–324

The rate of change of the PSA over time has important diagnostic and prognostic implications. Carter and colleagues reviewed the outcomes of men participating in the Baltimore Longitudinal Study of Aging for up to 39 years. PSA velocity measured 10 to 15 years before a diagnosis of prostate cancer was correlated with prostate CSS 25 years later.³²⁵ The prostate CSS was 92% in men with a PSA velocity of 0.35 ng/mL/yr or less compared with 54% with a PSA velocity of more than 0.35 ng/mL/yr (p <.001). D’Amico and associates reported that the risk of prostate cancer death in men undergoing RP³²⁶ or EBRT³²⁷ for localized prostate cancer was higher if the PSA velocity was more than 2.0 ng/mL/yr.

The PSA doubling time (PSADT) is as a useful tool to estimate prognosis and guide treatment decisions. Derived from the same data as PSA velocity (PSAV), which measures the absolute rate of change, PSADT is calculated as the slope of PSA measurements over time. PSADT represents the relative rate of PSA change over time and is the time needed for the PSA value to double. PSADT takes into account the exponential nature of prostate cancer growth and requires logarithmic conversion using the formula:

A number of PSADT calculators are available for use on handheld computing devices or from websites (www.nomograms.org). Klotz and colleagues²⁵⁶ have proposed that PSADT be used as a factor in the selection of patients for active surveillance. A PSADT of 3 years or less is associated with an 8.5-fold greater risk of biochemical failure after definitive treatment than longer PSADT. Hanks and colleagues³²⁸ used PSADT as a pretherapy predictive factor in patients treated with EBRT and found that a PSADT of less than 12 months adversely affected therapeutic outcome. Most reports have focused on post-treatment PSADT to predict the clinical outcome of patients with biochemical relapse after RT^{237,296,329–333} or surgery.^332–335 Although there is considerable individual variation, a local tumor recurrence tends to correlate with a longer PSADT (median, 12 to 14 months), whereas a shorter PSADT (median, 4 to 6 months) is observed for metastatic relapse.^314,336 The post-treatment PSADT in patients who have failed treatment with RT is similar to that observed in surgical patients.³³⁷

After RT, residual prostatic epithelial cells may produce PSA under androgenic stimulation and serum levels of PSA may be detected even in the absence of prostate cancer.³³⁸ The post-RT nadir PSA (nPSA) level is considered to be a predictive variable similar to pretherapy factors.^{235,331,339–341} The nPSA correlates with pretherapy PSA level, tumor grade, and clinical tumor stage.^339,342 Nonetheless, nPSA is an independent predictor associated with biochemical and/or clinical relapse,^{314,339,340,342,343} and with local and metastatic tumor control.^320,331,340 Alcantara and associates³⁴⁴ reported a 10-year distant metastasis rate of 4% versus 19% in men with a 12-month nPSA of less than or equal to 2 ng/mL versus greater than 2 ng/mL (p <.0001). Zelefsky and colleagues³⁴⁰ reported the incidence of distant metastases was 10.3% and 17.5% at 10 years in men who had a 2-year nadir value of less than or equal to 1.5 ng/mL versus greater than 1.5 ng/m (p <.002). Although nadir level can be used to predict the relative likelihood of failure after radiation, there is no absolute level that can determine success or failure after this therapy. Kuban and colleagues²³⁵ in a multi-institutional study of 4839 patients treated by EBRT for prostate cancer demonstrated that progressively lower nadir levels were associated with incremental improvement in PSA disease-free survival.

PSA Failure Definitions after Local Therapy

PSA is commonly being used to assess outcome after local therapy, because clinical failure, especially after treatment of localized disease, takes years to develop. In evaluating new treatment methods and techniques it is desirable to obtain comparative efficacy information as soon as possible. Therefore a consistent, reliable, and accurate method of using PSA as a marker for failure is necessary.

PSA failure definitions after surgery or RT must use different criteria because of the inherent effect on the prostate by each of these modalities. Because the goal of RP is to remove all prostatic tissue, a detectable PSA level after surgery has been correlated with outcome. Although an undetectable PSA level is desirable, the possibility of residual benign glands capable of producing low levels of serum PSA has been suggested. The PSA level is very low, in the 0.1 to 0.4 ng/mL range, and is representative of a small amount of prostate tissue left behind.^315,345,346 Although these levels have been correlated to postoperative relapse, further proof of progressive disease is evidenced by a subsequent rising PSA value. Cutpoints of 0.2 ng/mL and 0.4 ng/mL have been proposed to report biochemical failure after surgery. In each case, patients with PSA values greater than the cutpoint have higher rates of subsequent rising PSA or clinical failure.^315,345,346 The AUA has recommended defining biochemical recurrence as an initial serum PSA greater than or equal to 0.2 ng/mL with a confirmatory PSA greater than 0.2 ng/mL.³⁴⁷ For irradiated patients, a rising PSA level is the criterion for relapse that is often used, because a very low target PSA value, especially soon after treatment, is not very specific in predicting clinical failure.^348,349 This is because the prostate remains intact in irradiated patients and it takes a median of nearly 2 years for the PSA value to reach the nadir.³⁵⁰ Simultaneous with this gradual declining process, a PSA “bounce” can occur, mimicking failure because the “bounce” is a PSA rise but one secondary to a benign cause, likely prostatitis.^351–353 Therefore the definition of PSA relapse after RT must take both of these factors into account to minimize false-positive failure reporting.

The American Society of Radiation Oncology (ASTRO) held a Consensus Conference in 1996 for the purpose of devising a PSA-based failure definition after EBRT that would standardize outcome reporting. The ASTRO definition for biochemical failure is characterized by three consecutive PSA rises with the time of failure backdated to halfway between the PSA nadir and the first PSA rise.³⁵⁴ After several additional years’ experience in the PSA era and more sophisticated statistical analysis, criticisms of the ASTRO definition were brought forward and alternative definitions that were more sensitive and specific were advocated.^348,355 Problems with the original PSA definition included biases associated with back dating and the time necessary to collect three consecutive rises. The original ASTRO consensus also did not consider the use of neoadjuvant or adjuvant hormone therapy, nor did it include data from alternative radiation techniques such as permanent or temporary brachytherapy.

A second Consensus Conference sponsored by both ASTRO and RTOG was held in 2005 at which several investigators presented analyses with more long-term follow-up than was available at the first conference.³⁵⁶ A panel representing RT and surgical specialties as well as clinical trials groups formulated a revised consensus statement, commonly referred to as the “Phoenix” or nadir+2 definition. The revised definition states that a PSA rise of 2 ng/mL or more above the nadir PSA (defined as the lowest PSA achieved) is a biochemical failure. The date of failure is the date at which that PSA value is determined “at call” and not backdated.³⁵⁶ Patients not meeting these criteria but who undergo salvage therapy such as androgen deprivation, prostatectomy, brachytherapy, or cryosurgery are considered as having treatment failure at the time of a positive biopsy or when salvage therapy is initiated. Based on clinical outcomes reported by Kestin and colleagues,³⁵⁵ the original ASTRO definition had a sensitivity of 51%, specificity of 78%, positive predictive value of 31%, and negative predictive value of 88%. The Phoenix definition has a sensitivity of 64%, specificity of 78%, positive predictive value of 36%, and negative predictive value of 92%.³⁵⁶ Another important point stressed at both consensus conferences is the minimum follow-up necessary to use these definitions. The stated date of control should not be more than 2 years short of the median follow-up time. For example, to report a 5-year biochemical control rate, the median follow-up should be at least 7 years. The original ASTRO definition still has important utility because it can be used to compare results to prior publications. The consensus panel encouraged reporting results using both the Phoenix and ASTRO definitions.³⁵⁶

Predictive Models

The previous sections on prognostic factors emphasized the interactions many of them have with each other regarding patient outcomes. Furthermore, comparing treatment between modalities or between institutions and series is complicated by the fact that the patient populations vary considerably between them. It has become increasingly important to utilize predictive models to account for variable prognostic factors and allow more meaningful comparisons between heterogeneous populations of patients.

Numerous predictive models exist to assist decision making in many aspects of prostate cancer screening, diagnosis, staging, and treatment. These models include nomograms, risk groupings, probability tables, and artificial neural networks. Each of these predictive models vary on their levels of discrimination, calibration, generalizability, and complexity. Discrimination describes the ability to identify patients that will or will not experience the outcome of interest. Calibration refers to the ability to generate predictions that closely approximate observed outcomes, that is, given a risk of X% experiencing an event then close to X% actually experience it. Although there are methods to internally validate them, ideally predictive models should be externally validated by a dataset other than that used to create the model. This validation step speaks to the generalizability of the model—how likely the prediction is to be accurate in similar groups of patients at different centers or at different points in time.

Risk groupings are relatively easy to use and are based on a limited number of clinical variables that allow categorization of patients by their likelihood of recurrence. One of the most commonly used risk grouping was described by D’Amico and associates to estimate the biochemical outcome of men after RP, EBRT, or interstitial brachytherapy.³¹⁸ Patients are classified into three categories based on their presenting PSA, Gleason score, and clinical stage: low risk (stage T1c or T2a or PSA ≤10 ng/mL or Gleason score ≤6), intermediate risk (stage T2b or Gleason score = 7 or PSA >10 and ≤20 ng/mL) or high risk (stage ≥T2c or PSA >20 ng/mL or Gleason score ≥8). The NCCN has adopted a similar classification for treatment recommendations except stages T2b to T2c are considered intermediate risk and stages T3a to T4 represent high risk.³¹⁹ The University of California, San Francisco, group has described a risk grouping model that includes the percentage of positive biopsy cores and patient age. The Cancer of the Prostate Risk Assessment (CAPRA) score is calculated by assigning points to increasing risk factors. Unlike the groupings of D’Amico and associates, the PSA level is divided into five discrete categories. The CAPRA score may range from 0 to 10, although 0 to 1 and 7 to 10 groups are often combined. The CAPRA score has been validated as a useful tool to predict outcomes after RP.³⁵⁷ One problem with risk groupings is that each category may contain a heterogeneous population and continuous variables are converted into categorical variables, thereby removing the information about the actual value. It also assumes each variable carries equal weight in contributing to patient outcome. Probability tables have been used to estimate pathologic findings after RP and are commonly used in multidisciplinary settings to estimate the risks of extracapsular extension, SVI, or LNI. Partin and associates²⁰² popularized the use of probability tables based on a series of more than 5000 men undergoing RP at Johns Hopkins University. The “Partin tables” combine serum PSA, clinical stage, and biopsy Gleason score to predict the pathologic stage in four groups: organ-confined disease, established extracapsular extension, SVI, or LNI. The outcomes are considered mutually exclusive so the tables tend to underestimate the risk of extracapsular extension or SVI because a significant proportion of patients with LNI will also have these findings.^202,203 Nomograms are graphic calculation devices that incorporate at least two continuous or categorical variables. Risk points are determined by identifying the value of variables of interest on axes corresponding to a given prognostic factor. The risk points are then summed and used to estimate the combined effect of the predictors on the expected outcome. Nomograms often use Cox proportional hazards or modified logistic regression to correct for variable impact of incremental changes across a broad range of values to predict endpoints.³⁵⁸ Nomograms can be adapted to web-based or handheld personal data assistant applications to quickly and conveniently calculate predicted outcomes (see www.nomograms.org). Numerous nomograms have been developed to calculate the risk of positive biopsy results on screening examinations, estimate pathologic stage, and predict biochemical control before surgery,³⁵⁹ EBRT,³⁶⁰ or brachytherapy³⁶¹ as well as the risk of prostate cancer progression after local therapy.^362,363 Examples of nomograms for predicting biochemical DFS after RP and EBRT are presented in Figures 51-2 and 51-3.

Figure 51-2 Preoperative nomogram estimating the 1- to 10-year progression-free probability after radical prostatectomy alone.

Data from Stephenson AJ, et al: Preoperative nomogram predicting the 10-year probability of prostate cancer recurrence after radical prostatectomy, J Natl Cancer Inst 98:715-717, 2006.

Figure 51-3 Nomogram for 60-month probability of metastasis, recognizing the potential for death from other causes. PSA, prostate-specific antigen.

Data from Kattan MW, et al: Pretreatment nomogram that predicts 5-year probability of metastasis following three-dimensional conformal radiation therapy for localized prostate cancer, J Clin Oncol 21:4568-4571, 2003.

Clinical Symptoms and Signs

Most early-stage prostate cancer patients are asymptomatic because most cancers arise in the peripheral zone of the gland. However, a tumor may arise in the transition zone or may extend from a peripheral location to encroach on the prostatic urethra and produce obstructive symptoms. The resulting symptoms, commonly referred to as prostatism, initially consist of urinary hesitancy, diminished force of stream, intermittency, and a sense of incomplete voiding. Thereafter, the bladder detrusor muscle may lose compliance and irritative symptoms such as urgency, frequency, nocturia, dysuria, and incontinence may develop. Although acute urinary retention may ensue, it is no more likely to occur in the patient with prostate cancer than in the general population.^364,365 Indeed, the constellation of obstructive symptoms now referred to as lower urinary tract symptoms are common in the elderly male population³⁶⁶ and are often associated with benign prostatic hyperplasia or other mechanical or functional disorders of the urinary bladder. Hematuria, which may result from tumor involvement of the prostatic urethra or bladder trigone, and hematospermia are uncommon signs of prostate cancer and should prompt further investigation to identify their cause.

Local extraprostatic tumor spread is often asymptomatic and may not be apparent on physical or radiologic examination. However, extensive disease involvement of the periprostatic tissues may produce symptoms as a late manifestation of the disease process. Denonvilliers fascia is a barrier to rectal involvement, but symptoms similar to those of a primary rectal cancer, such as hematochezia, constipation or intermittent diarrhea, reduced stool caliber, or abdominopelvic pain, may occur. Renal impairment as a result of prolonged bladder outlet obstruction or from tumor involvement of the bladder trigone or periureteral tissues may cause hydronephrosis and symptoms or signs due to fluid retention or electrolyte imbalance. Invasion of the neurovascular bundles, urogenital diaphragm, or penile corporeal bodies may result in impotence, perineal pain, or priapism, respectively.

The pelvic, axial, and proximal regions of the appendicular skeleton are the most common sites of metastatic disease spread. Based on the distribution of uptake on radionuclide bone scan, the vertebral column (74%), ribs (70%), pelvis (60%), femora (44%), and shoulder girdle (41%) are the prevailing sites of involvement in patients with osseous metastases.³⁶⁷ Although approximately one third to one half of those with bone metastases may be asymptomatic, pain in these areas is commonly described or may be elicited through examination of the skeletal system. The extent of osseous metastasis and the presence of symptoms³⁶⁸ are associated with responsiveness to therapeutic intervention and survival. Pathologic fractures are uncommon but may affect the femora, humeri, and vertebral column. Extension of tumor into the epidural space or compression of nerve rootlets or peripheral nerves may cause neurologic deficits. Spinal cord compression is characterized by motor weakness, sensory changes, and bladder and bowel dysfunction. Its presence is considered an oncologic emergency that requires prompt recognition and treatment to avoid permanent disability.

Tumor involvement of pelvic or para-aortic lymph nodes is usually asymptomatic, and clinical signs, such as edema of the abdominal wall, genitalia, and lower extremities, are rarely apparent unless bulky nodal disease is present. Although metastases to other anatomic sites (e.g., adrenal glands, liver, lung, and skin) are described,³⁶⁹ symptoms or signs due to functional compromise are uncommon. Paraneoplastic syndromes, including Cushing syndrome, the syndrome of inappropriate antidiuretic hormone secretion, and abnormalities of serum calcium levels, and hematologic manifestations (e.g., anemia, disseminated intravascular coagulation) may also be present in the patient with advanced metastatic disease.³⁷⁰

History and Physical Examination

A thorough history and physical examination are fundamental components of patient evaluation and may be instrumental in directing further diagnostic and staging procedures. In particular, detailed inquiry regarding urinary status, sexual function, and skeletal symptoms identifies the patient’s baseline condition and may prompt further investigation and therapy. Assessment of comorbid illness may have relevance to therapeutic decision making, and the family medical history may identify other family members at high risk for developing prostate cancer.

A complete physical examination is essential and may reveal metastatic disease or other clinically important conditions. Careful DRE of the prostate forms the cornerstone of the physical examination and is instrumental in tumor staging. The examination may be performed in the lithotomy, knee-chest, lateral prone (Sims), or standing position with a well-lubricated, gloved index finger. The examiner should measure the craniocaudad and transverse dimensions of the gland and determine its overall consistency, sensitivity, and mobility, as well as the presence and size of any firm or elevated areas. Typically, an area of prostate cancer is quite hard, may or may not be raised above the surface of the gland, and is surrounded by compressible prostatic tissue. However, an area of prostatic induration per se is not pathognomonic of cancer, as induration may be the result of prostatic calculi, infection, infarction, granulomatous prostatitis, or nodularity owing to benign prostatic hyperplasia. Although DRE is important in staging, it is not particularly accurate in screening. In a meta-analysis of 13 prostate cancer screening studies, the pooled sensitivity, specificity, and positive predictive value for DRE were 53.2%, 83.6%, and 17.8%, respectively.³⁷¹ The lateral rectal sulci and the urogenital diaphragm are also examined to determine whether extraprostatic tumor extension is present. Although the normal seminal vesicles are too soft to be palpated as discrete structures, firmness extending superior to the prostate gland suggests seminal vesicle involvement, and further investigation may be warranted.

Staging

Whitmore³⁷² is credited with developing the first widely accepted stage groupings for prostate cancer, and the general concepts set forth were later incorporated into the AUA system.³⁷³ Subsequently, the Organ Systems Coordinating Center of the National Cancer Institute initiated a process to align the Whitmore-based groupings and the TNM system.This and other efforts led to development of the AJCC TNM classification and stage groupings,³⁷⁴ which were unified with the UICC in 1987.¹⁷⁸ This system retains the general organizational scheme of prior systems and conforms to the TNM rules applied to tumors originating in other sites.

The AJCC TNM classifications and the stage groupings for prostate cancer are shown in Tables 51-3 and 51-4, respectively. Clinical staging involves primary tumor assessment with DRE, complete general examination to determine disease extent, serum PSA determination, and diagnostic imaging (including TRUS). The regional lymph nodes are defined as those confined to the true pelvis (i.e., those lying below the bifurcation of the common iliac arteries) and include the obturator, external and internal iliac, and sacral lymph node groups. All information available before definitive therapy may be used for clinical staging. Pathologic staging requires histologic examination of the prostatoseminovesiculectomy and pelvic lymph node specimens; presence of EPE, SVI, and margin status, as well as the extent and location of involved lymph nodes should be specified.³⁷⁵

Diagnostic Imaging

Nuclear Medicine Imaging

Bone Scintigraphy

Clinically apparent metastatic disease is limited to bone in 80% to 85% of patients with prostate cancer metastases and is observed in a pattern that correlates with the distribution of hematopoietic marrow in the adult. The radionuclide bone scintigraphic appearance of metastases may range from a solitary focus of radionuclide uptake to the “super scan,” in which the entire skeleton is uniformly affected. Recognition of the latter condition, distinguished by the lack of renal uptake and excretion of the radionuclide, is important, because the scan may otherwise be interpreted as normal. Although bone scintigraphy is a sensitive imaging tool, various other conditions, such as arthritis, bone fracture, and Paget disease, may also result in uptake of the radionuclide. In certain situations, standard bone radiography or tomography, CT, or MRI may be required to determine the nature of the findings noted on bone scintigraphy.

Since the advent of serum PSA testing, Chybowski and associates³⁹⁷ were among the first to evaluate the utility of obtaining a routine bone scintiscan in the pretherapy evaluation of the patient with newly diagnosed prostate cancer. These investigators conducted a retrospective evaluation of 521 randomly selected patients and correlated clinical tumor stage, tumor grade of the diagnostic biopsy specimen, acid phosphatase, PAP, and PSA serum levels with bone scintiscan findings. The probability of a bone scintiscan showing metastatic disease was most strongly associated with and was directly proportional to the serum PSA level, as illustrated in Figure 51-4. Only 0.3% of patients with a pretherapy PSA of 20 ng/mL or lower had a positive bone scintiscan at presentation.³⁹⁷

Figure 51-4 Results of bone scintiscan findings as a function of pretherapy serum prostate-specific antigen (PSA) level; 95% confidence intervals are shaded.

From Chybowski FM, Larson Keller JJ, Bergstralh EJ, et al: Predicting radionuclide bone scan findings in patients with newly diagnosed, untreated prostate cancer. Prostate-specific antigen is superior to all other clinical parameters, J Urol 145:313, 1991.

In a study of 852 consecutive patients with newly diagnosed untreated prostate cancer with a pretherapy serum PSA level of less than 20 ng/mL, Oesterling and colleagues³⁹⁸ reported that only 7 patients (0.8%) had scintigraphic evidence of bone metastases, and five of these patients were symptomatic in the area corresponding to the bone scan abnormality. The observed rates of false-negative bone scan results were 0.5% and 0.8% for patients with serum PSA levels of 10 ng/mL or less and less than 20 ng/mL, respectively. On the basis of these observations, the authors concluded that a bone scan was not routinely required in the staging evaluation of the patient without skeletal discomfort if the pretherapy serum PSA level was 10 ng/mL or less. Furthermore, this investigation confirmed the finding that serum PSA was most closely associated with the bone scintigraphy results and that neither clinical tumor stage nor tumor grade improved the predictive accuracy of PSA alone. Oesterling and colleagues³⁹⁸ suggest that considerable economic savings could be realized if the use of a radionuclide bone scan were reserved for the symptomatic patient or for those with PSA levels above 10 ng/mL or (possibly) a high-grade tumor. Based on a similar retrospective analysis of 853 consecutive patients, Briganti and associates³⁹⁹ developed a classification and regression tree model that can be used to predict the frequency of a positive bone scintiscan. According to this tool, a staging bone scintiscan should be considered only for patients with a Gleason score greater than 7 or with a PSA greater than 10 ng/mL and palpable disease.

Radionuclide bone scintigraphy may also be of value in the post-therapeutic assessment of the patient with prostate cancer. The onset of skeletal symptoms, particularly in association with a rising serum PSA profile, should prompt radioscintigraphic investigation. However, post-therapeutic serum PSA monitoring is likely to identify tumor relapse before the occurrence of symptoms, because biochemical relapse may antedate development of bone scintigraphic abnormalities by several months. Freitas and colleagues⁴⁰⁰ used receiver operating characteristic curve analysis to identify a serum PSA threshold level at which a post-therapeutic bone scintiscan might be considered. This analysis indicated that few patients (1.5%) with a post-therapy PSA value of 8 ng/mL or less had a positive bone scintiscan, whereas 38% of patients with a PSA level greater than this value had a bone scintiscan indicative of osseous metastases. In a series of 239 patients evaluated with bone scintigraphy for biochemical relapse after RP, Dotan and colleagues⁴⁰¹ reported an incidence of bone metastases of 4%, 36%, 50%, and 79% with trigger PSA values of 0 to 10, 10.1 to 20, 20.1 to 50, and above 50 ng/mL, respectively. In a multivariate analysis, PSA slope, PSA velocity, and trigger PSA were all significant factors for predicting a positive bone in the evaluation of biochemical failures after prostatectomy. The group developed a nomogram to predict the likelihood of a positive bone scintiscan using preoperative and pathologic tumor features, along with postoperative PSA kinetics.

Monoclonal Antibody Imaging

¹¹¹In-labeled capromab pendetide (ProstaScint) is an antibody directed against a prostate-specific membrane antigen.⁴⁰² The image performance of ProstaScint has been assessed in the pretherapy evaluation of patients with clinically localized prostate cancer who are considered at high risk for disease spread to regional nodes and in the post-therapy setting when occult residual or recurrent disease is suspected. Surgical series suggest the diagnostic accuracy in the 70% range.^403,404 In a series of 255 men undergoing RP, Raj and associates⁴⁰⁵ reported that ProstaScint was able to differentiate between locally confined and metastatic disease, to detect soft tissue defects as small as 5 mm, and to have a sensitivity, specificity, and positive predictive value of 73%, 53%, and 89%, respectively. In the setting of a rising PSA value after prostatectomy, ProstaScint imaging has been used to evaluate patients for the presence of local or lymph node recurrence of disease after prostatectomy. The reported value of ProstaScint imaging has been inconsistent, with some investigators reporting improved patient selection and outcomes for salvage therapy of a presumed localized recurrence^406,407 whereas others have reported no difference in outcome in patients with localized versus distant tracer uptake.^408–410

Positron Emission Tomography

Positron emission tomography (PET) uses positron-emitting isotopes conjugated to compounds to detect physiologic or pathologic processes. The most common PET agent used to date is ¹⁸F-fluorodeoxyglucose (FDG), a glucose analogue that is taken up by up-regulated glucose transporters in many cancers. FDG-PET in prostate cancer has been disappointing, with poor sensitivity and specificity for staging the local and regional disease owing to poor uptake of FDG in slowly proliferating prostate cancer as well as urinary excretion causing bladder uptake to obscure pelvic disease.^411,412 Newer PET agents of ¹¹C-labeled methionine, acetate, or choline are showing promise in the staging of high-risk prostate cancers.⁴¹³

Laboratory Evaluation

Staging Workup

The ACR³⁸⁹ and the NCCN have established guidelines for the pretherapy staging of patients with a recently established diagnosis of prostate cancer. The scheme proposed by the NCCN³¹⁹ is displayed in Figure 51-5 and may be used for guidance. However, individual clinical circumstances may require departure from these guidelines, because independent medical judgment is necessary to meet needs unique to a particular situation.

Figure 51-5 Diagnostic algorithm guidelines for the initial staging workup. The National Comprehensive Cancer Network (NCCN) guidelines constitute a consensus statement by its authors on the accepted approaches to treatment. Any clinician seeking to apply or consult any NCCN guideline is expected to use independent medical judgment in the context of individual clinical circumstances to determine any patient’s care or treatment. The NCCN makes no warranties of any kind regarding the guidelines’ content, use, or application and disclaims any responsibility for their application or use in any way.

From National Comprehensive Cancer Network: Prostate Cancer NCCN Clinical Practice Guidelines for Oncology, NCCN 2010.

Expectant Management

Expectant management is a clinical strategy in which immediate therapeutic intervention is not pursued when the diagnosis of prostate cancer is first established. With this approach, treatment is withheld until the time of disease progression or the development of symptoms. Expectant management is used throughout the world, and it is considered a satisfactory alternative to the use of primary RT or RP by its proponents. Under a program of watchful waiting, noncurative treatment would be offered at the time of progression and would consist primarily of hormonal therapy to induce tumor regression and alleviate symptoms.^224,425,426 Increasingly, an alternative expectant management strategy, active surveillance, is replacing watchful waiting. Whereas the goal of watchful waiting is to avoid treatment, the goal of active surveillance is to individualize patient care and treat only those men who have a clinically significant cancer and avoid the risk of treatment-related morbidity in men whose cancers are unlikely to progress.

The rationale for expectant management is related to the natural history of the condition, the age and comorbidity of the incident population, the perceived need (or lack thereof) for therapeutic intervention, and health-related quality of life issues. PSA screening has resulted in an increased rate of prostate cancer “overdiagnosis.” Overdiagnosis refers to the lead-time bias in establishing a diagnosis of cancer that would otherwise not be detected during a man’s expected life span. Furthermore, incidental (or latent) prostate cancer is often found on autopsy examination of the prostate gland and, although particularly common in the elderly,^9,10 is observed in as many as one third of men in their fourth and fifth decades.⁸ The key to a successful expectant management strategy is the proper selection of patients who have indolent disease and/or comorbidities that would impact their expected survival relative to the natural history of the prostate cancer.

Several studies demonstrate that many localized prostate cancers have an indolent course that may support the use of expectant management for this condition. In particular, the prognosis appears quite favorable for the patient with a small-volume, low-grade tumor identified incidental to TURP (stage T1a) for benign prostatic hyperplasia.^427,428 However, the distinction between stage T1a and stage T1b is important, because the risk of disease progression and disease-specific mortality is highly dependent on tumor volume and grade.^428–430

One of the largest studies of expectant management was reported by Chodak and colleagues,²²⁴ who collected patient and disease characteristics from several nonrandomized institutional series and from a population-based study to evaluate the outcome of expectant management in patients with localized prostate cancer. The expectant management approach in this study was watchful waiting with no attempt to implement curative local therapy in these patients. The investigators identified 10 studies published during the preceding decade and obtained original data on 828 patients from 6 of them. In this meta-analysis, multivariate regression analysis identified tumor grade as the most significant factor associated with DSS. Based on their observations, the authors concluded that localized prostate cancer is a steadily progressive disorder but that observation with delayed hormonal therapy is a reasonable approach for some men with low-grade disease.

Although single-institution series include patients selected for expectant management on the basis of advanced age or life-limiting comorbidity, population-based studies may minimize this potential source of bias. Johansson and colleagues⁴²⁵ described a group of 300 consecutive patients diagnosed with stage T1 to T2 prostate cancer at the Örebro Medical Centre, a hospital serving a geographically well-defined population. Among these patients, 223 patients with predominantly low-grade tumors were treated with exogenous estrogen or orchiectomy if symptomatic tumor progression occurred. By the time of last evaluation, EPE or metastatic disease had developed in 36% and 17%, respectively. Although the 15-year PFS (45%) and OS (21%) were rather low, the high 15-year DSS (79%) appeared consistent with the results of several single-institution investigations.^431,432 However, in an update of this study, the prostate cancer mortality rate increased from 15/1000 person-years during the first 15 years to 44/1000 person-years beyond 15 years of follow-up.⁴²⁵ Although most cancers had an indolent course during the first 10 to 15 years after diagnosis, follow-up from 15 to 20 years revealed a substantial decrease in PFS (from 45% to 36%), survival without metastases (from 77% to 51%), and prostate CSS (from 79% to 54%). Therefore these findings support early curative therapy for patients with an estimated life expectancy of more than 15 years.

Other population-based studies from Sweden have not confirmed the high 10- to 15-year DSS rate observed in the series from the Örebro Medical Centre.⁴²⁵ For example, Aus and colleagues⁴³³ identified all patients with an antemortem diagnosis of nonmetastatic prostate cancer treated with primary hormonal (63%) or deferred (36%) therapy who died in Göteborg, Sweden, between 1988 and 1990. Among 301 such patients, the cause of death was related to prostate cancer in 149 (50%), and for those surviving at least 10 years, 63% died from this disease. The initial tumor stage and grade were associated with the prostate cancer–related death rate. Although only 10% of patients with stage T1a tumors died of prostate cancer, 47% to 53% of patients with stage T1b to T3 tumors and 70% of patients with stage T4 lesions died secondary to their malignant disease. Forty-three percent to 48% of patients with grade 1 and grade 2 tumors and 60% of patients with grade 3 lesions died of prostate cancer. Furthermore, 58% of patients underwent TURP, 22% underwent an upper urinary tract procedure (e.g., pyelostomy) due to urinary obstruction, and 35% received palliative irradiation.⁴³⁴ Norlén⁴³⁵ noted similar findings in an evaluation of 44,300 patients with prostate cancer, representing 99% of all newly diagnosed cases, reported to the Swedish Cancer Registry between 1960 and 1978. Survival rates were adjusted for mortality in the general population and were expressed as relative survival, which estimates the likelihood of dying of prostate cancer. Among patients with prostate cancer, the relative survival was 51%, 34%, and 17% after 5, 10, and 20 years of observation, respectively, representing an excess death rate of approximately 8% per year.

An important randomized study from Sweden comparing RP to watchful waiting has been reported.¹¹⁷ Although the effect of treatment in relative terms was substantial, a decrease in disease-specific mortality of approximately 50% with 8.2 years median follow-up, in absolute terms the impact was more modest. The disease-specific mortality decreased from 14.4% with watchful waiting to 8.6% with prostatectomy.¹¹⁷ Although after 5 years’ follow-up the overall mortality was similar, 7.8% for prostatectomy patients versus 9.8% for the watchful waiting group, at 10 years’ follow-up there was a significant difference (p = .04) in favor of the treated patients, 27% versus 32%, but still an absolute difference of only 5%. Furthermore, because the men in this study were largely recruited before widespread PSA screening, the lead time introduced with earlier detection may erode the benefit of definitive treatment in contemporary populations, especially those with less than a 10- to 15-year life expectancy.

Active Surveillance

Unlike watchful waiting, active surveillance strategies attempt to individualize the decision to recommend treatment based on a patient’s risk of cancer progression in his lifetime. The key to a successful active surveillance strategy is the accurate identification of patients with clinically insignificant cancer, a rational follow-up plan, and early intervention criteria. Epstein and colleagues⁴³⁶ introduced biopsy criteria to predict an insignificant cancer: clinical stage T1c, PSA density less than 0.15 ng/mL, no Gleason pattern 4 or 5, fewer than three positive cores, and less than 50% cancer per involved core. In an analysis of 237 nonpalpable (stage T1c) prostate cancer patients treated at Johns Hopkins University, the Epstein criteria predicted for organ-confined disease in 91.6% of cases.⁴³⁷ The other 8.4% who fulfilled the Epstein criteria had evidence of extraprostatic disease. In a European validation study of the Epstein criteria, Jeldres and colleagues⁴³⁸ examined a cohort of 366 patients. In that series, 20% of patients who fulfilled the Epstein criteria had unfavorable findings at RP, such as Gleason score 7 or non–organ-confined disease.⁴³⁸ Consequently, as many as 20% to 25% of patients predicted to have insignificant cancer by the Epstein criteria may be expected to have non–organ-confined disease.⁴³⁹ The risk of missing a clinically significant cancer and avoiding necessary treatment in a potentially curative patient is the primary reason that active surveillance has not been more broadly adapted.

To accurately identify truly clinically insignificant cancers, many active surveillance series have included additional selection factors to the Epstein criteria. These have included an absolute PSA value less than 10 ng/mL^440,441 or less than 15 ng/mL^442,443 and/or fewer than 33% positive cores.⁴⁴⁰ Although clinical validation has not been established, Shukla-Dave and associates⁴⁴⁴ have reported the utility of incorporating MRI and MRS into nomograms to predict for clinical significant disease. The receiver operating characteristic area under the curve for the nomogram was 0.54 compared with only 0.726 for the best clinical nomogram that did not include MRI data (p <.0001).

While on active surveillance, clear criteria need to be described for patient follow-up. All published series to date include repeated DRE, determination of PSA level, and repeated prostate biopsies. The interval for DRE varies from every 3 to 6 months for the first 2 years to every 6 months after 2 years. PSA testing frequency ranges from every 1 to 3 months for the first 1 to 2 years and then every 6 months thereafter. Recommended biopsy intervals vary from every 6 or 12 months during the first year, then annually or biannually thereafter. Likewise, the indications for treatment should be well defined before advising an active surveillance strategy. Common criteria for implementing therapy include a change in PSA kinetics, such as an increase in PSA velocity or a decrease in the PSA doubling time to less than 2 or 3 years.²⁵⁶ Clinical progression of a palpable or ultrasound-detected nodule and/or a rebiopsy pathology with any Gleason pattern 4 or 5 with more than two cores involved or more than 50% of any core involved with prostate cancer is also used as a trigger to implement treatment.

Several large institutional series have reported excellent outcomes with active surveillance strategies. The largest and most mature prospective series to date is that of the Princess Margaret Hospital with 450 patients and a median follow-up of 6.8 years.²⁵⁶ The criteria for active surveillance initially included all favorable-risk patients (i.e., Gleason score ≤6, PSA ≤10 ng/mL), and for patients older than 70 years a PSA level up to 15 ng/mL or Gleason scores up to 3 + 4 were allowed. After 2000 the study strictly limited enrollment to favorable-risk patients. The criteria for implementing treatment were PSADT of less than 3 years, histologic upgrading and repeat biopsy, and/or clinical progression on DRE. Initially, a PSADT of less than 2 years was used but this was increased to 3 years when it became apparent that only l0% of patients were identified as higher risk, necessitating treatment. Patients had a PSA test every 3 months for the first 2 years and every 6 months thereafter, as long as the PSA value was stable. A repeat biopsy was performed at 6 to 12 months after enrollment and repeated every 3 to 4 years until the patient reached the age of 80 years. Overall, 83% of patients had a Gleason score 6 or less and 13% had a Gleason score 3 + 4 = 7 at study entry. The 10-year prostate cancer–free survival rate was 97.2%. Overall, 30% of patients had evidence of progression to higher-risk categories and were offered therapy. Of 117 patients treated, the PSA failure rate from treatment was 50%, but this represented only 13% of the entire study cohort.²⁵⁶

From the Royal Marsden Hospital, van As and colleagues⁴⁴³ reported the results of 326 men managed with active surveillance. Inclusion criteria for their study were clinical stage T1 to T2a, Gleason score less than 7 (3 + 4), PSA less than 15 ng/mL, and less than 50% of biopsy cores positive. Treatment criteria included PSA velocity of greater than 1 ng/mL per year, Gleason score greater than or equal to 4 + 3, or greater than or equal to 50% cancer involvement in any involved core on repeat biopsy. With a median follow-up of 22 months, these researchers have reported no prostate cancer deaths and 20% of their patients have moved to active treatment.

Dall’era and colleagues⁴⁴⁰ reported the University of California, San Francisco, experience of 321 men selected as having PSA less than or equal to 10 ng/mL, no Gleason pattern 4 or 5, clinical stage T2a or lower, PSAD of less than 0.75 ng/mL, and fewer than 33% positive cores involved with cancer. Patients were offered treatment if they had a repeat biopsy Gleason score greater than or equal to 7 or any increase in cancer volume on repeat biopsy or a rising PSA value. With a median follow-up of 3.6 years, 37% of patients met at least one criterion for treatment. Overall, 38% had higher-grade cancer on repeat biopsy and 26% had a PSA velocity of more than 0.75 ng/mL. Seventy-eight of the 321 men (24%) received treatment, and there have been no prostate cancer deaths reported to date. These investigators reported that 13% of patients chose therapy despite no evidence of disease progression.

The Johns Hopkins University experience of 407 men with a median age of 65.7 years has been reported by Carter and colleagues.⁴⁴⁵ All patients met the Epstein criteria for clinically insignificant cancers. With a median follow-up of 3.4 years, 59% remain on active surveillance and 25% underwent active treatment with curative intent. No patient has died of prostate cancer in this series.

Several single-institution studies suggest a role for active surveillance in the management of men with low-risk and low-volume prostate cancer. The optimal selection criteria for active surveillance remain to be defined, as well as the frequency of follow-up examinations, such as DRE, PSA testing, and repeat biopsy. The criteria for which to initiate active therapy vary from series to series, and the long-term impact on prostate CSS and quality of life remains to be determined. It is worth noting that men often experience anxiety due to their diagnosis and fear of disease progression. A randomized clinical trial from Sweden comparing treatment to watchful waiting showed no difference in quality of life in men at 5 years. Worry, depression, and anxiety were equally common in each group.⁴⁴⁶

Increasingly, expectant management strategies are being endorsed by oncologic specialty societies and networks. The NCCN strongly recommends active surveillance for men with a short life expectancy.³¹⁹ The NCCN cites the observation by Lu-Yao and associates⁴⁴⁷ that the prostate cancer–specific death rate for men on an active surveillance strategy has dropped by 74% in the decade between 1992 and 2002 largely owing to earlier diagnosis and the identification of more indolent cancers through PSA screening. The AUA guidelines⁴⁴⁸ and the ACS⁴⁴⁹ both recommend that patients be informed of active surveillance as an alternative to a curative intervention. Currently, three randomized clinical trials are underway to evaluate the role of active surveillance relative to treatment. The Standard Treatment Against Restrictive Treatment (START)⁴⁵⁰ trial in North America, the Prostate Cancer Research Interactive: Active Surveillance (PRIAS)⁴⁴¹ trial in Europe and North America, and the PROstate TEsting for Cancer and Treatment (PROTECT)⁴⁵¹ trial in the United Kingdom are actively recruiting patients (START, PRIAS) or are currently in follow-up (PROTECT). These studies, along with the U.S.-based Prostate Cancer Intervention Versus Observation Trial (PIVOT) will help bring clarity to the issues surrounding expectant management and early-stage prostate cancer.⁴⁵²