Prostate Cancer

Worldwide, more than 900,000 cases of prostate cancer were diagnosed in 2008, making its incidence second only to lung cancer in men.1 Incidence rates vary approximately 25-fold worldwide, with the highest rates being observed in North America, Australia and New Zealand, and Western and Northern Europe. It is believed that a large part of this variation reflects differences in the use of prostate specific antigen (PSA) screening.1 Excluding skin cancer, prostate cancer is the most common cancer in American men. In 2010, it was estimated that almost a quarter of a million new cases of prostate cancer were diagnosed in North America, and more than 36,000 men died from the disease.2,3 The risk for prostate cancer increases with age; the median age of diagnosis in the United States during 2004–2008 was 67 years.4 With the aging population, prostate cancer will present a significant burden to health care services. In data from the Surveillance, Epidemiology, and End Results Program, more men were diagnosed with prostate cancer at a younger age and earlier stage in 2004–2005 than in the mid- to late 1990s, and the disparity between ethnic groups in cancer stage at diagnosis decreased.5

Risk Factors

Apart from age, ethnic group, and family history, the risk factors associated with prostate cancer are unclear,6 which makes primary prevention difficult.

Ethnic Group

Striking differences in incidence have been observed for different ethnic groups and populations. A high incidence has been observed in populations of African descent in several countries,7 including Brazil, the Caribbean, and France.8 In parts of sub-Saharan Africa, the incidence of prostate cancer in black populations lies in the range of 14 to 25 per 100,000 per year, compared with 40 to 70 per 100,000 per year in white populations in these areas, although it is noted that the black population does not have access to diagnostic and screening facilities that are available to the white population in these areas.9 These observations are complicated by differences in the use of PSA screening and/or access to care, which may result in differential ascertainment. Migrant studies suggest that prostate cancer incidence increases when men move from a lower to a higher incidence population. Many epidemiological studies have suggested a wide range of risk factors for prostate cancer, but controlled trials have either not been conducted, or have shown negative results.

Hereditary Factors

First-degree relatives of men with prostate cancer have a two- to threefold increased risk for developing the disease.6,10,11 In addition, the risk of relatives developing prostate cancer increases with an increase in the number of affected individuals in the family and with a decrease in the age at diagnosis of the index prostate cancer case.12 High concordance rates have been observed in monozygotic twins. In a combined analysis of data from three Scandinavian countries, the estimated heritability for prostate cancer was the highest of all the types of cancer investigated.13

A subset of familial prostate cancer cases show patterns of familial aggregation consistent with an autosomal dominant mode of inheritance of a susceptibility gene, but this accounts for no more than 15 percent of prostate cancer.14,15 Prostate cancer is currently considered to be a complex, multifactorial disease with the vast majority of familial clustering attributed to the interaction of multiple shared moderate to low penetrance susceptibility genes as well as shared environmental factors within these families.

Other Risk Factors

Compared with other common types of cancer, the risk factors associated with prostate cancer are unclear.6 Many epidemiological studies have suggested a wide range of risk factors for prostate cancer, but controlled trials have either not been conducted, or have shown negative results.

An analysis of individual patient data from 12 studies of the association between insulin-like growth factors (IGFs) and IGF binding proteins and prostate cancer suggests that higher levels of serum IGF1 are associated with a higher risk for prostate cancer.16 Several studies have investigated the possible association between diabetes mellitus and the risk for prostate cancer. Meta-analyses indicate an inverse relationship.17,18

Observational studies have suggested that diet may be important in the etiology of prostate cancer, but these have not translated into effective preventive interventions. An analysis of the Alpha-Tocopherol Beta-Carotene Intervention Trial of heavy smokers in Finland showed a 40 percent decrease in incidence and mortality in prostate cancer in men taking alpha-tocopherol compared with those taking placebo.19 Analysis of further randomized controlled trials (RCTs) that included prostate cancer as a secondary end-point have also indicated a possible protective effect of alpha-tocopherol.20 However, in a large, long-term trial of male physicians, neither vitamin E nor C supplementation reduced the risk of prostate or total cancer,21 and in another long-term trial, it was concluded that dietary supplementation with vitamin E significantly increased the risk of prostate cancer among healthy men.22 While observational studies have suggested a protective role for selenium, this was not confirmed in a large RCT.23 Inverse associations with consumption of tomatoes/lycopene24,25 and soy products26,27 have been reported. Positive associations with the consumption of dairy products and calcium have been reported.24,28,29 The evidence of association with alcohol,24,30 coffee,31 dietary fiber,32 fish consumption,33 and beta-carotene supplementation34 has been interpreted as null.

Other risk factors that have been considered include androgens,35 anthropometric measures,24,36 physical activity,6 sexual behavior,37 sexually transmitted infection,35,38,39 vasectomy,40,41 occupation as flight personnel,42,43 agricultural pesticide applications,44 use of nonsteroidal anti-inflammatory drugs,45 statin use,46,47 smoking,25,48 use of smokeless tobacco,49 sun exposure,50 and serum 25-hydroxyvitamin D level.51,52

Natural History

The natural history of prostate cancer is highly variable.53 In studies of autopsy series, histologically proven prostate cancer was found in approximately 30 to 40 percent of men over 50 years of age who died of other causes.54-60 This is three to four times higher than the lifetime risk of prostate cancer diagnosis in American men (approximately 11 percent),53 which suggests that the disease is indolent in a large proportion of affected men. However, it is difficult to predict the aggressiveness of the disease in individual men. The most commonly used scheme to grade prostate cancer is the Tumor, Nodes, Metastases (TNM) scheme, which evaluates the size and histological features of the tumor, the extent of involved lymph nodes, and the presence of metastasis. This information is used to classify the tumor into one of four categories: Stage I–small, localized focus within prostate, typically found when prostatic tissue is removed for other reasons such as benign prostatic hyperplasia; Stage II–more of the prostate is involved and a lump can be palpated (by digital rectal examination [DRE]) within the gland; Stage III–the tumor has broken through the prostatic capsule and the lump can be palpated on the surface of the gland; Stage IV–the tumor has invaded nearby structures, or has spread to lymph nodes or other organs.

The Gleason score is based on histopathological assessment of the glandular architecture of prostate tissue samples, usually obtained by transurethral ultrasound (TRUS) guided biopsy.61 The assessment involves determination of: the most prevalent pattern of growth and differentiation; and, the most aggressive pattern, each of which is assigned a score (range 1 to 5), which is then summed to give the overall Gleason score. The Gleason scoring system was modified,62 which resulted in a shift of the most commonly found score from six to seven.61 This has implications for the comparison of subgroup analyses by Gleason scores over time.

Several studies have sought to provide an estimate of the long-term risk of death from prostate cancer in men whose disease was clinically localized at diagnosis and who were managed solely by observation (watchful waiting), with or without androgen withdrawal therapy.53,63-72 Most of these studies were carried out before the advent of PSA testing, which is thought to have increased the detection of clinically indolent disease and extended lead time.73-78 Only a small proportion of men with prostate cancer diagnosed at an early clinical stage (Gleason scores ≤4) die from prostate cancer within 10 to 15 years of diagnosis. Men with poorly differentiated tumors frequently die within 5 to 10 years of diagnosis.66,69 The greatest variation in outcome is for men with moderately differentiated tumors (Gleason scores 5 to 7).53,66,69 The natural history over longer periods of observation is uncertain. A study in Sweden,69 observed an increase in prostate cancer mortality among a relatively small number of men who were alive more than 15 years after diagnosis of localized prostate cancer, but this was not observed in a larger study in Connecticut, United States.66 Numerous differences between these cohorts could account for this inconsistency.79 A modeling study in the United States projected that 20 to 33 percent of men have preclinical onset (i.e., asymptomatic, but diagnosed as a result of a routine PSA test) of whom, 38 to 50 percent would be clinically diagnosed, and 12 to 25 percent would die of the disease in the absence of screening and primary treatment.80

Treatment in Men With Clinically Localized Prostate Cancer

The value of aggressive management for localized prostate cancer is also debated, and only a small proportion of men with early stage prostate cancer die from the disease within 10 to 15 years of diagnosis. In the United States, African-American men have a poorer prognosis, which does not appear to be fully explained by comorbidity, PSA screening, or access to free health care, although the variation in the measurement of these factors complicates the interpretation.7

Two RCTs have compared the efficacy of radical prostatectomy and watchful waiting in men with clinically localized prostate cancer, almost all of which were detected by methods other than PSA testing. A small trial showed no differences in survival between these two management strategies.81 A larger trial by the Scandinavian Prostate Cancer Study Group showed a small reduction in the risk of progression or death from prostate cancer in the men treated with radical prostatectomy, but also noted the potential harms that resulted from surgery.70,71 Two further RCTs are ongoing, one in the UK82,83 and one in the United States.84

PSA Screening

PSA was discovered in the 1960s and 1970s,85 and the work identifying it as a serum marker for adenocarcinoma of the prostate was published in 1987.86 It was first approved by the U.S. Food and Drug Administration (FDA) in 1986 for monitoring progression in patients with prostate cancer, and later approved for the detection of the disease in symptomatic men (but not for screening asymptomatic men).87 Since 1986, it is estimated that more than a million additional men in the United States have been diagnosed and treated for prostate cancer because of PSA screening than would otherwise have been the case, the most dramatic increase observed being for those under the age of 50.88 The increase in incidence following the introduction of PSA screening has never returned to prescreening levels, and has been accompanied by an increase in the relative fraction of early stage cancers, but not a decrease in the rate of regional or metastatic disease.89

Seven randomized trials (12 publications) of screening using PSA testing alone, or in combination with DRE, have been reported, in the United States,90,91 Canada,92-94 and Europe,95-101 with conflicting results.

Meta-analysis of these trials indicates that prostate cancer screening did not result in a statistically significant decrease in all-cause or prostate cancer-specific mortality,102,103 and that overdiagnosis resulted in harms that are frequent, often persist, and are at least moderate in severity.103 The individual trials and meta-analyses have generated substantial debate, with many commentaries arguing for the development of more accurate markers to use in screening or a risk stratification approach.104-112 Investigation of genetic variants associated with prostate cancer has been considered a promising route to the identification of such markers.

Single Nucleotide Polymorphisms

Single nucleotide Polymorphisms (SNPs) are minute variations in the DNA sequence that are passed on from parents to children. They are the most common type of genetic variation in humans. Formally, an allele, that is, a variation in DNA sequence, is defined to be “polymorphic” if it occurs in at least 1 percent of a population.113 Therefore, although overall humans are very similar at the DNA sequence level, because the genome is large there is substantial latitude for individual genetic variation. SNPs occur about once in every 800 base pairs.114 The Human Genome Project and advances in related technologies have fostered the investigation of the relationship between genetic variation and many health outcomes, including prostate cancer.

Since 2001, about 1,000 publications have reported associations between prostate cancer and SNPs and other genetic variants. The vast majority of the studies have related to candidate genes, in which the genes and variants, usually SNPs, have been specifically selected for investigation based on biological and physiological information regarding the involvement of gene products in early developmental pathways, biochemical and cellular process of progression, and/or clinical manifestations (a “candidate gene” approach). For prostate cancer, the most intensively investigated associations have related to genes in the following pathways: adhesion molecules (CDH1115); androgen metabolism (AR,116,117 ESR2,118 SRDA2119,120); angiogenesis (VEGF121) angiotensin conversion (ACE122,123); base-excision repair (XRCC1124,125); inflammation and immune response (IL8, IL10,126-128 MSR1,129 PTGS2,130 TNF131); inhibition of cell growth (FGFR4,132,133 TGFB1,134 TGFBR1135); insulin-like growth factor metabolism (IGF1,136 IGFBP3137); one carbon metabolism (MTHFR,138 diverse genes139); oxidative response (MnSOD,140 hOGG1141); substrate metabolism (CYP1A1,142 CYP3A4,143 CYP17,144,145 GSTM1, GSTT1, GSTP1,146 NAT1 and NAT2,124 UGT2B17145); vitamin D metabolism (VDR147); and, common variants of genes for which rare mutations are associated with increased cancer risk (ELAC/HPC2,148 RNASEL,149,150 TP53,151,152 MDM2153). In general, the results of candidate gene studies have been inconclusive, for reasons discussed in many commentaries.154,155 However, when associations have been confirmed, they have been modest, with odds ratios (ORs) in the range of 1.1 to 2.2.156 Thus, the proportion of individuals carrying any one of these variants that also developed the health outcome under investigation is low (i.e., these variants are of low penetrance).

The HapMap Project, completed in 2005, has shown that SNPs are often correlated with their neighboring SNPs, which has provided a methodology for investigating the associations between genetic variation and health outcomes on a genome-wide scale.114 In genome-wide association (GWA) studies, a dense array of genetic markers that capture a substantial proportion of common variation in genome sequence, are typed in a set of DNA samples and tested for association with the trait of interest without specific prior hypotheses.157 In most investigations of this type, the ability to validate findings in independent samples is built in to the study.157 As of 31 January 2012, GWA studies have identified replicated associations between prostate cancer and more than 50 specific SNPs (Table 1),158-164,164-171 all of which appear to be of low penetrance at best.

Table 1. Replicated associations between prostate cancer and SNPs in GWA studies.

Table 1

Replicated associations between prostate cancer and SNPs in GWA studies.

It is generally accepted that screening based on single low penetrance alleles is of little value,172-175 and may in fact be harmful when psychosocial factors are considered. In contrast, it has been suggested that combinations of a small to moderate number of common, low penetrance variants may account for a high proportion of disease in a population173,176,177 and may be useful in predicting risk for disease.178 For example, for a common disease with a 5 percent lifetime risk, for which three hypothetical gene variants at different loci and one environmental exposure are modest risk factors (risk ratios 1.5 to 3.0), the positive predictive value of information for subjects with a variant allele at two to three loci could be 50 to 100 percent in the presence of a modifiable exposure.173 Thus, there has been mounting interest in the possibility that panels comprising combinations of germline genetic variants (SNPs) might be of value in screening for common chronic diseases,179,180 including prostate cancer. The aim of this review is to assess the evidence as to the possible value of SNP panels in the detection of, and prediction of risk for, prostate cancer.

Scope and Purpose of This Review

The Centers for Disease Control and Prevention (CDC), through the office of Public Health Genomics, and the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) project, requested a review of the evidence on the use of SNP-based genotyping panels to assess risk of prostate cancer. The overall goal of EGAPP is to facilitate the use of evidence-based decisionmaking that will assist health care providers, consumers, policymakers, and payers in distinguishing genetic tests that are safe and useful, and guiding their appropriate application in clinical practice. Within the “ACCE framework” (see Table 2), the EGAPP working group has developed approaches to evaluating, synthesizing, and grading evidence.181 This synthesis will be used by EGAPP to develop evidence-based recommendations on the application of SNP-based panels to prostate cancer. The overarching goal of the use of such panels is to facilitate early detection of, and enhance the ability to target men at increased risk for prostate cancer, as well as to assist in targeting invasive interventions at those men with diagnosed prostate cancer who are most likely to have an unfavorable prognosis.

Table 2. Elements and key components of evaluation framework for SNP-based panels in prostate cancer risk assessment.

Table 2

Elements and key components of evaluation framework for SNP-based panels in prostate cancer risk assessment.

An initial set of questions was proposed by the EGAPP to guide the development of the evidence report, focusing on all aspects of the use of these panels. The intent of the original questions was to encompass all areas of evaluation, including analytic and clinical validity of panels and associated algorithms for prostate cancer risk assessment, their clinical utility in bringing about change in clinical decisionmaking, and their potential for harm.

Objectives of This Review

The primary objectives of the review were to identify, synthesize, and appraise the literature on the use of SNP-based panels in men who may be at risk of prostate cancer, encompassing all relevant areas of test evaluation as proposed by the ACCE framework. Anticipating a limited evidence base for some of the key questions, an objective of this review was also to characterize the knowledge gaps and provide targeted recommendations for future research.

Key Questions of This Review

The original key questions articulated in the Task Order were revised and rearticulated for the purposes of clarity. Thus, the three Key Questions (KQs) encompassing broad aspects of the analytic validity, clinical validity, and clinical utility of SNP-based panels were developed with the input of a Technical Expert Panel (TEP) whose membership was nominated by the Evidence-based Practice Center and approved by the Agency for Healthcare Research and Quality (AHRQ).

Note: for the purposes of the review, the term ‘SNP-based panels’ is used to indicate any risk assessment system designed to assess risk of prostate cancer, which incorporates one or more defined SNPs alone or in combination with other indicators.

KQ1. What is the analytic validity of currently available SNP-based panels designed for prostate cancer risk assessment?

  1. What is the accuracy of assay results for individual SNPs in current panels?
  2. What is the analytic validity of current panels whose purpose is, or includes, predicting risk of prostate cancer?
  3. What are the sources of variation in accuracy or analytical validity across different panels?

KQ2. What is the clinical validity of currently available SNP-based panels designed for prostate cancer risk assessment?

  1. How well do available SNP-based genotyping platforms predict the risk of prostate cancer in terms of
    1. stratifying future risk and/or screening for current disease?
    2. distinguishing between clinically important and latent/asymptomatic prostate cancer?
    3. How well do available SNP-based genotyping panels predict prognosis in individuals with a clinical diagnosis of prostate cancer?
  2. What other factors (e.g., race/ethnicity, gene-gene interaction, gene-environment interaction) affect the predictive value of available panels and/or the interpretation of their results?

KQ3. What is the clinical utility of currently available SNP-based panels for prostate cancer risk assessment, in terms of the process of care, health outcomes, harms, and economic considerations?

Process of care
  1. Does the use of panels alter processes of care and behavior, in terms of
    1. screening or management decisions, and the appropriateness of these decisions, by patients and/or providers
    2. alteration in health-related behaviors of patients (e.g., adherence to recommended screening interventions and/or other lifestyle changes)?
Health outcomes

Does the use of panels lead to changes in health outcomes, in terms of

  1. all-cause mortality
  2. cancer-specific mortality
  3. morbidity, and do any such changes vary by race or ethnicity?

Does the use of panels lead to harms in terms of

  1. psychological harms
  2. other negative individual impacts (e.g., discrimination), and do any such harms vary by race or ethnicity?

What is known about the costs, cost-effectiveness, and/or cost-utility of using SNP-based panels for prostate cancer risk assessment, compared to current practice?