Prostate Cancer Screening Clinical Practice Guideline. Approved by the National Guideline Directors November, 2013

Transcription

1 Prostate Cancer Screening Clinical Practice Guideline This guideline is informational only. It is not intended or designed as a substitute for the reasonable exercise of independent clinical judgment by practitioners, considering each patient s needs on an individual basis. Guideline recommendations apply to populations of patients. Clinical judgment is necessary to design treatment plans for individual patients. Approved by the National Guideline Directors November,

2 Table of Contents Recommendations 3 Rationales 4 Effectiveness and safety of prostate cancer screening 4 Age to discontinue screening 7 Frequency of prostate cancer screening 9 Referral to urology 10 Systematic Review 11 Executive summary 11 Key questions 16 Benefits of prostate cancer screening 18 Harms of prostate cancer screening 35 Treatment of localized prostate cancer 38 Ages to initiate and discontinue screening 54 Screening frequency 64 PSA cut-points for referral to urology 76 External guidelines 98 References 101 Appendix A 105 Determination of Evidence grade and Recommendation strength 105 Appendix B 107 About the National Guideline Program 107 2

3 Recommendations Offering PSA-based 1 prostate cancer screening is an option for average-risk men ages who have at least a 10 year life expectancy. (Weak recommendation) [approved 11/2013] Offering PSA-based prostate cancer screening is an option for higher- risk men (Black/African American descent or family history 2 of prostate cancer in at least one first-degree relative) ages who have at least a 10-year life expectancy. (Weak recommendation) [approved 11/2013] If PSA-based screening is offered, it should be done in the context of shared decision-making. (Strong recommendation) [approved 11/2013] Prostate cancer screening is not recommended for men age 70 and older. (Strong recommendation) [approved 11/2013] For men who elect to have PSA-based prostate cancer screening, a screening interval of every 2 years is an option. (Weak recommendation) [approved 11/2013] When PSA values exceed the age-specific thresholds in Table 1, consider repeating the PSA test within one month 3. If the repeat PSA value still exceeds the age-specific thresholds, referral to Urology is recommended. [re-approved 11/2013] Table 1 Age range(years) PSA threshold >2.5 ng/ml >3.5 ng/ml >4.5 ng/ml (if tested) >6.5 ng/ml 1 While there is evidence that DRE may detect some cancers that are not detected by PSA alone, the randomized trials demonstrating a potential benefit of screening are based on PSA testing alone. 2 Elements associated with higher risk from family history of prostate cancer include: multiple first degree family members, family member(s) diagnosed at advanced stages, age <65 years at diagnosis 3 Normal variation of PSA values of ~20% have been demonstrated in some studies 3

4 Rationale TOPIC Recommendation Effectiveness and safety of prostate cancer screening A. Offering PSA-based 1 prostate cancer screening is an option for averagerisk men ages who have at least a 10 year life expectancy. (Weak recommendation) B. Offering PSA-based prostate cancer screening is an option for higher-risk men (i.e., Black/African American descent or family history 2 of prostate cancer in at least one first-degree relative) ages who have at least a 10- year life expectancy. (Weak recommendation) C. If PSA-based screening is offered, it should be done in the context of shared decision-making. (Strong recommendation) Basis of Recommendation Prostate cancer is the most common non-skin cancer in men. Despite conflicting evidence regarding the balance of potential benefits vs. harms, epidemiological data and cultural expectations since the PSA test was introduced make consideration of screening for this cancer reasonable. The 5-year relative survival rate for prostate cancer for those with localized and regionally spread prostate cancer approaches 100%. Prostate cancer specific symptoms do not typically arise until distant metastases are present, and the 5-year survival rate for those with distant metastases is low, approximately 28%. Because some men may wish to screen in spite of potential risks, it is appropriate to offer screening in the context of a shared decision-making discussion. Shared decision-making is appropriate when the balance of benefits and harms is unclear or controversial. The shared decision-making approach for prostate cancer screening is supported by multiple guidelines, including those published by the American Urological Association (AUA), the American Cancer Society (ACS), the National Comprehensive Cancer Network (NCCN) and the American College of Physicians (ACP). While the USPSTF generally recommends against prostate cancer screening, it suggests that if screening is considered, it should be done in the context of shared decision-making. The recommendation of age 50 as the lower end of the range for a screening discussion in average-risk men is based primarily on evidence from the European Randomized Study of Screening for Prostate Cancer(ERSPC). The study showed a 17% relative reduction (0.1% absolute reduction) in prostate cancer-specific mortality in the entire cohort (50-74 years) and a 21% relative reduction (0.08% absolute reduction) in the core group (55-69 years). As well, a single randomized trial in men ages years from ERSPC s Swedish site (Goteborg), the only site of ERSPC that included men under the age of 55, demonstrated a 44% relative reduction (0.4% absolute reduction) in prostate cancer-specific mortality. The recommendation of age 45 as the lower end of the range for higher-risk men was based in part on observational studies indicating that higher risk men have the same prevalence of prostate cancer mortality as average-risk men beginning about 5 years earlier than averagerisk men.. As well, SEER data shows that mortality from prostate cancer in men younger than 45 is exceedingly low regardless of risk factors. 1 While there is evidence that DRE may detect some cancers that are not detected by PSA alone, the randomized trials demonstrating a potential benefit of screening are based on PSA testing alone. 2 Elements associated with higher risk from family history of prostate cancer include: multiple first degree family members, family member(s) diagnosed at advanced stages, age <65 years at diagnosis 4

5 Balance of Desirable and Undesirable Effects Current evidence for effectiveness of prostate cancer screening is limited. After a follow-up of years, absolute prostate cancer-specific mortality reduction from screening in randomized studies has ranged from 0 to 0.4%, and relative risk ranged from a 21% decrease to a 9% increase. ERSPC, which showed a prostate cancer-specific mortality reduction, also demonstrated a reduction in metastatic disease in the screening arm compared to the control arm (absolute risk of metastatic disease was 0.67% per 1000 men in the screening arm vs. 0.86% per 1000 men in the control arm ( ARR=0.31%; HR = 0.70 ( )). All-cause mortality was reported from the ERSPC and Stockholm trials. Compared with no screening, PSA-based screening resulted in no difference in all-cause mortality in either of these trials (pooled RR (95%CI) = 0.99( ). The results of two good quality RCTs of treatment for localized prostate cancer suggest that prostatectomy does not result in reduction in all-cause mortality compared with observation, but may provide protection against distant metastasis. The impact on prostate cancerspecific mortality is unclear but may suggest a benefit in younger men with intermediate or high risk disease. Harms of the screening exam and biopsy include pain, fever, bleeding, and transient urinary difficulties associated with prostate biopsy. Observed rates for transient urinary difficulties, severe urinary retention and infection leading to hospitalization are 22-50%, 0.4% and 0.5%, respectively. Unnecessary biopsy is also associated with psychological harm from falsepositive test results, overdiagnosis and over-treatment. In men with localized prostate cancer treated with prostatectomy, results from the 2 treatment RCTs reported absolute increases in erectile dysfunction of ~ 28.0% and in urinary incontinence of 10.8%-27.5% (compared to watchful waiting or active surveillance).serious perioperative events included mortality (~0.5%) and cardiovascular events(3%). On a population basis, some may argue that the known harms of screening and subsequent treatment of localized prostate cancer outweigh the potential mortality benefit. Nonetheless, prostate cancer is the second leading cause of cancer death in American men, and PSAbased screening is the only means available for early detection. Individual men are likely to weigh the potential benefits and risks differently and to make different personal decisions whether to participate in prostate cancer screening. 5

6 Quality of Evidence SCREENING (vs. no screening) Category (overall assessment) Benefit(low quality) Harms (low quality) Outcome Importance #/study design Assessment Explanation critical 5 RCTs low quality high risk of bias, indirectness) critical 2 RCTs moderate indirectness quality important 2 RCTs moderate indirectness quality All-cause mortality False-positive rates Hospitalizations important 1 RCT low quality Imprecision indirectness Urinary important 1 RCT low quality Imprecision retention indirectness TREATMENT (vs. watchful waiting/active surveillance) Category (overall assessment) Outcome Importance #/study design PROSTATECTOMY Benefit (moderate quality) Harms (high quality) Prostatecancer mortality Prostatecancer mortality All-cause mortality Urinary incontinence Erectile dysfunction RADIATION THERAPY Benefit (very Prostatecancer low quality) mortality Harms (very low quality) All-cause mortality Urinary incontinence Erectile dysfunction Bowel dysfunction Assessment Critical 2 RCTs Moderate quality Critical 2 RCTs High quality Critical 2 RCTs High quality Critical 2 RCTs High quality Critical 4 cohort Critical 5 cohort Very low quality Very low quality Explanation imprecision Study design Risk of bias Study design Risk of bias Critical 1 RCT Low quality Risk of bias Imprecision Critical 6 Very low Study cohort quality design Critical 2 cohort Very low quality Risk of bias Study design Risk of bias Values and Preferences Resource Implications This recommendation places a high value on individual patient choice. Variability of values and preferences with regard to the balance of benefits and harms of screening for asymptomatic prostate cancer is likely to be high. Uncertainty regarding these values and preferences is estimated to be high. Although the cost of an individual PSA test is low, the burden of screening all eligible men is high. Considerable clinician time is needed for shared decision-making discussions and considerable resources are required for biopsy and treatment of localized prostate cancers. 6

7 TOPIC Recommendation Basis of Recommendation Balance of Desirable and Undesirable Effects Age to discontinue screening Prostate cancer screening is not recommended for men age 70 and older. (strong recommendation) The KP National Prostate Cancer Guideline Development team endorses the following statement from the AUA 2013 Guidelines for Early Detection of Prostate Cancer as the basis for this KP recommendation: The rationale for this recommendation is based on the absence of evidence of a screening benefit in this population with clear evidence of harms. In the ERSPC randomized trial of screening, there was no reduction in mortality among men age 70 years or older. Although men in this age group have a higher prevalence of prostate cancer and a higher incidence of fatal tumors, they also have increased competing mortality compared to younger men, and no compelling evidence of a treatment benefit, especially in men with a limited life expectancy below 10 to 15 years. Therefore, given the lack of direct evidence for benefit of screening beyond age 70 years, and especially beyond age 74 years, as well as higher quality data regarding harms, the Panel discourages routine screening in this age group. Post-hoc sub-group analyses of prior randomized trials of prostate cancer screening (ERSPC and PLCO) and treatment (SPCG-4 and PIVOT) were used to weigh the balance of potential benefit versus harm of prostate cancer screening in men ages 70 and older. In a sub-group analysis from the ERSPC randomized trial of screening, there was no reduction in either all-cause or prostate cancer-specific mortality among PSA-screened men age 70 years or older(all-cause: RR=1.03( ); prostate cancer-specific: RR=1.18( ). In a sub-group analysis from the PLCO randomized trial of screening, there was no reduction in prostate cancer-specific mortality among all PSA-screened men, and when stratified by younger (55-64 years; RR=1.19( )) and older (65-74 years; RR=1.02( )) men. A sub-group analysis of the SPCG-4 trial of radical prostatectomy vs. watchful waiting for the treatment of localized prostate cancer showed no difference between groups in either all-cause or prostate cancer-specific mortality in men ages 65 years and older (all-cause: RR=1.04( ); prostate cancer specific: 0.87( ). Post-hoc subgroup analyses of the PIVOT trial of radical prostatectomy vs. watchful waiting or active surveillance in the treatment of localized prostate cancer showed no difference between groups in either all-cause or prostate cancer-specific mortality for men ages 65 years or older (all-cause: RR=0.89 ( ); prostate cancer-specific: RR=0.63( ). More than 90% of the study population was older than 65 years of age. There is a lack of evidence for mortality reduction from screening for or treating localized prostate cancer in men ages 70 and over. Given the known harms of screening, which are no less common in men of this age range, screening men ages 70 and over is not recommended. 7

8 Quality of Evidence Values and Preferences Resource Implications The evidence for benefit in men >70 years consists of post-hoc analyses of two RCTs of screening and two RCTs of treatment. The quality of evidence for harms of screening and treatment is described in the rationale for prostate cancer screening. Benefits of screening in men >70 years Category (overall assessment) Benefit(very low quality) Benefit ( low quality) Outcome critical critical Benefit (low quality) Benefit(low quality) Assessment Prostatecancer mortality All-cause mortality Benefits of treatment in men >70 years Category (overall assessment) Outcome Importance Prostatecancer mortality All-cause mortality #/study design 2 RCTs (subgroup analysis) 1 RCT (subgroup analysis) #/study design critical 2 RCTs(subgrou p analysis) critical 2 RCTs(subgrou p analysis) Importance Very low quality low quality Assessme nt Very low quality Very low quality Explanation high risk of bias, indirectness imprecision high risk of bias, indirectness Explanation high risk of bias, imprecision high risk of bias, imprecision This recommendation places a high value on avoiding the harms of screening and treatment over very low likelihood of prostate cancer mortality benefit in men screened for asymptomatic prostate cancer at age 70 years or older. Variability of values and preferences is estimated to be low. Uncertainty regarding values and preferences is estimated to be high as values as preferences were derived by guideline development team members. Although the cost of an individual PSA test is low, the resource burden of screening men age 70 years or older would be high. Considerable clinician time is needed for shared decisionmaking discussions and considerable resources are required for biopsy and treatment of localized prostate cancers. 8

9 TOPIC Recommendation Basis of Recommendation Balance of Desirable and Undesirable Effects Quality of Evidence Values and Preferences Resource Implications Frequency of prostate cancer screening For men who elect PSA-based prostate cancer screening, a screening interval of every 2 years is an option. (weak recommendation) As opposed to annual screening, the 2 year screening interval option is based on the natural history of prostate cancer, on very low quality observational evidence, and on the recommendations of external guidelines. These different resources indicate that moving from annual to biennial screening intervals results in a similar benefit-to-risk profile, measured by risk of prostate cancer death vs. risk of harm due to unnecessary treatment in patients with indolent cancers. Key external guidelines recommend a 2-year or 4-year screening interval for similar reasons. The decision to recommend biennial screening over 4-year screening is based upon a study comparing two randomized trials with 2- and 4-year screening intervals, with the biennial screening site showing a more significant reduction in prostate cancerrelated deaths, albeit with a concomitant slightly increased risk of over-diagnosis. Evidence to inform appropriate screening intervals is limited and based on indirect comparisons, observational data and predictive modeling. Nonetheless, these studies consistently predict a small reduction in over-diagnosis (~0.5% reduction in predicted risk over a lifetime) with very little increased risk of prostate cancer death (~0.1% increase in predicted risk over a lifetime) for biennial screening, as compared to annual screening. Overall quality = very low (study design, imprecision) No direct RCT evidence exists. 1 vs. 2-year interval: very low quality [observational study, imprecision] Advanced cancer at diagnosis: low [observational study] Prostate cancer mortality: very low [observational study, imprecision] 2 vs. 4-year interval: very low quality Advanced cancer at diagnosis: very low [observational, high risk of bias, indirect] This recommendation attempts to balance the importance to patients of reducing prostate cancer death while minimizing over-diagnosis. As no patient-specific information is available, uncertainty in values and preferences is high and variability in acceptance of a biennial screening interval is unknown. Extending screening intervals to two years reduces cost with minimal impact on risk of prostate cancer death. 9

10 TOPIC Recommendation 3 Referral to urology When PSA values exceed the age-specific thresholds in Table 1, consider repeating the PSA test within one month 4. If the repeat PSA value still exceeds the age-specific thresholds, referral to Urology is recommended. Table 1 Age range(years) PSA threshold >2.5 mg/l >3.5 mg/l >4.5 mg/l (if tested) >6.5 mg/l Basis of Recommendation The age-specific PSA thresholds recommended above (Table 1) were agreed to by the GDT on a consensus basis. Although there are no studies that evaluate the direct effect of the using screening with age-specific reference ranges on prostate cancer mortality, the GDT decided on the particular age-specific cutoffs after considering estimates from observational data (Oesterling et al.(1993,1995) Jacobsen et al.(1996)) suggesting that age-specific thresholds increase the sensitivity of the PSA test in younger men and decrease it in older men (compared to using a single cutoff regardless of age). The GDT believes that using agespecific ranges would create a balance of benefits and harms that is appropriate for the Kaiser Permanente population 3 This recommendation was approved in 3/2011 and deemed to be current as of 11/ Normal variation of PSA values of ~20% have been demonstrated in some studies 10

11 Systematic Review Executive Summary Uncertain evidence In 2012, the United States Preventive Services Task Force (USPSTF) recommended against systematically screening all men for prostate cancer, due to the lack of definitive evidence of screening benefit, combined with a moderate level of evidence for potential harm from the treatments. In the clinical considerations, the USPSTF laid out talking points for explaining the balance of benefit and harm to age-appropriate men. The American Academy of Family Practitioners (AAFP) has endorsed the USPSTF recommendations. The American Cancer Society (ACS), American College of Physicians (ACP), American Urological Association (AUA), and National Comprehensive Cancer Network (NCCN), recommend an informed discussion describing potential benefits and risks of screening. These organizations interpret the implications of the evidence differently, resulting in moderately disparate recommendations. Prostate cancer screening and incidence Prostate cancer is the most commonly diagnosed cancer, and the second most common cause of cancer-related death for men in the United States. 1 Randomized trials demonstrate that screening increases the detection of prostate cancer by 4-12%. Epidemiologic studies reflect the increased cancer detection resulting from PSA screening programs. Incidence rates rose rapidly in the early 1990s in the United States, Australia, Canada and the Nordic countries soon after the introduction of PSA testing, followed by a sharp decline, presumably due to a smaller pool of prevalent cases. The relationship between screening, treatment, and prostate cancer-related mortality is less clear. Death rates for prostate cancer have been declining in the last 20 years in many developed countries, including Australia, Canada, the United Kingdom, the United States, Italy, and Norway. It is unclear how much of this decline is due to screening, and how much is due to improved treatment techniques. Brachytherapy and hormonal treatment of localized prostate cancer have become more common during the past two decades and higher doses are being used in external beam radiotherapy. 2 In contrast to the trends in Western countries, mortality rates appear to be rising in some Eastern European countries, such as Slovenia and Croatia. 3 It is unclear whether this is due to increased life expectancy or to biologic or cultural factors that are increasing the incidence of prostate cancer in these countries. Natural history The high survival rate from prostate cancer partly reflects a slow growth rate of most prostate tumors. Evidence from observational studies suggests most men with PSAdetected, localized untreated prostate cancer have excellent health outcomes for up to 11

12 ten years after diagnosis. 4 The 5-year relative survival rate for prostate cancer for those with localized and regionally spread prostate cancer approaches 100%, whereas the 5- year survival for those with distant metastases is approximately 28%. 5 Screening effectiveness An ideal screening program would identify those men with disease most likely to result in prostate cancer-related morbidity and/or mortality. Three fair-quality randomized trials of prostate cancer screening with low risk of bias have reported mortality outcomes (PLCO, ERSPC, and the Swedish [Goteborg] site of ERSPC). Absolute prostate cancerspecific mortality reduction (ARR) from screening in these studies ranged from 0 to 0.4% after a follow-up of years (RR(95%CI)=0.87( )). 6 It is unclear whether these results reflect contamination or bias in the studies, inadequate length of follow-up or rather that PSA (leading to current standard treatments) is ineffective in distinguishing prostate cancer that is likely to cause premature death. ERSPC also demonstrated a lower rate of metastatic disease in the screening arm compared to the control arm (0.67% vs.0.86% per 1000 men; ARR=0.31%; HR = 0.70 ( )). Future directions in prostate cancer screening refinement There are no new or ongoing trials of prostate cancer screening, aside from those that have recently reported results. Given that PSA-based screening is the standard of care for patients who wish to be screened, designing a trial without any utilization of screening in the control group would not likely be feasible. Unless a new test for prostate cancer screening emerges, the refinement of prostate cancer screening may depend on better identification of those men with PSA-detected prostate cancer who are most likely to benefit from treatment. The informed discussion A less controversial element of informed discussion relates to the harms of screening and biopsy procedures needed to identify prostate cancer. The screening tests themselves are associated with rare bleeding or pain from the DRE (0.03%) and bruising or fainting due to venipuncture (0.26%). 6 After 4 screening sessions, the chance of having a false positive PSA test approaches 13%. 7 A false positive test result has been associated with increased worry about prostate cancer and perceived increased risk for prostate cancer in observational studies. 4 In large trials, a prostate biopsy was relatively safe, leading to urinary retention or infection in less than 1% of patients, with about 0.5% of men developing a complication requiring hospitalization (e.g., sepsis). Deciding whether to screen for prostate cancer must also take into account whether the patient would want to receive treatment for localized prostate cancer, if it were detected. Treatments are effective but also have a risk of harm. The most common excess risks associated with treatment of localized prostate cancer with prostatectomy are erectile dysfunction (absolute risk increase (ARI) = 35.6%-37.0%) and urinary incontinence (ARI= 10.8%-27.5%). 8;9 In the 30-day perioperative period, PIVOT reported cardiovascular events (0.2%), deep venous thrombosis (0.7%) and death (<0.1%). 12

13 Watchful waiting or active surveillance of prostate cancer may also be recommended options for localized prostate cancer, which would involve additional PSA tests and/or prostate biopsies at intervals. There is a paucity of current evidence to guide treatment decisions about prostate cancer. Patients currently decide to treat or monitor screen-detected localized prostate cancer based primarily on their individual judgment and the recommendations of their urologist. Only consensus-based recommendations about treatment thresholds exist currently. Under current practices in the United States, 77-80% of those diagnosed with screen-detected prostate cancer receive treatment. 6 Given the uncertainties about effectiveness of PSA-based screening, it is important for men to understand that a positive screening result would more likely lead to treatment than to active surveillance or watchful waiting. Screening discussion age range: average-risk men ages years, high-risk men ages years There is no strong, comparative evidence to inform the age range for offering prostate cancer screening to average-risk or higher-risk men. Recommending a screening discussion starting at age 50 in average-risk men is based largely on the positive findings of the Swedish site of the ERSPC trial (Goteborg), which is the only site in the multi-site study that included men years old. The randomized trial at this site demonstrated a 44% relative reduction and 0.40% absolute reduction in prostate cancer-specific mortality among those in the screening group. Recommending a screening discussion to start specifically at age 45 in higher risk men is based primarily on observational data. SEER data indicate that starting screening at age 45 would miss few aggressive prostate cancers. The mortality from prostate cancer in men aged is low (0.0006% for black men, who have the highest incidence and mortality by race). 10 As well, observational studies using comprehensive disease registries in Sweden indicate that prostate cancer incidence and mortality for men with a family history of prostate cancer (whether one affected relative or more) reach comparable levels to those of average-risk men beginning about 5 years earlier than average-risk men. External recommendations make a distinction between a high risk family history (a single first-degree relative affected) and a very high risk family history (multiple family members affected). There is no existing evidence to support a screening discussion at age 40 in men with a very high risk family history of prostate cancer. Because of this, and to improve ease of implementation, we collapsed the definition of family history into one higher risk family history category. Similarly, the selection of a cut-point for age of the affected relative by external guidelines is consensus-based, and no evidence exists to inform that decision. Because of concern about discouraging screening for those with a family history that may not meet an arbitrary definition of age 65 or less (e.g., a patient with a father who was found to have metastatic prostate cancer at age 66), we elected to not specifically define an age cut-point in the body of the recommendation. 13

14 The variability in the current recommendations of professional societies regarding prostate cancer screening age ranges by risk status illustrates the existing low level of evidence available: Professional Association USPSTF AAFP Average-risk Do not recommend screening Do not recommend screening ACS Informed discussion > 50 years, >10 year life expectancy ACP Informed discussion 50-69, >10-15 year life expectancy AUA Informed discussion years, >10-15 year life expectancy High-risk (by family history or Black race) Do not recommend screening Do not recommend screening Informed discussion > 45 years Informed discussion Decisions should be individualized. Very high-risk family history Do not recommend screening Do not recommend screening Informed discussion > 40 years Informed discussion years Decisions should be individualized. PSA-based screening The decision to offer screening primarily with PSA, without the addition of digital rectal exam, is based largely on the fact that randomized trials demonstrating a potential benefit of screening are based on PSA testing alone. As well, PSA as a standalone screening strategy is intended to reduce barriers to screening. This recommendation emphasizes patient choice to screen, based upon a face-to-face, relatively complex discussion between healthcare provider and patient about screening pros and cons. If the embarrassment or anticipated discomfort associated with the digital rectal examination appears to be a factor in the patient s decision about screening, it should be kept in mind that DRE contribute proportionally little additional sensitivity to the screening sensitivity of PSA alone. PSA threshold for biopsy The currently recommended PSA thresholds for biopsy represent PSA levels that are greater than the 95 th percentile for the respective ages: Age range (years) PSA threshold >2.5 g/l >3.5 g/l >4.5 g/l (if tested) >6.5 g/l 14

15 This recommendation was developed for the 2011 national prostate cancer guideline and remains current. We did not update this recommendation in PSA-based screening interval The new optional recommendation to screen all men biennially, regardless of PSA level, is based on two retrospective cohort studies and a modeling study. Two of these studies suggest that there is no significant increase in prostate cancer-specific mortality from screening biennially instead of annually; screening biennially in both studies was found to mitigate overdiagnosis as well. An additional low-quality observational study suggested that screening every 4 years instead of every 2 years may result in an unacceptable increase in prostate cancer specific mortality. Based on these three studies, the current recommendation is to offer biennial screening to all men, and to refer to urology if the PSA reaches the age-specific threshold (as above). Recommendations from major professional societies regarding PSA threshold for biopsy and screening interval are below. Professional Association Recommended PSA threshold for biopsy Screening Interval ACS No specific recommendation Every year if PSA >2.5 ng/ml Every 2 years if PSA <2.5 ng/ml ACP No specific recommendation Every year if PSA >2.5 ng/ml Every 4 years if PSA <2.5 AUA PSA of 3-4 ng/ml, taking into account other factors Every 2 years if PSA >1.0 ng/ml Longer interval if PSA<1.0 ng/ml Summary and conclusion Prostate cancer is relatively common, and screening detects more prostate cancer cases than usual care. Men who wish to be screened for prostate cancer should understand that if they do screen positive, watchful waiting or active surveillance is often an appropriate decision. Through a comprehensive shared decision-making discussion with his healthcare provider, the patient should understand the benefits and risks of screening as well as the benefits and risks of treatment when deciding whether to be screened. 15

16 Key Questions This evidence update uses the analytic framework and search strategy from AHRQ s Evidence Review AHRQ Publication No EF-1 October 2011 Prostate- Specific Antigen-Based Screening for Prostate Cancer: An Evidence Update for the U.S. Preventive Services Task Force 11 and AHRQ Publication No EF-1 October 2011 Treatments for Localized Prostate Cancer: Systematic Review to Update the 2002 U.S. Preventive Services Task Force Recommendation. 12 In the analytic framework below, the overarching clinical question addresses the effect of screening for prostate cancer on reduction in morbidity and mortality, through screening s effect on identification of prostate cancer at an early, treatable stage. To address this, the Evidence Practice Centers (EPCs) examined both the direct effect of screening and the effect of treatment in men with screen-detected or early-stage prostate cancer. Of note, they limited examined outcomes to mortality only, in spite of the mention of morbidity in the main clinical question. Analytic framework 5 KQ#1 No prostate cancer Population at risk SCREENING KQ#2 Harms Early detection of prostate cancer KQ#3 TREATMENT Prostatectomy Radiation therapy Androgen deprivation therapy Cryotherapy Ultrasonography Active surveillance Watchful waiting Reduced mortality and/or morbidity KQ#4 Harms Screening 1. What is the direct evidence that screening for prostate cancer with prostatespecific antigen (PSA), as a single-threshold test or as a function of multiple tests over time, decreases morbidity and/or mortality? 2. What are the harms of PSA-based screening for prostate cancer? 5 Adapted from: AHRQ Publication No EF-1 October 2011 Treatments for Localized Prostate Cancer: Systematic Review to Update the 2002 U.S. Preventive Services Task Force Recommendation 16

17 Treatment 3. What are the benefits of treatment of early-stage or screen-detected prostate cancer? 4. What are the harms of treatment of early-stage or screen-detected prostate cancer? 17

18 PSA screening Benefits of prostate cancer screening Clinical Question: Does PSA-Based Screening Decrease Prostate Cancer-Specific or All- Cause Mortality? Population: Intervention: Men without a prior diagnosis of prostate cancer and without bone pain or constitutional symptoms PSA-based screening* Comparison Critical Health Outcomes No prostate cancer screening Prostate cancer mortality All-cause mortality Lower urinary tract symptoms and erectile dysfunction are not indications for ordering a PSA test. *PSA-based screening is defined as a screening program for prostate cancer in asymptomatic men that incorporates one or more PSA measurements, with or without additional modalities such as digital rectal examination or trans-rectal ultrasonography. The literature search conducted by the EPC examined studies published up to July, Because updates to the two major screening trials (PLCO 7, ERSPC 13 ) as well as a major treatment study (PIVOT 8 ) were published subsequent to the EPC review, a bridge search was done in November 2012 to identify all relevant studies published subsequent to July 2011.In addition, a literature search to inform the definition of family history of prostate cancer was performed. Methodology The search strategy and other methodological details from the EPC s evidence synthesis may be found here, with details of search strategy found here. Study design was limited to RCTs, systematic reviews and meta-analyses. PSA-based screening is defined by the EPC as a screening program for prostate cancer in asymptomatic men that incorporates one or more PSA measurements, with or without additional modalities such as digital rectal examination or transrectal ultrasonography. Asymptomatic is defined as without symptoms that are highly suspicious for prostate cancer. We consider both prostate-specific mortality and all-cause mortality to be critical outcomes and are equally important for different reasons. While all-cause mortality may be the most germane outcome, it generally is not as robust due to inadequate sample size. 18

19 Search Results A QUOROM figure describing the EPC s search results may be found here. Among 379 unique articles retrieved, 7 publications met inclusion criteria and are included in the review: Andriole 2009 (PLCO); Schroder 2009 (ERSPC); Hugosson 2010 (ERSPC- Goteborg center); Kjellman 2009 (Stockhom South Hospital); Sandblom 2011 (Norrkoping); Djulbegovic 2010 (BMJ meta-analysis); and Ilic 2010 (Cochrane Review). Literature search update A bridge search was conducted to update the EPC evidence. The search strategies are consistent with that of the EPC. Search results are shown in Table 1, and study selection in Figure 1. Among 102 unique articles retrieved, 5 studies met inclusion criteria and are included in this review: 4 ERSPC updates and 1 PLCO update for screening studies. In addition, 2 relevant treatment studies were identified. 14, 15 Table 1: Screening for prostate cancer: search strategy and results Dates of search: 7/1/ /1/2012 Database: PubMed Search Name Search String Query translation Number of hits outcome-prostate cancer intervention- PSA screening Database: PubMed 1. Prostatic Neoplasms[Mesh] 2. Screening OR prostate-specific antigen[mesh] prostatic neoplasms[mesh Terms] ("diagnosis"[subheading] OR "diagnosis"[all Fields] OR "screening"[all Fields] OR "mass screening"[mesh Terms] OR ("mass"[all Fields] AND "screening"[all Fields]) OR "mass screening"[all Fields] OR "screening"[all Fields] OR "early detection of cancer"[mesh Terms] OR ("early"[all Fields] AND "detection"[all Fields] AND "cancer"[all Fields]) OR "early detection of cancer"[all Fields]) OR "prostate-specific antigen"[mesh Terms] outcome qualifier 3. Early diagnosis[mesh] early diagnosis[mesh Terms] PSA qualifier 4. PSA velocity[all Fields] PSA velocity[all Fields] 390 PSA qualifier PSA qualifier 5. Prostate specific antigen velocity[title/abstract] 6. PSA doubling time[title/abstract] Prostate specific antigen velocity[title/abstract] 128 PSA doubling time[title/abstract]

20 PSA qualifier Combination searches 7. Prostate specific antigen doubling[title/abstract] 8. #2 OR #3 OR #4 OR #5 OR #6 OR #7 Prostate specific antigen doubling[title/abstract] 152 #2 OR #3 OR #4 OR #5 OR #6 OR # Combination searches 9. #1 AND #8 #1 AND # (#1 AND #8) AND 10. Limit 9 to English[lang] AND (("2011/07/01"[PDAT] : Randomized Controlled Trial[ptyp] "2012/11/01"[PDAT]) AND AND Publication Date from Randomized Controlled Limits 2011/07/01 to 2012/11/01 Trial[ptyp] AND English[lang]) 102 Total Number of Hits

21 Figure 1: Selection of studies from bridge search of screening for prostate cancer PubMed search (n=102) Rejected (n=85): P=31 I=40 C=2 O=6 Study design=6 Full-text review (n=17) Accepted references (n=7) Q1(screen): n=5 PLCO(1) ERSPC(4) Q3(treat): n=2 Wilt Johannson Reject (n=10) I=2 C=5 O=1 Study design=2 A literature search to inform the definition of family history of prostate cancer was performed on 6/8/2013 (see Table 2). From the 1594 studies retrieved, 44 studies were selected for further review; of these, 4 studies provided relevant information (see Figure 2). 16, 17, 18, 19 21

22 Table 2: Literature search: definition of family history Search Query Items found #12 Search Prostatic neoplasms [MeSH] #13 Search Prostate cancer [tiab] #14 Search Prostatic cancer [tiab] 6379 #15 Search Prostate neoplasm [tiab] 29 #16 Search Prostatic neoplasm [tiab] 58 #17 Search Prostate adenocarcinoma [tiab] 1636 #18 Search Prostatic adenocarcinoma [tiab] 2224 #19 Search (#12 OR #13 OR #14 OR #15 OR #16 OR #17 OR 18) #20 Search Family Health [MeSH] #21 Search Genetic predisposition to disease [MeSH] #22 Search Family history [tiab] #23 Search First-degree relative [tiab] 1595 #24 Search First-degree relatives [tiab] 7418 #25 Search (#20 OR #21 OR #22 OR #23 OR #24) #26 Search Death [tiab] #27 Search Deaths [tiab] #28 Search Survival [tiab] #29 Search Mortality [tiab] #30 Search Mortality [MeSH] #31 Search Survival Rate [MeSH] #32 Search Survival Analysis [MeSH] #33 Search (#26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32) #34 Search (#19 AND #25 AND #33)

23 Figure 2: Selection of studies: definition of family history PubMed search (n=1594) High-level rejection (n=1550) Full-text review (n=44) Reject (n=35) P=14 I=8 C=0 O=10 Study design=2 Duplicate=1 Accepted references (n=9) 23

24 EVIDENCE SYNTHESIS Key Study Characteristics Five RCTs examining the effect of prostate cancer screening on mortality are included in the EPC review (Quebec, Norrkoping, Stockholm, ERSPC, PLCO), and are synthesized in a meta-analysis by The Cochrane Collaboration. 20 Studies were published between Subsequent to the EPC review, updated results from the PLCO and ERSPC trials were published in Study sizes ranged from 9, ,387; median follow-up time ranged from years. Studies differed in screening protocol, screening interval and biopsy threshold. The EPC report rated PLCO and ERSPC as fair quality evidence; the remaining 3 RCTs were considered poor quality. Key study characteristics of individual studies are found here; narrative of study characteristics and results are found here. Although 5 trials were included in the analysis, special attention was given to the PLCO and ERSPC trials, as they were significantly larger and of higher quality than the other screening RCTs. These two RCTs differed in several important characteristics, including setting, screening protocol, screening interval and biopsy threshold. Table 3 compares characteristics of the PLCO and ERSPC trials. Table 3: Comparison of ERSPC and PLCO Characteristic Study ERSPC PLCO Sample size 162,388 (core group ages 55-69) 76, ,999 (entire group ages 50-74) Trigger for biopsy PSA 3 ng/ml, with some variation >4 ng/ml among individual sites Screening interval Q 4 years for 87%; Q 2 years for Q yearly 13%) Median follow up 11 years 13 years Screening protocol PSA Years 1-4: PSA+DRE; Years 5-6: PSA only Primary endpoint Prostate cancer-specific mortality Cause-specific mortality for each of the PLCO cancers Age Range: (core group) (entire group) Range: Base population Population registries in 8 EU 10 centers in US countries Inclusion/Exclusion Not well defined. Exclusion: History of a PLCO cancer Current cancer treatment Having had more than one PSA blood test in the previous 3 years (starting in 24

25 Compliance (Screened at least once) Contamination 1995). Mean (range) = 83% ( ) PSA: 85% DRE: 86% It has been estimated that in the In the control group, the rate of control group, ~20% of men per PSA testing was 40% in the first year underwent PSA screening. year and increased to 52% in the sixth year. The rate of screening by digital rectal examination in the control group ranged from 41 to 46%. Results Results of the key prostate cancer screening trials are organized by data reported prior to 2012 (and therefore included in the Cochrane meta-analysis and EPC review for the USPSTF), and additional trial updates (for ERSPC and PLCO) reported in We performed an original meta-analysis, incorporating the more recent publications, below. Updated ERSPC and PLCO results Table 4 shows detailed results of these trial updates published in ERSPC results showed a 21% relative decrease (RR [95% CI] = 0.79[ ]) and a 0.1% absolute decrease in prostate cancer-specific mortality for a pre-specified subgroup of men, aged Results for the entire study population were similar, with a 17% relative decrease (RR [95% CI] = 0.83 [ ]) and a 0.08% absolute decrease in prostate cancer-specific mortality. The most recent PLCO results failed to show any statistically significant difference in prostate cancer mortality between the screening and control groups (RR[95%CI]=1.09[ ]) meta-analysis results We acknowledged the existence of clinical heterogeneity, as the design of some RCTs differed from each other with respect to screening frequency and biopsy threshold. It was decided that a meta-analysis was still appropriate, and chose to acknowledge clinical heterogeneity in our GRADE assessment of quality of evidence by downgrading for indirectness. We chose to include all 5 RCTs (using the updated results from the ERSPC and PLCO studies) in our primary meta-analysis since this was reflective of the most current body of evidence and used a random-effects model to account for heterogeneity between studies. A sensitivity analysis was conducted comparing a fixed effect to a random effects model. Further sensitivity analyses were done on the basis of evidence quality, performing separate analyses for low vs. high risk of bias studies. With the exception of using a fixed effects model with the updated results of the two studies with the lowest 25

26 risk of bias, the sensitivity analyses consistently showed an absence of a statistically significant reduction in prostate cancer-specific mortality. Table 5 compares meta-analyses of the 5 screening studies reported in the EPC 6 and Cochrane 20 reviews to those incorporating 2012 ERSPC and PLCO updates (for the endpoint of prostate cancer-specific mortality). We report results separately by fixed vs. random effects modeling techniques, and also report the different results for pooling only the two most high-quality trials (ERSPC and PLCO), as opposed to pooling all 5 trials. A meta-analysis of the 2012 ERSPC and PLCO results alone (using fixed effects) showed a 13% relative decrease in prostate cancer mortality and 0.07% absolute decrease; however, no statistically significant effect was seen when using a random effects analyses. There was considerable statistical heterogeneity, which may be partially explained by the study differences highlighted in Table 3 (above). As both the PLCO and ERSPC studies suffered from contamination (40% and 20% of those in the controlled group received screening, respectively), it is possible these results underestimate the true effect of screening on prostate cancer mortality. No statistically significant reduction in prostate cancer mortality was seen when all 5 (most recent) studies were included in a meta-analysis. Figures 3 and 4 provide individual and pooled results in the meta-analyses, including Forest plots. An updated (2012) meta-analysis of the 2 studies reporting all-cause mortality is shown in Figure 5. No effect of screening on all-cause mortality was seen (RR(95% CI)=0.99( )). Pre-2012 results The Cochrane review also reported the effect of screening on Prostate cancer mortality stratified by risk of bias and age, as well as all-cause mortality and Prostate cancer diagnosis. A summary of these analyses are found here. To summarize, there were no differences in Prostate cancer mortality when stratified by risk of bias or age, and no effect on all-cause mortality. An updated analysis of all-cause mortality including 2012 ERSPC data showed similar results as the Cochrane analysis (see meta-analysis below). The Cochrane review considered the ERSPC and PLCO trials to be low risk of bias, with the remaining trials, high risk of bias. Details of the risk of bias assessment are found here. 26

27 Table 4: Recent ERSPC and PLCO results (2012) Study (Follow-up) Outcome Screen Control Relative risk (95% CI) Absolute risk (95% CI) ERSPC core group- ages years (excludes France) (11 yr F/U) Schroder Screen (n= 72,891) Control (n=89,352) ERSPC entire group-ages years (excludes France) (11 yr F/U) Prostate cancer Mortality Prostate cancer Mortality CI= 0.39/1000 personyears CI = 0.42/1000 personyears CI= 0.50/1000 personyears CI = 0.50/1000 personyears MR= 0.79( ) (unadjusted) MR= 0.71 ( ) (adjusted for compliance) MR = 0.83 ( ) ARR= 1.07 /1000 men RD= (-0.17,- 0.04) per 1000 person-years NNI= 936 to prevent 1 death over 11 years NND = 33 to prevent 1 death over 11 years ARR = 0.8/1000 men RD = (-0.14,-0.02) per 1000 person years Screen (n=82,816) Control (n=99,183) NNI=1,250 to prevent 1 death over 11 years ERSPC Sweden 21 (Goteborg)- ages years Prostate cancer Mortality CI = 0.53/1000 personyears CI = 0.95/1000 personyears MR = 0.56( ) ARR = 4.2/1000 men RD = per 1000 person-years Median 14 years of F/U Screen: (n=10,000 randomized,, 9952 evaluated) Control: (n=10,000 randomized,, 9952 evaluated) NNI=293( ) to prevent 1 death over 14 years 27

28 PLCO (13 yr F/U) Andriole Screen (n=38,340) Control (n=38,345) Prostate cancer Mortality Prostate cancer incidence CI = 0.37/1000 personyears CI= 108.4/ 10,000 personyears CI = 0.34/1000 personyears CI= 97.1/10,000 personyears Key: ARR: Absolute risk reduction ARI: Absolute risk increase CI: Cumulative incidence MR: Mortality ratio NND: number of cancers needed to detect NNI: number needed to invite to Prostate cancer screening RD = Risk difference MR=1.09 ( ) RR=1.12 ( ) ARI = 0.3/1000 men RD = per 1000 person-years Table 5: Meta-analyses: Comparison of prostate cancer-specific mortality metaanalyses previously reported in 2011 with subsequently calculated meta-analyses incorporating 2012 ERSPC and PLCO trial results Evidence N Fixed effects-mh Existing evidence (pre-2012) weighting Random effects-iv weighting (Not published) ERSPC+PLCO 258, ( ) 0.93 ( ) (low RoB) Norrkoping + Quebec + Stockholm 82, ( ) 1.05 ( ) (high RoB) All existing evidence (5 studies) 341, ( ) 0.95 ( ) Updated evidence (2012 ERSPC, PLCO updates) ERSPC*+PLCO 238, ( ) 0.92 ( ) All 5 studies including updated results 321, ( ) 0.97 ( ) *Includes core ages for ERSPC (55-69 yrs) 28

29 Figure 3: Meta-analysis of 2012 ERSPC+PLCO results only: (a) random effects; (b) fixed effects (a) ERSPC+PLCO Prostate cancer mortality Random effects IV weighting 29

30 (b) ERSPC+PLCO Prostate cancer mortality Fixed effects MH weighting 30

31 Figure 4: All studies of Prostate cancer screening, incorporating updated ERSPC & PLCO results (a) Random effects IV weighting 31

32 (b) Fixed effects MH weighting 32

33 Figure 5: All-cause mortality with updated ERSPC results 33

34 Quality Assessment Appendix 3 describes the critical appraisal of individual RCTs included in the EPC review. Both the ERSPC and PLCO trials were considered fair quality by AHRQ and low risk of bias by Cochrane; the remaining 3 trials (Quebec, Norrkoping, Stockholm) were considered poor quality by AHRQ and high risk of bias by Cochrane. Using GRADE methodology, the evidence for the effect of PSA-based screening on Prostate cancer mortality, including all 5 RCTs, was considered low quality (Table 6). If studies with high risk of bias were eliminated, evidence would be considered of moderate quality, with downgrades only for indirectness due to differences from Kaiser Permanente practices in screening protocols and biopsy threshold. As previously noted, a meta-analysis of most recent results of ERSPC + PLCO trials showed statistical significance only when using fixed-effects modeling techniques. Evidence was considered moderate quality for the 2 RCTS reporting all-cause mortality. Table 6: Quality assessment of evidence examining the benefits of PSA screening No of studie s Design Prostate cancer specific mortality 5 randomiz ed trials All-cause mortality 2 randomiz ed trials Quality assessment No of patients Effect Risk of Inconsisten Indirectne Imprecisi bias cy ss on seriou s 1 no seriou s risk of bias 4 no serious serious 2 inconsisten cy no serious serious 2 inconsisten cy no serious imprecisi on no serious imprecisi on Other consideratio ns Screening for prostate cancer none 3 683/ (0.47%) none / (19.8%) Control 1285/ (0.73%) 27584/ (24.2%) Relativ e Absolut (95% e CI) RR 0.97 (0.82 to 1.14) RR 0.99 (0.97 to 1.01) 0 fewer per 1000 (from 1 fewer to 1 more) Quality LOW 2 fewer per MODERA 1000 TE (from 7 fewer to 2 more) Importan ce CRITICA L CRITICA L 1 At least 3 of 5 studies had high Risk of Bias; the remaining 2 studies had significant contamination issues. 2 Some studies differed from KP practices with respect to screening frequency and biopsy threshold. Screening frequency ranged from 1 to 7 years and biopsy thresholds ranged from PSA of 3.0 to10.0 ng/ml. Kaiser Permanente uses age-specific biopsy thresholds, and 1 to 2 year screening intervals. 3 Due to limited # of studies, it is difficult to assess publication bias; therefore cannot be ruled out. 4 Although Stockholm trial has high risk of bias, ~90% of weight in meta-analysis contributed by ERSPC with low RoB 34

35 HARMS OF PROSTATE CANCER SCREENING Clinical Question: What Are the Harms of PSA-Based Screening for Prostate Cancer? Population: Intervention: Comparison Critical Health Outcomes Men without a prior diagnosis of prostate cancer and without bone pain or constitutional symptoms PSA-based screening* No prostate cancer screening Critical outcomes: False-positive rates Harms associated with biopsy: urinary retention, hospitalizations Minor harms: Physical harms associated with screening test Psychological harms of testing Lower urinary tract symptoms and erectile dysfunction are not indications for ordering a PSA test. *PSA-based screening is defined as a screening program for prostate cancer in asymptomatic men that incorporates one or more PSA measurements, with or without additional modalities such as digital rectal examination or transrectal ultrasonography. Evidence on harms of screening was derived from the same RCTs used to evaluate screening benefits. Two RCTs (ERPSC, PLCO) reported results on harms of screening. False positive rates False positive (FP) rates were available from both the ERSPC and PLCO trials. The definition of a false positive screen differed across trials, and false positive rates varied widely across ERSPC centers. Some of the variation may be attributed to differences in screening protocol. The harms associated with a false positive result include unnecessary biopsies and associated physical harms resulting from biopsy (e.g. hospitalizations, urinary retention) as well as psychological harms (e.g. anxiety). A study published subsequent to the EPC report, Kilpelainen reported false positive rates from 5 ERPSC centers with 61,604 participants who had at least one prostate cancer screening. Screening interval ranged from 2-7 years with a PSA threshold of ng/ml. A false positive result was defined as a positive screen with no histologically confirmed Prostate cancer by biopsy within 1 year. For men who were screened at least once, 17.8% had one or more false positive results, ranging from 10.5% to 22.3% across the 5 centers. For those who participated in every screening round (n=22,068), 19.0% had at least one false result. The absolute risk of a screen- 35

36 detected Prostate cancer in the next round following a previous false positive (at first or second round) was 10%. The most recent PLCO publication (Andriole 2012) 7 did not report false positive results; however the 2008 EPC publication reported false positive results 4. After four PSA tests, men had a 12.9% cumulative risk of receiving at least one false-positive result (defined as a PSA level of 4.0 ng/ml and no prostate cancer diagnosis after 3 years), and a 5.5% risk of having at least one biopsy as a direct consequence of a false-positive screening test. Harms of biopsy The PLCO study reported a biopsy complication rate of 68 events per 10,000 evaluations, including infection, bleeding, and urinary difficulty. 23 Statistics for individual events were not available, and no information regarding severity of adverse events was provided. The Rotterdam center of ERSPC defined major complications of biopsy as adverse effects causing significant discomfort, disability, or requiring additional treatment. 24 These events were self-reported by patient and recorded by an urologist 2 to 3 weeks after biopsy. Hospitalization for prostatitis or urosepsis and urinary retention were rare: 0.5% and 0.4% for hospitalization and urinary retention, respectively. Other serious events can be found in Table 1. Minor events not requiring intervention include: Hematuria > 3 days post biopsy (22.6%); and hematospermia> 3 days post biopsy (50.4%). Physical harms of screening Non-serious physical harms of screening reported in the PLCO trial included bleeding or pain from digital rectal examination (0.3 events per 10,000 men screened), and bruising or fainting due to venipuncture (26.2 events per 10,000 men screened), Psychological harms of screening The EPC report informing the 2008 USPSTF prostate cancer screening guidelines found evidence that false-positive results were associated with adverse psychological events; however, the authors could not determine the magnitude of psychological harm. Quality Assessment The overall quality of harms of screening was considered low; individually, the evidence on false positive results is graded as moderate quality, with downgrading for indirectness only (Table 7). The evidence for both biopsy-related urinary retention and hospitalizations was also downgraded for imprecision. 36

37 Table 7: Quality assessment of evidence examining the harms of PSA screening No of studies Design False positive results 2 randomized trials Quality assessment Risk of bias Inconsistency Indirectn ess no serious risk of bias Biopsy related urinary retention 1 randomized trials no serious risk of bias Biopsy related hospitalizations 1 randomized trials no serious risk of bias Imprecis ion no serious inconsistency no serious inconsistency Other considerat ions No of patients PSA base d testin g Contr ol Relati ve (95% CI) none not pooled serious 2 serious 4 none not pooled serious 2 serious 4 none not pooled 1 Heterogeneity is explained by differences in protocols 2 Differs from KP with respect to biopsy thresholds 3 Unable to determine, but no reason to suspect existence of publication bias 4 Low event rates Effect Absolu te not pooled not pooled not pooled Qualit y MOD- ERATE LOW LOW Importan ce no serious inconsistency 1 serious 2 no serious imprecision IMPOR- TANT CRITICAL CRITICAL 37

38 Treatment of localized prostate cancer Clinical Question: What are the benefits and harms of treatment of early-stage or screen-detected prostate cancer? Population: Include: Men with screen-detected or early-stage prostate cancer (defined as stage I or II) Exclude: Men with later-stage prostate cancer, men with refractory, hormone refractory, or recurrent prostate cancer Intervention: Most important interventions: Surgery (radical prostatectomy, including different surgical techniques, such as nerve sparing, robotic) Radiation therapy (external-beam radiation therapy, brachytherapy, and combination therapies) Less important interventions: Androgen deprivation therapy (androgen deprivation therapy via luteinizing hormone-releasing hormone agonists, antiandrogen therapy, and/or orchiectomy) Cryotherapy Ultrasonography (high-intensity focused ultrasonography- HIFU) Comparison Critical Health Outcomes Exclude: Chemotherapy Active surveillance Watchful waiting No treatment Benefits Mortality (overall and disease-specific) Serious harms: Erectile dysfunction: prostatectomy, radiation therapy Urinary incontinence: prostatectomy, radiation therapy Bowel dysfunction: radiation therapy Surgical complications: prostatectomy Non-serious harms: Quality of life (overall and disease-specific) 38

39 Function (overall and disease-specific) Psychological effects (e.g., mental status, depression, cognitive dysfunction) Endocrinological effects (e.g., bone health, hot flashes, gynecomastia) Methodology The literature search conducted by the EPC examined studies published up to July, A bridge search was done in November 2011 to identify all relevant studies published subsequent to July The search strategy and other methodological details from the EPC s evidence synthesis may be found here, with details of search strategy found in Appendix B1. Acceptable study designs included RCT, cohort and large case-series (n 1000) for harms data. Smaller case-series studies were acceptable if no larger studies were available. To be included, studies need to have follow-up of 30 days for perioperative complications and >12 months for other harms. Included studies reported all-cause mortality, prostate cancer-specific mortality, or pre-specified harms and compared radical prostatectomy, radiation therapy (EBRT or brachytherapy), ADT, cryoablation, or HIFU with watchful waiting or active surveillance. Although a narrative of the harms of less common treatments were included (ADT, cryoablation, HIFU), this report concentrates on major harms of the 2 most common treatments, prostatectomy and radiation therapy. Search Results A QUOROM figure describing the EPC s search results may be found in Appendix B3. From a potential 7,920 abstracts, 2 RCTs and 9 cohort studies were included for benefit, and 2 RCTs, 14 cohort and 11 case-series were included for harms of treatment. Literature search update To update the EPC evidence, a bridge search was conducted of OVID and Cochrane databases. The search strategies are consistent with that of the EPC s systematic review. Search results are shown in Table 8, and study selection in Figure 6. Among 459 unique articles retrieved, one new RCT 8 (PIVOT), one follow-up to a previously 39

40 reported RCT 25 (SPCG-4) and one cohort study 26 were identified and included in this review. No systematic reviews were identified in the bridge Cochrane search. 40

41 Table 8: Literature search: Treatment of localized prostate cancer Dates of search: 7/1/ /1/2012 Database: Ovid Search String Database: Ovid Medline Number of hits 1 Prostatic Neoplasms/dh, dt, rt, su, th, us prostate cancer.mp. or Prostatic Neoplasms/ Treatment Outcome/ and or (ae or co or de or mo).fs (adverse and (effect$ or event$)).mp (safe$ or harm$ or side effect$).mp or/ Quality of Life/ Anxiety/ Depression/ px.fs or/ and (9 or 14) and (2011$ or 2012$).ed not (case reports or comment or editorial or letter).pt limit 17 to (English language and humans) 939 remove duplicates from

42 Figure 6: Selection of studies from bridge search of treatment for prostate cancer Ovid search (n=459) Full-text review (n=107) Rejected (n=352): P=110 I=118 C=8 O=23 Study design=93 Reject (n=104) P=4 I=6 C=29 O=2 Study design=63 Accepted references (n=3) Abdollah Wilt(PIVOT)-duplicate Johannson-duplicate Date of search: 11/29/12 Database: Cochrane Search terms: prostate cancer Search dates: Cochrane reviews: n=27 Relevant to screening/treatment published after 7/2011: 0 Other reviews: n=16 Relevant to screening/treatment: 0 42

43 EVIDENCE SYNTHESIS Key Study Characteristics Eleven studies on benefits of treatment (2 RCT, 9 cohort) and 16 harms studies (2 RCTs, 14 cohort) were included in the EPC review. Sample sizes ranged from 72 to 44,630 and duration of follow-up from 1 to 23 years. The evidence for both benefits and harms of prostatectomy and radiation therapy were considered fair (by AHRQ criteria); two cohort studies reporting benefits of androgen deprivation therapy were considered of poor quality; no controlled studies of cryotherapy or HIFU meeting inclusion criteria were identified. A high-level overview of study characteristics are found in Table 4, with more detailed information in Appendix C1 (benefits of treatment) and Appendix C2 (harms of treatment). Studies comparing radical prostatectomy (RP) to watchful waiting (WW) identified subsequent to EPC report include the PIVOT trial 14 and one cohort study using SEER data 26 as well as an update on harms from the SPCG-4 trial. 15 Study characteristics are found in Table 9. 43

44 Table 9: Study characteristics of publications identified subsequent to AHRQ systematic review (a) Study Characteristics Relevant Population Characteristics Study/Desig n Wilt 2012 (PIVOT) N Length (followup) RCT 731 Median = 10 yr Intervention (# events/[%]) Critical Outcome All-cause mortality Prostate cancer mortality Bone metastasis Periop complications Interventions n Mean Age (yrs) % Black Median PSA (ng/ml) Mean Gleaso n grade 44 Low risk* (%) Moderat e risk (%) Radical Prostatecto my (RP) (5.2) (1.5) Watchful Waiting (WW) (5.6) (1.6) Compariso n (# events/[%]) 47.0% 49.9% HR=0.88 ( ) 5.8% 8.4% HR=0.63 ( ) 4.7% 10.6% HR=0.40 ( ) 21.4% HR (95%CI) ARR Sensitivity analyses ARR=-2.9 % (-4.1,+10.3) ARR= -2.6% (-1.1, +6.5) ARR = -5.9% Prostate cancer mortality: No interaction: age, self-reported performance, urinary incontinence. Non statistically significant trend toward treatment benefit in men with PSA>10 mg/dl and higher tumor risk Primary Objective: The primary aim of PIVOT is to compare all-cause mortality between the radical prostatectomy and watchful waiting groups. Secondary aims include comparison of prostate cancer mortality, progression and disease specific quality of life. Primary Outcome(s):The primary endpoint is all-cause mortality. Secondary endpoints include prostate cancer specific mortality; and biochemical (PSA), local and metastatic progression. Additional outcomes are general and disease specific quality of life and harms of therapy including: 30-day perioperative mortality and morbidity; and longer-term urinary, bowel and erectile dysfunction. High risk (%)

45 Inclusion/Exclusion: Eligible men had to have biopsy proven clinically localized prostate cancer (T1 T2, N M0) of any histologic grade, diagnosed within the past 12 months, PSA =50 ng/ml, age 75 years, bone scan negative for metastatic disease, an estimated life expectancy of at least 10 years and judged to be medically and surgically fit for radical prostatectomy. Additional Treatment Info: Type of radical prostatectomy (e.g. retropubic, transperineal, use of lymph node dissection) was left to the discretion of the operating surgeon. Additional follow-up and interventions were also at the discretion of the participant's physician but the type and reason recorded. Men randomized to watchful waiting were offered palliative (noncurative) therapies (e.g. transurethral resection of the prostate for local progression causing urinary obstruction, androgen deprivation and/or targeted radiation therapy for evidence of distant spread). Risk Of Bias: Sequence Generation: Low: Details about randomization technique not described fully, but no reason to believe ineffective randomization, i.e. groups were well-balanced Allocation Concealment: Low: Randomization was stratified by site and was implemented via a central interactive telephone system. Blinding: Unclear: Study design precludes blinding of clinician and participant; however primary outcomes are objective and are not likely to be impacted by knowledge of treatment group. Third party histology assessment performed on biopsies and prostatectomy samples. Incomplete Outcome Data: unclear: Intention to treat (ITT) analysis performed, Cox models account for Person-Time on study, however no description of missing outcome data. Selective Outcome Reporting: Low: Based on methodology publication Other: Potential under-estimation of effect due to contamination: 21% in RP group did not receive treatment, 10% in WW group received RP. * Tumor risk category (based on PSA, Gleason grade and tumor stage) 45

46 (b) Study Characteristics Relevant Population Characteristics Study/Desig n Johansson 2011 (extension of SPG4) N Length (follow-up time) Median = 12.2 y Interventions Radical Prostatectom n Mean Age (range) RP Androgen deprovatio ntx Antiantidroge n tx y (RP) (63-85) 94% 19% 12% % 28% 23% (61-88) Watchful waiting (WW) Critical Outcome RP (# events/total [%] WW (# events/total [%] High QoL 62/179 (35%) 55/160 (34%) 0.98( ) Anxiety 77/178 (43%) 69/161 (43%) 0.97 ( ) ED 146/ /153 Not reported (84%) (80%) Depression 85/180(47%) 82/159 (52%) 0.92 ( ) in physical 45% 60% Not reported symptoms Reduction in 61% 64% Not reported QoL RR Primary Objective: To assess long-term quality of life and functional outcomes in men with localized prostate cancer treated with radical prostatectomy or watchful waiting Primary Outcome(s): Quality of Life; functional status Inclusion/Exclusion: All Swedish and Finnish men enrolled in the SPG4 trial randomized to radical prostatectomey or watchful waiting alive at 12 year follow-up. Risk Of Bias: High this should be analyzed as an observational trial. No multivariable analysis reported, therefore effect estimates are likely confounded. Selective outcome reporting possible as primary outcomes were not pre-specified or well defined. 46

47 (c) Study Characteristics Relevant Population Characteristics Study/design N Length (follow -up time) Abdollah Cohort (SEER Medicare data) Critical Outcome Prostate cancer mortality Interventions 44, yr Radical Prostate ctomy (RP) Watchful waiting (WW) RP WW % (95%CI) 2.0% ( ) % (95%CI) 5.8% ( ) n Mean Age (range ) 22, (65-80) Clinical stage T1 (%) T2 a,b (%) T2c (%) Tumor Grade (Gleason score) 2-5 (%) 6-7 (%) 8-10 (%) , (65-80) RRRunivariate RRR = 0.5% (p<0.001) HRmultivariate HR=0.48 ( ) ARR ARR=3.0% Primary Objective: To examine cancer-specific mortality (CSM) after accounting for other-cause mortality (OCM) in Prostate cancer patients treated with either RP or observation. Primary Outcome(s): Cancer specific mortality (CSM) and other causes of mortality (OCM) Inclusion/Exclusion: Patients were eligible if they had non-metastatic Prostate cancer and both Medicare Part A and Part B claims available and who were not enrolled in a health maintenance organization throughout the duration of the study. Patients were not included if Prostate cancer was diagnosed at autopsy or on death certificate only, or if their original or current reason for Medicare entitlement was listed as disability or a Medicare status code including disability. Patients with T3/T4 tumors, anaplastic or unknown grade, unknown stage, aged 80 years at diagnosis or missing socioeconomic information were excluded. Risk Of Bias: Risk of bias (RoB) due to selection bias and/or confounding is minimized by use of propensity score matching as well as multivariate analysis. However, there is the potential for residual confounding due to unmeasured confounders as data is derived from claims. Selection bias /confounding by indication is likely as those in the observation group were more likely to be diagnosed at an earlier clinical stage. 47

48 Results A summary of the evidence included in the EPC report is found in the report s Table 11. Prostatectomy Benefits Two RCTs reported mortality outcomes. At 12 year follow-up, the Scandinavian Prostate Cancer Group-4 (SPCG-4) found that prostatectomy was associated with decreased risk of prostate cancer-specific mortality (RR (95% CI) = 0.65 (0.45 to 0.94), ARR (95%CI)= -5.4% (-0.2,- 11.1)), and distant metastasis (RR=0.65( ), ARR=-6.7%(-0.2,-13.2)),but not all-cause mortality (RR= 0.82 (0.65 to 1.03, ARR=- 7.1 % (-0.5, +14.7)). Based on subgroup analysis, benefits were seen only in men younger than 65 years. At 12 years the PIVOT trial did not find a statistically significant mortality benefit for prostatectomy over observation: Prostate cancer mortality (HR (95%CI)=0.63 ( ), ARR= -2.6% (-1.1, +6.5)); all-cause mortality (HR=0.88 ( ), ARR=-2.9 (- 4.1,+10.3)). Of note, the PIVOT trial reported that those treated with prostatectomy were less likely to develop bone metastasis (HR= 0.40 ( )). In addition, age did not modify the effect of treatment on mortality in PIVOT, however only 10% of men were less than 60 years. Results suggest higher PSA levels (>10 ng/ml) and higher risk tumors 9 were associated with a significant benefit of prostatectomy over observation. Compared with SPCG-4, PIVOT enrolled a higher percentage of men with non-palpable tumors (stage T1c, 50% vs. 12%) and with PSA values 10 ng/ml. The majority of men enrolled in the SPCG-4 trial had more advanced stage tumors (T2 tumors =75%). Based on these two good quality RCTs, overall, the results suggest prostatectomy does not result in reduction in all-cause mortality, but may provide protection against distant metastasis. The impact on Prostate cancer-specific mortality is unclear but may suggest a benefit in younger men with intermediate or high risk disease. More studies are needed to determine which men with localized prostate cancer would most benefit from surgical intervention. Results from cohort studies consistently showed a mortality benefit of prostatectomy; however confounding by indication cannot be ruled out, that is, those who were more likely to benefit from prostatectomy may have more often received prostatectomy. 9 D Amico tumor risk score [low, intermediate, or high], which was based on tumor stage, histologic score, and PSA level 48

49 Harms Urinary Incontinence: In the two treatment RCTs, prostatectomy increased the absolute risk of urinary incontinence by 10.8% to 27.5% in men with localized prostate cancer compared with similar men who were being followed with watchful waiting. In the watchful waiting group, 21.3% (SPCG-4) and 6.3% (PIVOT) reported urinary incontinence, compared with 48.8% (SPCG-4) and 17.1% (PIVOT) in the prostatectomy group. Pooled results from cohort studies showed similar trends (EPC Figure 2). Risk of urinary incontinence is reported from two RCTs and four cohort studies: SPCG-4: RR=2.29( ), ARI = +27.5% PIVOT: ARI = +10.8% (p<0.001) [published after the EPC report] Meta-analysis of four cohort studies: RR=3.68( ), ARI = +19.1% Erectile dysfunction: In the two treatment RCTs, prostatectomy increased the risk of erectile dysfunction by approximately 37.0% in men with localized prostate cancer compared with similar men who were being followed with watchful waiting. Results from the two RCTs (SPCG-4, PIVOT) show ~44% risk of erectile dysfunction in the watchful waiting group, compared with ~81% in the prostatectomy group. Pooled results from cohort studies showed similar trends (EPC Figure 3). Risk of erectile dysfunction is reported from two RCTs and five cohort studies: SPCG-4: RR=1.79( ), ARI = +35.6% PIVOT: ARI = +37.0% (p<0.001) Meta-analysis of five cohort studies: RR=1.56( ), ARI = +24.6% Surgical complications (30-day perioperative period): A perioperative mortality risk of ~0.5% and 0.6% to 3% risk of cardiovascular events were reported in several caseseries studies. Serious rectal or ureteral injury attributed to surgery ranged from 0.3% to 0.6%. Details of study characteristics reporting adverse outcomes are found in Appendix C5. In the PIVOT trial, 21.4% (n=60) of patients undergoing prostatectomy reported 84 perioperative events. Although the study did not define criteria for serious adverse events, 39% of events would be considered serious, including rare death, MI, stroke, pulmonary embolism, sepsis, renal failure, surgical repair, urinary catheter required >30 days post-surgery, and bleeding requiring transfusion. A detailed discussion of evidence for prostatectomy from the EPC report may be found here: Benefits, harms. 49

50 Radiation therapy Benefits No RCTs examining the benefits of radiation therapy were identified. Results from 5 cohort studies showed radiation therapy was associated with decreased prostate cancer mortality (median adjusted HR= 0.66, range=0.63 to 0.70)and all-cause mortality (median adjusted HR = 0.68, range = 0.62 to 0.81). Refer to Table 6 and Appendix C1 for more details. Harms Urinary incontinence (EPC Figure 4): Results from one small RCT suggested radiation therapy increased the risk of urinary incontinence by 15% in men with localized prostate cancer compared with similar men who were being followed with watchful waiting. In the watchful waiting group, 2.0% reported urinary incontinence, compared with 17.0% in the radiation group. One RCT showed increased urinary incontinence, while pooled results of four cohort studies did not show a statistically significant increase in urinary incontinence: RCT 27 : RR = 8.31 ( ), ARI = +14.9% Meta-analysis of four cohort studies: RR= 1.29( ), ARI = +3.8% Studies were small and effect estimates were imprecise. Erectile dysfunction (EPC Figure 5): Pooled results from five cohort studies show radiation therapy is associated with a greater risk of erectile dysfunction, compared to watchful waiting: RR = 1.33 ( ), ARI = +15.3%. In pooled results of the watchful waiting group, 42.3% reported erectile dysfunction, compared with 57.6% in the radiation group. Bowel dysfunction: Six cohort studies found that radiotherapy was associated with worse PCI (UCLA prostate cancer index) bowel bother 10 (median difference, -8 points [range, -15 to -3]) and function (median difference, -6 points [range, -10 to -2]) compared with watchful waiting (EPC Table 9). In studies that evaluated bowel function serially, effects appeared most pronounced in the first few months after radiation therapy and gradually improved. Data from one study suggest that low-dose brachytherapy may be associated with fewer harms compared with high-dose brachytherapy or EBRT. There were no clear adverse effects related to general health-related quality of life. 10 Bowel bother refers to the degree of distress or annoyance caused by any impairments in bowel function (from UCLA prostate cancer index- J Urol Aug; 172(2):515-9) 50

51 A detailed discussion of evidence for radiation therapy from the EPC report may be found here: Benefits, harms Androgen deprivation therapy No benefits from androgen deprivation therapy were found. Two poor-quality cohort studies found a 30% to 80% increase in relative risk of prostate cancer mortality. Cohort studies found an increased risk of erectile dysfunction, as well as androgen deprivation related effects of gynecomastia and hot flashes. A detailed discussion of evidence for androgen deprivation therapy from the EPC report may be found here: mortality, harms Cryotherapy No studies examining mortality benefits of cryotherapy were identified. One small, poorquality cohort study reported adverse effects of cryotherapy. Men in cryotherapy group were more likely to experience erectile dysfunction and men over 70 years were found to have increased urinary incontinence. A detailed discussion of the evidence regarding cryotherapy may be found here: harms High-intensity focused ultrasonography (HIFU) No studies examining mortality benefits of HIFU were identified. No RCTs or cohort studies reported adverse effects. Case-series studies reported rates of adverse effects: ~50% rate of erectile dysfunction (two studies) and urinary incontinence rates of 2%- 11%. More details may be found here: harms Quality assessment Quality assessments of individual studies included in the EPC report are found in Appendices c3 (RCTs) and c4 (cohort studies). Of three RCTs, one (SPCG-4) was considered good quality; the other two were rated poor to fair. One additional RCT reported subsequent to EPC report (PIVOT) is considered low risk of bias. Of 24 cohort studies, 3 were considered good quality; the remainder was rated as fair. The overall quality for the 2 RCTs examining the benefits and harms of prostatectomy (PIVOT, SPCG-4) was considered moderate for studies examining benefits of treatment (all-cause and prostate cancer-specific mortality, metastasis), downgrading only the outcome prostate cancer mortality for imprecision (Table 10). The evidence for harms of prostate cancer treatment was considered high quality (urinary incontinence, erectile dysfunction). 51

52 Table 10: Quality assessment of evidence examining effect of prostatectomy vs. watchful waiting on localized prostate cancer No of studies all-cause mortality 2 randomized trials Quality assessment Design Risk of bias Inconsistency Indirectness Imprecision no serious risk of bias prostate cancer mortality 2 randomized trials no serious risk of bias metastasis 2 randomized trials Uninary incontinence 2 randomized trials no serious risk of bias no serious risk of bias no serious inconsistency no serious inconsistency no serious inconsistency no serious inconsistency 3 no serious indirectness no serious indirectness no serious indirectness no serious indirectness no serious imprecision Other considerations No of patients RP WW none not pooled serious 2 none not pooled no serious imprecision no serious imprecision Erectile dysfunction 2 randomized no serious risk no serious no serious no serious trials of bias inconsistency indirectness imprecision 1 Not enough studies to examine publication bias, cannot rule out the potential for pub bias 2 Event rate low - total # events = Heterogeneity explained by difference in protocol/study population none not pooled none not pooled none not pooled Effect Relative (95% CI) Absolute not pooled Quality HIGH not pooled MODERATE not pooled not pooled not pooled HIGH HIGH HIGH Importance CRITICAL CRITICAL IMPORTANT CRITICAL CRITICAL 52

53 Table 11: Quality assessment: Radiation therapy vs. watchful waiting No of studies Design Risk of bias Prostate cancer-mortality (follow-up 4-13 years) 4 observational studies serious 1 no serious inconsistency All-cause mortality 5 observational studies serious 1 Quality assessment No of patients Effect Inconsistency Indirectness Imprecision no serious inconsistency urinary incontinence 1 randomized trials serious 1,3 no serious inconsistency Erectile dysfunction 6 observational studies Bowel dysfunction 2 observational studies serious 1 serious 1 no serious inconsistency no serious inconsistency no serious indirectness no serious indirectness no serious indirectness no serious indirectness no serious indirectness 1 Rated 'fair' by AHRQ criteria 2 Unable to evaluate for publication bias, although no reason to suspect 3 Inadequate documentation, high LTFU 4 Very low event rates, wide confidence intervals no serious imprecision no serious imprecision Other considerations Radiation therapy Control Relative (95% CI) Absolute none not pooled none not pooled serious 4 none not pooled no serious imprecision no serious imprecision none not pooled none not pooled not pooled not pooled not pooled not pooled not pooled Quality Importance VERY LOW VERY LOW LOW VERY LOW VERY LOW CRITICAL CRITICAL CRITICAL CRITICAL CRITICAL 53

54 Ages to initiate and discontinue screening Clinical Question Population Health Intervention Most important health outcomes In men who desire prostate cancer screening, what is the appropriate age to initiate and discontinue screening? Men without a prior diagnosis of prostate cancer and without bone pain or constitutional symptoms Begin screening for prostate cancer with PSA at age: 40 vs. 45 vs. 50 Stop screening for prostate cancer with PSA at age: 65 vs. 70 vs. 75 vs. 85 Prostate cancer-specific mortality All-cause mortality Lower urinary tract symptoms and erectile dysfunction are not indications for ordering a PSA test. Search Strategy The target study that could directly address this clinical question would be a RCT of PSA screening, randomizing men by different age groups. As no such direct evidence is available, the indirect evidence to support this recommendation comes from evaluation of benefits of screening in the age groups included in existing RCTs as well as SEER incidence and mortality data. A search of PubMed using the following search strategy was performed on 5/7/2013 (Table 12). No study design or publication date limits were imposed. Table 12: Search strategy: ages to initiate and discontinue prostate cancer screening Search Query Items found 1 Prostatic Neoplasms [MeSH] Prostate/pathology [MeSH] Prostate cancer [TIAB] Prostatic cancer [TIAB] Prostate neoplasm [TIAB] 29 6 Prostatic neoplasm [TIAB] 58 7 (#1 OR #2 OR #3 OR #4 OR #5 OR #6) prostate-specific antigen[mesh] Tumor markers, Biological/blood* [MeSH] (#8 OR #9) Early diagnosis [MeSH] Mass screening [MeSH]

55 13 Early detection of cancer [MeSH] Prospective Studies [MeSH] Longitudinal Studies [MeSH] Retrospective Studies [MeSH] (#11 OR #12 OR #13) (#14 AND #17) (#15 AND #17) Search (#16 AND #17) (#18 OR #19 OR #20) Age factors [MeSH] Time factors [MeSH] Predictive value of tests [MeSH] Risk factors [MeSH] Risk assessment [MeSH] Neoplasm Staging [MeSH] (#22 OR #23 OR #24 OR #25 OR #26 OR #27) (#7 AND #10 AND #21 AND #28)

56 Search results Of the 170 studies retrieved, no additional relevant information beyond the five RCTs of screening vs. no screening were identified (see Figure 7). Figure 7: Selection of studies: ages to initiate and discontinue prostate cancer screening PubMed search (n=170) High-level rejection (n=148) Full-text review (n=22) Reject (n=22) P=2 I=12 C=0 O=3 Study design=1 Duplicate = 4 Accepted references (n=0) Evidence Synthesis The target RCT to inform ages to initiate and discontinue prostate cancer screening would be a multi-arm RCT, randomizing groups to different ages to initiate and discontinue screening, with a non-screening arm as control. As no direct evidence exists, indirect comparisons of age groups included in the RCTs examining the benefits 56

57 of screening and descriptive statistics were used to inform this clinical question. Results are described below and summarized in Table 13. Table 13: Summary of Results: Ages to Initiate and Discontinue Prostate Cancer Screening Outcome: prostate cancer mortality Data source Age range (years) Prostate cancer mortality SEER data Risk per 100,000 men: whites: blacks: 2.37 Risk of bias NA. (Descriptive data) ERSPC Goteborg site ERSPC core group ERSPC subgroup PLCO subgroup RR(95% CI): 0.56( ) low RR(95% CI): 0.79( ) low 70+ RR(95% CI): 1.18( ) high RR(95% CI): 1.02( ) high Age to initiate prostate cancer screening Details of the screening RCTs are found in previous sections of this document. One ERSPC site (Sweden-Goteborg) included men under age 55, ranging from 50 to 64 years, with a median of 57 years. The rest of the ERSPC sites recruited men at least 55 years at enrollment. PLCO recruited men years. The ERSPC reported prostate cancer specific reduction in the total cohort, ages (RR (95%CI)=0.83( ), the core age group, ages (RR=0.79( ), and in the Swedish site (Goteborg), ages years (RR=0.56( ). No prostate cancer-specific mortality benefit was seen in the PLCO trial or the other three poor quality RCTs. Although results from ERSPC suggest younger men may benefit most from screening, the overall number of cancers in the age range was too low to provide significant results. The Goteborg site is the only site to have included men as young as age 50, whereas the other sites included men aged 55 and above. Age to discontinue prostate cancer screening Screening trials In the ERSPC trial, subgroup analysis of men randomized at age 70 years or older, found no prostate cancer-specific mortality benefit (RR (95%CI) = 1.18 ( ); ARI 57

58 (95%CI) = 0.20(-0.26, +0.66). These results should be interpreted with caution as imprecision and potential confounding factors limit validity: Event rates were low, and therefore estimates of effect were imprecise. As men were not randomized by age group, there is a potential for imbalance in important characteristics that may have confounded the effect of screening on prostate cancer mortality. In the PLCO trial, no prostate cancer-specific mortality benefit was seen in men ages 55-74, with results similar in sub-group analysis of men ages (RR=1.19( ) and (RR=1.02( ). Validity of the sub-group analyses is limited due to high rate of contamination, imprecision and potential confounding factors. Treatment trials A sub-group analysis of the SPCG-4 trial of radical prostatectomy vs. watchful waiting for the treatment of localized prostate cancer showed no difference between groups in either all-cause or prostate cancer-specific mortality in men ages 65 years and older (all-cause: RR=1.04( ); prostate cancer specific: 0.87( ). Post-hoc subgroup analyses of the PIVOT trial of radical prostatectomy vs. watchful waiting or active surveillance in the treatment of localized prostate cancer showed no difference between groups in either all-cause or prostate cancer-specific mortality for men ages 65 years or older (all-cause: RR=0.89 ( ); prostate cancer-specific: RR=0.63( ). More than 90% of the study population was older than 65 years of age. Prostate cancer incidence and mortality The National Cancer Institute s Surveillance, Epidemiology and End Results (SEER) Program report incidence and mortality from selected cancer sites. Rates from SEER prostate cancer incidence and US Mortality data are shown in Table 14 below. Prostate cancer incidence begins to increase after age 40 years and is about 3 times greater in African-Americans compared to white men. Mortality from prostate cancer also increases with age, but is rare under age 50 years, at and per 100,000 for whites and African-Americans, respectively. 58

59 Table 14: SEER prostate cancer incidence and mortality rates Prostate cancer incidence rates per Prostate cancer mortality rates per 100,000 men 100,000 men (SEER data ) (U.S. mortality data ) Age At Diagnosis AA (med=64y) White (med=67 y) All (med=67 y) Age AtDeath AA White All <1 ~ ~ ~ <1 ~ ~ ~ 1-4 ~ ~ ~ 1-4 ~ ~ ~ 5-9 ~ ~ ~ 5-9 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

60 Quality Assessment Age to initiate screening Indirect evidence to inform this question is derived from results of the ERSPC-Goteborg site. 21 The overall quality, based on results from the ERSPC-Goteborg center is considered moderate, downgrading for indirectness only (Table 15). Age to discontinue screening Indirect evidence to inform this question is derived from the ERSPC and PLCO trials. Overall evidence quality is considered very low, downgrading for risk of bias, imprecision and indirectness (Table 16). 60

61 Table 15: Quality assessment of studies evaluating age to initiate prostate cancer screening No of studies Design Risk of bias Quality assessment No of patients Effect Inconsistency Indirectness Imprecision prostate cancer-specific mortality(follow-up median 14 years) 1 randomized trials no serious risk of bias 1 no serious serious 3 no serious inconsistency 2 imprecision Other considerations Screening for prostate cancer begin at age 50 none 44/9952 (0.44%) Control 78/9952 (0.78%) Relative (95% CI) HR 0.56 (0.39 to 0.82) Absolute Quality 3 fewer per 1000 (from 1 fewer to MODERATE 5 fewer) Importance CRITICAL 1 randomization: low RoB; allocation concealment: unclear RoB blinding: low RoB, however only outcome assessors were blind. As outcome is objectively assessed, lack of blinding unlikely to result in significant bias; Incomplete outcome data: Low RoB-A small # of randomized were not analyzed (0.5%, similar in treatment groups). This is unlikely to result in significant bias. Selective reporting: low RoB - outcomes reported as pre-specifed in protocol; Other: low RoB: Although results analyzed one year prior to protocol's designation, unlikely to result in significant positive bias due to large magnitude of effect and increasing differences in effect over time. However, effect may be underestimated due to 24% lack of compliance in screening group. 2 Only one RCT included in this assessment 3 Indirect comparison: Not directly comparing different ages to initiate screening Indirect population: Ethnicity of Swedish population differs substantially from US. Biopsy protocols differ from KP standards. 61

62 Table 16: Quality assessment of studies evaluating age to discontinue prostate cancer screening (a) Screening trials No of studies Design Risk of bias Quality assessment No of patients Effect Inconsistency Indirectness Imprecision Other considerations Screened Control Relative (95% CI) Absolute prostate cancer-specific mortality 2 randomized serious 1 no serious serious 3 serious 4 none - - not pooled not pooled trials inconsistency 2 0% not pooled All-cause mortality 1 randomized trials serious 1 NA-only one trial serious 3 No serious imprecision none 2246/44402 PY (50.58 per 1000 PY) 2215/45285 PY (48.91 per 1000 PY) 1.03 ( ) 9 more per 1000 (-9, +35) Quality Importance VERY LOW LOW CRITICAL CRITICAL 1 Sub-group analysis; potential for confounding was not addressed in the analyses 2 Inconsistencies can be explained by study differences 3 Indirect population/protocol 4 Low event rates 62

63 (b) Treatment trials No of studies Design Risk of bias Quality assessment No of patients Effect Inconsistency Indirectness Imprecision treatment-pca mortality (follow-up median 12 years) 2 randomised trials serious 1 no serious inconsistency no serious indirectness treatment-all-cause mortality 2 randomised trials serious 1 no serious inconsistency no serious indirectness 1 Sub-group analysis; potential for confounding was not addressed in the analyses 4 Low event rates Other considerations Discontinuing PCa Relative Control screening at age 70 years (95% CI) Absolute serious 4 none - - not pooled serious 4 none - - not pooled 0% 0% not pooled not pooled not pooled not pooled Quality Importance LOW LOW CRITICAL CRITICAL 63

64 PSA screening frequency Clinical Question In men who desire prostate cancer screening, what is the optimal PSA screening interval/frequency? Population Men without a prior diagnosis of prostate cancer and without bone pain or constitutional symptoms Health Intervention PSA Screening. every 2 years every 3 years every 4 years conditional screening (given a specific PSA result, determine optimal screening interval) Comparison Annual screening Critical health outcomes Mortality from prostate cancer Proportion of men with advanced cancer diagnosis staget3 Lower urinary tract symptoms and erectile dysfunction are not indications for ordering a PSA test. Search Strategy The target study that could directly address this clinical question would be a RCT of PSA screening, randomizing men into different screening intervals. As no direct evidence is available, indirect evidence to support this recommendation includes: Comparison of screening intervals across RCTs Natural history (descriptive) studies The evidence to support this question was derived from a search of Pubmed. (Table17). The ERSPC study of Goteborg vs. Rotterdam (van Leeuwen 2012) 28 was identified during the bridge search of prostate cancer screening (Figure 1). A search was then performed to identify related articles to the van Leeuwen study (Table 17a). In addition a de novo search was created to identify studies informing optimal screening intervals (Table 17b). Table 17: Search strategies: Frequency of prostate cancer screening (a) Database: PubMed related article search (van Leeuwen 2012) Date of search: 6/4/2013 Limitations: no limitations imposed #1 Related Citations for PubMed (Select ) #2 Select 20 document(s) 20

65 (b) Database:PubMed Date of search: 6/12/2013. Limitations: No limits Search Query Items found #33 Search (Prostatic Neoplasms/prevention and control 3430 [Mesh]) #34 Search Mass screening/methods* #36 Search Time Factors [MeSH] #37 Search Screening interval [tiab] 423 #38 Search Screening frequency [tiab] 166 #39 Search (#34 OR #35) #40 Search (#36 OR #37 OR #38) #43 Search Prostate cancer [tiab] #44 Search Prostatic cancer [tiab] 6379 #45 Search Prostate neoplasm [tiab] 29 #46 Search Prostatic neoplasm [tiab] 58 #47 Search Prostate adenocarcinoma [tiab] 1637 #48 Search Prostatic adenocarcinoma [tiab] 2224 #49 Search (#33 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48) #50 Search (#49 AND #40) 3704 #51 Search (#49 AND #39 AND #40) 104 Search Results Of the 230 articles identified by related article search, 20 were selected for further review, and 2 new articles were identified and included in this review (Yao 2001) 29 (Gulati 2013). 30 The study selection process is documented in Figure 9.Of the 104 articles identified by search strategy #3, no unique articles were identified and included in this review. The study selection process is documented in Figure

66 Figure 9: study selection search #2: frequency of screening PubMed search (n=230) High-level rejection (n=210) Full-text review (n=20) Accepted references (n=2) Reject (n=18) I=7 O=4 Study design=3 Not a clinical study=4 66

67 Figure 10: study selection search #3: frequency of screening PubMed search (n=104) High-level rejection (n=99) Full-text review (n=5) Reject (n=5) Not a clinical study=3 Duplicate = 1 Not English=1 Accepted references (n=0) 67

68 Evidence Synthesis Results Key Study Characteristics No RCTs of PSA screening intervals were identified. One cohort study used linked Medicare and SEER data of 36,422 men 65 years diagnosed with prostate cancer 1989 to A second cohort study compared a subset of results from each of two ERSPC sites (total n=13,301), only including men ages years from each site, 31 but not matching the two populations in any other way.this indirect comparison for the two sites, Goteborg (which had a two year screening interval) with Rotterdam (4 year interval) reported reduction in prostate cancer mortality and prevalence of localized and advanced prostate cancer. A third study used predictive modeling of 35 different screening strategies (employing different PSA thresholds for biopsy and different screening intervals) to compare calculated rates of detecting localized cancer (a proxy for overdiagnosis) vs. calculated rates of prostate cancer mortality among strategies. 32 Characteristics of included studies are found in Table 18. Yao 2001 A group of 36,422 men diagnosed with prostate cancer was identified from the SEER database. The authors calculated the screening interval from the date of the last PSA test prior to diagnosis, which they had identified from Medicare claims. The authors continued to track outcomes from the SEER database for each included subject for a mean of 70 months after diagnosis. Results showed 30.5% of prostate cancers were diagnosed at stage 3. The proportion of advanced cancer detected among those who had screened 3 months to 1 year prior was similar to those who had screened 2-3 years prior to diagnosis. Compared with a PSA testing interval of 3 months to 1 year prior to diagnosis, with screening intervals up to 3 years the proportion of advanced cancer was consistent (19.1%-19.4%), adjusted RR(95%CI)=1.00( ) and 1.02( ) for 1 to 2 year and 2 to 3 year intervals, respectively. At an interval of 3 to 4 years, diagnosis of advanced cancer increased to 27.9%, and at an interval of 4 to 5.3 years, advanced cancer increased to 30.8%. The adjusted RR for year interval = 1.73( ). Among those who were never screened prior to cancer diagnosis, 31.7% [RR=1.76( )] were diagnosed with advanced cancer. Approximately 28% of study participants died during the study period, with 9.6% dying from prostate cancer; 10.2% of patients who had never undergone PSA testing and 30.5% of men diagnosed with advanced cancer died from the disease. Compared with 1 year from last screening, there was no statistically significant difference in Prostate cancer mortality up to 64 months since last PSA, although estimates were imprecise. 68

69 Adjusted RR of Prostate cancer mortality for 1-2 years, 2-3 years and years was 0.90( ), 1.41( ), and 1.10 ( ), respectively. In men who were never screened, RR=1.85( ). These analyses were adjusted for age, year of diagnosis, geographic region, and race. The analytic team believes the potential for significant residual confounding is low, and no other threats to internal validity were identified. However, follow-up time to detect effect of screening interval on mortality may not be adequate (mean follow-up = 70 months) and results were imprecise (low event rate, wide confidence intervals). Further, results are restricted to diagnosis in men 65 years, and may not be generalizable to younger men. Van Leeuwen 2012 This study compared results from 2 ERSPC sites as a proxy for comparing the effectiveness of 2-year (Goteborg, n=4,202) vs. 4-year screening intervals (Rotterdam, n=13,301). Authors included a sub-set of study participants ages 55 to 64 years from each study site, and compared detection rates of prostate cancer (total, advanced, and localized). Follow-up was censored at 12 years to increase comparability of the 2 centers. Comparing the two sites, (and indirectly comparing 2-year to 4-year intervals), there was a higher incidence of total and localized prostate cancer at Goteborg (2 year interval), and a lower incidence of advanced prostate cancer. From screening round 2 until the end of follow-up, the proportional total prostate cancer incidence was 3.64 in Goteborg and 3.08 in Rotterdam, RR(95%CI)=1.18( ). The proportion of advanced cancer incidence was 0.40 in Goteborg and 0.69 in Rotterdam (RR= 0.57( ). For localized cancer, RR= 1.46 ( ). This study indirectly compared 2-year vs. 4-year screening intervals. Although analyses were performed to account for baseline differences in the two ERSPC sites, residual confounding cannot be ruled out, as differences between the two sites with respect to screening protocols, follow-up interval, and randomization protocol exist. Gulati et al This study used microsimulation modeling to predict negative outcomes (number of tests, false-positive results, localized cancer (as a proxy for overdiagnosis), and prostate cancer deaths) and positive outcomes (cancer detected, lives saved, and months oflife saved) of prostate cancer screening. Summary of the model components and source datasets may be found here. Screening protocols included variations in age to initiate and discontinue screening (40-74, 40-69, 50-74, 50-69), screening interval (annual, biennial), and biopsy threshold (PSA >2.5, PSA>4.0, PSA+PSA velocity combinations, age specific PSA). 69

70 Among the different PSA thresholds evaluated, age specific biopsy referral strategies are aligned with KP practices, and so were the constant factor in the excerpted results described below. Refer to table 18 for study specifics. Under no screening, the model projects a lifetime chance of a prostate cancer diagnosis of 12.0% and a lifetime chance of dying of prostate cancer of 2.86%. Annual screening of ages aligns with KP s current practices. Although predicted results for biennial screening in this age group are not available, comparisons of biennial vs. annual screening in ages 50-74, 40-69, and suggest a small reduction in the lifetime probability of overdiagnosis (0.4%-0.6%) with very little change in probability of cancer death over a lifetime (increase of 0.1%). 70

71 Table 18: Characteristics of studies evaluating prostate cancer screening frequency (a) Study Characteristics Relevant Population Characteristics Study/ design Yao 2001 Retrospe ctive cohort Critical Outcome Adv Prostate cancer at diagnosis N 36,4 22 Prostate cancer mortality Followup Interventions n Mean=70 months 2 yr (# events/total [%] 258/1347 (19.2%) 61/1310 (4.7%) (%) (%) Ages (%) 80+ (%) White (%) Race Black (%) 2 year PSA screening interval year PSA screening interval Never screened 32, yr RR (95%CI) ARR RoB (# events/total [%] 346/ ( ) 0 fewer per low (19.1%) 1000 (from 31 fewer to 38 97/1766 (5.5%) RR 0.90 (0.66 to 1.24) more) 5 fewer per 1000 (from 19 fewer to 13 more) low Primary Objective: To examine the effect of the PSA testing interval on prostate cancer specific survival and the risk of non-localized cancer. Primary Outcome(s): Advanced cancer at diagnosis; prostate cancer mortality Inclusion/Exclusion: Men age 65 years with Medicare, diagnosed with prostate cancer Additional Treatment Info: Interval was defined by last PSA test prior to diagnosis of prostate cancer 71

72 (b) Study Characteristics Relevant Population Characteristics Study/ design N Followup (yrs) Interventions n Age range (yrs) Van Leeuwen 2012 Compara -tive retrospec -tive cohort Critical Outcome 17, year PSA testing interval Censored at 12 years 4-year PSA testing interval year 4-year RR (95%CI) RoB Proportional Proportional incidence incidence Advanced cancer at diagnosis Total Prostate cancer Localized Prostate cancer 0.40 ( ) ( ) High not adjusted for confounding 3.64 ( ) 3.08 ( ) 1.18( ) 9.44 ( ) 6.47 ( ) 1.46 ( ) Primary Objective: Assess the effectiveness of Prostate cancer screening programs using a 2- or 4- yr screening interval. Primary Outcome(s): incidence of advanced Prostate cancer; Total Prostate cancer, localized Prostate cancer Inclusion/Exclusion: Men aged yr were participants at two centers of the European Randomized Study of Screening for Prostate Cancer: Gothenburg, Sweden (2-yr screening interval, n = 4202), and Rotterdam, the Netherlands (4-yr screening interval, n = ). Additional: We followed participants until the date of Prostate cancer, the date of death, or the last follow-up at December 31, 2008, or up to a maximum of 12 yr after initial screening. Potentially lifethreatening (advanced) cancer was defined as cancer with at least one of following characteristics: clinical stage T3a, M1, or N1; serum prostate- specific antigen (PSA) >20.0 ng/ml; or Gleason score 8 at biopsy. 72

73 (c) Study/ design Study variables Gulati 2013 microsim ulation modeling Interval: annual vs. biennial Age groups: 40-74, 40-69, 50-74, Criterion for biopsy referral (fixed): PSA level >95 th percentile for age Predicted probabilities based on age-specific PSA thresholds (PSA level > 95 th percentile for age) Screening age Screenin g interval Probability * of cancer detected (%) Probability * of overdiagnosis (%) Probability * of cancer death (%) Probability * of life saved (%) Mean time life saved, mo No screening 12.0 NA 2.86 Ref Ref NA Annual Annual Biennial Annual Biennial Annual Biennial *Life-time probability of event **NND: Number needed to detect to prevent one prostate cancer death NND** Primary Objective: To evaluate comparative effectiveness of alternative PSA screening strategies. Primary Outcome(s):PSA tests, false-positive test results, cancer detected, overdiagnoses, prostate cancer deaths, lives saved, and months of life saved. Additional Info: Intervention: 35 screening strategies that vary by start and stop ages, screening intervals, and thresholds for biopsy referral. Data Sources: National and trial data on PSA growth, screening and biopsy patterns, incidence, treatment distributions, treatment efficacy, and mortality 73

74 Quality Assessment Two year screening intervals One retrospective cohort study (Yao) 29, examining two year screening intervals is considered overall very low quality (Table 19). Table 19: Quality assessment of studies examining 2- vs. 1- year screening intervals No of studies Design Risk of bias Quality assessment No of patients Effect Inconsistency Indirectness Imprecision advanced prostate cancer diagnosis (follow-up median 70 months) 1 observational studies no serious no serious risk of inconsistency bias no serious indirectness Other considerations 2 year screening interval no serious none 2 258/1347 imprecision 1 (19.2%) prostate cancer mortalilty (follow-up median 70 months) 1 observational no serious no serious no serious serious 3 none 2 61/1310 studies risk of bias inconsistency indirectness (4.7%) 1 Adequate event rate, CI intervals not too wide for comparison of annual vs biennial intervals 2 Unable to assess, but no reason to suspect 3 low event rate, wide CI for comparison of annual vs. biennial intervals 1 year screening intervall 346/1814 (19.1%) 97/1766 (5.5%) Relative (95% CI) RR 1.00 (0.84 to 1.20) RR 0.90 (0.66 to 1.24) Absolute 0 fewer per 1000 (from 31 fewer to 38 more) 5 fewer per 1000 (from 19 fewer to 13 more) Quality Importance LOW VERY LOW CRITICAL CRITICAL 74

75 Four year screening intervals One indirect comparison of two ERSPC sites 28, examining four vs. two-year screening intervals is considered overall very low quality (Table 20). Table 20: Quality assessment of studies examining 4-year screening intervals Quality assessment No of patients Effect No of studies Design Risk of bias advanced prostate cancer diagnosis 1 observational studies serious 1 no serious inconsistency Inconsistency Indirectness Imprecision serious 2 no serious imprecision Other considerations 1 Inadequate adjustment for confounding 2 Indirect comparison using data from 2 separate RCTs 3 Unable to assess, but no reason to suspect 4 proportional advanced cancer incidence-- 2 yr interval: 0.40; 4 yr interval: year PSA screening interval 2 year PSA screening interval Relative (95% CI) none RR 0.57 (0.33 to 0.99) Absolute - Quality Importance VERY LOW CRITICAL 75

76 PSA Cut-points for urology referral Clinical Question Population Health Intervention/ Comparison Critical health outcomes What is the optimal cut-point for screening with the PSA test? Men without a prior diagnosis of prostate cancer and without bone pain or constitutional symptoms Fixed PSA cut-point (3.0 vs. 4.0) vs. Age-specific cut-points vs. Age and Race-specific cut-points Sensitivity Specificity Positive predictive value Lower urinary tract symptoms and erectile dysfunction are not indications for ordering a PSA test. Search Strategy Table 21: Search strategy PSA cut-points Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * PubMed (digital[all Fields] AND rectal[all Fields] AND exam[all Fields] AND PSA[All Fields]) AND ("prostatic neoplasms"[mesh Terms] OR ("prostatic"[all Fields] AND "neoplasms"[all Fields]) OR "prostatic neoplasms"[all Fields] OR ("prostate"[all Fields] AND "cancer"[all Fields]) OR "prostate cancer"[all Fields]) AND ("diagnosis"[subheading] OR "diagnosis"[all Fields] OR "screening"[all Fields] OR "mass screening"[mesh Terms] OR ("mass"[all Fields] AND "screening"[all Fields]) OR "mass screening"[all Fields] OR "screening"[all Fields] OR "early detection of cancer"[mesh Terms] OR ("early"[all Fields] AND "detection"[all Fields] AND "cancer"[all Fields]) OR "early detection of cancer"[all Fields]) AND accuracy[all Fields] Thru 4/21/10 0/8 * Note: No. Included refers to studies that are relevant to the problem formulation and, therefore, are included in this analysis of the evidence. Total Retrieved refers to the number of studies retrieved in the search, regardless of relevance. Because individual studies can be captured in multiple databases, they may be counted more than once in the number included. 76

77 Database: Search Terms: Article Type and Other Limits: Search Date No. Included / Total Retrieved * PubMed (("digital rectal examination"[mesh Terms] OR ("digital"[all Fields] AND "rectal"[all Fields] AND "examination"[all Fields]) OR "digital rectal examination"[all Fields]) AND psa[all Fields]) AND ("prostatic neoplasms"[mesh Terms] OR ("prostatic"[all Fields] AND "neoplasms"[all Fields]) OR "prostatic neoplasms"[all Fields] OR ("prostate"[all Fields] AND "cancer"[all Fields]) OR "prostate cancer"[all Fields]) AND ("diagnosis"[subheading] OR "diagnosis"[all Fields] OR "screening"[all Fields] OR "mass screening"[mesh Terms] OR ("mass"[all Fields] AND "screening"[all Fields]) OR "mass screening"[all Fields] OR "screening"[all Fields] OR "early detection of cancer"[mesh Terms] OR ("early"[all Fields] AND "detection"[all Fields] AND "cancer"[all Fields]) OR "early detection of cancer"[all Fields]) Systematic reviews/ meta-analyses Clinical trials Thru 4/21/10 Thru 4/21/10 3/24 17/198 ("2000/08"[Publication Date] : "2010/01"[Publication Date]) AND (("prostatespecific antigen"[title/abstract] OR "prostate specific antigen"[title/abstract]) OR "PSA"[Title/Abstract]) AND ("prostate cancer"[title/abstract] OR "prostatic neoplasms"[title/abstract]) AND (sensitivity OR specificity OR "predictive value" OR "predictivevalue" OR "false-positive rate" OR "false-negative rate") Note: Studies were included provided they reported data on sensitivity/specificity/ppv of the PSA and/or DRE test and/or reported on data pertaining to the application of specific PSA thresholds to patient populations. Evidence Synthesis An overview of the key studies evaluating the sensitivity/specificity of the PSA test using a fixed cut-point is provided below. Studies evaluating the sensitivity/specificity associated with age and/or race-specific PSA reference ranges are also included (see Table 22). Sensitivity and Specificity of the PSA Test Use of a Fixed PSA Cutpoint (e.g., PSA = 3 or 4 ng/ml) Two systematic reviews (Harvey and Mistry ) and six additional epidemiological studies reporting on the sensitivity, specificity, and predictive value of the PSA test at individual PSA cut-points were identified and included in this review. Most of the studies were subject to significant selection bias. Moreover, because many studies were conducted among patients identified at urology clinics, results may not be 77

78 generalizable to screening populations. In most studies, the reference standard (prostate biopsy) was not properly applied to all appropriate study participants. No clear pattern emerges from the evidence regarding which PSA cut-point should be used. The studies were qualitatively heterogeneous and sensitivity/specificity calculations also varied significantly between studies. Harvey et al. (2009), 33 conducted a systematic review (search date 1998 to 2008) to assess the sensitivity and specificity of the PSA test in the diagnosis of prostate cancer. They identified ten studies (seven cohort, three case-control; N = 5,373), all of which were conducted in populations of European men. No pooling of data was performed. Sensitivities ranged from 0.78 to 1.00 and specificities ranged from 0.06 to 0.66 (Table 23). The study populations in this review were limited to European men attending urology clinics because of referral due to clinical evidence of primary or secondary care. As such, the results may not be applicable to screening populations or generalizable to American men. In addition, the systematic review did not report whether an individual PSA cutpoint or age-specific reference point was used for the study. Auvinen et al. (2009) 35 estimated the sensitivity of the PSA test using data from the ERSPC trial 36 centers in Finland, Netherlands and Sweden (total N screened = 46,618). The sensitivity estimate was derived by using the interval cancer incidence rate among 39,389 men with negative (PSA < 4 ng/ml ) screening rates (an indicator of the false-negative rates), compared to the incidence among 79,525 men in the control arm of the trial (an indicator of the expected rates). Sensitivity in Finland was 0.87 (95% CI: 0.83 to 0.92), 0.93 (95% CI: 0.90 to 0.96) in the Netherlands, and 0.87 (95% CI: 0.62 to 1.00) in Sweden. Because the interval cancer incidence method was used to estimate sensitivity, the reference standard of biopsy was not consistently applied to all appropriate study participants. Using data from the Prostate Cancer Prevention Trial, Thompson et al. (2004) 37 evaluated the prevalence of prostate cancer among men who had PSA levels of 4.0 ng/ml or less. Among 2,950 men who never had a PSA level of more than 4.0 ng/ml or abnormal DRE, 449 (15.2%) were diagnosed with prostate cancer and 67 of 449 (14.9%) had a Gleason score of 7 or higher. Hence, high grade prostate cancer was not uncommon even at very low PSA levels. The sensitivity and specificity of various PSA cutpoints below 4.0 ng/ml are described in Table 24. In 2005, Thompson et al. 38 again used data from the Prostate Cancer Prevention Trial to evaluate sensitivity and specificity for PSA among 18,882 healthy men aged 55 years or older without prostate cancer and with PSA levels less than or equal to 3.0 ng/ml and normal DRE. If PSA level exceeded 4.0 ng/ml or rectal examination result was abnormal, a prostate biopsy was recommended. After seven years of study participation, an end-of-study prostate biopsy was recommended in all cancerfree men. Of 8,575 men in the placebo group with at least one PSA measurement and digital rectal examination in the same year, 5,587 (65.2%) had had at least one 78

79 biopsy; of these, 1,225 (21.9%) were diagnosed with prostate cancer. Of 1,213 cancers with Gleason grade recorded, 250 (20.6%) were Gleason grade 7 or greater and 57 (4.7%) were Gleason grade 8 or greater. For detecting any prostate cancer, PSA cutoff values of 1.1, 2.1, 3.1, and 4.1 ng/ml yielded sensitivities of 83.4%, 52.6%, 32.2%, and 20.5%, and specificities of 38.9%, 72.5%, 86.7%, and 93.8%, respectively. These data suggest that there is no cutpoint of PSA with simultaneous high sensitivity and high specificity for monitoring healthy men for prostate cancer, but rather a continuum of prostate cancer risk at all values of PSA. See Table 25. Määttänen et al. (2007) 39 estimated PSA specificity for detecting prostate cancer by using data from the Finnish Randomized Prostate Cancer Screening Trial. The reference standard, biopsy, was not applied to everyone in the test population. Verification bias likely exists because the decision to perform biopsy was dependent on results of initial PSA test only negative results were sent for biopsy. Table 26 provides a summary of specificity data reported in the study. A meta-analysis by Mistry et al. (2003) 34 examined sensitivities, specificities, and positive predictive values of the PSA test from 13 studies (search date 1966 to 1999) of asymptomatic men older than 50. The meta-analysis included only those studies where a prostate biopsy was performed as a reference standard and a PSA cutoff of > 4 ng/ml was used. The pooled analysis showed that among 47,791 sampled men, the sensitivity, specificity, and positive predictive values were 25.1%, 72.1% and 93.2% respectively (Tables 27 and 28). As with many studies that examine accuracy of PSA screening, this study was limited because of the studies included lacked biopsy results (the gold standard) for patients with normal PSA tests (i.e., < 4 ng/ml) and lacked control groups. Hakama et al. (2001), 40 reported the sensitivity and specificity of the PSA test based on data obtained from Finnish cancer registries and blood serum banks. Using a PSA cutoff of 4 ng/ml, this study reported a sensitivity and specificity of 44% and 94%, respectively, among men between 55 to 79 years of age who were followed between 1968 to Among patients who were diagnosed five years after their blood was drawn, the sensitivity was 86%, while specificity remained at 94%. In this study the sensitivity depended on the interval between the drawing of the sample and the diagnosis of cancer. The sensitivity reported in this study is considerably lower than the sensitivity of the PSA reported in the paper by Morgan et al. 41 Thus if anything, this new information weakens the case for screening with PSA. Age and Race-Specific PSA Screening Cut-Points There are no studies that directly evaluate the effect of using different age and/or race specific PSA cut-points on prostate cancer morbidity/mortality. A review of the evidence identified nine epidemiological studies reporting on sensitivity and specificity associated with using age/race specific cutoffs for PSA testing. The studies were all subject to significant selection bias. In addition, most studies (with exception of Jacobsen et al., 79

80 ) were conducted among study subjects who were identified at urology clinics in North America, and may not be generalizable to screening populations. Ultimately, although no direct evidence exists, there are studies to suggest that use of age and/or race-specific cutoffs may enhance the sensitivity in younger men while increasing the specificity in older men. Key studies are described briefly below. Age-Specific Reference Ranges Oesterling et al. ( & ) evaluated the usefulness of age-specific PSA reference ranges by examining the medical records of 2,988 men age 60 or older who underwent PSA testing of whom 1,686 of whom underwent subsequent biopsy. Compared to using a PSA cutoff of 4.0 ng/ml, age-specific reference ranges detected 20% more cancers among men 60 to 89 years of age. Age-specific cutoffs were slightly less sensitive and more specific than the standard cutoff of 4.0 ng/ml (85% vs. 74% -sensitivity, 50% vs. 69% -specificity). If the age-specific reference ranges had been used, 92 prostate biopsies (5.5%) performed could have been avoided, while 19 men in the study population (0.6%) would not have had prostate cancer diagnosed. Of the non-detected cancers 13 (67%) occurred in men 70 years old or older and 18 (95%) were small tumors with favorable pathological status and not likely to be of clinical importance. The study was not conducted in a screening population, but rather was based on subjects attending urology clinics. In a case-control study by Jacobsen et al. (1996) 42 PSA sensitivities and specificities according to the age-specific PSA thresholds vs. PSA 4.0 ng/ml threshold were calculated from a population-based cohort men age 50 to 79. Compared to the 4.0 ng/ml cutoff, the age-specific thresholds were more specific and less sensitive in older men age 70 to 79, reducing the likelihood of detecting insignificant cancers in this age group. Results are reported in Tables 29 and 30. Partin et al. (1996) 45 conducted a retrospective review (N = 4,597 men with clinically localized prostate cancer) to evaluate the effect of using of age-specific PSA references ranges ((0 to 2.5 ng/ml serum PSA (40 to 49 years), 0 to 3.5 ng/ml (50 to 59 years), 0 to 4.5 ng./ml (60 to 69 years) and 0 to 6.5 ng/ml (70 to 79 years)) have on prostate cancer detection rates in younger vs. older men. Overall, 18% of the men had PSA levels less than the standard PSA reference range (4.0 ng/ml) compared to 22% when using the age-specific ranges. There were 74 more cancers detected in men younger than 60 years with the use of age-specific ranges, of which 81% had favorable pathological results. Among the men 60 years or older, 191 of 252 cancers (76%) not detected by using age specific ranges less than 3% were also stage T1c and 95% of these undetected T1c cancers were of favorable pathological status. Age-specific PSA reference ranges increased the potential for detection of prostate cancer by 18% in the younger men and decreased the detection by 22% in the older men. This data suggests that age-specific PSA references ranges may increase the detection of more potentially curable tumors in young men and decreased the detection of less advanced tumors in the older men 80

81 compared to the standard reference range of 4.0 ng/ml Among older men with nonpalpable (stage T1c) tumors age-specific PSA references ranges would have detected fewer tumors. The study was not conducted in a screening population, but rather was based on subjects attending urology clinics. Borer et al. (1998) 46 conducted a retrospective review of prostate biopsy and PSA data among 1,046 men (age 60 to 79) at a VA hospital. The study reported that, use of age-specific PSA reference ranges 73 of 1,280 biopsies (5.7%) would have been avoided. Of those 73 avoided biopsies 15 (20.5%) had cancer that would have gone undetected and 9 of 15 (60%) undetected cancers had unfavorable histology. Results were not statistically significantly different among the four age specific PSA reference ranges. Regarding race, cancer detection rates were significantly higher for black/african-american men compared with white men but there was no statistically significant difference for missed cancers or missed cancers with unfavorable histology. Although limitations associated with retrospective chart reviews apply, the study suggests that use of age-specific reference ranges does not eliminate the possibility of missing significant cancers. The study was not conducted in a screening population, but rather was based on subjects attending urology clinics who underwent treatment for prostate cancer. In a retrospective case-series study, Wolff et al. (2000) 47 found that the use of agespecific reference ranges (defined by Oesterling et al. 43;44 ) would not safely eliminate the need for biopsy among a sample of 60 year-old men. If age-specific cutoffs had been used, 56 of 271 (or 20.6%) cancer cases would have been missed; 30 of 56 (54%) missed cancers had an unfavorable pathology. Actual identification of the 271 prostate cancer patients was done using a PSA cutoff of 4.0 ng/ml. 81

82 Age and Race Specific Reference Ranges One epidemiological study (Morgan et al.) 41 reported on sensitivity and specificity for age and race-specific PSA reference ranges. One subsequent study (Cooney et al.) 48 evaluated whether use of age/race specific ranges were warranted. Morgan et al. (1996) 41 conducted a prospective study measuring PSA in 1,783 men with prostate cancer (1,372 white and 411 black/african-american) and 3,475 controls with no prostate cancer (1,802 white and 1,673 black/african-american) in order to evaluate sensitivity and specificity of the PSA test at different ranges for white and black/african-american men. They reported that if traditional age-specific reference ranges (defined by Oesterling et al. 43;44 ) were used in screening black/african-american men, with the test specificity kept at 95%, 41% of cases of prostate cancer would be missed. For the test to have 95% sensitivity among black/african-american men, the normal reference ranges shown in Table 31 should be used. Also see Table 32. A study by Cooney et al. (2001) 48 compared the PSA levels from a random sample of 350 black/african-american men with those of men from other studies (Morgan, 41 Oesterling et al. 43;44 ) and found that, the 95th percentile PSA values were estimated to range from 2.36 ng/ml for men in the fifth decade to 5.59 ng/ml for men in the eighth decade. The 95th percentile values for age-specific PSA were comparable to those observed in a similar study of white men in Olmsted County, Minnesota. The authors suggest that minor differences in PSA reference ranges between black/african-american men and white men may not be significant enough to recommend the use of race-specific PSA reference ranges for screening. 82

83 Table 22: Age-Specific Reference Ranges for PSA Screening Test Study Name Design Population Oesterling et al Serum Prostatespecific antigen in a community-based population of healthy men: Establishment of age-specific reference ranges USA Supported by grant AR30582 from the NIG and a grant from Merck Research Laboratories as part of the Benign Prostatic Hyperplasia Natural History Study Group Clinical Series Blinding N/a Determined the number of prostate biopsies that could have been avoided and the number and pathological characteristics of cancers not detected if the age-specific reference ranges had been used instead of the ng/mL. Men, years between Dec and Mar All patients screened w/ serum PSA, DRE, TRUS All reside in Olmsted Co., Minn. No evidence of prostate cancer after screening w/ tests Screening Groups Size Compliance Groups: Rx1: y.o. Rx2: y.o Rx3: y.o. Rx4: y.o. Initial: 537 pts. Final: 471 pts. Rx1: 165 Rx2: 144 Rx3: 94 Rx4: 68 Rx: 471/537 (88.0%) Breakdown of drop-outs 61 pts. either refused to participate or did not complete all three diagnostic tests. 5 men were diagnosed with prostate cancer after TRUS screening. Follow- Up Results Bias N/a PSA > 4.0 ng./ml: Rx1: 0 (0%) Rx2: 3 (2%) Rx3: 12 (13%) Rx4: 13 (19%) Determined that Serum PSA concentration is correlated directly w/ age (r=0.43, P<0.0001) Hypothesized that serum PSA concentration increases ~3.2%/year. recommended the following age specific thresholds for a postive PSA test: Age PSA Range < 2.5 ng/ml < 3.5 ng/ml < 4.5 ng/ml < 6.5 ng/ml Population bias 83

84 Study Name Design Population Oesterling et al The use of agespecific reference ranges for serum prostate specific antigen in men 60 years old or older USA Clinical Series Blinding N/a Determined the number of prostate biopsies that could have been avoided as well as the number and pathological characteristics of cancers not detected if the age-specific reference ranges had been used instead of the ng/mL. Age PSA Value Standard: ng./ml ng./ml Age-specific: ng./ml ng./ml Men, years between Jan and Mar All patients screened w/ serum PSA, DRE, TRUS All at single urological practice in Mobile, Alabama Screening Groups Size Compliance Groups*: Sx1: year olds Sx2: 70+ years old *Study compares the rates of prostate detection using the standard 4.0 cutoff versus the age specific cutoffs on the same population so the results are reported by standard = 4.0 ng./ml PSA cutoff age specific = age specific cutoffs Initial: 2,988 Sx1: 1,583 Sx2: 1,405 Follow- Up Results Bias N/A N/A Prostate cancer detected Standard: 608/2,988 (20.0%) Age specific: 471/2,988 (42.0%) Sensitivity: Standard: % % Age-specific: 83% % Specificity: Standard: % % Age-specific: 66% % Selection Bias Positive Predictive Value: Standard: % % Age-specific: 29% % 84

85 Borer et al Study Name Design Population Age specific prostate specific antigen reference ranges: population specific USA Case Review Blinding N/a Comparison of results using standard reference range of ng/ml and age specific PSA reference ranges provided by Oesterling et al. Age PSA Range ng/ml ng/ml Men, years Had PSA assay, DRE, and prostate biopsy for an abnormal rectal exam &/or PSA >4.o ng/ml All at Brooklyn VA Med Ctr. From Jan August 1995 Screening Groups Size Compliance Groups: C: > 4.0 ng/ml Rx1: age-specific Rx2: age-specific Focused on men who would not have undergone biopsy w/ agespecific reference range but did w/ the > 4.0 ng/ml cutoff. Initial & Final N: Men: 1,046 Prostate biopsies 1,280 Follow- Up Results Bias N/a N/a Prostate cancer detected C: 483/1280 (37.7%) Rx1: 230/667 (34.0%) Rx2: 180/613 (29.0%) Age-specific: 410/1280 (32.0%) Prostate cancers missed Rx1: 3 Rx2: 12 Age-specific: 15/73 (20.5%) Prostate cancers missed w/ unfavorable histology: Rx1: 2/3 (66.0%) Rx2: 7/12 (58.0%) Age-specific: 9/15 (60.0%) Positive Predictive Value: Standard: 45% 48% Age-specific: % % Conclusion: In marked contrast to previous reports of unfavorable histological characteristics in only 5% of missed cancers using age specific PSA reference ranges, 60% of missed cancers in our patients had unfavorable histology. Selection Bias 85

86 Littrup et al Study Name Design Population Cost-effective prostate cancer detection reduction of low-yield biopsies USA Supported by a Cancer Control Grant from the American Cancer Society Partin et al 1996 Standard versus agespecific prostate specific antigen reference ranges among men with clinically localized prostate cancer: a pathological analysis USA American Cancer Society National Prostate Cancer Detection Project Cohort Study Blinding N/a Analyzed DRE, TRUS, PSA, and age-specific PSA, glandvolume adjusted PSA levels, and longitudinal PSA changes Cost-effectiveness of screening, biopsy, and follow-up of patients for each indicator was determined Age PSA Range > 2.5 ng/ml > 3.5 ng/ml > 4.5ng/mL > 6.5 ng/ml Retrospective Case Study Blinding N/a Determined # of cancers diagnosed with age specific PSA cutoff values vs. standard PSA values Men, years No history of prostate cancer All at Brooklyn VA Med Ctr. From Jan August 1995 Men, years old All races Had radical prostatectomy following diagnosis of clinically localized prostate cancer Underwent radical prostatectomy between 10/84 10/94 Had PSA done Did not have TRUS Screening Groups Size Compliance Groups: All patients screened for PSA and PSA density Sx: screening group Sx1: w/dre, TRUS Sx2: w/dre Sx3: w/ TRUS Sx4: no DRE nor TRUS Sx5: repeat PSA Group Age I II III IVS PSA age-specific cutoff values I <2.5ng./ml II <3.5ng./ml III <4.5ng./ml IVS. <6.5ng./ml Initial: 2,999 Final: 2,930 N: 4,957 I: 180 II: 1,146 III: 2,482 IV: 789 Number of PSA results available by year of follow up: 5 th yr.: 433/2,930 (14.8%) 4 th yr.: 622/2,930 (21.2%) 3rd yr.: 677/2,930 (23.1%) 2 nd yr.: 406/2,930 (13.9%) 1 st yr.: 792/2,930 (27.0%) Follow- Up Results Bias Annually for up to five years Total Positive Predictive Value (PSA > 4.0 ng./ml): Sx1: 28.9% Sx2: 17.8% Sx3: 13.7% Sx4: N/R Sx5: 50.0% Sensitivity Specificity Total: 62% 84% Positive Predictive Value: Total: 50% Negative Predictive Value: Total: 40% N/A N/A % Cancer Detected Age 4.0ng./ml Age-specific Net Diff. I 63% 77% +14% II 77% 82% +5% III 84% 80% -4% IVS. 86% 67% -19% Total 82% 78% -4% 18% increase in tumors detected using age-specific cutoffs in the over 60 years age group 22% decrease using age-specific cutoffs in younger than 60 years Selection Bias Selection Bias 86

87 Study Name Design Population Cooney et al. Age-specific distribution of serum prostate-specific antigen in a community-based study of African- American men. Urology, 2001 Clinical series, based on random sampling of AA men PSA levels obtained from this random sample of African-American (AA) were compared to PSA levels of White men and other AA men recorded in separate studies. A random sample of African-American men from Flint, MI aged years w/o clinically diagnosed prostate cancer or prior surgery on prostate gland Screening Groups Size Compliance All participants received PSA test and DRE. Age Range N Total 350 Follow- Up Results Bias N/a N/a 95 th Percentile PSA Levels (ng/ml) by age in US White & African- American men without clinically evident prostate cancer: review of studies. Oester. Cooney DeAntoni Morgan Weinrich et al* et al. et al. et al. et al.. White AA AA White White AA White AA N , *based on Oesterling studies (1993,1995). Selection Bias only half of those identified by random sample completed clinical prototcol 87

88 Morgan et al Study Name Design Population Age-specific reference ranges for serum prostatespecific antigen in black men USA Prospective case control Blinding N/a Total Population: White (W) and Black (B) males, 40 to 79 years old Screened at Walter Reed Army Med Ctr. Control Group (C): Tested from Jan to Aug on finasteride at time of PSA testing prostate cancer diagnosed by biopsy following PSA test performed 6 months prior Case Group (Rx): Tested between Jan to May 1995 No known prostate cancer Could have abnormal DRE and PSA > 4.0 ng./ml 20 on finasteride at the time of PSA testing Screening Groups Size Compliance Groups: C: w/ prostate cancer Rx: w/o evidence of prostate cancer Both groups were positive if PSA level was more than 4.0ng/mL and DRE was abnormal Initial & Final: C: 3,475 Rx: 1,783 Follow- Up Results Bias N/a N/a Controls Age Race PSA level of 95% pop W 2.1 B W 3.6 B W 4.3 B W 5..8 B 12.5 Cases Age Race PSA level of 5% pop W 2.6 B W 3.5 B W 3.3 B W 3.3 B 5.5 Sensitivity Specificity W/Standard PSA cutoff B: 48.6% 98.6% W: 62.0% 100.0% B: 94.7% 88.4% W: 74.8% 97.4% B: 98.1% 79.0% W: 91.1% 93.0% B: 98.6% 73.1% W: 92.1% 81.2% W/Age-Specific PSA cutoff B: 87.9% 94.9% W: 91.2% 97.3% B: 98.2% 81.9% W: 86.3% 95.3% B: 98.1% 79.0% W: 81.1% 95.1% B: 95.1% 77.5% W: 77.0% 97.2% Population bias (population might not be representative to all men) 88

89 Table 23: True Positives (TP), False Positives (FP), True Negatives (TN), False Negatives (FN), Sensitivity and Specificity for All Studies With 95% Confidence Intervals. (Harvey 2009) 89

90 Table 24: Relationship of the PSA Level to the Prevalence of Prostate Cancer and High-Grade (Gleason > 7) Disease (Thompson 2004) PSA Level # of Men Men with Prostate Cancer Sensitivity Specificity 0.5 ng/ml (6.6%) to 1.0 ng/ml (10.1%) to 2.0 ng/ml (17.0%) to 3.0 ng/ml (23.9%) to 4.0 ng/ml (26.9%)

91 Table 25: Sensitivity and Specificity for Prostate Cancer and High-Grade Disease, by Cutpoints of Prostate-Specific Antigen (PSA) and by Age. * (Thompson 2005) * Twelve cancer cases did not have Gleason grade recorded and are omitted from the grade-related comparisons. For any cancer vs. non cancer, n = 2,956 for age < 70 years and n = 2,631 for age 70 years, for the grade-related comparisons, n = 2,950 for age < 70 years and n = 2,625 for age 70 years. 91

92 Table 26: The Frequency of TN and FP Screening Findings by Age in the First Screening Round of the Finnish Prostate Cancer Screening Trial (Määttänen 2007) Age TN a FP b Specificity (95% CI) 55 6, (0.962 to 0.971) 59 5, (0.943 to 0.954) 63 4, (0.902 to 0.919) 67 3, (0.871 to 0.892) Total 18,825 1, (0.929 to 0.936) Abbreviations: CI ¼ confidence interval; FP ¼ false positive; TN ¼ true negative. a) TN: Number of men with negative screening result (serum PSA < 3.0 ng/ml or PSA 3.0 to 3.9 ng/ml with a negative ancillary examination (benign finding at digital rectal examination or free/total PSA ratio times 0.16)). b) FP: No. of men with positive screening result (serum PSA 4.0 ng/ml or PSA 3.0 to 3.9 ng/ml, with a positive ancillary examination (suspicious finding at digital rectal examination or free/total PSA ratio) minus number of screen-detected cancers. Note: men refusing biopsy (N = 102) were excluded. c) Specificity: TN/(TN+FP). 92

93 Table 27: Detection Rate and Biopsies Performed to Detect Prostate Carcinoma (PCA). (Mistry 2003) 93

94 Table 28: Positive Predictive Value, Sensitivity, and Specificity of Prostate-Specific Antigen (PSA). (Mistry 2003) 94

95 Table 29: Population-based Estimates of Sensitivity, Specificity for Age Specific Reference Ranges 13 (Jacobsen 1996) Age Threshold Sensitivity (%) 95% CI: for Sensitivity Specificity (%) 95% CI: for Specificity 45 to 49 > 2.5 ng/ml not reported not reported not reported not reported 50 to 59 > 3.5 ng/ml to to 69 > 4.5 ng/ml to to to 79 > 6.5 ng/ml to to 99 Table 30: Population-Based Estimates of Sensitivity, Specificity for PSA Cut-Off of 4.0 ng/ml (Jacobsen 1996) Age Threshold Sensitivity (%) 95% CI: for Sensitivity Specificity (%) 95% CI: for specificity 45 to 49 not reported not reported not reported not reported 50 to to to 100 > 4.0 ng/ml 60 to to to to to to 89 Table 31: Ages-Specific Reference Ranges for the PSA Test, Based on the 5 th Percentile of the Distribution of PSA levels in the Patients According to Race (Morgan 1996) Age (yr) Whites ng/ml Blacks/African-Americans ng/ml 40 to to to to to to to to to to to to Age specific PSA reference ranges according to Oesterling et al.: 50 to 59 (3.5 ng/ml), 60 to 69 (4.5 ng/ml), and 70 to 79 (6.5 ng/ml). 95

96 Table 32: Sensitivity and Specificity of PSA Test, By Age, PSA Value, and Race (Morgan 1996) 96

97 Quality Assessment No quantitative GRADE system exists for diagnostic studies, and no risk of bias tools for observational studies have been evaluated. Based on qualitative assessment of the body of literature, the quality of evidence to determine performance characteristics (sensitivity, specificity, etc.) of PSA and DRE is very low. This is also true for studies which examined the use of age-specific PSA thresholds. Due to the high degree of missing biopsies among those with normal PSA results, accurate measures of diagnostic accuracy cannot be calculated. As such, there continues to be a high-risk of bias in these studies, resulting in the quality of evidence likely being very low. 97

98 External Guidelines NCCN, ACS, ACP and AUA recommend a shared decision-making approach to discuss the benefits and risk of prostate cancer screening. NCCN outlines an algorithm for baseline PSA testing at 40 years, ACS continues to recommend discussing screening at 45 years for high-risk and 50 years for average risk, AUA recommends a shared decision-making approach in men years, and ACP recommends a shared decision-making approach in men ages years. USPSTF recommends against prostate cancer screening, citing harms outweighing the potential benefits of screening. Nonetheless, the USPSTF also outlines points to use in an informed decision-making discussion. NCCN 2012 The NCCN prostate cancer early detection CPG presents a series of algorithms for PSA-based screening in the context of a shared decision-making discussion: Start risk and benefit discussion about offering baseline digital rectal examination (DRE) and PSA at age 40. ACS The American Cancer Society recommends that men make an informed decision with their doctor about whether to be tested for prostate cancer. Research has not yet proven that the potential benefits of testing outweigh the harms of testing and treatment. The American Cancer Society believes that men should not be tested without learning about what we know and don t know about the risks and possible benefits of testing and treatment. Starting at age 50, men should talk to a doctor about the pros and cons of testing so they can decide if testing is the right choice for them. If they are African American or have a father or brother who had prostate cancer before age 65, men should have this talk with a doctor starting at age 45. If men decide to be tested, they should have the PSA blood test with or without a rectal exam. How often they are tested will depend on their PSA level. AUA 2013 The current AUA guidelines for early detection of prostate cancer address average-risk men in 4 age groups: <40, 40-54, and 70 years. AUA recommends against screening in men under 40, 70 years or greater, or men with <10 to 15 year life expectancy. AUA recommends against routine screening in average risk men ages 40-98

99 54. For men ages years, AUA recommends a shared decision-making approach in men who are considering PSA screening. The guideline statements are as follows (reported verbatim): 1. The Panel recommends against PSA screening in men under age 40 years. (Recommendation; Evidence Strength Grade C) In this age group there is a low prevalence of clinically detectable prostate cancer, no evidence demonstrating benefit of screening and likely the same harms of screening as in other age groups. 2. The Panel does not recommend routine screening in men between ages 40 to 54 years at average risk. (Recommendation; Evidence Strength Grade C) For men younger than age 55 years at higher risk (e.g. positive family history or African American race), decisions regarding prostate cancer screening should be individualized. 3. For men ages 55 to 69 years the Panel recognizes that the decision to undergo PSA screening involves weighing the benefits of preventing prostate cancer mortality in 1 man for every 1,000 men screened over a decade against the known potential harms associated with screening and treatment. For this reason, the Panel strongly recommends shared decision-making for men age 55 to 69 years that are considering PSA screening, and proceeding based on a man s values and preferences. (Standard; Evidence Strength Grade B) The greatest benefit of screening appears to be in men ages 55 to 69 years. 4. To reduce the harms of screening, a routine screening interval of two years or more may be preferred over annual screening in those men who have participated in shared decision-making and decided on screening. As compared to annual screening, it is expected that screening intervals of two years preserve the majority of the benefits and reduce overdiagnosis and false positives. (Option; Evidence Strength Grade C) Additionally, intervals for rescreening can be individualized by a baseline PSA level. 5. The Panel does not recommend routine PSA screening in men over age 70 years or any man with less than a 10 to 15 year life expectancy. (Recommendation; Evidence Strength Grade C) Some men over age 70 years who are in excellent health may benefit from prostate cancer screening. 99

100 American College of Physicians 2013 (ACP) Guidance Statement 1: ACP recommends that clinicians inform men between the age of 50 and 69 years about the limited potential benefits and substantial harms of screening for prostate cancer. ACP recommends that clinicians base the decision to screen for prostate cancer using the prostate specific antigen test on the risk for prostate cancer, a discussion of the benefits and harms of screening, the patient s general health and life expectancy, and patient preferences. ACP recommends that clinicians should not screen for prostate cancer using the prostate-specific antigen test in patients who do not express a clear preference for screening. Guidance Statement 2: ACP recommends that clinicians should not screen for prostate cancer using the prostate-specific antigen test in average-risk men under the age of 50 years, men over the age of 69 years, or men with a life expectancy of less than 10 to 15 years. USPSTF The U.S. Preventive Services Task Force (USPSTF) recommends against prostatespecific antigen (PSA)-based screening for prostate cancer (D recommendation). In their rationale, USPSTF goes on to state: Although the precise, long-term effect of PSA screening on prostate cancer specific mortality remains uncertain, existing studies adequately demonstrate that the reduction in prostate cancer mortality after 10 to 14 years is, at most, very small, even for men in what seems to be the optimal age range of 55 to 69 years. There is no apparent reduction in all-cause mortality. In contrast, the harms associated with the diagnosis and treatment of screen-detected cancer are common, occur early, often persist, and include a small but real risk for premature death. Many more men in a screened population will experience the harms of screening and treatment of screen-detected disease than will experience the benefit. The inevitability of overdiagnosis and overtreatment of prostate cancer as a result of screening means that many men will experience the adverse effects of diagnosis and treatment of a disease that would have remained asymptomatic throughout their lives. Assessing the balance of benefits and harms requires weighing a moderate to high probability of early and persistent harm from treatment against the very low probability of preventing a death from prostate cancer in the long term. The USPSTF concludes that there is moderate certainty that the benefits of PSA-based screening for prostate cancer do not outweigh the harms. 100

101 Reference List (1) ACS. Cancer Facts and Figures American Cancer Society [serial online] (2) Cooperberg MR, Broering JM, Carroll PR. Time trends and local variation in primary treatment of localized prostate cancer. J Clin Oncol 2010;28: (3) Bray F, Lortet-Tieulent J, Ferlay J, Forman D, Auvinen A. Prostate cancer incidence and mortality trends in 37 European countries: an overview. Eur J Cancer 2010;46: (4) Lin K LRMTJS. Benefits and Harms of Prostate-Specific Cancer Screening: An Evidence Update for the U.S. Preventive Services Task Force. Evidence Synthesis No AHRQ Publication No EF-1. Rockville, Maryland: Agency for Healthcare Research and Quality. (5) ACS. Cancer Treatment and Survivorship Facts and Figures American Cancer Society [serial online] (6) Lin K CJKHLCMA. Prostate-Specific Antigen-Based Screening for Prostate Cancer: An Evidence Update for the U.S. Preventive Services Task Force. Evidence Synthesis No. 90. AHRQ Publication No EF AHRQ. (7) Andriole GL, Crawford ED, Grubb RL, III et al. Prostate cancer screening in the randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: mortality results after 13 years of follow-up. J Natl Cancer Inst 2012;104: (8) Wilt TJ, Brawer MK, Jones KM et al. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med 2012;%19;367: (9) Johansson E B-AAHLeal. Time, symptom burden, androgen deprivation, and selfassessed quality of life after radical prostatectomy or watchful waiting: the Randomized Scandinavian Prostate Cancer Group Study Number 4 (SPCG-4) clinical trial. Eur Urol 2009;55: (10) SEER Stat Fact Sheets: Prostate. National Cancer Institute [serial online] (11) Kenneth Lin M, et al. Prostate-Specific Antigen-Based Screening for Prostate Cancer: An Evidence Update for the U.S. Preventive Services Task Force AHRQ Publication No EF-1. (12) Roger Chou M, et al. Treatments for Localized Prostate Cancer: Systematic Review to Update the 2002 U.S. Preventive Services Task Force Recommendation AHRQ Publication No EF

102 (13) Schroder FH, Hugosson J, Roobol MJ et al. Prostate-cancer mortality at 11 years of follow-up. N Engl J Med 2012;366: (14) Wilt TJ, Brawer MK, Jones KM et al. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med 2012;%19;367: (15) Johansson E, Steineck G, Holmberg L et al. Long-term quality-of-life outcomes after radical prostatectomy or watchful waiting: the Scandinavian Prostate Cancer Group-4 randomised trial. Lancet Oncol 2011;12: (16) Hemminki K. Familial risk and familial survival in prostate cancer. World J Urol 2012;30: (17) Brandt A, Bermejo JL, Sundquist J, Hemminki K. Age at diagnosis and age at death in familial prostate cancer. Oncologist 2009;14: (18) Brandt A, Bermejo JL, Sundquist J, Hemminki K. Age-specific risk of incident prostate cancer and risk of death from prostate cancer defined by the number of affected family members. Eur Urol 2010;58: (19) Zeegers MP, Jellema A, Ostrer H. Empiric risk of prostate carcinoma for relatives of patients with prostate carcinoma: a meta-analysis. Cancer 2003;97: (20) Ilic D ODGSWTJ. Screening for prostate cancer Cochrane Database of Systematic Reviews. (21) Hugosson J, Carlsson S, Aus G et al. Mortality results from the Goteborg randomised population-based prostate-cancer screening trial. Lancet Oncol 2010;11: (22) Kilpelainen TP, Tammela TL, Roobol M et al. False-positive screening results in the European randomized study of screening for prostate cancer. Eur J Cancer 2011;47: (23) Andriole GL, Crawford ED, Grubb RL, III et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med 2009;360: (24) Raaijmakers R, Kirkels WJ, Roobol MJ, Wildhagen MF, Schrder FH. Complication rates and risk factors of 5802 transrectal ultrasound-guided sextant biopsies of the prostate within a population-based screening program. Urology 2002;60: (25) Johansson E, Steineck G, Holmberg L et al. Long-term quality-of-life outcomes after radical prostatectomy or watchful waiting: the Scandinavian Prostate Cancer Group-4 randomised trial. Lancet Oncol 2011;12: (26) Abdollah F, et al. Cancer-Specific and Other-Cause Mortality After Radical Prostatectomy Versus Observation in Patients with Prostate Cancer: Competing-Risks 102

103 Analysis of a Large North American Population-Based Cohort. European Urology 60, (27) Fransson P, Damber JE, Tomic R, Modig H, Nyberg G, Widmark A. Quality of life and symptoms in a randomized trial of radiotherapy versus deferred treatment of localized prostate carcinoma. Cancer 2001;92: (28) van Leeuwen PJ, Roobol MJ, Kranse R et al. Towards an optimal interval for prostate cancer screening. Eur Urol 2012;61: (29) Yao SL, Lu-Yao G. Interval after prostate specific antigen testing and subsequent risk of incurable prostate cancer. J Urol 2001;166: (30) Gulati R, Gore JL, Etzioni R. Comparative effectiveness of alternative prostate-specific antigen--based prostate cancer screening strategies: model estimates of potential benefits and harms. Ann Intern Med 2013;158: (31) van Leeuwen PJ, Roobol MJ, Kranse R et al. Towards an optimal interval for prostate cancer screening. Eur Urol 2012;61: (32) Gulati R, Gore JL, Etzioni R. Comparative effectiveness of alternative prostate-specific antigen--based prostate cancer screening strategies: model estimates of potential benefits and harms. Ann Intern Med 2013;158: (33) Harvey P, Basuita A, Endersby D, Curtis B, Iacovidou A, Walker M. A systematic review of the diagnostic accuracy of prostate specific antigen. BMC Urol 2009;9:14. (34) Mistry K, Cable G. Meta-analysis of prostate-specific antigen and digital rectal examination as screening tests for prostate carcinoma. J Am Board Fam Pract 2003;16: (35) Auvinen A, Raitanen J, Moss S et al. Test sensitivity in the European prostate cancer screening trial: results from Finland, Sweden, and the Netherlands. Cancer Epidemiol Biomarkers Prev 2009;18: (36) Schroder FH, Hugosson J, Roobol MJ et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med 2009;360: (37) Thompson IM, Pauler DK, Goodman PJ et al. Prevalence of prostate cancer among men with a prostate-specific antigen level < or =4.0 ng per milliliter. N Engl J Med 2004;350: (38) Thompson IM, Ankerst DP, Chi C et al. Operating characteristics of prostate-specific antigen in men with an initial PSA level of 3.0 ng/ml or lower. JAMA 2005;294:

104 (39) Maattanen L, Hakama M, Tammela TL et al. Specificity of serum prostate-specific antigen determination in the Finnish prostate cancer screening trial. Br J Cancer 2007;96: (40) Hakama M, Stenman UH, Aromaa A, Leinonen J, Hakulinen T, Knekt P. Validity of the prostate specific antigen test for prostate cancer screening: followup study with a bank of 21,000 sera in Finland. J Urol 2001;166: (41) Morgan TO, Jacobsen SJ, McCarthy WF, Jacobson DJ, McLeod DG, Moul JW. Agespecific reference ranges for prostate-specific antigen in black men. N Engl J Med 1996;335: (42) Jacobsen SJ, Guess HA, Oesterling JE. Re: Selection of optimal prostate specific antigen cutoffs for early detection of prostate cancer: receiver operating characteristic curves. J Urol 1996;155: (43) Oesterling JE, Jacobsen SJ, Chute CG et al. Serum prostate-specific antigen in a community-based population of healthy men. Establishment of age-specific reference ranges. JAMA 1993;270: (44) Oesterling JE, Jacobsen SJ, Cooner WH. The use of age-specific reference ranges for serum prostate specific antigen in men 60 years old or older. J Urol 1995;153: (45) Partin AW, Criley SR, Subong EN, Zincke H, Walsh PC, Oesterling JE. Standard versus age-specific prostate specific antigen reference ranges among men with clinically localized prostate cancer: A pathological analysis. J Urol 1996;155: (46) Borer JG, Sherman J, Solomon MC, Plawker MW, Macchia RJ. Age specific prostate specific antigen reference ranges: population specific. J Urol 1998;159: (47) Wolff JM, Brehmer B, Borchers H, Rohde D, Jakse G. Are age-specific reference ranges for prostate specific antigen population specific? Anticancer Res 2000;20: (48) Cooney KA, Strawderman MS, Wojno KJ et al. Age-specific distribution of serum prostate-specific antigen in a community-based study of African-American men. Urology 2001;57:

105 Appendix A Table A-1. Criteria for Assigning Grades of Evidence for Each Outcome 14 Type of evidence Decrease grade Increase grade Randomized trial Observational study Any other evidence Serious limitation to study quality -1 Very serious limitation to study quality -2 Important inconsistency -1 Some uncertainty about directness -1 Major uncertainty about directness -2 Imprecise or sparse data -1 High probability of reporting bias -1 Strong evidence of association: Significant RR of >2 (<0.5) based on consistent evidence from two or more observational studies, with no plausible confounders Very strong evidence of association: Significant RR of >5 (<0.2) based on direct evidence with no major threats to validity All plausible confounders would have reduced the effect High Low Very low Grade High Moderate Low Very low Table A-2. GRADE Levels of Evidence across outcomes Definition Further research is very unlikely to change our confidence in the estimate of effect. Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Any estimate of effect is very uncertain. 14 GRADE Working Group. Grading quality of evidence and strength of recommendations. BMJ. 2004;328:p

106 Table A-3: Determinants of strength of recommendation Guyatt, GH et al. GRADE: going from evidence to recommendations. BMJ :