Systematic Review: Comparative Effectiveness and Harms of Treatments for Clinically Localized Prostate Cancer

By T. Wilt, R. MacDonald, I. Rutks, Tatyana A. Shamliyan, B. Taylor, Robert Kane. Annals of Internal Medicine, 2008

625 citations recorded by the source corpus.

Context Sorting through the proven benefits and harms of the multiple strategies available to treat clinically localized prostate cancer is difficult. Contribution This systematic review of 18 randomized trials and 473 observational studies found little high-quality evidence that established the superiority of one therapy over another. All treatments, including androgen deprivation, radical prostatectomy, and radiotherapy, caused urinary, bowel, or sexual dysfunction; the frequency, duration, and severity of these adverse events varied among treatments. Implication Available data insufficiently characterize the relative benefits of various treatments for clinically localized prostate cancer, and all therapies cause some harms. The Editors In 2007, prostate cancer was diagnosed in an estimated 218900 men in the United States, and approximately 27050 men died of the disease. Approximately 90% of men have disease confined to the prostate gland (clinically localized disease). Prostate cancer incidence increased and disease-specific mortality rates decreased after introduction of the prostate-specific antigen (PSA) blood test and with greater use of early interventions (1). Treatment goals are to prevent death and disability from prostate cancer while minimizing intervention-related complications. Common treatments include watchful waiting (expectant management or active surveillance), surgery to remove the prostate gland (radical prostatectomy), external-beam radiation therapy (EBRT) and interstitial radiation therapy (brachytherapy), and androgen deprivation. Patient treatment decisions incorporate physician recommendations; estimated likelihood of cancer progression without early intervention; and treatment-related convenience, costs, and potential for eradication and adverse effects (2). Little is known about how patient and tumor characteristics modify treatment outcomes. This report summarizes evidence from a review prepared for the Agency for Healthcare Research and Quality (3). We determined the comparative short- and long-term benefits and harms of therapies for clinically localized prostate cancer and how patient and tumor characteristics affect the outcomes of these therapies, overall and differentially. Methods Data Sources and Selection of Randomized, Controlled Trials We identified randomized, controlled trials (RCTs) published through mid-September 2007 by using the Cochrane Library (through Issue 3, 2007) and the Cochrane Review Group in Prostate Diseases and Urologic Malignancies specialized registry (through November 2007) (Appendix Figure 1). We included studies if they enrolled men with clinically localized disease (tumor stage T1 or T2) and randomly allocated them to any prostate cancer treatment, including usual care or watchful waiting. We included RCTs that enrolled men with advanced disease (tumor stage T3 or T4) if outcomes were reported separately for localized disease. Appendix Figure 1. Study flow diagram. AUA = American Urological Association; RCT = randomized, controlled trial; RP = radical prostatectomy. Data Sources and Selection of Observational Studies Very few RCTs compared major treatment options, especially for PSA-detected prostate cancer (Appendix Figure 1). Therefore, we reanalyzed a database that primarily comprised nonrandomized reports that we had previously extracted under a separate contract with the American Urological Association (AUA) Prostate Cancer Clinical Guideline Panel (4). The AUA panel used PubMed to identify English-language articles published from 1991 through April 2004. We rejected articles if authors did not report or stratify outcomes for localized disease or reported on fewer than 50 patients. Our review indicated that this database varied greatly in effectiveness estimates, outcome reporting, lack of controls or risk adjustment, and likelihood that it contained results from studies using identical or nearly identical samples. Our technical expert panel recommended against updating this search because the panel did not believe that additional data from nonrandomized studies would provide reliable evidence for assessing comparative effectiveness. We used results from the AUA database to evaluate and demonstrate the range of specific outcomes reported with each intervention in nonrandomized studies and to further assess harms. We obtained additional data on harms and patient satisfaction from the Prostate Cancer Outcomes Study because it is a large, nationally representative prospective survey of men with localized prostate cancer (5). Investigators designed the Prostate Cancer Outcomes Study to assess the long-term health-related quality-of-life outcomes in a large, diverse population-based sample of men in whom prostate cancer was diagnosed in 1994 or 1995. Eligible men were treated in community-based settings and resided in 1 of 6 U.S. Surveillance, Epidemiology, and End Result geographic regions. No RCT evaluated emerging technologies, and the AUA database contained little information on these therapies. Therefore, we used MEDLINE (between April 2004 and September 2007) and contact with Endocare (a manufacturer of cryotherapy devices) to identify and include English-language nonrandomized trials, reviews, and case series of cryotherapy, laparoscopic or robotic-assisted radical prostatectomy, high-intensity focused ultrasonography, proton-beam radiation, and intensity-modulated radiation therapy that reported patient outcomes regardless of size or duration. Study Selection for Effect of Patient and Tumor Characteristics We addressed the effect of patient and tumor characteristics on outcomes of therapies by reviewing RCTs for comparative effectiveness according to age, race, comorbid condition or PSA level, tumor stage, histologic grade, and tumor risk strata. We extracted any study from the AUA database or U.S. populationbased studies that had outcomes stratified according to age or race, looking for comparative effectiveness between treatments according to these factors. We used U.S. populationbased observational studies to evaluate the effect of tumor characteristics on the natural history and survival of men with prostate cancer treated with watchful waiting. Data Extraction and Quality Assessment Two researchers independently abstracted information on study, patient, tumor, and intervention characteristics and outcomes. Outcomes included all-cause and disease-specific mortality, biochemical and clinical progression, adverse events, and patient satisfaction. We assessed the study quality of RCTs on the basis of allocation concealment (6), intention-to-treat-analysis, length of follow-up, and dropouts or loss to follow-up. We rated the strength of the evidence as high (consistent results from 2 high-quality studies with long-term follow-up), medium (data from <2 high-quality studies or studies that did not have long-term follow-up), or low (inconsistent results, studies of low quality or from populations with little relevance to current patients or practice) (3). We considered evidence from nonrandomized trials, case series, and meta-analyses of these studies as low strength. We evaluated applicability of patient populations, clinical settings, and length of follow-up and adjustment for confounding. Data Synthesis and Statistical Analysis Only 3 trials of radical prostatectomy plus neoadjuvant androgen deprivation had data that permitted pooling. We analyzed overall mortality, disease-specific mortality, and biochemical progression or recurrence by using Cochrane Collaboration Review Manager software, version 4.2 (The Cochrane Collaboration, Oxford, United Kingdom) (7). A chi-square test with a P value less than 0.100 and I 2 value greater than 50% indicated statistical heterogeneity (8). We calculated absolute risk differences, odds ratios, and 95% CIs by using random-effects models (9). We present outcomes individually according to study size and duration of follow-up. Role of the Funding Source This project was prepared by the Minnesota Evidence-based Practice Center, Minneapolis, Minnesota, with funding from the Agency for Healthcare Research and Quality. The funding source suggested the initial questions and provided copyright release for this manuscript but did not participate in the literature search, data analysis, or interpretation of the results. Results Summary of Evidence on Treatment Effectiveness The AUA panel search and our additional searches identified 14045 potentially relevant articles. Eighteen RCTs and 473 observational studies met the inclusion criteria. No treatment option had consistent results from at least 2 high-quality RCTs with adequate follow-up and statistical power (Table 1). Only 3 RCTs (1012) compared effectiveness between major treatment categories (radical prostatectomy vs. radiation therapy or watchful waiting). None enrolled men with primarily PSA-detected disease. Many RCTs were inadequately powered to provide long-term survival outcomes. Most reported biochemical progression or recurrence as the main outcomes. No RCT evaluated cryotherapy, laparoscopic or robotic-assisted radical prostatectomy, primary androgen deprivation, high-intensity focused ultrasonography, proton-beam radiation, or intensity-modulated radiation therapy. Table 1. Summary of Evidence on the Comparative Effectiveness and Harms of Therapies for Localized Prostate Cancer Nonrandomized studies varied widely in treatment effectiveness and harms (Figures 1 and 2). The definitions of outcomes (for example, the criteria used to define biochemical progression) and their reporting varied considerably. Investigators rarely stratified outcomes according to patient and tumor characteristics. Many studies included patients with locally advanced disease but did not analyze outcomes on the basis of tumor stage. Figure 1. Overall survival at time points, by treatment, from nonrandomized studies. The symbol size is proportional to the number of patients:

Subfields: Pulmonary and Respiratory Medicine, Urology

View Paper: https://www.semanticscholar.org/paper/78d94c461214031ca9b65aaadbef8095ee548e21

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

  1. M. Cooperberg
  2. P. Carroll
  3. J. Broering
  4. H. S. de Bruijn
  5. Demian van Straten
  6. D. Robinson
  7. Vida Mashayekhi
  8. S. Oliveira

Citations in the corpus