[PubMed] [Google Scholar] 52. primary treatment modality should be considered in future studies examining associations between statins and oncologic outcomes. and impede progression to metastatic disease [18, 19]. Recent observational studies have explored whether statins may have a role in reducing progression after diagnosis. However, these analyses were ill-equipped to evaluate the potentially modifying effects of primary treatment as most only examined outcomes following either radical prostatectomy or radiation therapy. Both radical prostatectomy and radiation therapy (either external beam or brachytherapy) are commonly employed as definitive treatment modalities for early-stage prostate cancer, and outcomes after each treatment may be affected by statin use differentially, especially given the hypothesis that statins may act as radiosensitizers [3]. As has been observed with androgen deprivation therapy, an adjuvant treatment that impacts outcome after radiotherapy (RT) may not affect the outcome after radical prostatectomy [20C22]. Thus, in order to comprehensively evaluate the association of statin use with prostate cancer recurrence and the potentially modifying effects of primary treatment, we undertook a meta-analysis of the available data. methods search methods Search terms were designed by five authors (JDS, HP, RM, MS and RO) to include all studies that investigated the association of statin use with prostate cancer outcomes, using all relevant synonyms for prostate cancer, genitourinary malignancies, and both trade and generic drug GSK429286A names for all statins in clinical use. These search terms are fully detailed in the Appendix. The search was applied to PubMed (1965 to present) and EMBASE (1974 to present), with the last search run on 2 August 2012. All publications, including abstracts, were eligible for retrieval, with duplicate publications removed. In addition, six authors (JDS, HP, RM, MS, RH and RO) conducted a manual review of the reference sections SRSF2 of the retrieved articles in order to identify additional relevant studies. selection criteria All unpublished, published, in press, and in progress studies were initially targeted for review if they were identified in the PubMed or EMBASE search, reported primary data and investigated the association between statins and outcomes after diagnosis among men with initially localized, non-metastatic prostate cancer. Both full-text articles and abstracts were eligible. Epidemiological studies that did not report disease outcomes after diagnosis such as mortality, biochemical failure and disease-specific mortality according to statin use were excluded. Studies including metastatic prostate cancer patients at diagnosis or nonhuman subjects were also excluded. Language selection was limited to articles written in English, Spanish, Italian, Portuguese and French. Each citation was assessed for inclusion independently by at least two out of six authors (JDS, HP, RM, MS, RH and RO), and any discrepancies were arbitrated by all authors. In studies that recorded outcomes for similar or GSK429286A overlapping cohorts, data from the publication with the longest follow-up time were utilized. data extraction and analysis Two independent reviewers extracted data from each study. Hazard ratio (HR) effect estimates were used to assess potential associations between statin use and prostate cancer recurrence following treatment. Biochemical recurrence-free survival (RFS) estimates were used when available; however, progression-free survival estimates were also utilized if biochemical data were not reported separately. We attempted to contact study authors by e-mail to obtain these data if not available from the published reports. Adjusted multivariate estimates were used in all cases. In one study [23], the multivariate result obtained by personal communication was discordant with the univariate analysis; these data were verified before inclusion. Summary HR measurements were calculated with 95% confidence intervals (CIs) using both fixed-effects and DerSimonian and Laird random-effects modeling, which accounted for both within-study and between-study variation [24]. The standard error of the HRs was determined by dividing the difference of the upper and lower 95% CI values on the log scale by 3.92 (twice the value of the test to test for heterogeneity between studies, with a value 0.10 considered statistically significant [26]. We also used the values were two-tailed with = 0.05 to establish statistical significance. This work was carried out in accordance with the guidelines proposed by the Meta-Analysis of Observational Studies in Epidemiology group [31] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) group [32]. STATA 11.1 GSK429286A was used to conduct all statistical analyses (STATA, College Station, TX). results study characteristics Our.