259Archivio Italiano di Urologia e Andrologia 2017; 89, 4 ORIGINAL PAPER The association between prostatitis and prostate cancer. Systematic review and meta-analysis Gianpaolo Perletti 1, 2, Elena Monti 1, Vittorio Magri 3, 4, Tommaso Cai 5, Anne Cleves 6, Alberto Trinchieri 7, Emanuele Montanari 4 1 Department of Biotechnology and Life Sciences, Università degli Studi dell'Insubria, Busto A., Italy; 2 Department of Human Structure and Repair, Faculty of Medicine and Medical Sciences, Ghent University, Ghent, Belgium; 3 Urology Secondary Care Clinic, ASST-Nord, Milan, Italy; 4 Department of Urology, University of Milan Fondazione Ca' Granda IRCCS - Ospedale Maggiore Policlinico, Milan, Italy; 5 Department of Urology, Santa Chiara Regional Hospital, Trento, Italy; 6 Velindre NHS Trust Library, Cardiff University, Velindre Cancer Centre, Cardiff, UK; 7 Urology Unit, A. Manzoni Hospital, Lecco, Italy. Objective: The main outcome of this review was the association between a history of clinical chronic prostatitis (NIH category II or III) and a histo- logically confirmed diagnosis of prostate cancer. Materials and methods: Crude odds ratios and 95% confidence intervals (CI) were calculated to analyze dichotomous data. For analysis of pooled data we adopted a random-effects model and the inverse variance weighing method. Heterogeneity was assessed by calculating the I2 value. Results: Out of 2794 screened records, we retrieved 16 full-text articles written in English, reporting the data of 15 case-control studies, involving 422.943 patients. Pooled analysis resulted in a significant crude odds ratio of 1.83 (95% CI: 1.43 to 2.35; P < 0.00001). The total set of data showed considerable hetero- geneity (I2 = 91%). Both the Egger’s test and the Begg's test for funnel plot asymmetry did not reach statistical significance. The ‘trim and fill’ method applied to the funnel plot imputed 3 missing studies and the resulting adjusted estimate of the odds ratio was 2.12 (95% CI: 1.38 to 3.22). According to GRADE cri- teria, the overall quality of the meta-analysis data is low, mainly due to the presence of bias, confounders and extreme effect size outliers. Five among the included studies reported data assessed in 8015 African-American subjects. Pooled analysis resulted in a non-significant crude odds ratio of 1.59 (95% CI: 0.71 to 3.57; P = 0.26), and considerable heterogeneity (I2 = 90%). Conclusions: Meta-analysis of 15 case-control studies shows that a history of clinical chronic prostatitis can significantly increase the odds for prostate cancer in the general population, whereas such association in African-American individuals remains uncertain. KEY WORDS: Prostate cancer; Prostatitis; Chronic prostatitis; Chronic pelvic pain syndrome; Meta-analysis; Case-control study. Submitted 13 July 2017; Accepted 20 October 2017 Summary No conflict of interest declared. ulcerative colitis, H. pylori gastritis, acid reflux-related esophagitis/Barrett’s syndrome and hepatitis can signifi- cantly increase the risk of developing malignant neo- plasms. In the urological setting, the association between inflammation and urothelial bladder cancer has been recently demonstrated (3), and the interest for the role of inflammation in urolo-genital oncogenesis is increasing. In the last two decades, considerable effort has been devoted to investigate the linkage between inflammation and prostate cancer. For example, at the molecular and cellu- lar level a model has been proposed whereby overexpres- sion of the Vav3 oncogene plays a key role in the trans- duction of aberrant signals leading to both chronic prosta- titis and prostate cancer (4). At the tissue level, inflamma- tion in the prostate can modify the organization of the glands and generate early cancer precursor lesions. Post- atrophic hyperplasia, a variety of ‘proliferative inflammatory atrophy’, is believed by some authors to be ‘fertile ground’ for development of preneoplastic lesions, as it appears to be implicated in lethal prostate cancer (5). At the clinical level, a number of studies investigated whether a history of clinical prostatitis may increase the risk of developing prostate cancer. The results of these studies have been meta-analyzed by Jiang et al. (6). This work, based on a lit- erature search performed up to July 2012, confirmed pre- vious findings (7) and indicated that clinical prostatitis may be significantly associated with prostate cancer. In subsequent years and up to the present day, new case-con- trol studies have been performed on large patient popula- tions. This systematic review is aimed to update and com- plement the meta-analytic data so far produced, focusing on the relationship between a diagnosis of prostate cancer (any grade) and previous exposure to clinical prostatitis. PATIENTS AND METHODS No funding was received to support the present research. Eligibility criteria We included only full-text articles written in English, reporting case-control studies evaluating with various DOI: 10.4081/aiua.2017.4.259 INTRODUCTION Advanced age, family history, BRCA gene mutations and African descent are established risk factors for prostate cancer (1, 2). Inflammation is known to be a major risk factor for vari- ous types of cancer. Strong epidemiological evidence demonstrates that chronic inflammatory diseases like Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 259 Archivio Italiano di Urologia e Andrologia 2017; 89, 4 G. Perletti, E. Monti, V. Magri, T. Cai, A. Cleves, A. Trinchieri, E. Montanari 260 epidemiological and statistical approaches the relation- ship between a history of ‘prostatitis’ and a subsequent diagnosis of prostate cancer. Self-reported or physician-assessed clinical ‘prostatitis’ might include different symptomatic inflammatory condi- tions, characterized by chronic pain in the pelvic region, lower urinary tract symptoms (LUTS) and sexual dysfunc- tion. These conditions are currently classified as NIH cat- egory II Chronic Bacterial Prostatitis (CBP) and NIH cate- gory III Chronic Prostatitis/Chronic Pelvic Pain Syndrome (CP/CPPS, formerly prostatodynia) (8). Studies based exclusively on acute bacterial prostatitis were not included in this review and meta-analysis, as this condition is characterized by short duration and by prompt post-therapy remission. Subgroup data of patients affected by acute prostatitis were excluded from odds-ratio calculation and meta-analysis. Studies focusing exclusive- ly on the assessment of inflammatory cell infiltrates in biopsy or radical prostatectomy specimens (‘histological prostatitis’, NIH category IV) were not included in the present analysis. Patients of any ethnicity with a history of prostate cancer of any grade were eligible for the present study. Prostate cancer diagnosis could be documented by inspection of patient records or could be retrieved from community, hospital, medicare or other national private or public health insurance program databases/registers. Outcomes The main and single outcome considered for this review is the association between a history of clinical chronic prostatitis and prostate cancer of any grade. Search strategy and study selection Search of published reports was performed by an infor- mation specialist (AC). Records were identified by search- ing international databases and trial registers including Medline, PreMedline, Embase, Cochrane Library, Web of Science, LILACS; Scopus, OpenGrey, WHO International Clinical Trial Registry and clinicaltrials.gov. All searches were performed starting from January 1st, 2000, and were assessed as up to date on January 31st, 2017. This time frame has been chosen to minimize the use of different definitions of prostatitis in included studies, as a new clas- sification for clinical prostatitis was implemented in year 1999 and almost universally adopted thereafter (8). In the text of the present review, included studies are referred to by the first author and year of publication. Quality assessment The risk of bias (ROB) of included studies was assessed independently by two researchers (GP, EM). The quality of individual studies was rated using the case-control study version of the Newcastle-Ottawa Scale (NOS) (9), as recommended in chapter 13 of the Cochrane collaboration handbook, addressing the inclu- sion of non-randomized studies in systematic reviews and meta-analyses (Cochrane Handbook for Systematic Reviews of Interventions, version 5.1.0) (10). The thresholds for converting the NOS scores to Agency for Healthcare Research and Quality (AHRQ) standards were: Good quality: 3 or 4 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in out- come/exposure domain. Fair quality: 2 stars in selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in out- come/exposure domain. Poor quality: 0 or 1 star in selection domain OR 0 stars in comparability domain OR 0 or 1 stars in outcome/ exposure domain. The NOS allowed to evaluate ascertainment bias/recall bias (exposure item of the NOS) and hospital control bias (hos- pital control section of the selection item of the NOS), according to Sutton-Tyrrel (11). Detection bias was evaluat- ed separately, according to Cochrane guidelines (10). Publication bias and small-study effect were investigated by visually assessing funnel plots and by performing both the Egger’s regression test and the Begg's rank cor- relation analysis (12, 13). The quality of the evidence resulting from analysis of pooled data was evaluated according to GRADE criteria, modified as recommended in Chapter 13 of the Cochrane handbook (10). Briefly, GRADE recommends rating as ‘low’ all meta-analytic evidence generated by pooling non- randomized studies. Evaluation may be upgraded to ‘moderate’ only in the presence of a large magnitude of effects or of lack of concern about confounders. Data collection and statistical analysis Data extraction was performed by two independent researchers (GP, EM). To analyze dichotomous data we calculated crude (unad- justed) odds ratios (OR) and log-odds ratios. Analysis included the calculation of 95% confidence intervals (CI), and Z statistics. For meta-analysis we adopted a random- effects model and the inverse variance weighing method. Heterogeneity was assessed by calculating the I2 value. We performed Galbraith’s plot analysis to identify out- liers contributing substantially to heterogeneity. Heterogeneity was tentatively investigated by excluding from meta-analysis studies showing small sample sizes, major effect estimate outliers, or specific study design characteristics deviating from the rest of included studies. The ‘trim and fill’ missing study imputation approach was applied to funnel plots, and adjusted overall effect sizes were calculated according to Duval and Tweedie (14). In the presence of a prevalence of exposure in controls (Pctr) higher than 10%, a risk-ratio (RR) approximation was calculated according to vanRhee and Suurmond (15). Pooled analysis was performed using the RevMan 5.3 software. Funnel plots, Galbraith’s plots, the Egger’s and Begg's tests, and ‘trim and fill’ effect size adjustments were performed with the Meta-Essentials Excel workbook 1.0 (Erasmus Research Institute of Management, Erasmus University, Rotterdam, The Netherlands). Optimal infor- mation size for meta-analysis was calculated with the G*Power 3.1 software (assuming an α level equal to 0.05 and a 1-β error equal to 0.95). RESULTS A PRISMA flow-chart of the search and screening process is shown in Figure 1 of Supplementary Materials. A total of 2794 de- duplicated records were identified using our Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 260 261Archivio Italiano di Urologia e Andrologia 2017; 89, 4 Prostate cancer and prostatitis, meta-analysis search strategy. From 29 potentially relevant articles selected by two independent reviewers on the basis of title and abstract content, 16 articles met the inclusion criteria for the present review (16-31). These articles report the results of 15 case-control studies including a total population of 422.943 subjects. The Cheng 2010 and Chao 2010 articles (20, 21) contained essential data from a single study (the California Men’s health Study), whereas it was not sure whether the Rosenblatt 2001 study and the Rothman 2004 study focused on the same case and control populations (28, 30). The sample size of our meta-analysis complied “optimal information size” criteria, as recommended by the GRADE guidelines (32, not shown). Table 1 of Supplementary Materials summarizes the main characteristics of the 15 included studies and the data extracted for the present systematic review. Table 2 of Supplementary Materials presents the quality and bias assessments for the present review. The median score and mode of the Newcastle-Ottawa scale were 5* and 5*, respectively (NOS range: 0 to 9). NOS scores were converted to AHRQ standards. 11 stud- ies were rated as ‘poor’, 2 studies were rated as ‘fair’ and 2 studies were rated as ‘good’ (is shown in Table 2 of Supplementary Materials). Detection bias, evaluated separately, was present in vir- tually all included studies, mainly due to the increased probability of prostate cancer detection in prostatitis patients subjected to intensive follow-up assessments. The prostate cancer cases were 13.942, of which 1,806 were previously exposed to clinical chronic prostatitis;, whereas controls were 409.001, of which 57.203 had a history of clinical chronic prostatitis. Pooled analysis resulted in a significant crude odds ratio of 1.83 (95% CI: 1.43 to 2.35; P < 0.00001). Figure 1 (panel A) shows the forest plot, study data and statistics. The total set of data shows ‘considerable’ heterogeneity (Cochrane handbook, chapter 9) (10), as the calculated value of I2 was 91%. The Hosseini 2010 study (22) appeared to be the major determinant of heterogeneity. This study yielded an odds ratio equal to 32.3, and thus appeared to be an extreme outlier in our analysis. A Galbraith plot was generated, and the Hosseini 2010 study was confirmed to be an effect size outlier (is shown in Figure 2 of Supplementary Materials). The authors of the study were asked to verify whether any error in data analysis/reporting had occurred, or whether exposure vs. non-exposure data had been acci- dentally swapped. Hosseini et al. collaborated fully in this investigation by re-assessing the study database and statis- tics, and confirmed the original results of their study. To assess to which extent the pooled effect size might have been inflated by the presence of extreme outliers, sensitiv- ity analysis was performed by excluding the Hosseini 2010 study. The resulting pooled effect size (crude odds ratio) was 1.55, and retained statistical significance (95% CI: 1.30 to 1.85, P < 0.00001) (Figure 1, panel B). Exclusion of Hosseini 2010 from meta-analysis decreased the I2 value to 81% (‘substantial’ heterogeneity). The Weinmann 2010 study differed from all other thir- teen studies, as it included only lethal prostate cancer cases (19). However, tentative exclusion of this study from the meta-analysis did not modify the I2 value (I2 = 91%). To further explore heterogeneity, the Pelucchi 2006 study was excluded, as data were collected as early as 1985, and exposure might be partly based on a dated defi- nition and understanding of prostatitis (26). The Sutcliffe study was also tentatively excluded, as analy- ses in this study included participants with missing pro- statitis exposure information. In addition, case data were collected as early as 1985, and might be partly based on a pre-1999 definition of prostatitis (24). Exclusion of indi- vidual studies (Pelucchi 2006 or Sutcliffe 2006) did not sub- stantially alter heterogeneity, as the I2 values were 92% and 91%, respectively. The Sarma 2006 study was also tentatively excluded as it included exclusively African-American patients (25), but also in this case heterogeneity was not substantially decreased (I2 = 90%). Five among the included studies reported data assessed in African-American men (17, 20, 23, 25, 27), and meta- analysis in this specific ethnic subgroup was attempted. The total population included 8015 subjects; prostate can- cer cases were 1066, of which 135 had a history of clini- cal chronic prostatitis, whereas the controls were 6949, of which 436 had been previously exposed to the disease. Pooled analysis resulted in a non-significant crude odds ratio of 1.59 (95% CI: 0.71 to 3.57; P = 0.26, Figure 2). When a fixed-effect model was applied to this analysis, the resulting odds ratio was 1.58 (95% CI: 1.23 to 2.03; P = 0.0003). In the frame of the present meta-analysis, the sta- tistical significance of the odds-ratio calculated with such model should be interpreted conservatively. Heterogeneity was ‘considerable’ (I2 = 90%), and appeared to be mainly generated by the Sarma 2006 study (25), as its exclusion yielded a I2 value of 73% (‘substantial’ het- erogeneity). In this study, age distributions differed sub- stantially between cases and controls, with older patients being present in the case cohort. This may imply that con- founding factors (e.g., BPH confounder symptoms or higher number of ‘historical’ sexual partners in the cases cohort) might have played a role in the generation of the outlier odds ratio assessed in this study (crude OR: 5.02). Heterogeneity was not further explored, due to the small number of included studies. Funnel plots were generated to analyze publication bias and small-study effects. The funnel plot (Figure 3, panel A) suggested a certain degree of asymmetry of the data distribution, though such visual impression was not con- firmed by the Egger’s test or by the Begg's rank correla- tion analysis, as neither test reached statistical signifi- cance (Egger’s, P = 0.631; Begg’s, P = 0.125). The ‘trim and fill’ method applied to the funnel plot imputed 3 missing studies (Figure 3, panel A), and the resulting adjusted estimate of the overall effect size was 0.75 log-odds ratio (95% CI: 0.32 to 1.17), whose natu- ral anti-logarithm is 2.12 (95% CI: 1.38 to 3.22). Such adjusted effect size is greater compared to the original finding of the pooled analysis (Log 1.83 = 0.60). Again, the Hosseini 2010 study (22) was a significant out- lier in the funnel plot (Figure 3, panel A). Since Hosseini 2010 shows the smallest sample size among all included studies (137 cases and 137 controls), a small-study effect may account for such a drift, in addition to other unique Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 261 Archivio Italiano di Urologia e Andrologia 2017; 89, 4 G. Perletti, E. Monti, V. Magri, T. Cai, A. Cleves, A. Trinchieri, E. Montanari 262 Figure 2. Subgroup analysis performed on patients of African ethnicity (African-American). Data are plotted according to the increasing weight of each study (top to bottom). Figure 1. Meta-analysis of case-control studies investigating the association between prostate cancer and a previous history of clinical chronic prostatitis. A, pooled analysis of the general population of included prostate cancer cases and controls; B, sensitivity analysis performed by excluding the Hosseini 2010 study (22) from the pooled effect size estimate. The number of subjects allocated to cases or control groups, crude odds ratios, the 95% confidence intervals, the Z value for the overall effect, the significance of the pooled comparisons and heterogeneity data (Chi2, I2), are presented. Data to the right of the vertical no-effect line of forest plots represent increased odds for prostate cancer in patients exposed to prostatitis. Diamonds represent overall effect sizes extending to the limits of the 95% confidence intervals of odds ratios. Data are plotted according to the increasing weight of each study (top to bottom). Total Study Population _______________________________A B Sensitivity analysis (Hosseini 2010 study excluded) _______________________________ Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 262 263Archivio Italiano di Urologia e Andrologia 2017; 89, 4 Prostate cancer and prostatitis, meta-analysis Figure 3. Funnel plot for publication bias analysis. A, ‘trim and fill’ method (14) applied to the analysis of the total study population. The combined effect size (CES, green) and the adjusted estimate of the combined effect size (red) resulting from the imputation of three additional studies (orange) are shown. B, funnel plot analysis performed after excluding the Hosseini 2010 study (22). In these plots the effect sizes are expressed as the natural logarithms of the odds ratios. !"!! !"#! !"$! !"%! !"&! '"!! '"#! (#"!! ('")! ('"!! (!")! !"!! !")! '"!! '")! #"!! #")! *"!! !" #$ %# &% '( && )& '*+ ), '- %% .' /# "0) 1 233(4"'.05('6(#.7&('*+),'-%%.'/#"0)1 +,-./01 2345/60.78990:,7+/;0 <.=-1,0.728+ >4?-,0.7.@,@7?3/6,1 !"!! !"#! !"$! !"%! !"&! '"!! '"#! (#"!! ('"!! !"!! '"!! #"!! )"!! $"!! !" #$ %# &% '( && )& '*+ ), '- %% .' /# "0) 1 233(4"'.05('6(#.7&('*+),'-%%.'/#"0)1 *+,-./0 1234.5/-6788/9+6*.:/ ;-<,0+/-617* =3>,+/-6-?+?6>2.5+0 Hosseini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ini 2esosH A ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 - ,)+*&)&&( %&#%$# !" ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 %!!" &!!" !!'" ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 5.70 0.60 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 !!!" ( !!#" ( )!'" ( !!'" (!" ."4(323 '" #!'" 0/.-,*+ 7-/5.4312 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 )!!" !!!" )!!" !!'" )0"#/.%%-,)+*(&7.#(6(50. /:.*+9/887 *71-/+0,<;- ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 )!'" !!#" )!#" !!*" 1 0+5.2>?+?--/+,>=3 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 1)0"#/ .%% - ,)+ ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 #!!" $!!" ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 B ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 +*&)&&( %&#%$# !" ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 %!!" &!!" !!'" ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 9.40 4.40 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 #!'" 10/.-+, 8.06/5423 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 0;/+,:0998 +82.0,1-=<. ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 1,6/3?@,@..0,-?>4 ' ' ' ' ' ' ' ' ' 7 7 7 7 7 ' ' ' ' ' ' ' ' ' 6 6 6 6 6 Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 263 Archivio Italiano di Urologia e Andrologia 2017; 89, 4 G. Perletti, E. Monti, V. Magri, T. Cai, A. Cleves, A. Trinchieri, E. Montanari 264 features of the study. Exclusion of Hosseini 2010 from the funnel plot analysis (Figure 3, panel B) confirmed the non- significance of the plot asymmetry tests (Egger’s, P = 1.0; Begg’s, P = 0.352). The ‘trim and fill’ analysis imputed 2 missing studies, and the adjusted estimate of the overall effect size was 0.49 log-odds ratio (95% CI: 0.25 to 0.74), whose natural anti-logarithm is 1.63 (95% CI: 1.28 to 2.09) (Figure 3, panel B). Such adjusted odds ratio is slightly higher compared to the original finding of the pooled analysis (without Hosseini 2010, Log 1.55 = 0.44). We did not assess for publication bias in the subgroup analysis of African-American patients, due to the small number of included studies. According to GRADE criteria (32), the overall quality of the meta-analysis data is ‘low’, mainly due to the presence of bias, confounders and extreme outliers. Moreover, the magnitude of the effect size generated from meta-analysis and the assessed heterogeneity did not justify upgrading the quality evaluation to ‘moderate’. DISCUSSION The present meta-analysis of fifteen case-control studies performed between year 2000 and January 31st, 2017 shows that a history of clinical chronic prostatitis can sig- nificantly increase the odds for prostate cancer of any grade (OR = 1.83, 95% CI: 1.43 to 2.35). Our results support and update previous findings, pointing to a sig- nificant association between prostate cancer and expo- sure to prostatitis (Dennis et al., OR = 1.6, 95% CI 1.0 to 2.4; Jiang et al., OR = 1.64, 95% CI: 1.36 to 1.98) (6, 7). Although calculation of the odds ratio is the most appro- priate strategy to retrospectively quantify the association between a disease and a hypothetical risk factor, its inter- pretation is not always straightforward, and the perception of risk can be often overestimated by readers not familiar with its underlying statistics. Thus, we converted the 1.83 odds ratio resulting from our meta analysis to a risk-ratio estimate equal to 1.63 (95% CI: 1.23 to 2.17) (15). Subgroup meta-analysis focusing on men of African descent did not yield significant results when a random- effects model was adopted. It is indeed crucial to assess whether prostatitis may be a risk factor for prostate can- cer in this population, as its incidence is approximately 60% higher and the mortality rate is 2-3 times greater compared with caucasian men (33). Thus, additional studies performed on large patient populations are war- ranted to provide unequivocal evidence in this respect. Included studies were characterized by high risk of bias, mainly due to the presence a number of confounding fac- tors, such as comorbidities (e.g., BPH) in both cases and controls, different intra-study or inter-study proportions of African-American subjects, difficulty in obtaining medical documentation of previous exposure to risk factors, differ- ent inter- and intra-study age ranges, issues in the selection of control populations, population size issues, etc. Increased digital rectal examination, PSA or ultrasound assessment rates may expose patients affected by prostatitis or sexually-transmitted diseases to increased detection of indolent, clinically irrelevant cancers, thus potentially gen- erating detection biases between cases and controls. In addition, recalling bias, due to subjective reporting of pro- statitis exposure, might be universally present in the stud- ies included in this review, also because prostatitis and BPH (likely prevalent in older subjects) are known to be cross-confounders, due to partial symptom overlap (34). These biases are intrinsically present in most case-control investigations, independently of the rigorousness of the study design. Clinical diagnosis of chronic prostatitis presumes the pres- ence of chronic inflammation of the prostatic tissue in diagnosed patients. However, the presence of inflammato- ry mononuclear cells is a very common finding in histo- logical prostate specimens (up to 77%), especially in men beyond the age of 50 (35). Moreover, it has been demon- strated that the distribution of prostatic inflammation is similar for patients with and without chronic prostatitis- like symptoms (36). However, histological chronic inflam- mation has been associated to increased prostate cancer risk in several studies (e.g., 37), though this issue is con- troversial, as other studies have shown that inflammation may actually decrease the risk for prostate cancer (38-40). Thus, from our point of view clinical, symptomatic chron- ic prostatitis and histological evidence of chronic inflam- mation of the prostate should be provisionally investigat- ed as separate entities, whose impact on prostatic oncoge- nesis may be based on distinct mechanisms of action at the tissue, cellular or molecular levels. Clinical implications and key points The prognostic and therapeutic implications of our find- ings, together with the findings of Dennis et al. and Jiang et al. (6, 7), may be of considerable importance. In this respect, it might be interesting to investigate whether aggressive therapeutic management of chronic prostatitis syndromes may have cancer-preventive potential. CONCLUSIONS Meta-analysis of 15 case-control studies shows that a his- tory of clinical chronic prostatitis can significantly increase the odds for prostate cancer in the general pop- ulation, whereas such association in African-American individuals remains uncertain. REFERENCES 1. Modena A, Iacovelli R, Scarpa A, et al. Investigating BRCA Mutations: A Breakthrough in Precision Medicine of Castration- Resistant Prostate Cancer. Target Oncol. 2016; 11:569-577. 2. Rebbeck TR. Prostate Cancer Genetics: Variation by Race, Ethnicity, and Geography. Semin Radiat Oncol. 2017; 27:3-10. 3. Nesi G, Nobili S, Cai T, et al. Chronic inflammation in urothelial bladder cancer. Virchows Arch. 2015; 467:623-33. 4. Liu Y, Mo JQ, Hu Q, et al. Targeted overexpression of vav3 onco- gene in prostatic epithelium induces nonbacterial prostatitis and prostate cancer. Cancer Res. 2008; 68:6396-406. 5. Davidsson S, Fiorentino M, Andrén O, et al. Inflammation, focal atrophic lesions, and prostatic intraepithelial neoplasia with respect to risk of lethal prostate cancer. Cancer Epidemiol Biomarkers Prev. 2011; 20:2280-7. 6. Jiang J, Li J, Yunxia Z, et al. The role of prostatitis in prostate can- cer: meta-analysis. PLoS One 2013; 8:e85179. Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 264 265Archivio Italiano di Urologia e Andrologia 2017; 89, 4 Prostate cancer and prostatitis, meta-analysis Correspondence Gianpaolo Perletti, PhD (Corresponding author) gianpaolo.perletti@uninsubria.it Università degli Studi dell’Insubria Department of Biotechnology and Life Sciences Via A. Da Giussano, 12 21052 Busto A., Italy Elena Monti, PhD Università degli Studi dell’Insubria Department of Biotechnology and Life Sciences Via A. Da Giussano, 12 21052 Busto A., Italy Vittorio Magri, MD Urology Secondary Care Clinic, ASST-Nord, Milan, Italy Tommaso Cai, MD Department of Urology, Santa Chiara Regional Hospital, Trento, Italy Anne Cleves, Mrs Velindrencer NHS Trust Library, Cardiff University, Velindre Cancer Centre, Cardiff, UK Alberto Trinchieri, MD Urology Unit, A. Manzoni Hospital, Lecco, Italy Emanuele Montanari, MD Department of Urology, University of Milan Fondazione Ca' Granda IRCCS - Ospedale Maggiore Policlinico, Milan, Italy 7. Dennis LK, Lynch CF, Torner JC. Epidemiologic association between prostatitis and prostate cancer. Urology. 2002; 60:78-83. 8. Krieger JN, Nyberg L Jr, Nickel JC. NIH consensus definition and classification of prostatitis. JAMA 1999; 282:236-7. 9. http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp 10. http://handbook.cochrane.org/ 11. Sutton-Tyrrell K. Assessing bias in case-control studies. Proper selection of cases and controls. Stroke. 1991; 22:938-42. 12. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta- analysis detected by a simple, graphical test. BMJ. 1997; 315:629-34. 13. Begg CB, Mazumdar M. Operating characteristics of a rank cor- relation test for publication bias. Biometrics. 1994; 50:1088-101. 14. Duval S, Tweedie R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analy- sis. Biometrics. 2000; 56:455-63. 15. https://www.erim.eur.nl/fileadmin/erim_content/images/meta- essentials/Meta-analyze_dichotomous_data.pdf 16. Nair-Shalliker V, Yap S, Nunez C, et al. Adult body size, sexual history and adolescent sexual development, may predict risk of devel- oping prostate cancer: Results from the New South Wales Lifestyle and Evaluation of Risk Study (CLEAR). Int J Cancer. 2017; 140:565-574. 17. Rybicki BA, Kryvenko ON, Wang Y, et al. Racial differences in the relationship between clinical prostatitis, presence of inflamma- tion in benign prostate and subsequent risk of prostate cancer. Prostate Cancer Prostatic Dis. 2016; 19:145-50. 18. Boehm K, Valdivieso R, Meskawi M, et al. Prostatitis, other gen- itourinary infections and prostate cancer: results from a population- based case-control study. World J Urol. 2016; 34:425-30. 19. Weinmann S, Shapiro JA, Rybicki BA, et al. Medical history, body size, and cigarette smoking in relation to fatal prostate cancer. Cancer Causes Control. 2010; 21:117-25. 20. Cheng I, Witte JS, Jacobsen SJ, et al. Prostatitis, sexually trans- mitted diseases, and prostate cancer: the California Men's Health Study. PLoS One 2010; 5:e8736. 21. Chao C, Haque R, Van Den Eeden SK, et al. Red wine con- sumption and risk of prostate cancer: the California men's health study. Int J Cancer. 2010; 126:171-9. 22. Hosseini M, Seyed Alinaghi S, Mahmoudi M, McFarland W. A case-control study of risk factors for prostate cancer in Iran. Acta Med Iran. 2010; 48:61-6. 23. Huang WY, Hayes R, Pfeiffer R, et al. Sexually transmissible infections and prostate cancer risk. Cancer Epidemiol Biomarkers Prev. 2008; 17:2374-81. 24. Sutcliffe S, Giovannucci E, De Marzo AM, et al. Gonorrhea, syphilis, clinical prostatitis, and the risk of prostate cancer. Cancer Epidemiol Biomarkers Prev. 2006; 15:2160-6. 25. Sarma AV, McLaughlin JC, Wallner LP, et al. Sexual behavior, sexually transmitted diseases and prostatitis: the risk of prostate cancer in black men. J Urol. 2006; 176:1108-13. 26. Pelucchi C, Talamini R, Negri E, et al. Genital and urinary tract diseases and prostate cancer risk. Eur J Cancer Prev. 2006; 15:254-7. 27. Patel DA, Bock CH, Schwartz K, et al. Sexually transmitted dis- eases and other urogenital conditions as risk factors for prostate cancer: a case--control study in Wayne County, Michigan. Cancer Causes Control. 2005; 16:263-73. 28. Rothman I, Stanford JL, Kuniyuki A, Berger RE. Self-report of prostatitis and its risk factors in a random sample of middle-aged men. Urology. 2004; 64:876-9. 29. Roberts RO, Bergstralh EJ, Bass SE, et al. Prostatitis as a risk factor for prostate cancer. Epidemiology. 2004; 15:93-9. 30. Rosenblatt KA, Wicklund KG, Stanford JL. Sexual factors and the risk of prostate cancer. Am J Epidemiol. 2001; 153:1152-8. 31. Wright JL, Lin DW, Stanford JL. Circumcision and the risk of prostate cancer. Cancer. 2012; 118:4437-43. 32. Guyatt GH, Oxman AD, Kunz R, et al. GRADE guidelines 6. Rating the quality of evidence - imprecision. J Clin Epidemiol. 2011; 64:1283-93. 33. Kheirandish P, Chinegwundoh F. Ethnic differences in prostate cancer. Br J Cancer. 2011; 105:481-5. 34. Collins MM, Meigs JB, Barry MJ, et al. Prevalence and corre- lates of prostatitis in the health professionals follow-up study cohort. J Urol. 2002; 167:1363-6. 35. Nickel JC, Roehrborn CG, O'Leary MP, et al. The relationship between prostate inflammation and lower urinary tract symptoms: examination of baseline data from the REDUCE trial. Eur Urol. 2008; 54:1379-84. 36. Nickel JC, Roehrborn CG, O'leary MP, et al. Examination of the relationship between symptoms of prostatitis and histological inflammation: baseline data from the REDUCE chemoprevention trial. J Urol. 2007; 178:896-900. 37. Benedetti I, Bettin A, Reyes N. Inflammation and focal atrophy in prostate needle biopsy cores and association to prostatic adeno- carcinoma. Ann Diagn Pathol. 2016; 24:55-61. 38. Porcaro AB, Novella G, Mattevi D, et al. Chronic Inflammation in Prostate Biopsy Cores is an Independent Factor that Lowers the Risk of Prostate Cancer Detection and is Inversely Associated with the Number of Positive Cores in Patients Elected to a First Biopsy. Curr Urol. 2016; 9:82-92. 39. Moreira DM, Nickel JC, Gerber L, et al. Baseline prostate inflammation is associated with a reduced risk of prostate cancer in men undergoing repeat prostate biopsy: results from the REDUCE study. Cancer. 2014; 120:190-6. 40. Kryvenko ON, Jankowski M, Chitale DA, et al. Inflammation and preneoplastic lesions in benign prostate as risk factors for prostate cancer. Mod Pathol. 2012; 25:1023-32. Perletti print_Stesura Seveso 03/01/18 09:35 Pagina 265