Association of Clinical Benign Prostate Hyperplasia with Prostate Cancer Incidence and Mortality Revisited: A Nationwide Cohort Study of Men

Transcription

1 EUROPEAN UROLOGY 60 (2011) available at journal homepage: Platinum Priority Prostate Cancer Editorial by Ryan P. Kopp, Stephen J. Freedland and J. Kellogg Parsons on pp of this issue Association of Clinical Benign Prostate Hyperplasia with Prostate Cancer Incidence and Mortality Revisited: A Nationwide Cohort Study of Men David D. Ørsted, Stig E. Bojesen, Sune F. Nielsen, Børge G. Nordestgaard * Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Faculty of Health Sciences, University of Copenhagen, Denmark Article info Article history: Accepted June 8, 2011 Published online ahead of print on June 20, 2011 Keywords: Clinical benign prostate hyperplasia Prostate cancer Incidence Mortality Epidemiology Please visit europeanurology to read and answer questions on-line. The EU-ACME credits will then be attributed automatically. Abstract Background: Although benign prostate hyperplasia (BPH) and prostate cancer (PCa) share features such as hormone-dependent growth and response to treatment with antiandrogen therapy, BPH is generally not considered a premalignant lesion. Objective: To determine whether clinical BPH is associated with an increased risk of PCa incidence and mortality. Design, setting, and participants: Using designs with individual participant data from five national registries, we studied the entire Danish male population from 1980 through 2006, a total of Danish men. We collected PCa diagnoses (n = ), information on PCa mortality (n = ), and ascertained clinical BPH (not histologically proven BPH) through hospitalization (n = ) and/or surgery (n = ) from 1980 to 2006 and the use of a-adrenergic receptor antagonists (n = ) and/or the use of 5a-reductase inhibitors (5-ARIs) (n = ) from 1995 to Measurements: PCa incidence and mortality was assessed for each category of clinical BPH using Kaplan-Meier plots of cumulative incidence and Cox proportional hazard ratios (HRs) adjusted for potential confounders. Results and limitations: For the entire cohort studies, multivariate-adjusted HRs for PCa incidence were 2.22 (95% confidence interval, ) in men hospitalized and 3.26 ( ) in men operated on for clinical BPH versus general population controls. Corresponding HRs for PCa mortality were 2.00 ( ) for hospitalization and 7.85 ( ) for surgery. For age-matched cohort studies, corresponding HRs for PCa incidence were 3.04 ( ) for hospitalization, 2.60 ( ) for surgery, 4.49 ( ) for a-adrenergic receptor antagonist use, and 2.54 ( ) for 5-ARI use. Each category of clinical BPH has limitations, but limitations differ between the categories and therefore are unlikely to explain the results. Conclusions: In Danish men followed for up to 27 yr, clinical BPH was associated with a two- to three-fold increased risk of PCa incidence and with a two- to eight-fold increased risk of PCa mortality. These data should not be used to infer causality. # 2011 European Association of Urology. Published by Elsevier B.V. All rights reserved. * Corresponding author. Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Herlev Ringvej 75, DK-2730 Herlev, Denmark. Tel ; Fax: address: brno@heh.regionh.dk (B.G. Nordestgaard). 1. Introduction Prostate cancer (PCa) is among the most common cancers worldwide [1] and a leading cause of cancer death in men in the developed world [2]. Benign prostate hyperplasia (BPH) is present in 70% of men >70 yr of age [3]. Despite the fact that the two diseases share features such as hormone-dependent growth and response to treatment /$ see back matter # 2011 European Association of Urology. Published by Elsevier B.V. All rights reserved. doi: /j.eururo

2 692 EUROPEAN UROLOGY 60 (2011) with antiandrogen therapy, BPH is generally not considered a premalignant lesion [4]. Nevertheless, due to the common features of BPH and PCa, previous studies have attempted to determine an association between the two [5]. Two studies suggested association [6,7]; the only population-based study showed an increased risk of PCa up to 5 yr after BPH [8] but found no association after >5 yr of follow-up. Importantly, the latter study [8] ended follow-up in 1989 and did not include the period after the introduction of prostate-specific antigen (PSA) testing into clinical practice. One study examining the association between BPH and PCa mortality did not show an association [9]. Thus it is unknown whether a diagnosis of BPH is associated with increased PCa incidence and mortality under current clinical practice. Such data may help physicians determine how closely they should follow men with BPH for possible symptoms of PCa. We tested the hypothesis that clinical BPH (not histologically proven BPH) is associated with an increased risk of PCa incidence and PCa mortality; clinical BPH was hospitalization, surgery, use of a-blockers, and/or 5a-reductase inhibitors (5-ARIs). For this purpose, we followed the entire Danish male population from 1980 through 2006, comprising men. We used information from four nationwide registries: the Danish Patient Registry, Danish Causes of Death Registry, Danish Civil Registration System, and Statistics Denmark; these registries were all complete from 1980 through We used information from the national Danish Register of Medicinal Products Statistics from 1995 through 2006, which was complete during this period. 2. Methods 2.1. Setting and participants The entire male Danish population from 1980 through 2006, comprising individuals, was studied. Using the civil registration number from the Danish Civil Registration System unique to every person in Denmark [10], we drew information on mortality as well as information from four additional nationwide registries, as cited earlier. Details on these registries are described in the supplementary material. The studies were approved by Herlev Hospital, Copenhagen University Hospital, and by Statistics Denmark; in Denmark, approval by scientific ethical committees of anonymous nationwide studies is not necessary End points: prostate cancer incidence and mortality Records on PCa diagnosis and mortality were drawn from the Danish Patient Registry [11,12] and the Danish Causes of Death Registry [13], defining event and date of event. PCa was classified according to the International Classification of Diseases, 8th revision (ICD-8) code 185 and 10th revision (ICD-10) code C61.9 [14,15]. PCa death was the primary, second, or third main cause of death Predictive variable: clinical benign prostate hyperplasia We assessed the presence of clinical BPH in four different ways: through hospitalization, surgery, use of a-adrenergic receptor antagonists, and/or use of 5-ARIs. Records of hospital discharge diagnoses of clinical BPH (ICD-8 code 600; ICD-10 code N40) and surgery for clinical BPH [16] were collected from the Danish Patient Registry. The group of surgical cases was included in the total hospitalized group of clinical BPH. Furthermore, records on pharmacologic treatment for BPH were drawn from the Danish Register of Medicinal Products Statistics [17] and used to categorize men into having clinical BPH according to the use of a- adrenergic receptor antagonists (Anatomical Therapeutic Chemical Classification codes G04CA and C02CA) and/or 5-ARIs (code G04CB). General population controls without clinical BPH, that is, without hospitalization, surgery, use of a-adrenergic receptor antagonists, and/or use of 5-ARIs, were included from the Danish Civil Registration System [10] Other covariates Statistics Denmark records information on ethnicity, highest obtained educational level, and geographic residency. Information on age and gender was provided by the Danish Civil Registration System [10] Statistical analyses and study design First, we examined the association between clinical BPH and PCa incidence and mortality in the entire cohort of all men living in Denmark from 1980 through 2006 (Fig. 1). Second, four age-matched cohort studies were carried out (Fig. 1). In each study, men were identified by different evidence of clinical BPH as described previously. Each man with clinical BPH was matched by birth year with up to five general population controls without evidence of the clinical BPH category examined, creating matched subsets of up to six men; controls may have subclinical BPH. If no controls were available, the men with clinical BPH were excluded from the analysis. For each matched subset, follow-up began for all six at the date of diagnosis of BPH. Third, in sensitivity analyses we repeated age-matched analyses excluding matched subsets where any man in the subset had a diagnosis of PCa within 1, 5, or 10 yr after start of follow-up. Finally, in further sensitivity analyses we calculated hazard ratios (HRs) for the risk of PCa before and after availability of prostate-specific antigen testing from the beginning of All analyses were performed on individual participant data. Statistical analyses were performed using Stata v.11.0 (StataCorp, College Station, TX, USA). A two-sided p value < 0.05 was considered significant. We used Kaplan-Meier curves, log-rank tests, and Cox regression models with delayed entry and left-truncated age as time scale, which implies that age is automatically adjusted for. Cox regression models were used to calculate HRs with 95% confidence intervals (CIs) for PCa incidence and mortality as a function of clinical BPH, and for total mortality as a function of PCa. Competing risks due to any death were accounted for in the analysis by censoring at date of any death (or emigration) for individual participants, information that is 100% complete in the Danish registries. For entire cohort studies from 1980 to 2006 for hospitalization and surgery for clinical BPH (Fig. 1), follow-up began at the 20th birthday, January 1, 1980, date of immigration (whichever came last), and participants were censored at the first date of diagnosis of PCa, death, permanent emigration, or December 31, 2006 (whichever came first). Thus participants were censored at end of follow-up. For calculation of 5-yr risk of PCa and 5-yr mortality as a function of calendar year, men were censored after 5 yr of follow-up. Tests of the proportional hazards assumption were done graphically and showed no major violations up until 80 yr of age in the entire cohort. We therefore calculated HRs for PCa before 80 yr of age in the entire cohort; in these calculations men were censored at the 80th birthday. In the age-matched cohort studies, tests of the proportional hazards assumption showed no major violations up until 100 yr of age. Multivariate models were adjusted for age, year of birth, ethnicity, geographic residency, educational level, and calendar year at the start of follow-up.

3 [(Fig._1)TD$FIG] EUROPEAN UROLOGY 60 (2011) Danish men n = ~3 million Men with prostate cancer: Men dead from prostate cancer: Note: A total of 1964 men with prostate cancer diagnosis before benign prostate hyperplasia diagnosis were excluded Entire cohort studies Men hospitalized with benign prostate hyperplasia: Men operated on for benign prostate hyperplasia: General population controls: Age-matched cohort studies Hospitalization: Men hospitalized with benign prostate hyperplasia: Age-matched general population controls: Surgery: Men operated for benign prostate hyperplasia: Age-matched general population controls : Age-matched cohort studies α-adrenergic receptor antagonists: Men treated: Age-matched general population controls: α-reductase inhibitors: Men treated : Age-matched general population controls : Fig. 1 Overall study design and cohort sizes. Numbers of men with clinical benign prostate hyperplasia are lower in age-matched cohort studies than in entire cohort studies because some of these men were excluded due to lack of controls. 3. Results We studied the entire Danish male population from 1980 through 2006, including men. Table 1 shows the Table 1 Baseline characteristics of the entire Danish male population, baseline characteristics of the cohort. Median age at diagnosis was 72 yr for PCa, 75 yr for BPH hospitalization, 72 yr for BPH-related surgery, 69 yr for a-adrenergic receptor antagonist use, and 72 yr for 5-ARI use. Median age of death from PCa was 78 yr. Median time to diagnosis of PCa after operation for clinical BPH was 3 yr (range: 0 27 yr). The 1964 men with a diagnosis of PCa before a diagnosis of clinical BPH were excluded (Fig. 1). Covariate n 3.1. Entire cohort studies Age group, yr Geographic residency Copenhagen Zealand Funen and Southern Jutland Central Jutland Northern Jutland Descent Danish Other Highest level of education obtained Unknown Primary school High school Labor work Short academic education Medium academic education Long academic education Cumulative incidence of PCa among men hospitalized for clinical BPH (n = ) was higher than among general population controls (n = ) (log-rank p < ) (Fig. 2). Similarly, cumulative incidence of PCa among men operated on for clinical BPH (n = ) was higher than among general population controls (n = ) (log-rank p = ). Multivariate-adjusted HRs for PCa before age 80 yr were 2.22 (95% CI, ) for men hospitalized and 3.26 ( ) for men operated for clinical BPH, versus general population controls (Fig. 2). The 5-yr HR for PCa for men hospitalized for clinical BPH versus general population controls increased gradually from 2.48 (95% CI, ) in 1980 to 4.23 ( ) in 1993 (Fig. 3). This was followed by a sharp increase to a HR of 7.63 ( ) in 1995 coinciding with the introduction of prostate-specific antigen (PSA) testing in clinical practice in Denmark [18]. After this sudden increase, the 5-yr HR decreased somewhat to 5.78 ( ) in The 5-yr HR for total mortality for men with a PCa diagnosis versus general population controls was relatively constant, around a four-fold increased risk from 1980 through 2002, with a slight increase after 1995 (Fig. 3).

4 694 [(Fig._2)TD$FIG] EUROPEAN UROLOGY 60 (2011) Fig. 2 Cumulative incidence of prostate cancer (PCa) for entire cohort studies, The hazard ratios were adjusted for age, year of birth, calendar year of start of follow-up, ethnicity, geographic residency, and level of education. The entire Danish male population, comprising approximately individuals, was studied. The red line shows cumulative incidence of PCa for men hospitalized or operated on for benign prostate hyperplasia. The blue line shows cumulative incidence of prostate cancer for general population controls. CI = confidence interval. We calculated multivariate-adjusted HRs for PCa mortality after hospitalization (n = ), surgery (n = ), and use of a-adrenergic receptor antagonists (n = ) and/or 5-ARIs (n = ) for clinical BPH. HRs were 2.00 (95% CI, ) for men hospitalized and 7.85 ( ) for men operated for clinical BPH, versus general population controls from 1980 through 2006 (Table 2). Corresponding HRs for men treated from 1995 through 2006 with a-adrenergic Table 2 Prostate cancer mortality in entire cohort studies Risk of PCa mortality n PCa deaths HR (95% CI) Hospitalized with BPH, General population controls Men with BPH ( ) Operated on for BPH, General population controls Men with BPH ( ) Prescription of 5a-reductase inhibitors, General population controls Men with BPH ( ) Prescription of a-adrenergic receptor antagonists, General population controls Men with BPH ( ) PCa = prostate cancer; HR = hazard ratio; CI = confidence interval; BPH = benign prostate hyperplasia.

5 [(Fig._3)TD$FIG] EUROPEAN UROLOGY 60 (2011) Hazard ratio for prostate cancer 5 yr after a benign prostate hyperplasia diagnosis Hazard ratio for prostate cancer mortality 5 yr after a prostate cancer diagnosis Calendar year at beginning of 5-yr follow-up Fig. 3 The 5-yr hazard ratios (HRs) for prostate cancer and prostate cancer mortality in entire cohort studies, The solid line represents the HR and the shaded area represents the 95% confidence interval. receptor antagonists and 5-ARIs were 2.58 ( ) and 2.10 ( ) Age-matched cohort studies In an attempt to further exclude influence of age and calendar year on PCa incidence, we also performed nested age-matched studies. For these calculations we included men with clinical BPH matched with up to five general population controls born in the same year. For agematched cohort studies from 1980 to 2006, multivariateadjusted HRs for PCa before 80 yr of age were 3.04 (95% CI, ) for BPH hospitalization and 2.60 ( ) for BPH surgery, versus general population controls (Table 3). Furthermore, to also include men with clinical BPH from general practice in the period where measurement of PSA was readily available in Denmark, we compared men with a prescription for treatment of clinical BPH with up to five age-matched controls. For men treated with a-adrenergic receptor antagonists (n = ) and 5-ARIs (n = ), multivariate-adjusted HRs for PCa before age 80 yr were 2.54 (95% CI, ) and 4.49 ( ) versus general population controls (Table 3) Sensitivity analyses Exclusion of matched subsets of men with clinical BPH, together with controls if any man in a matched subset had a diagnosis of PCa within 5 yr after start of follow-up, resulted in a multivariate-adjusted HR for PCa before age 80 yr for BPH hospitalization from 1980 to 2006 of 1.54 (95% CI, ) (Table 4). The corresponding HR for BPH operation was 1.74 ( ). Table 4 also lists similar HRs for clinical BPH

6 696 EUROPEAN UROLOGY 60 (2011) Table 3 Prostate cancer incidence in age-matched cohort studies Risk of PCa < 80 yr of age Risk of PCa < 100 yr of age n Cancer n HR (95% CI) n Cancer n HR (95% CI) Hospitalized with BPH, General population controls Men with BPH ( ) ( ) Operated on for benign prostate hyperplasia, General population controls Men with BPH ( ) ( ) Prescription of 5a-reductase inhibitors, General population controls Men with BPH ( ) ( ) Prescription of a-adrenergic receptor antagonists, General population controls Men with BPH ( ) ( ) PCa = prostate cancer; HR = hazard ratio; CI = confidence interval; BPH = benign prostate hyperplasia. Table 4 Sensitivity analyses of prostate cancer incidence in age-matched cohort studies HR (95% CI) for PCa < 80 yr of age BPH by: Follow-up: Hospitalization Operation a-adrenergic receptor antagonists a-reductase inhibitors Period of follow-up PCa after start of follow-up All 3.04 ( ) 2.60 ( ) >1 yr 1.61 ( ) 1.90 ( ) >5 yr 1.54 ( ) 1.74 ( ) >10 yr 1.55 ( ) 1.84 ( ) All 2.14 ( ) 2.37 ( ) >1 yr 1.46 ( ) 2.08 ( ) >5 yr 1.54 ( ) 2.03 ( ) >10 yr 1.69 ( ) 2.18 ( ) All 6.77 ( ) 3.48 ( ) 4.49 ( ) 2.54 ( ) >1 yr 2.77 ( ) 2.02 ( ) 3.11 ( ) 1.91 ( ) >5 yr 2.29 ( ) 1.76 ( ) 2.75 ( ) ( ) >10 yr 4.23 ( ) 2.47 ( ) 2.36 ( ) 1.84 ( ) PCa = prostate cancer; HR = hazard ratio; CI = confidence interval; BPH = benign prostate hyperplasia. HRs were adjusted for age, ethnicity, geographic residency, calendar year of start of follow-up, and level of education. *** No information on prescriptions is available for this period. assessed as the use of a-adrenergic receptor antagonists or 5-ARIs from 1995 to For BPH hospitalization versus age-matched controls, multivariate-adjusted HRs for PCa were 2.14 (95% CI, ) with follow-up in and 6.77 ( ) in , respectively, without and with PSA testing readily available (Table 4). Corresponding HRs for BPHrelated surgery were 2.37 ( ) and 3.48 ( ). 4. Discussion The main finding in this study of the entire Danish male population with up to 27 yr of follow-up is that clinical BPH is associated with a two- to three-fold increased risk of PCa incidence and a two- to eight-fold increased risk of PCa mortality. These data should not be used to infer causality. Armenian et al also observed an increased risk of PCa incidence in BPH patients [6]. Conversely, the studies performed by Greenwald et al [9] and Simons et al [19] found no association between BPH and PCa incidence and mortality. Importantly, these studies were all of limited size and not population based. The only previous populationbased study did not test the association between BPH and PCa mortality; in support of our findings, that study found an increased incidence of PCa among Swedish men up to 5 yr after either being hospitalized or operated on for BPH [8]. However, in that study, follow-up ended in 1989 before PSA testing became readily available, which reduces the applicability of these results to current clinical practice. The presence of PCa may lead to a codiagnosis of BPH as a result of the examination or treatment, and vice versa. Increased disease awareness in patients with BPH might also affect the likelihood of being diagnosed with PCa [20]. Yet when we excluded age-matched subsets where any man in the subset had a diagnosis of PCa within 1, 5, or 10 yr after start of follow-up, we still observed a positive

7 EUROPEAN UROLOGY 60 (2011) association. Furthermore, there was a positive association between clinical BPH and PCa mortality, documenting that surveillance bias due to increased disease awareness and codiagnosis cannot likely explain our findings. Our study has several strengths. First, we were able to examine the association between clinical BPH and PCa before and after introduction of PSA testing. Interestingly, even though PSA testing is not recommended for the general population in Denmark, the risk of PCa increased markedly after the introduction of PSA testing, with only a slight concurrent elevation in total mortality after PCa. The lack of general screening with PSA in Denmark may also explain the relatively low incidence of PCa and the relatively higher mortality following a PCa diagnosis because on average more severe PCa cases may be diagnosed in Denmark compared with other countries. Second, we had no losses to follow-up; that is, because of the civil registration number unique for every person living in Denmark, for practical purposes all men were accounted for in the entire follow-up period. Third, we were able to examine the association between clinical BPH and PCa mortality, enabling us to avoid major influence from surveillance bias on risk estimates. Potential limitations of our study include that results are mainly for men of Danish descent living in Denmark, and therefore the results may not necessarily be applicable to other ethnic groups or populations with a health care system or lifestyle different from the Danish. Another limitation is that we cannot adjust for potential risk factors such as smoking and metabolic risk factors [21 23]. Also, because we do not have any information on pathologic findings, we are unable to stratify or adjust for disease grade. Another potential limitation may be misclassification of clinical BPH. For hospitalization for BPH, it is possible that some men hospitalized with lower urinary tract symptoms might receive a diagnosis of BPH without sufficient diagnostic procedures. Nevertheless, a study examining the validity of discharge diagnoses in the Danish Patient Registry found that most of the discharge diagnoses are valid [11],thus reducing the risk of misclassification influencing our results when BPH is assessed through hospitalization. Also, this study examines clinical, not histologically proven BPH. Thus some men may harbor subclinical BPH and still be classified as population controls. Similarly, use of a-adrenergic receptor antagonists may have been misclassified if other conditions (eg, hypertension) led to a prescription of this type of treatment. However, in Denmark a-adrenergic receptor antagonists are generally not recommended for the treatment of hypertension. In any case, this type of misclassification is likely to be nondifferential and would only tend to reduce risk estimates toward the null hypothesis. Thus misclassification of clinical BPH is unlikely to explain our findings. Furthermore, we obtained similar results on the association between clinical BPH and PCa for all four categories of clinical BPH. Yet another potential limitation may be that 5-ARIs influence the risk of PCa. However, in Denmark 5-ARIs are not recommended for the chemoprevention of PCa [24,25]. Furthermore, the current type of study does not allow us to examine the effect of 5-ARIs but merely assesses clinical BPH as use of 5-ARIs. Taken together, each category of clinical BPH has limitations, but limitations differ between the categories individually and therefore are unlikely to explain our results. Finally, a potential limitation is possible misclassification of the end points (ie, PCa incidence and mortality). Some men with PCa might have a false-negative biopsy. Also, for some men PCa might have been assigned as cause of death if there was a history of clinical BPH. In any case, such misclassification may also be nondifferential and if so tend to bias results toward the null hypothesis. Nonetheless, misclassification of PCa mortality cannot explain the positive association between clinical BPH and PCa incidence. 5. Conclusions In the Danish male general population, clinical BPH is associated with a two- to three-fold increased risk of PCa and a two- to eight-fold increased risk of PCa mortality in the period , whereas the risk of PCa incidence was even higher in the period after the introduction of PSA testing. However, these data should not be used to infer causality. Due to the possible effect of other factors, additional studies with detailed information on PSA testing, analysis of biopsies, number of biopsies, number of visits to a urologist, number of digital rectal examinations, familial risk of PCa, and staging of PCa may be useful before guidelines regarding screening of BPH patients for PCa are revised. Author contributions: Børge G. Nordestgaard had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Ørsted, Bojesen, Nielsen, Nordestgaard. Acquisition of data: Ørsted, Nielsen Analysis and interpretation of data: Ørsted, Bojesen, Nielsen, Nordestgaard. Drafting of the manuscript: Ørsted. Critical revision of the manuscript for important intellectual content: Ørsted, Bojesen, Nielsen, Nordestgaard. Statistical analysis: Ørsted, Bojesen, Nordestgaard. Obtaining funding: None. Administrative, technical, or material support: None. Supervision: Nordestgaard. Other (specify): None. Financial disclosures: I certify that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/ affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: This work was supported by Herlev Hospital, Copenhagen University Hospital, the Jascha Foundation, and the University of Copenhagen. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi: /j.eururo

8 698 EUROPEAN UROLOGY 60 (2011) References [1] Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN Int J Cancer 2010;127: [2] Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, CA Cancer J Clin 2010;60: [3] McVary KT. BPH: epidemiology and comorbidities. Am J Manag Care 2006;12:S [4] Bostwick DG, Burke HB, Djakiew D, et al. Human prostate cancer risk factors. Cancer 2004;101: [5] Guess HA. Benign prostatic hyperplasia and prostate cancer. Epidemiol Rev 2001;23: [6] Armenian HK, Lilienfeld AM, Diamond EL, Bross ID. Relation between benign prostatic hyperplasia and cancer of the prostate. A prospective and retrospective study. Lancet 1974;2: [7] Hammarsten J, Hogstedt B. Calculated fast-growing benign prostatic hyperplasia a risk factor for developing clinical prostate cancer. Scand J Urol Nephrol 2002;36: [8] Chokkalingam AP, Nyren O, Johansson JE, et al. Prostate carcinoma risk subsequent to diagnosis of benign prostatic hyperplasia: a population-based cohort study in Sweden. Cancer 2003;98: [9] Simons BD, Morrison AS, Young RH, Verhoek-Oftedahl W. The relation of surgery for prostatic hypertrophy to carcinoma of the prostate. Am J Epidemiol 1993;138: [10] Pedersen CB, Gotzsche H, Moller JO, Mortensen PB. The Danish Civil Registration System. A cohort of eight million persons. Dan Med Bull 2006;53: [11] Nickelsen TN. Data validity and coverage in the Danish National Health Registry. A literature review [in Danish]. Ugeskr Laeger 2001; 164:33 7. [12] Andersen TF, Madsen M, Jorgensen J, Mellemkjoer L, Olsen JH. The Danish National Hospital Register. A valuable source of data for modern health sciences. Dan Med Bull 1999;46: [13] Juel K, Helweg-Larsen K. The Danish registers of causes of death. Dan Med Bull 1999;46: [14] International classification of diseases and related health problems, 8th revision. Geneva, Switzerland: World Health Organization; [15] International classification of diseases and related health problems, tenth revision. Geneva, Switzerland: World Health Organization; [16] Classification of surgical procedures, ed. 1. Copenhagen, Denmark: Nordic Medico-Statistical Committee; [17] Gaist D, Sorensen HT, Hallas J. The Danish prescription registries. Dan Med Bull 1997;44: [18] Mukai TO, Bro F, Pedersen KV, Vedsted P. Use of prostate-specific antigen testing [in Danish]. Ugeskr Laeger 2010;172: [19] Greenwald P, Kirmss V, Polan AK, Dick VS. Cancer of the prostate among men with benign prostatic hyperplasia. J Natl Cancer Inst 1974;53: [20] Jacobs SJ, Girman CJ, Guess HA, et al. Do prostate size and urinary flow rates predict health care-seeking behavior for urinary symptoms in men? Urology 1995;45:64 9. [21] Kristal AR, Arnold KB, Schenk JM, et al. Dietary patterns, supplement use, and the risk of symptomatic benign prostatic hyperplasia: results from the prostate cancer prevention trial. Am J Epidemiol 2008;167: [22] Platz EA, Till C, Goodman PJ, et al. Men with low serum cholesterol have a lower risk of high-grade prostate cancer in the placebo arm of the prostate cancer prevention trial. Cancer Epidemiol Biomarkers Prev 2009;18: [23] Schenk JM, Kristal AR, Neuhouser ML, et al. Serum adiponectin, C-peptide and leptin and risk of symptomatic benign prostatic hyperplasia: results from the Prostate Cancer Prevention Trial. Prostate 2009;69: [24] Thompson IM, Goodman PJ, Tangen CM, et al. The influence of finasteride on the development of prostate cancer. N Engl J Med 2003;349: [25] Tindall DJ, Rittmaster RS. The rationale for inhibiting 5alphareductase isoenzymes in the prevention and treatment of prostate cancer. J Urol 2008;179: