Sexually Transmitted Infections, Prostatitis, Ejaculation Frequency, and the Odds of Lower Urinary Tract Symptoms

Transcription

1 American Journal of Epidemiology Copyright ª 2005 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A. Vol. 162, No. 9 DOI: /aje/kwi299 Advance Access publication September 21, 2005 Original Contribution Sexually Transmitted Infections, Prostatitis, Ejaculation Frequency, and the Odds of Lower Urinary Tract Symptoms Siobhan Sutcliffe 1, Edward Giovannucci 2, Angelo M. De Marzo 3,4, Walter C. Willett 2, and Elizabeth A. Platz 1,3 1 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD. 2 Departments of Nutrition and Epidemiology, Harvard School of Public Health, and the Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women s Hospital, Boston, MA. 3 James Buchanan Brady Urological Institute and the Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins Medical Institutions, Baltimore, MD. 4 Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, MD. Received for publication February 24, 2005; accepted for publication June 10, Early life sexual factors, including histories of sexually transmitted infections, young-onset prostatitis, and frequency of ejaculation, were investigated in relation to lower urinary tract symptoms (LUTS) in a large case-control study nested within the Health Professionals Follow-up Study. In 1992, study participants were asked to provide information on their histories of sexually transmitted infections, prostatitis, ejaculation frequency, surgery for an enlarged prostate, and LUTS. Information on prostate surgery and LUTS was updated every 2 years. LUTS cases were defined as men who reported surgery for an enlarged prostate or high-moderate to severe LUTS (15 points on the American Urological Association symptom index) on any study questionnaires (n ¼ 4,608). Controls were men who did not report surgery for an enlarged prostate and who scored 0 7 points on the American Urological Association symptom index on all questionnaires (n ¼ 17,967). (adjusted odds ratio ¼ 1.76, 95% confidence interval: 1.43, 2.15) or young-onset prostatitis (adjusted OR ¼ 1.55, 95% confidence interval: 1.22, 1.96) was positively associated with LUTS. No association was observed between ejaculation frequency in early adulthood and LUTS. These results suggest that early genitourinary infections may contribute to later development of LUTS, although confirmation in additional population settings is warranted. ejaculation; gonorrhea; prostatic hyperplasia; prostatitis; sexually transmitted diseases; syphilis Abbreviations: BPH, benign prostatic hyperplasia; CI, confidence interval; LUTS, lower urinary tract symptoms; OR, odds ratio. Benign prostatic hyperplasia (BPH), a common condition among middle-aged and older men, is characterized clinically by lower urinary tract symptoms (LUTS) and a palpably enlarged prostate and histologically by hyperplastic nodules in the periurethral region and transition zone. By 70 years of age, over 40 percent of men may have LUTS (1), and over 70 percent may have histologic evidence of BPH (2). Despite its high frequency, relatively little is known regarding the etiology of BPH. The underlying pathology of BPH is believed to begin before 30 years of age, although symptoms do not typically occur until later ages (2). Thus, exogenous and endogenous factors that affect prostatic growth early in life may be important for the initial development of BPH. Potential early life factors include sexually transmitted infections, which are typically acquired during youth and early adulthood, young-onset prostatitis Correspondence to Dr. Elizabeth A. Platz, Department of Epidemiology, Room E6138, Johns Hopkins Bloomberg School of Public Health, 615 North Wolfe Street, Baltimore, MD ( eplatz@jhsph.edu). 898

2 Early Life Factors and Benign Prostatic Hyperplasia 899 (inflammation of the prostate due mainly to Escherichia coli infection (3)), and frequency of early life sexual activity. Several sexually and nonsexually transmitted agents, such as Neisseria gonorrhoeae, the causative agent of gonorrhea, and E. coli, have been shown to infect and induce inflammation in the prostate (3, 4). Chronic inflammation has long been believed to contribute to the development of BPH nodules (5) and is frequently observed in prostate tissue sections from BPH patients (6 9). Studies of the rat prostate further suggest that inflammation may precede or induce the development of hyperplastic lesions similar to those observed in men with BPH (10, 11). In an argument similar to that proposed for prostate cancer (12), early life sexual activity may contribute to the development of BPH by determining the frequency with which prostatic secretions are expelled from the prostate. These secretions may contain cytotoxic constituents and, when stagnant, may form irritative precipitates capable of inducing an inflammatory immune response. Early life sexual activity may also serve as a marker of androgenicity or exposure to sexually transmitted agents. Few studies have investigated early life sexually transmitted infections and sexual activity in relation to BPH (13 18), and none, to our knowledge, have investigated a history of young-onset prostatitis and BPH. Although the results of these studies are suggestive of an association between sexually transmitted infection history and BPH (but inconsistent for early life sexual activity), all of these studies were small in size and, with the exception of one (17), assessed sexual history retrospectively in the context of a prostate disease case-control study, allowing for potential recall and interviewer biases. To further explore early life sexual factors and BPH, we investigated self-reported histories of sexually transmitted infections, young-onset prostatitis, and early life sexual activity as they relate to subsequent surgery for an enlarged prostate and LUTS in a large study nested within the Health Professionals Follow-up Study. A history of any form of prostatitis and BPH has previously been investigated in this cohort and observed to be positively associated with LUTS, likely because of symptom overlap (19); in the present study, we focused on early life prostatitis. Men in this cohort were unselected for medical concerns, were unaware of study hypotheses at the time of exposure and outcome assessment, and provided study information by mailed questionnaires, thereby reducing the potential for recall and interviewer biases. MATERIALS AND METHODS Study population The Health Professionals Follow-up Study is an ongoing, prospective cohort study of 51,529 male health professionals aged years at enrollment in At baseline, participants completed a mailed questionnaire on demographics, lifestyle, and medical history, as well as a semiquantitative food frequency questionnaire. Since this time, participants have completed questionnaires biennially to update and expand exposure and disease information. For this analysis, we excluded men who did not return the 1992 questionnaire (n ¼ 11,129), which requested information on sexually transmitted infections, prostatitis, ejaculation frequency, and LUTS, and those who did not provide complete information on LUTS (n ¼ 1,938). Additionally, we excluded participants who did not adequately complete the 1986 food frequency questionnaire, had a history of cancer (except nonmelanoma skin cancer) or surgery for an enlarged prostate prior to 1992, or were diagnosed with prostate cancer through 2000 (to increase the specificity of LUTS classification), leaving 30,123 participants in the analysis. This study was approved by the Human Subjects Committee at the Harvard School of Public Health and the Committee on Human Research at the Johns Hopkins Bloomberg School of Public Health. Assessment of a history of sexually transmitted infections, prostatitis, and frequency of ejaculation On the 1992 questionnaire, participants were asked whether they had ever had a diagnosis of syphilis, gonorrhea, or neither; whether they had ever had prostatitis or prostatic infection; how long their symptoms had persisted (< 1, 1 2, 3 5, 6 10, or >10 years); whether they had ever been treated for prostatitis; and at what age they were first treated (<30, 30 39, 40 49, 50 59, or 60 years). Participants were further asked how many ejaculations they had on average per month at ages and years and during the past year (none, 1 3, 4 7, 8 12, 13 20, or >20). We categorized gonorrhea and syphilis separately, as no, yes, or missing (but responded to the 1992 questionnaire), and prostatitis as no, young-onset prostatitis (defined as symptoms of prostatitis of less than 1 year s duration and treatment before the age of 30 years), other prostatitis (defined as any other reported prostatitis), and missing. Ejaculation frequency from ages years was used to describe early life sexual activity. We combined the lower two ejaculation frequency categories (none and 1 3 times/month) because of small numbers and included a missing category. Ejaculation frequencies from ages years and in the year prior to completion of the 1992 questionnaire were categorized in the same manner. Assessment of LUTS LUTS was assessed by a history of 1) surgery for symptomatic prostatic enlargement, 2) specific LUTS, and 3) use of medications to treat LUTS. On the 1988 and each subsequent questionnaire, participants were asked whether they had ever had surgery for an enlarged prostate (typically transurethral resection of the prostate), the responses to which have been previously validated (20). For the present study, only surgery occurring after the date of return of the 1992 questionnaire was included. On the 1992, 1994, 1998, and 2000 follow-up questionnaires, participants were asked to indicate the frequency (0, 10, 25, 50, 75, or almost 100 percent of the time) with which they experienced the following six LUTS during the previous month (modified slightly from the American Urological Association symptom index (21) to fit the constraints of our mailed questionnaire): 1) sensation of incomplete bladder

3 900 Sutcliffe et al. TABLE 1. Age-standardized characteristics by histories of gonorrhea, prostatitis, and frequency of ejaculation, Health Professionals Follow-up Study, 1986* y,z No (n ¼ 26,988) Yes (n ¼ 832) No (n ¼ 25,650) emptying (incomplete emptying), 2) having to urinate again after less than 2 hours (frequency), 3) stopping and starting several times during urination (intermittency), 4) difficulty postponing urination (urgency), 5) weak urinary stream (weak stream), and 6) having to push or strain to begin urination (hesitancy). Participants were further asked how many times they typically had to get up to urinate during the night (zero, one, two, three, four, five, or six or more times) over the past month (nocturia). Each of the above symptoms (with the exception of nocturia) was assigned a score of 0 5 corresponding to the percentage of time that the symptom was reported. The scores for each of the six LUTS and nocturia (with six or more times assigned a score of 5) were then summed to obtain a total symptom score that ranged from 0 to 35. Participants who did not answer any of the first six LUTS questions on the 1992 questionnaire y,z Young-onset prostatitis{ (n ¼ 658) Other prostatitis (n ¼ 3,603) were excluded from the analysis, with the exception of men who reported surgery for an enlarged prostate. Men who responded to at least one of the first six questions, but left others blank, were assumed not to have those symptoms. On the 1998 and 2000 follow-up questionnaires, participants were also asked whether they currently used (two or more times per week) finasteride or alpha-blockers to treat LUTS. American Urological Association symptom index scores were not summed for men who reported surgery for an enlarged prostate or use of LUTS medications during the follow-up period. Case and control definitions Monthly ejaculation frequency at years of agey, 0 3 (n ¼ 865) 8 12 (n ¼ 7,742) >20 (n ¼ 6,663) Total prevalent LUTS# (%)** Mean age (years)y Race/ethnicity (%) Southern European Scandinavian Other White African American Asian Other (%)y History of syphilis (%)y (%)y,yy No Young-onset prostatitis{ Other prostatitis Ejaculation frequency at years of age (%)yy 0 3/month /month /month /month >20/month Alcohol intake at years of age (%)yy,zz None <1 1 drink/week drinks/week drinks/week drinks/week Table continues Several case definitions were used: 1) surgery for an enlarged prostate, defined as a report of surgery for prostatic enlargement between the date of return of the 1992

4 Early Life Factors and Benign Prostatic Hyperplasia 901 TABLE 1. Continued y,z No (n ¼ 26,988) Yes (n ¼ 832) questionnaire and January 31, 2000; 2) prevalent LUTS, defined as an American Urological Association symptom index score of 15 or more points on any of the follow-up questionnaires or use of medications to treat LUTS among participants who never reported surgery for an enlarged prostate; and 3) total prevalent LUTS, consisting of surgery for an enlarged prostate and prevalent LUTS. Prevalent LUTS cases were further classified as having high-moderate to severe LUTS (15 of 35 possible points), severe LUTS (20 of 35 possible points), severe irritative LUTS (9 of 15 possible points for frequency, urgency, and nocturia), and severe obstructive LUTS (12 of 20 possible points for incomplete emptying, intermittency, weak stream, and hesitancy). Parallel incident case definitions were used for men No (n ¼ 25,650) y,z Young-onset prostatitis{ (n ¼ 658) who reported a low American Urological Association symptom index score (0 7 points) on the 1992 questionnaire and on the 1994 questionnaire, if completed, and LUTS (defined as above) on subsequent questionnaires. Controls were defined as participants who never reported surgery for an enlarged prostate, who scored 0 7 points on the American Urological Association symptom index on all completed questionnaires, and who did not report use of medications to treat LUTS. Statistical analysis Other prostatitis (n ¼ 3,603) Monthly ejaculation frequency at years of agey, 0 3 (n ¼ 865) 8 12 (n ¼ 7,742) >20 (n ¼ 6,663) Cigarette smoking before age 30 years (%)yy None >0 4 pack-years pack-years pack-years pack-years Mean body mass index at 21 years of age (kg/m 2 ) Lack of strenuous physical activity while in school (%)y High school College Mean total alcohol intake (g/day) Cigarette smoking status (%) Never Former Current Mean leisure-time physical activity (MET#-hours/week) Mean total energy intake (kcal) 2,006 1,989 2,003 2,008 2,000 1,947 1,977 2,075 Mean body mass index (kg/m 2 ) Vasectomy (%) Diabetes mellitus type 2 (%) * Among men who responded to the 1992 questionnaire on which there were questions to elicit histories of sexually transmitted infections, prostatitis, and ejaculation frequency. y Assessed in z Numbers do not sum to 30,123 because of missing participant responses. Only three of six ejaculation frequency categories are shown. { Defined as prostatitis treated before the age of 30 years that was of less than 1 year s duration. # LUTS, lower urinary tract symptoms; MET, metabolic equivalent. ** Defined as surgery for an enlarged prostate, high-moderate to severe LUTS, or use of medications to treat LUTS. yy Percentages may not sum to 100% because of missing participant responses. zz Assessed in Age-standardized means and proportions were calculated for demographic and lifestyle factors by histories of gonorrhea,

5 902 Sutcliffe et al. TABLE 2. Odds ratios and 95% confidence intervals of total lower urinary tract symptoms in relation to histories of gonorrhea and prostatitis, Health Professionals Follow-up Study, No. of cases Age-adjusted odds ratio 95% confidence interval Multivariable odds ratio* 95% confidence interval Total prevalent LUTSy z No 4, Referent 1.00 Referent Yes , , 2.15 z No 3, Referent 1.00 Referent Young-onset prostatitis , , 1.96 Other prostatitis , , 2.57 Surgery for an enlarged prostate No 1, Referent 1.00 Referent Yes , , 2.08 No 1, Referent 1.00 Referent Young-onset prostatitis , , 2.28 Other prostatitis , , 2.85 Prevalent high-moderate to severe LUTS No 2, Referent 1.00 Referent Yes , , 2.37 No 2, Referent 1.00 Referent Young-onset prostatitis , , 1.97 Other prostatitis , , 2.49 Prevalent severe LUTS No Referent 1.00 Referent Yes , , 3.67 No Referent 1.00 Referent Young-onset prostatitis , , 3.05 Other prostatitis , , 3.44 Table continues syphilis, prostatitis, and ejaculation frequency. Logistic regression was used to calculate age- and multivariableadjusted odds ratios and 95 percent confidence intervals for each case definition and its association with each exposure. All age- and multivariable-adjusted models included missing exposure terms for comparison with the referent categories. When odds ratios for missing exposure differed from the null, sensitivity analyses were performed, combining men with missing exposure with men in the referent category (because of similarities in demographic and lifestyle distributions) to investigate possible biases due to missing information. None of the results from sensitivity analyses differed from the main results. Multivariable models also included terms for age, race/ethnicity, alcohol consumption, cigarette smoking, leisure-time vigorous physical activity, energy intake, body mass index, vasectomy, and diabetes mellitus type 2. Covariate values at baseline in 1986 were used because they were ascertained prior to assessment of LUTS and, thus, should be less influenced by the outcome. Stratified analyses were performed to determine whether associations between histories of gonorrhea and young-onset prostatitis and the odds of LUTS were modified by factors postulated to influence the presence and duration of prostatic inflammation early in life. Factors considered were aspirin use at the ages of years and age (<70, 70 years) as a surrogate measure for opportunity to have acquired an infection in the pre-antibiotic era before 1937 when acute prostatitis was a common sequela of sexually transmitted

6 Early Life Factors and Benign Prostatic Hyperplasia 903 TABLE 2. Continued infections (4). The statistical significance of stratum-specific differences was assessed by including a cross-product term along with the main effect terms in the logistic regression models and evaluating the cross-product term coefficient by the Wald test. Statistical analyses were performed using SAS version 8.2 software (SAS Institute, Cary, North Carolina). All reported p values are two sided. RESULTS No. of cases Age-adjusted odds ratio Of the 30,123 eligible participants, 1,757 reported surgery for an enlarged prostate between 1992 and January 31, 2000, and among those who did not report surgery, 95% confidence interval Multivariable odds ratio* 95% confidence interval Total incident LUTS{ No 1, Referent 1.00 Referent Yes , , 2.33 No Referent 1.00 Referent Young-onset prostatitis , , 1.90 Other prostatitis , , 1.77 Incident high-moderate to severe LUTS{ No Referent 1.00 Referent Yes , , 2.72 No Referent 1.00 Referent Young-onset prostatitis , , 2.09 Other prostatitis , , 1.48 Incident severe LUTS{ No Referent 1.00 Referent Yes , , 4.38 No Referent 1.00 Referent Young-onset prostatitis , 2.81 # Other prostatitis , 2.21 * Adjusted for age (in 3-year intervals), race/ethnicity (southern European, Scandinavian, other White, African American, Asian, and other), baseline alcohol intake (g/day), cigarette smoking status (never, former, current), leisure-time physical activity (metabolic equivalent-hours/week), total energy intake (kcal), body mass index (kg/m 2 ), vasectomy, and diabetes mellitus type 2. y LUTS, lower urinary tract symptoms (defined as surgery for an enlarged prostate, high-moderate to severe LUTS, or use of medications to treat LUTS). z Participants with missing history of gonorrhea (n ¼ 2,303) or prostatitis (n ¼ 212) information were included in the analyses; results not shown. Defined as prostatitis treated before the age of 30 years that was of less than 1 year s duration. { Among men with low American Urological Association symptom index scores on the 1992 questionnaire and on the 1994 questionnaire, if completed. # Too few cases to perform multivariable adjustment. 2,851 reported high-moderate to severe LUTS. Three percent of eligible participants reported a history of gonorrhea, 0.2 percent reported a history of syphilis, 2.2 percent reported a history of young-onset prostatitis, and 12.0 percent reported a history of other prostatitis. At ages years, the most common frequency of ejaculation was ejaculations/month (32.1 percent). In general, participants who reported a history of gonorrhea, syphilis (data not shown), young-onset prostatitis, or higher frequencies of ejaculation at ages years were younger and reported higher levels of alcohol consumption and cigarette smoking during their youth and in 1986 (table 1). Additionally, gonorrhea, syphilis, young-onset prostatitis, and higher ejaculation frequency were each associated with each other.

7 904 Sutcliffe et al. Sexually transmitted infections In age- and multivariable-adjusted analyses, a history of gonorrhea was positively associated with each prevalent and incident definition of LUTS (table 2). The odd ratios for prevalent severe irritative and obstructive LUTS were similar in magnitude to each other and to that of prevalent LUTS (data not shown). Further adjustment for early life factors, including ejaculation frequency at ages years, alcohol consumption at ages years, cigarette smoking before age 30 years, body mass index at age 21 years, strenuous physical activity in high school and college, and histories of syphilis, young-onset prostatitis, or other prostatitis, did not appreciably alter the results (data not shown). Restriction to men who completed at least three follow-up questionnaires (for more complete case ascertainment) or who did not report other prostatitis or diabetes mellitus type 2 (conditions with potentially overlapping symptoms with BPH) over the course of follow-up did not influence the results (data not shown). The age-adjusted odds ratio of total prevalent LUTS comparing a history of syphilis with none was 1.45 (95 percent confidence interval (CI): 0.77, 2.73), which was not explained by a concurrent history of gonorrhea. The lifetime reported prevalence of syphilis was too low to perform multivariable adjustment or stratified analyses. The observed association between a history of gonorrhea and total prevalent LUTS was nonsignificantly stronger among men who did not use aspirin at ages years (multivariable-adjusted odds ratio (OR) ¼ 2.15, 95 percent CI: 1.56, 2.95) than among aspirin users (OR ¼ 1.39, 95 percent CI: 1.01, 1.91; p interaction ¼ 0.09). Age as a surrogate measure for opportunity to have acquired gonorrhea in the pre-antibiotic era did not modify the association between a history of gonorrhea and total prevalent LUTS (p interaction ¼ 0.96). Prostatitis A history of young-onset prostatitis was statistically significantly associated with each prevalent definition of LUTS and nonsignificantly associated with incident highmoderate to severe LUTS in age- and multivariable-adjusted analyses (table 2). Odds ratios for prevalent severe obstructive LUTS were slightly greater in magnitude than those for prevalent severe irritative LUTS (data not shown). Further adjustment for early life factors did not appreciably alter the results, nor did restriction to participants who completed at least three follow-up questionnaires or who did not report diabetes mellitus type 2 over the course of follow-up (data not shown). Among men who did not use aspirin at ages years, the association between a history of young-onset prostatitis and total prevalent LUTS (multivariable-adjusted OR ¼ 1.23, 95 percent CI: 0.82, 1.85) was nonsignificantly weaker than among aspirin users (OR ¼ 1.90, 95 percent CI: 1.37, 2.62; p interaction ¼ 0.08). Age did not modify the association between a history of young-onset prostatitis and total prevalent LUTS (p interaction ¼ 0.70). Frequency of ejaculation With the exception of a slightly lower occurrence of LUTS among men who reported 0 3 ejaculations/month (for total prevalent LUTS: multivariable-adjusted OR ¼ 0.84, 95 percent CI: 0.68, 1.05), no association was observed between ejaculation frequency at ages years and LUTS in age- or multivariable-adjusted analyses (data not shown). Further adjustment for early life factors did not alter the results, nor did restriction to men who completed at least three follow-up questionnaires or who did not report other prostatitis or diabetes mellitus type 2 during follow-up (data not shown). Overall, similar results were observed for ejaculation frequency at ages years as for ejaculation frequency at ages years (data not shown). A statistically significant trend of increasing odds of each prevalent case definition was observed for men with lower frequencies of ejaculation in the year prior to completion of the 1992 questionnaire (data not shown). However, no trend was present for incident case definitions. DISCUSSION In this large, well-characterized cohort of older male health professionals, we observed statistically significant positive associations for a history of gonorrhea and young-onset prostatitis with several different definitions of LUTS. These associations persisted after adjustment for early life sexual and behavioral factors, as well as other previously identified risk factors for BPH in this cohort. No association was observed between early life ejaculation frequency and LUTS, with the exception of a slight, nonsignificant inverse association between very low frequencies of ejaculation at ages years and LUTS, possibly because of lower androgenicity or confounding by unmeasured lifestyle factors associated with a lower risk of LUTS. A statistically significant inverse association was observed between lower frequencies of ejaculation in the year prior to completion of the 1992 questionnaire and each prevalent, but not incident, definition of LUTS, suggesting that prevalent results were due to symptom-related reductions in men s frequencies of ejaculation. This potential for reverse causation precluded an investigation of lifetime ejaculation frequency and LUTS. Our finding of a positive association between a history of gonorrhea and BPH is consistent with findings from four small previous case-control studies (14 16, 18) but differs from those of two additional, small case-control studies, one of which observed no association between a self-reported history of sexually transmitted infections and BPH (13), while the other observed a nonstatistically significant inverse association between a history of medical record-confirmed gonorrhea and BPH (17). To our knowledge, no studies have investigated a history of young-onset prostatitis as it relates to BPH. Findings from previous studies of early sexual activity and BPH are inconsistent and difficult to compare with our null study findings, because of the varied measures of sexual activity used in previous studies (13 16). One possible mechanism by which gonorrhea and other urogenital infections (e.g., E. coli infection) may contribute

8 Early Life Factors and Benign Prostatic Hyperplasia 905 to the development of BPH is by infecting and inducing a chronic inflammatory immune response within the prostate. Although acute prostatitis was a common sequela of sexually transmitted infections (particularly gonorrhea) in the pre-antibiotic era (4), it is now relatively rare because of treatment at an earlier stage of infection before progression to the prostate. However, asymptomatic infection and consequent inflammation of the prostate likely still occur in an unknown proportion of men. Infection-mediated intraprostatic inflammation may potentially contribute to the development of BPH nodules by the secretion of growth factors in excess of that required to replace cells damaged by infection or inflammation (11, 22). This hypothesis is supported by the frequent observation of chronic inflammatory infiltrates in and around BPH nodules (6 9), the findings of greater prostate volume (6) and epithelial cell proliferation (23) in BPH tissue sections with chronic inflammation than in those without inflammation, the demonstration of platelet-derived growth factor receptors on cells derived from BPH tissue similar to those observed in other hyperplastic diseases (22), and the induction of hyperplastic lesions similar to BPH nodules in the rat prostate model after administration of immunostimulatory compounds (10, 11). All of the exposures and outcomes considered in this study were assessed by self-report. Although, relative to other sexually transmitted infections, gonorrhea and syphilis lend themselves well to assessment by self-report, because of their frequent symptomatic presentation, laboratory confirmation, and traditional recognition as sexually transmitted infections, they may have been underreported because of a general reluctance to report stigmatizing diseases. Underreporting is unlikely, however, to have been differential by LUTS status. Young-onset prostatitis is less likely to have been misreported than a history of sexually transmitted infections, because it is not a stigmatized disease. Ejaculation frequency may potentially have been under- or overreported, depending on participants interpretation of ejaculation. However, this concern is unlikely to have caused profound nondifferential misclassification, because a previous study conducted in this cohort observed a significant inverse association between higher ejaculation frequency and prostate cancer (24). Despite self-reported assessment, the outcome of LUTS is likely to have been highly sensitive, because it was assessed in multiple ways and at four separate points in time. Possible inclusion of other medical conditions with urinary symptoms (e.g., prostatitis, diabetes mellitus type 2, and urethral stricture, a previously common sequela of gonorrhea (4)) is unlikely to have altered study inferences, because the observed associations persisted for all case definitions, including more specific definitions, such as surgery for an enlarged prostate, which requires a physician s diagnosis of BPH, and severe LUTS, and because restriction to men without a history of other prostatitis or diabetes mellitus type 2 did not appreciably alter the results. Associations between sexually transmitted infections and later diseases are difficult to investigate because of the potential for confounding by early life biologic (e.g., sexual drive) and lifestyle (e.g., alcohol consumption) factors strongly associated with acquisition of sexually transmitted infections. We adjusted for several of these factors, none of which altered the association between gonorrhea and LUTS. A further difficulty in investigating associations between sexually transmitted infections and later diseases is the potential for confounding by other sexually transmitted infections. Participants were not asked to report their histories of any additional sexually transmitted infections beyond gonorrhea and syphilis in the Health Professionals Follow-up Study cohort. Therefore, it is possible that the observed association between a history of gonorrhea and LUTS may be the result of confounding by a history of other sexually transmitted infections. Alternatively, this association may not be specific to any one particular sexually transmitted infection or to sexually transmitted infections in general, but it may reflect a causal relation between infection of the prostate and BPH. These explanations are also relevant for a history of syphilis, because syphilis rarely infects the prostate and is, thus, unlikely to be causally associated with LUTS. Our findings suggest that genitourinary infections acquired during youth and early adulthood may contribute to later development of clinical BPH. Further studies are needed to confirm these findings in other population settings and to investigate the specific pathogens and biologic mechanisms involved. ACKNOWLEDGMENTS Siobhan Sutcliffe was supported by a Doctoral Research Award from the Canadian Prostate Cancer Research Initiative/Canadian Institutes of Health Research. This work was supported by research grants CA55075, HL35464, and DK45779 (Harvard) and by grant P50CA58236 (Hopkins) from the National Institutes of Health. The authors thank Jill Arnold, Elizabeth Frost-Hawes, Mira Kaufman, Laura Sampson, Alvin Wing, and Mildred Wolff for their continued help in conducting the Health Professionals Follow-up Study. They are grateful to Drs. Stephen R. Cole, Charlotte A. Gaydos, and Jonathan M. Zenilman for discussion related to this manuscript. Conflict of interest: none declared. REFERENCES 1. Garraway WM, Collins GN, Lee RJ. High prevalence of benign prostatic hypertrophy in the community. Lancet 1991; 338: Berry SJ, Coffey DS, Walsh PC, et al. The development of human benign prostatic hyperplasia with age. J Urol 1984; 132: Lepor H. Prostatic diseases. Philadelphia, PA: WB Saunders Company, Pelouze PS. Gonorrhea in the male and female: a book for practitioners. Philadelphia, PA: WB Saunders Company, Moore RA. Benign hypertrophy of the prostate: a morphological study. J Urol 1943;50: Di Silverio F, Gentile V, De Matteis A, et al. Distribution of inflammation, pre-malignant lesions, incidental carcinoma

9 906 Sutcliffe et al. in histologically confirmed benign prostatic hyperplasia: a retrospective analysis. Eur Urol 2003;43: Nickel JC, Downey J, Young I, et al. Asymptomatic inflammation and/or infection in benign prostatic hyperplasia. BJU Int 1999;84: Kohnen PW, Drach GW. Patterns of inflammation in prostatic hyperplasia: a histologic and bacteriologic study. J Urol 1979; 121: Anjum I, Ahmed M, Azzopardi A, et al. Prostatic infarction/ infection in acute urinary retention secondary to benign prostatic hyperplasia. J Urol 1998;160: Kessler OJ, Keisari Y, Servadio C, et al. Role of chronic inflammation in the promotion of prostatic hyperplasia in rats. J Urol 1998;159: Robinette CL. Sex-hormone-induced inflammation and fibromuscular proliferation in the rat lateral prostate. Prostate 1988; 12: Isaacs JT. Prostatic structure and function in relation to the etiology of prostatic cancer. Prostate 1983;4: Steele R, Lees RE, Kraus AS, et al. Sexual factors in the epidemiology of cancer of the prostate. J Chronic Dis 1971; 24: Araki H, Watanabe H, Mishina T, et al. High-risk group for benign prostatic hypertrophy. Prostate 1983;4: Oishi K, Okada K, Yoshida O, et al. A case-control study of prostatic cancer in Kyoto, Japan: sexual risk factors. Prostate 1990;17: Signorello LB, Tzonou A, Lagiou P, et al. The epidemiology of benign prostatic hyperplasia: a study in Greece. BJU Int 1999;84: Lees RE, Steele R, Wardle D. Arsenic, syphilis, and cancer of the prostate. J Epidemiol Community Health 1985;39: Joseph MA, Harlow SD, Wei JT, et al. Risk factors for lower urinary tract symptoms in a population-based sample of African-American men. Am J Epidemiol 2003;157: Collins MM, Meigs JB, Barry MJ, et al. Prevalence and correlates of prostatitis in the Health Professionals Follow-up Study cohort. J Urol 2002;167: Giovannucci E, Rimm EB, Chute CG, et al. Obesity and benign prostatic hyperplasia. Am J Epidemiol 1994;140: Barry MJ, Fowler FJJ, O Leary MP, et al. The American Urological Association symptom index for benign prostatic hyperplasia. J Urol 1992;148: Gleason PE, Jones JA, Regan JS, et al. Platelet derived growth factor (PDGF), androgens and inflammation: possible etiologic factors in the development of prostatic hyperplasia. J Urol 1993;149: Feneley MR, Span PN, Schalken JA, et al. A prospective randomized trial evaluating tissue effects of finasteride therapy in benign prostatic hyperplasia. Prostate Cancer Prostatic Dis 1999;2: Leitzmann MF, Platz EA, Stampfer MJ, et al. Ejaculation frequency and subsequent risk of prostate cancer. JAMA 2004;291: