ANDROLOGY
ISSN: 2047-2919
REVIEW ARTICLE
Correspondence: Zhigang Zhao, Department of Urology & Andrology, Minimally Invasive Surgery Center, Guangdong Provincial Key Laboratory of Urology, The First Affiliated Hospital of Guangzhou Medical University, No. 1-3, Kangda Road, Guangzhou, Guangdong Province 510230, China. E-mail: [email protected]
Vasectomy and risk of prostate cancer: a systematic review and meta-analysis of cohort studies L. H. Liu, R. Kang, J. He, S. K. Zhao, F. T. Li, S. P. Wan and Z. G. Zhao
Keywords: meta-analysis, prostate cancer, vasectomy
Department of Urology & Andrology, Minimally Invasive Surgery Center, Guangdong Key Laboratory of Urology, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China
Received: 15-Jan-2015 Revised: 13-Mar-2015 Accepted: 27-Mar-2015 doi: 10.1111/andr.12040
SUMMARY The results of published literature focusing on the association between vasectomy and the incidence of prostate cancer are often inconsistent. We conducted a meta-analysis to provide a quantitative assessment of the association between vasectomy and the risk of prostate cancer. We identified all cohort studies by searching the PubMed, Embase, and Cochrane Library before August 2014. The quality of the studies was evaluated using the Newcastle-Ottawa Scale checklist. Summary effect estimates with 95% confidence intervals (CI) were derived using a fixed or random effects model, depending on the heterogeneity of the included studies. Nine cohort studies that spanned across two continents involving 1 127 096 participants (ages 20–75) with 7539 cases of prostate cancer cases were included in the meta-analysis. The overall combined relative risks for men with the reference group were 1.08 (95% CI: 0.87–1.34) in a random effects, however, the association was not statistically significant (p = 0.48). Estimates of total effects were generally consistent in the sensitivity and subgroup analyses. No evidence of publication bias was observed. This meta-analysis indicated that vasectomy may not contribute to the risk of prostate cancer. The conclusion might have a far-reaching significance for the public health, especially in countries with high prevalence rates of vasectomy.
INTRODUCTION Prostate cancer is the second most frequently diagnosed cancer in men and the fifth most common cancer reported in the world (Ferlay et al., 2010). Nearly three-quarters of the prostate cancer occur in developed countries (Ferlay et al., 2010). Its incidence has increased worldwide in recent decades, mostly due to aging of the population and the wide use of serum PSA for prostate cancer screening. The etiology of prostate cancer is not wellknown. Epidemiological evidence suggests that biological, environmental, and social risk factors all play a role in the initiation and the progression of prostate cancer (Haas & Sakr, 1997; Miyahira et al., 2014). Vasectomy is the most frequently used method for male contraception in the United States with a prevalence rate of 15% (Schwingl & Guess, 2000; Eisenberg et al., 2009). The estimated prevalence rates of vasectomy ranged from 7 to 8% in the married couples in China and India (Hsing et al., 1994; Tripathy et al., 1994). In the late 1980s, Honda et al. (1988) first reported a positive association between vasectomy and the risk of © 2015 American Society of Andrology and European Academy of Andrology
developing prostate cancer. Since then, many new epidemiological studies had examined the relationship between vasectomy and the risk of developing prostate cancer. The results had been inconclusive or contradictory (Sidney et al., 1991; Nienhuis et al., 1992; Giovannucci et al., 1993; Moller et al., 1994a,b; Hiatt et al., 1994; Moller et al., 1994a,b; Lynge 2002; Goldacre et al., 2005; Rohrmann et al., 2005; Romero et al., 2012). The question about the long-term safety of vasectomy attracted widespread attention in recent years. A 24-year follow-up study of vasectomy found that this procedure was associated with a modest increase in the incidence of aggressive prostate cancer (Siddiqui et al., 2014). But there were still methodologic weaknesses and limitations in this updated cohort study (Sokal et al., 2015). Elucidation of the association between vasectomy and prostate cancer will have important clinical and public health implications. To our knowledge, there had been several reviews that were directly focused on the possible association between vasectomy and the risk of prostate cancer (DerSimonian et al., 1993; Andrology, 2015, 3, 643–649
643
ANDROLOGY
L. H. Liu et al.
Bernal-Delgado et al., 1998; Schwingl & Guess, 2000; Dennis et al., 2002). The results of these studies were inconsistent and inconclusive. Whether vasectomy is an independent risk factor associated with an increased incidence of prostate cancer remains unsettled. Recently there were additional publications addressing this issue. Therefore, we decided to conduct a broader meta-analysis to include studies up to 2014 to better evaluate the cause–effect and the spatial relationship between vasectomy and prostate cancer.
MATERIAL AND METHODS Data search and analysis strategy We searched the electronic databases of PubMed, Embase and Cochrane Library for relevant articles. We enlisted the help of an experienced librarian to identify all the appropriate literature using MESH heading, keywords and text words. The following search terms were used: prostate cancer, prostate neoplasm, prostate tumor, prostate adenocarcinoma, prostate carcinoma, vasectomy, deferentectomy, vasoligation, vasoligature, cohort studies, and prostate cancer risk factors. This search was repeated until no additional articles were found prior to August 2014. We also personally reviewed the reference lists for relevant publications to search for additional literature. The language was limited to English. Study selection We performed an initial screening based on the titles and abstracts. The second screening was a full-text review. Studies were included if they met the following criteria: (i) the study design was a cohort study; (ii) contained vasectomy and prostate cancer measures; (iii) studies that reported relative risk (RR), odds ratio (OR), or hazard ratio (HR) relating pre-existing vasectomy to prostate cancer outcome, and the corresponding 95% confidence interval (CI) (or sufficient data to calculate them); (iv) it was at the least age-adjusted or age-matched; (v) The subjects had a follow-up for ≥1 year. Reviews, meeting abstract, editorials, case reports and commentaries were excluded from our analysis. When two or more studies had overlapping study samples, only the most complete and the more recent publication was included in the metaanalysis. Data extraction Two investigators independently extracted the data from all the eligible studies. Any discrepancies were resolved by discussion or by a third reviewer for adjudication. Data extraction was performed using a standardized data collection form. The following information was sought in each of the studies: the first author’s last name, year of publication, data source, study period, characteristics for study population, subjects’ age at baseline, cases of prostate cancer, the total number of participants, follow-up time, effect estimates with 95% CIs for the association of vasectomy and prostate cancer, and statistical adjustments for confounding factors. Quality assessment Because our data analyses relied on the published results, the methodological quality was a vitally important factor of the included studies. The methodological quality of non-randomized 644
Andrology, 2015, 3, 643–649
studies was assessed by the Newcastle-Ottawa Scale (NOS). NOS consists of three broad perspectives: the selection of the studied cases, comparability of the studied groups, and the ascertainment of either the exposure or the outcome of participants (Stang, 2010). A score of 0–9 was allocated to each study. Studies scored greater than or equal to 7 were considered to be of high-quality information. Statistical analysis Major outcome variables were used for the quantitative measures of the association between vasectomy and prostate cancer. Adjusted RRs were used for the calculation of common effect measure. Both RRs adjusted only for the age variable and those that used multivariate analysis had been included. If multivariate RRs were provided, they were used for statistical analysis in preference to the univariate RRs. Unadjusted RRs were excluded from quantitative synthesis to minimize confounding. Before pooling the data, adjusted RRs from the included studies were transformed to their log (RRs) to stabilize the variance and normalize the distribution (Walter & Cook, 1991). The summary RRs and 95% CIs were used to assess the strength of association between vasectomy and prostate cancer. Because of the relatively low number of prostate cancer in these studies, HR and OR were regarded as RR. Statistical heterogeneity for studies reporting the same effect measures was evaluated with Chisquared test of the Cochrane Q statistics (p < 0.10 is considered statistically significant) and I2 statistics. The I2 statistics described the percentage of total variation across studies that was caused by heterogeneity and thus, was a quantitative measure of inconsistency across studies (Higgins et al., 2003). I2 ≥50% ratio indicated significant statistical heterogeneity. When there was no heterogeneity among studies (p > 0.10 or p < 0.10 but I2 0.05 indicated no publication bias (Begg & Mazumdar, 1994; Egger et al., 1997). All analyses were performed using STATA version 12.0 (Stata Corporation, College Station, TX, USA).
RESULTS Literature search The literature search was updated to August, 2014 in PubMed, Embase and Cochrane Library. It yielded a total of 440 articles. Of these, the majority were excluded after the first screening based on the titles and abstracts. They were either ineligible studies or irrelevant to our analysis. After full-text review of 46 potentially relevant studies, three articles (Moller et al., 1994a,b; Goldacre et al., 2005) were found to have used the same study population. Among them, only the most recently published study with the largest number of subjects was used (Goldacre et al., 2005). At the end of the search, we identified total nine © 2015 American Society of Andrology and European Academy of Andrology
ANDROLOGY
VASECTOMY AND PROSTATE CANCER
studies (Sidney et al., 1991; Nienhuis et al., 1992; Giovannucci et al., 1993; Hiatt et al., 1994; Lynge et al., 2002; Goldacre et al., 2005; Rohrmann et al., 2005; Romero et al., 2012; Siddiqui et al., 2014) that met the inclusion criteria. A flowchart showing the study selection was presented in Fig. 1.
Figure 1 Flowchart of study selection. 440 articles identified from PubMed, Embase, the Cochrane Library after initial search
394 trials excluded for: 34 duplicate documents
Study characteristics and quality assessment The characteristics of the included studies were presented in Table 1. The nine selected studies contained a total 1 127 096 participants (ranging from 2175 to 733 249) and 7539 cases of prostate cancer from different geographic regions (five studies in America, three studies in Europe, and one study in Brazil) with varying length of follow-up period (ranging from 1.8 to 24 years). Outcome assessments were from a variety of sources, including a follow-up study of health professionals, medical records and hospital database. Two studies only adjusted for age. The other studies included a group of conventional risks for prostate cancer, such as age, race, region, family history of prostate cancer, cigarette smoking history and so on. In this total nine points evaluation system, six studies were identified as relatively high
17 review and comment 343 screening of titles or abstracts 46 trials identified 37 trials excluded for: 34 trials do not meet inclusion criteria 3 from the same study population
9 trials included in meta-analysis
Table 1 Characteristics of nine cohort studies included in the meta-analysis References
Data source
Location/time period
Sidney et al. (1991)
the Northern California Kaiser Permanente Medical Care Program Six health districts in Oxford region the Health Professionals Follow-up Study the Kaiser Permanente Medical Care Program (KPMCP) in northern California (United States) Denmark
USA 1977–1987
Nienhuis et al. (1992) Giovannucci et al. (1993) Hiatt et al. (1994)
Lynge (2002)
Age (years) mean SD (range)
Diseased/ reference
Vasectomy RR
95% CI
Variable adjustment
6.8
45.8
135/20476
1.0
0.7–1.6
UK 1970–1986 USA 1986–1990 USA 1978–1985
6.6
25–49
6/35442
0.44
0.1–4.0
Age, race, marital status, and date and location of the examination Age
2
40–75
300/47855
1.66
1.25–2.21
4.6
47.4 13
238/43432
0.8
0.5–1.3
Denmark 1977–1995 UK 1963–1999 USA 1989–2004
12.7
25–49
46/733203
0.98
0.7–1.3
Age
12.7
20–59
656/184253
0.74
0.45–1.14
Age, place of residence
8
35.2 6.9
78/3373
2.03
1.24–3.32
1.8
>40
57/2118
0.26
0.06–1.04
1.10
1.04–1.17
Age, body mass index, cigarette smoking history, family history of prostate cancer, alcohol consumption, intake of food Age, race, ethnicity, family history, school level, increased blood pressure, diabetes mellitus, and urethritis Age, race, height, current body mass index, vigorous physical activity, smoking, diabetes, family history of prostate cancer, multivitamin use, intake of supplemental vitamin E and alcohol, and history of PSA testing
Goldacre et al. (2005) Rohrmann et al. (2005)
National Health Service (NHS) CLUE II cohort
Romero et al. (2012)
Brazil Health Care System
Brazil 2006–2011
Siddiqui et al (2014)
Health Professionals Follow-up Study
USA 1986–2010
Follow-up time (years)
24
© 2015 American Society of Andrology and European Academy of Andrology
40–75
6023/49405
Age, region, race, marital status Age, race, education level
Andrology, 2015, 3, 643–649
645
ANDROLOGY
L. H. Liu et al. Table 2 Newcastle-Ottawa Scale (NOS) assessment of the quality of the studies Study
Comparabilitya
Selection
Sidney (1991) Nienhuis (1992) Giovannucci (1993) Hiatt (1994) Lynge (2002) Goldacre (2005) Rohrmann (2005) Romero (2012) Siddiqui (2014)
Outcome
Total scores b
c
1
2
3
4
5
6
7
8
9
No No Yes No No Yes No No Yes
No Yes No Yes Yes No Yes Yes Yes
Yes No No No Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes No Yes Yes No Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes Yes No No Yes Yes Yes No Yes
Yes No Yes Yes Yes No Yes Yes No
7 5 6 6 7 7 8 7 8
1, Representativeness of the exposed cohort; 2, selection of the non-exposed cohort; 3, ascertainment of exposure; 4, demonstration that outcome of interest was not present at start of study; 5 and 6, comparability of cohorts on the basis of the design or analysis; 7, assessment of outcome; 8, was follow-up long enough for outcomes to occur; 9, adequacy of follow-up of cohorts. aStudies that controlled for age received one score, whereas studies that controlled for other important confounders received an additional score. bStudy with follow-up time >5 years was assigned one score. cStudy with follow-up rate >70% was assigned one score.
Figure 2 Forest plot of cohort studies examining the association between vasectomy and prostate cancer. Study
%
ID
RR (95% CI)
Weight
Sidney et al.
1.00 (0.70, 1.60)
12.13
Nienhuis et al.
0.44 (0.10, 4.00)
1.26
Giovannucci et al.
1.66 (1.25, 2.21)
15.95
Hiatt et al.
0.80 (0.50, 1.30)
10.52
Lynge et al.
0.98 (0.70, 1.30)
15.17
Goldacre et al.
0.74 (0.45, 1.14)
10.83
Rohrmann et al.
2.03 (1.24, 3.32)
10.19
Romero et al.
0.26 (0.06, 1.04)
2.03
Siddiqui et al.
1.10 (1.04, 1.17)
21.91
Overall (I2 = 66.7%, p = 0.002)
1.08 (0.87, 1.34)
100.00
NOTE: Weights are from random effects analysis 0.06
1
quality (≥7 points). The results of quality assessment according to NOS system were summarized in Table 2. Overall and sensitivity analyses The relationship between vasectomy and the risk of prostate cancer were explored in the nine cohort studies. Among them, three studies reported a positive association of vasectomy with prostate cancer (RR ranged from 1.10 to 2.03) and six studies reported negative association between vasectomy and the risk of prostate cancer (RR ranged from 0.26 to 0.98). Overall in random effects, the pooled RR of prostate cancer associated with vasectomy relative to that of no vasectomy was 1.08 (95% CI: 0.87– 1.34). This association was not statistically significant (p = 0.48). Substantial heterogeneity was observed (p = 0.002, I2 = 66.7%), Fig. 2. Our data thus supported the hypothesis that vasectomy was not associated with an increased incidence of prostate cancer. To further elicit the possible relationship between vasectomy and prostate cancer, subgroup analyses were performed. These 646
Andrology, 2015, 3, 643–649
16.7
analyses were done according to the stratifications of the time since vasectomy, publication year of the article, geographic location, and follow-up time. The RR of prostate cancer following vasectomy relative to those without vasectomy was 1.20 (95% CI: 0.83–1.74) after 10 years, 1.10 (95% CI: 0.83–1.46) after 10~19 years, and 1.25 (95% CI: 0.92–1.71) after 20 years. The CIs associated with these RRs were quite wide and included 1. The association between vasectomy and prostate cancer was more significant for studies conducted in America (RR = 1.18, 95% CI: 0.89–1.56) than in Europe (RR = 0.89, 95% CI: 0.69–1.14). The summary estimates were higher for those with follow-up ≤10 years than ones with follow-up for >10 years and for the studies published before 2002 than for the studies published in 2002 or later. Table 3 compares each of these stratifications. Sensitivity analysis was performed to assess the influence of individual studies on the overall risk of prostate cancer. After excluding any study that did not substantially influence the direction and magnitude of the cumulative estimates, we derived relatively stable results (Table 4). © 2015 American Society of Andrology and European Academy of Andrology
ANDROLOGY
VASECTOMY AND PROSTATE CANCER Table 3 Subgroup analysis of the association between vasectomy and prostate cancer Category of variables
Number of studies
The time since vasectomy ≤10 years 4 10–19 years 4 ≥20 years 6 Geographic location America 6 Europe 3 Follow-up time >10 years 3 ≤10 years 6 Year of publication
ISSN: 2047-2919
REVIEW ARTICLE
Correspondence: Zhigang Zhao, Department of Urology & Andrology, Minimally Invasive Surgery Center, Guangdong Provincial Key Laboratory of Urology, The First Affiliated Hospital of Guangzhou Medical University, No. 1-3, Kangda Road, Guangzhou, Guangdong Province 510230, China. E-mail: [email protected]
Vasectomy and risk of prostate cancer: a systematic review and meta-analysis of cohort studies L. H. Liu, R. Kang, J. He, S. K. Zhao, F. T. Li, S. P. Wan and Z. G. Zhao
Keywords: meta-analysis, prostate cancer, vasectomy
Department of Urology & Andrology, Minimally Invasive Surgery Center, Guangdong Key Laboratory of Urology, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China
Received: 15-Jan-2015 Revised: 13-Mar-2015 Accepted: 27-Mar-2015 doi: 10.1111/andr.12040
SUMMARY The results of published literature focusing on the association between vasectomy and the incidence of prostate cancer are often inconsistent. We conducted a meta-analysis to provide a quantitative assessment of the association between vasectomy and the risk of prostate cancer. We identified all cohort studies by searching the PubMed, Embase, and Cochrane Library before August 2014. The quality of the studies was evaluated using the Newcastle-Ottawa Scale checklist. Summary effect estimates with 95% confidence intervals (CI) were derived using a fixed or random effects model, depending on the heterogeneity of the included studies. Nine cohort studies that spanned across two continents involving 1 127 096 participants (ages 20–75) with 7539 cases of prostate cancer cases were included in the meta-analysis. The overall combined relative risks for men with the reference group were 1.08 (95% CI: 0.87–1.34) in a random effects, however, the association was not statistically significant (p = 0.48). Estimates of total effects were generally consistent in the sensitivity and subgroup analyses. No evidence of publication bias was observed. This meta-analysis indicated that vasectomy may not contribute to the risk of prostate cancer. The conclusion might have a far-reaching significance for the public health, especially in countries with high prevalence rates of vasectomy.
INTRODUCTION Prostate cancer is the second most frequently diagnosed cancer in men and the fifth most common cancer reported in the world (Ferlay et al., 2010). Nearly three-quarters of the prostate cancer occur in developed countries (Ferlay et al., 2010). Its incidence has increased worldwide in recent decades, mostly due to aging of the population and the wide use of serum PSA for prostate cancer screening. The etiology of prostate cancer is not wellknown. Epidemiological evidence suggests that biological, environmental, and social risk factors all play a role in the initiation and the progression of prostate cancer (Haas & Sakr, 1997; Miyahira et al., 2014). Vasectomy is the most frequently used method for male contraception in the United States with a prevalence rate of 15% (Schwingl & Guess, 2000; Eisenberg et al., 2009). The estimated prevalence rates of vasectomy ranged from 7 to 8% in the married couples in China and India (Hsing et al., 1994; Tripathy et al., 1994). In the late 1980s, Honda et al. (1988) first reported a positive association between vasectomy and the risk of © 2015 American Society of Andrology and European Academy of Andrology
developing prostate cancer. Since then, many new epidemiological studies had examined the relationship between vasectomy and the risk of developing prostate cancer. The results had been inconclusive or contradictory (Sidney et al., 1991; Nienhuis et al., 1992; Giovannucci et al., 1993; Moller et al., 1994a,b; Hiatt et al., 1994; Moller et al., 1994a,b; Lynge 2002; Goldacre et al., 2005; Rohrmann et al., 2005; Romero et al., 2012). The question about the long-term safety of vasectomy attracted widespread attention in recent years. A 24-year follow-up study of vasectomy found that this procedure was associated with a modest increase in the incidence of aggressive prostate cancer (Siddiqui et al., 2014). But there were still methodologic weaknesses and limitations in this updated cohort study (Sokal et al., 2015). Elucidation of the association between vasectomy and prostate cancer will have important clinical and public health implications. To our knowledge, there had been several reviews that were directly focused on the possible association between vasectomy and the risk of prostate cancer (DerSimonian et al., 1993; Andrology, 2015, 3, 643–649
643
ANDROLOGY
L. H. Liu et al.
Bernal-Delgado et al., 1998; Schwingl & Guess, 2000; Dennis et al., 2002). The results of these studies were inconsistent and inconclusive. Whether vasectomy is an independent risk factor associated with an increased incidence of prostate cancer remains unsettled. Recently there were additional publications addressing this issue. Therefore, we decided to conduct a broader meta-analysis to include studies up to 2014 to better evaluate the cause–effect and the spatial relationship between vasectomy and prostate cancer.
MATERIAL AND METHODS Data search and analysis strategy We searched the electronic databases of PubMed, Embase and Cochrane Library for relevant articles. We enlisted the help of an experienced librarian to identify all the appropriate literature using MESH heading, keywords and text words. The following search terms were used: prostate cancer, prostate neoplasm, prostate tumor, prostate adenocarcinoma, prostate carcinoma, vasectomy, deferentectomy, vasoligation, vasoligature, cohort studies, and prostate cancer risk factors. This search was repeated until no additional articles were found prior to August 2014. We also personally reviewed the reference lists for relevant publications to search for additional literature. The language was limited to English. Study selection We performed an initial screening based on the titles and abstracts. The second screening was a full-text review. Studies were included if they met the following criteria: (i) the study design was a cohort study; (ii) contained vasectomy and prostate cancer measures; (iii) studies that reported relative risk (RR), odds ratio (OR), or hazard ratio (HR) relating pre-existing vasectomy to prostate cancer outcome, and the corresponding 95% confidence interval (CI) (or sufficient data to calculate them); (iv) it was at the least age-adjusted or age-matched; (v) The subjects had a follow-up for ≥1 year. Reviews, meeting abstract, editorials, case reports and commentaries were excluded from our analysis. When two or more studies had overlapping study samples, only the most complete and the more recent publication was included in the metaanalysis. Data extraction Two investigators independently extracted the data from all the eligible studies. Any discrepancies were resolved by discussion or by a third reviewer for adjudication. Data extraction was performed using a standardized data collection form. The following information was sought in each of the studies: the first author’s last name, year of publication, data source, study period, characteristics for study population, subjects’ age at baseline, cases of prostate cancer, the total number of participants, follow-up time, effect estimates with 95% CIs for the association of vasectomy and prostate cancer, and statistical adjustments for confounding factors. Quality assessment Because our data analyses relied on the published results, the methodological quality was a vitally important factor of the included studies. The methodological quality of non-randomized 644
Andrology, 2015, 3, 643–649
studies was assessed by the Newcastle-Ottawa Scale (NOS). NOS consists of three broad perspectives: the selection of the studied cases, comparability of the studied groups, and the ascertainment of either the exposure or the outcome of participants (Stang, 2010). A score of 0–9 was allocated to each study. Studies scored greater than or equal to 7 were considered to be of high-quality information. Statistical analysis Major outcome variables were used for the quantitative measures of the association between vasectomy and prostate cancer. Adjusted RRs were used for the calculation of common effect measure. Both RRs adjusted only for the age variable and those that used multivariate analysis had been included. If multivariate RRs were provided, they were used for statistical analysis in preference to the univariate RRs. Unadjusted RRs were excluded from quantitative synthesis to minimize confounding. Before pooling the data, adjusted RRs from the included studies were transformed to their log (RRs) to stabilize the variance and normalize the distribution (Walter & Cook, 1991). The summary RRs and 95% CIs were used to assess the strength of association between vasectomy and prostate cancer. Because of the relatively low number of prostate cancer in these studies, HR and OR were regarded as RR. Statistical heterogeneity for studies reporting the same effect measures was evaluated with Chisquared test of the Cochrane Q statistics (p < 0.10 is considered statistically significant) and I2 statistics. The I2 statistics described the percentage of total variation across studies that was caused by heterogeneity and thus, was a quantitative measure of inconsistency across studies (Higgins et al., 2003). I2 ≥50% ratio indicated significant statistical heterogeneity. When there was no heterogeneity among studies (p > 0.10 or p < 0.10 but I2 0.05 indicated no publication bias (Begg & Mazumdar, 1994; Egger et al., 1997). All analyses were performed using STATA version 12.0 (Stata Corporation, College Station, TX, USA).
RESULTS Literature search The literature search was updated to August, 2014 in PubMed, Embase and Cochrane Library. It yielded a total of 440 articles. Of these, the majority were excluded after the first screening based on the titles and abstracts. They were either ineligible studies or irrelevant to our analysis. After full-text review of 46 potentially relevant studies, three articles (Moller et al., 1994a,b; Goldacre et al., 2005) were found to have used the same study population. Among them, only the most recently published study with the largest number of subjects was used (Goldacre et al., 2005). At the end of the search, we identified total nine © 2015 American Society of Andrology and European Academy of Andrology
ANDROLOGY
VASECTOMY AND PROSTATE CANCER
studies (Sidney et al., 1991; Nienhuis et al., 1992; Giovannucci et al., 1993; Hiatt et al., 1994; Lynge et al., 2002; Goldacre et al., 2005; Rohrmann et al., 2005; Romero et al., 2012; Siddiqui et al., 2014) that met the inclusion criteria. A flowchart showing the study selection was presented in Fig. 1.
Figure 1 Flowchart of study selection. 440 articles identified from PubMed, Embase, the Cochrane Library after initial search
394 trials excluded for: 34 duplicate documents
Study characteristics and quality assessment The characteristics of the included studies were presented in Table 1. The nine selected studies contained a total 1 127 096 participants (ranging from 2175 to 733 249) and 7539 cases of prostate cancer from different geographic regions (five studies in America, three studies in Europe, and one study in Brazil) with varying length of follow-up period (ranging from 1.8 to 24 years). Outcome assessments were from a variety of sources, including a follow-up study of health professionals, medical records and hospital database. Two studies only adjusted for age. The other studies included a group of conventional risks for prostate cancer, such as age, race, region, family history of prostate cancer, cigarette smoking history and so on. In this total nine points evaluation system, six studies were identified as relatively high
17 review and comment 343 screening of titles or abstracts 46 trials identified 37 trials excluded for: 34 trials do not meet inclusion criteria 3 from the same study population
9 trials included in meta-analysis
Table 1 Characteristics of nine cohort studies included in the meta-analysis References
Data source
Location/time period
Sidney et al. (1991)
the Northern California Kaiser Permanente Medical Care Program Six health districts in Oxford region the Health Professionals Follow-up Study the Kaiser Permanente Medical Care Program (KPMCP) in northern California (United States) Denmark
USA 1977–1987
Nienhuis et al. (1992) Giovannucci et al. (1993) Hiatt et al. (1994)
Lynge (2002)
Age (years) mean SD (range)
Diseased/ reference
Vasectomy RR
95% CI
Variable adjustment
6.8
45.8
135/20476
1.0
0.7–1.6
UK 1970–1986 USA 1986–1990 USA 1978–1985
6.6
25–49
6/35442
0.44
0.1–4.0
Age, race, marital status, and date and location of the examination Age
2
40–75
300/47855
1.66
1.25–2.21
4.6
47.4 13
238/43432
0.8
0.5–1.3
Denmark 1977–1995 UK 1963–1999 USA 1989–2004
12.7
25–49
46/733203
0.98
0.7–1.3
Age
12.7
20–59
656/184253
0.74
0.45–1.14
Age, place of residence
8
35.2 6.9
78/3373
2.03
1.24–3.32
1.8
>40
57/2118
0.26
0.06–1.04
1.10
1.04–1.17
Age, body mass index, cigarette smoking history, family history of prostate cancer, alcohol consumption, intake of food Age, race, ethnicity, family history, school level, increased blood pressure, diabetes mellitus, and urethritis Age, race, height, current body mass index, vigorous physical activity, smoking, diabetes, family history of prostate cancer, multivitamin use, intake of supplemental vitamin E and alcohol, and history of PSA testing
Goldacre et al. (2005) Rohrmann et al. (2005)
National Health Service (NHS) CLUE II cohort
Romero et al. (2012)
Brazil Health Care System
Brazil 2006–2011
Siddiqui et al (2014)
Health Professionals Follow-up Study
USA 1986–2010
Follow-up time (years)
24
© 2015 American Society of Andrology and European Academy of Andrology
40–75
6023/49405
Age, region, race, marital status Age, race, education level
Andrology, 2015, 3, 643–649
645
ANDROLOGY
L. H. Liu et al. Table 2 Newcastle-Ottawa Scale (NOS) assessment of the quality of the studies Study
Comparabilitya
Selection
Sidney (1991) Nienhuis (1992) Giovannucci (1993) Hiatt (1994) Lynge (2002) Goldacre (2005) Rohrmann (2005) Romero (2012) Siddiqui (2014)
Outcome
Total scores b
c
1
2
3
4
5
6
7
8
9
No No Yes No No Yes No No Yes
No Yes No Yes Yes No Yes Yes Yes
Yes No No No Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes No Yes Yes No Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes Yes No No Yes Yes Yes No Yes
Yes No Yes Yes Yes No Yes Yes No
7 5 6 6 7 7 8 7 8
1, Representativeness of the exposed cohort; 2, selection of the non-exposed cohort; 3, ascertainment of exposure; 4, demonstration that outcome of interest was not present at start of study; 5 and 6, comparability of cohorts on the basis of the design or analysis; 7, assessment of outcome; 8, was follow-up long enough for outcomes to occur; 9, adequacy of follow-up of cohorts. aStudies that controlled for age received one score, whereas studies that controlled for other important confounders received an additional score. bStudy with follow-up time >5 years was assigned one score. cStudy with follow-up rate >70% was assigned one score.
Figure 2 Forest plot of cohort studies examining the association between vasectomy and prostate cancer. Study
%
ID
RR (95% CI)
Weight
Sidney et al.
1.00 (0.70, 1.60)
12.13
Nienhuis et al.
0.44 (0.10, 4.00)
1.26
Giovannucci et al.
1.66 (1.25, 2.21)
15.95
Hiatt et al.
0.80 (0.50, 1.30)
10.52
Lynge et al.
0.98 (0.70, 1.30)
15.17
Goldacre et al.
0.74 (0.45, 1.14)
10.83
Rohrmann et al.
2.03 (1.24, 3.32)
10.19
Romero et al.
0.26 (0.06, 1.04)
2.03
Siddiqui et al.
1.10 (1.04, 1.17)
21.91
Overall (I2 = 66.7%, p = 0.002)
1.08 (0.87, 1.34)
100.00
NOTE: Weights are from random effects analysis 0.06
1
quality (≥7 points). The results of quality assessment according to NOS system were summarized in Table 2. Overall and sensitivity analyses The relationship between vasectomy and the risk of prostate cancer were explored in the nine cohort studies. Among them, three studies reported a positive association of vasectomy with prostate cancer (RR ranged from 1.10 to 2.03) and six studies reported negative association between vasectomy and the risk of prostate cancer (RR ranged from 0.26 to 0.98). Overall in random effects, the pooled RR of prostate cancer associated with vasectomy relative to that of no vasectomy was 1.08 (95% CI: 0.87– 1.34). This association was not statistically significant (p = 0.48). Substantial heterogeneity was observed (p = 0.002, I2 = 66.7%), Fig. 2. Our data thus supported the hypothesis that vasectomy was not associated with an increased incidence of prostate cancer. To further elicit the possible relationship between vasectomy and prostate cancer, subgroup analyses were performed. These 646
Andrology, 2015, 3, 643–649
16.7
analyses were done according to the stratifications of the time since vasectomy, publication year of the article, geographic location, and follow-up time. The RR of prostate cancer following vasectomy relative to those without vasectomy was 1.20 (95% CI: 0.83–1.74) after 10 years, 1.10 (95% CI: 0.83–1.46) after 10~19 years, and 1.25 (95% CI: 0.92–1.71) after 20 years. The CIs associated with these RRs were quite wide and included 1. The association between vasectomy and prostate cancer was more significant for studies conducted in America (RR = 1.18, 95% CI: 0.89–1.56) than in Europe (RR = 0.89, 95% CI: 0.69–1.14). The summary estimates were higher for those with follow-up ≤10 years than ones with follow-up for >10 years and for the studies published before 2002 than for the studies published in 2002 or later. Table 3 compares each of these stratifications. Sensitivity analysis was performed to assess the influence of individual studies on the overall risk of prostate cancer. After excluding any study that did not substantially influence the direction and magnitude of the cumulative estimates, we derived relatively stable results (Table 4). © 2015 American Society of Andrology and European Academy of Andrology
ANDROLOGY
VASECTOMY AND PROSTATE CANCER Table 3 Subgroup analysis of the association between vasectomy and prostate cancer Category of variables
Number of studies
The time since vasectomy ≤10 years 4 10–19 years 4 ≥20 years 6 Geographic location America 6 Europe 3 Follow-up time >10 years 3 ≤10 years 6 Year of publication
