Introduction

The term asbestos comprises a group of natural minerals that form long, thin fibers when they crystallize [1]. Asbestos fibers tend to possess good strength properties (e.g. high tensile strength, wear and friction characteristics), flexibility (e.g. the ability to be woven), excellent thermal properties (e.g. heat stability and thermal, electrical and acoustic insulation), absorption capacity and resistance to chemical, thermal and biological degradation. Owing to its properties asbestos has been widely used worldwide and the story of this mineral was one of progressive commercial success until the mid-twentieth century [2]. The range of applications in which asbestos has been used includes roofing, thermal and electrical insulation, concrete pipes and sheets, flooring, gaskets, friction materials, coating and compounds, plastics, textiles, paper, mastics, thread, fiber jointing, and millboard. As the health risks associated with asbestos became increasingly recognized, its use began to decline. Despite widespread knowledge of the hazards of asbestos and bans on any use of asbestos in many countries, an estimated 1 million tons of this mineral was used around the world in 2020 [3].

In addition to the well-known association of asbestos with mesothelioma [4] and lung cancer [5], asbestos minerals have also been associated with ovarian [6], laryngeal [7] and gastrointestinal tract cancers [8]. Regarding other cancer sites, some epidemiological studies have reported an association between occupational exposure to asbestos and increased incidence of and mortality from bladder cancer [9,10,11]. Furthermore, asbestos fibers have been detected in tissue samples of bladder cancer patients affected by pulmonary asbestosis [12].

To our knowledge, no previous systematic reviews have been conducted on occupational asbestos exposure and the risk of mortality and incidence of bladder cancer. Thus, we aimed to perform a systematic review and meta-analysis to investigate this association.

Materials and methods

This systematic review and meta-analysis was performed and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [13].

Search strategy

Three different databases were searched: Medline (PubMed), Scopus and Ovid (Embase). Firstly, a comprehensive search strategy was created using the following terms: "Neoplasms”, “Carcinoma”, “Asbestos”, “Amosite", “Crocidolite”, “Amphibole”, “Serpentine”, “Asbestosis". Next, a specific search strategy was performed adding the term “Bladder”. Results were restricted to studies conducted on humans, while no limits were applied for language. The databases were searched from inception to October 20, 2020. The resulting papers were hand screened and relevant references were evaluated to find any further relevant papers. Database searches were conducted with the aid of an expert librarian to ensure completeness and rigor. The search strategies for each database are given in Online Information 1.

Inclusion and exclusion criteria

Only full articles published in peer-reviewed journals were considered. Cohort-studies and nested case-control studies of workers with a respiratory occupational exposure to asbestos in all industries and occupations were included. Cohorts of workers who were exposed to asbestos through ingestion only and had doubtful exposure to asbestos were excluded. We included all studies conducted on workers employed in industries or occupations in which asbestos exposure was considered substantial, such as asbestos workers, miners and millers, and shipyards. For studies of workers for whom occupational exposure to asbestos was possible (e.g. electricity workers or chimney sweeps [10, 14]), we used the criterion of a standardized mortality ratio (SMR) or standardized incidence ratio (SIR) for mesothelioma > 2 as marker of significant exposure, leading to the retention of the study for the meta-analysis. Descriptive studies, other systematic reviews or meta-analyses, community-based studies (either of cohort or case-control design), as well as conference proceedings, theses, and letters to the editor were excluded. Articles for which the full text was not available either online or by direct request to the journal in which they were published were also excluded. Two reviewers independently assessed the papers against the inclusion and exclusion criteria. Disagreement was solved by discussion.

Data extraction

For each mortality study we extracted the SMR for bladder cancer and its 95% confidence interval (CI); when these measures were not directly available from publications, but raw data were reported, we calculated them. Similarly, we abstracted, or calculated if not specified in the text, the SIR and its corresponding CI for cancer incidence studies. If 90% CI was reported, it was converted to 95% CI. Results of internal analysis (e.g., based on hazard ratios) were used for dose-response. In cases of multiple reports from the same cohort, the most comprehensive results (i.e. those based on the largest number of cases) were used.

The following study characteristics were also extracted where available: publication year, study design, country, cohort size (or number of cases and controls), number of person-years, duration of employment, duration of follow-up, minimum time of exposure, duration of exposure, type of outcome (incidence or mortality). Data were extracted independently by two reviewers and any disagreement was solved by a third reviewer.

Quality assessment

The methodological quality of the included studies was assessed through the National Institute of Health quality assessment tool for each study design [15]. The tools evaluate the presence of potential sources of bias, confounding factors, study power, and the strength of the association between the exposure and the outcome. The quality of the articles was rated as poor (score <9), fair (score = 9) and good (score >9). Quality assessment was performed by two independent reviewers, and results were discussed with the other authors until reaching consensus.

Statistical analysis

The main analysis included results for ever vs. never asbestos exposure. Results across studies were combined, separately for studies of bladder cancer incidence and mortality, using random-effect models meta-analyses [16] based on the log-transformation of the SMR/SIR and its standard error. Inter-study heterogeneity was evaluated with the I2 test [17]. Stratified meta-analyses were conducted to explore potential sources of heterogeneity, by sex (>90% males, >90% females and <90% for both), period of first employment of cohort members (1908–1940, 1941–1949 and 1950–1993), type of asbestos (amphiboles, chrysotile, mixed and unspecified type), geographic region (Europe, UK, North America, Australia and Asia), and quality score (poor, fair and good). Finally, we assessed the presence of publication bias by visual inspection of the funnel plot and applying the test proposed by Egger et al. [18]. A meta-regression was performed to assess the association between the duration of asbestos exposure and bladder cancer. For studies reporting multiple SMR or SIR for different durations of exposure, a single meta-regression was performed and the results were then meta-analyzed.

Results

A total of 13,267 articles were retrieved from Medline, Scopus, and Embase databases. After reviewing the titles, 2948 articles were considered potentially relevant, and after duplicates were removed, 2379 articles remained. Of these, 1639 articles were discarded following a review of the abstract. The full texts of the remaining 740 articles were examined in detail and assessed against the inclusion and exclusion criteria; 643 did not meet the inclusion criteria as described, and the full text was not available for a further 38. A manual search of the reference lists of included articles did not reveal any additional pertinent studies. Thus, 59 articles [9,10,11, 14, 19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73] met the inclusion criteria for the systematic review (Online Information 2).

The number of workers in each cohort ranged from 88 [65] to 160,640 [41], the number of incidences of bladder cancer per cohort varied from four [45] to 1257 [14], and that of bladder cancer deaths from zero [65] to 310 [9]. Bladder cancer incidence was evaluated in 26 cohorts, bladder cancer mortality was evaluated in 38, and both outcomes were reported in four cohorts [23, 24, 29, 63, 64]. The total number of bladder cancer cases and deaths across all articles was 3596 (out of 525,585 subjects) and 1169 (out of 385,552 subjects), with a follow-up duration ranging from five [59] to 88 years [10]. Selected characteristics of the cohorts are reported in Table 1.

Table 1 Characteristics of the included cohorts

Meta-analysis

The results of the random-effects meta-analysis showed that bladder cancer incidence and mortality were not associated with occupational asbestos exposure. Results are shown in Figure 1a and b respectively (pooled SIR: 1.04, 95% CI: 0.95–1.13, P=0.000; pooled SMR: 1.06, 95% CI: 0.96–1.17, P=0.031). Both the pooled SIR and the pooled SMR showed significant heterogeneity (P<=0.001; I2=72.9%; P=0.031; I2=32.6%, respectively). One of the two cohorts analyzed in Tomioka et al. 2011 [65] could not be included in the meta-analysis because it reported no bladder cancer deaths (SMR: 0.00, 95%CI: 0.00–15.50).

Fig. 1
figure 1

a Forest plot of the pooled standardized incidence ratio and 95% confidence intervals of urinary bladder cancer incidence associated with occupational asbestos exposure, using random-effect models. b Forest plot of the pooled standardized mortality ratio and 95% confidence intervals of urinary bladder cancer mortality associated with occupational asbestos exposure, using random-effect models

Subgroup analysis stratified by first year of employment, industry, asbestos type, and geographic region showed that bladder cancer incidence increased significantly among workers employed between 1908 and 1940 (SIR: 1.15, 95% CI: 1.01–1.313) and bladder cancer mortality increased among women (SMR: 1.83, 95% CI: 1.22–2.75) and asbestos workers (SMR: 1.12, 95% CI: 1.06–1.30) (Table 2)

Table 2 Pooled SMR and SIR subgroup meta-analysis

Meta-analysis of meta-regressions was possible for 5 studies. The RRs for each year of exposure were 0.97 (CI: 0.905–1.041) and 1.019 (CI: 0.997–1.041) for mortality and incidence respectively.

Funnel plots indicated no obvious outliers, and no evidence of publication bias was observed for either bladder cancer incidence or mortality. No small-study effect was found for mortality (p = 0.984) and incidence (p = 0.824) (Online Information 3).

Discussion

We investigated the relation between occupational asbestos exposure and risk of bladder cancer with a meta-analysis of the results obtained from a wide systematic review. Incidence of and mortality from bladder cancer were not significantly increased among asbestos-exposed workers (pooled SIR: 1.04, 95% CI: 0.95–1.13, P=0.000; pooled SMR: 1.06, 95% CI: 0.96–1.17, P=0.031). These results were in line with the largest studies done on asbestos workers [14, 20, 21, 29, 47, 64], and were confirmed in subgroup analyses stratified by type of asbestos fibers, country and quality assessment.

Also, from a biological point of view, there is no evidence of a reasonable mechanism to elucidate how respiratory asbestos fibers could reach the bladder. In the literature, there are no established physio-pathological pathways nor pathological evidence that could explain increased bladder cancer incidence or mortality among workers exposed to respiratory asbestos fibers.

When stratifying by year of first exposure, an increased risk of bladder cancer was found for workers employed between 1908 and 1940. This result should be interpreted with caution because it derives from multiple stratified analyses.

Our results showed that female workers had a higher bladder cancer mortality rate. This result was based on five studies comprising a total of 27 observed deaths. The pooled result was greatly influenced by the article by Ferrante et al. [9], and was not confirmed in the parallel analysis of cancer incidence among women. Ferrante et al. reported that their preliminary analyses suggested that the risk of bladder cancer was concentrated in the industrial sectors where asbestos exposure was associated with combustion fumes and other agents related to metalworking and painting. These industries had been classified in relation to carcinogenic risk, including bladder cancer [74, 75]. Furthermore, in four out of five studies no data on smoking habits, the most important environmental risk factor for bladder cancer, were considered. Cigarette smoking prevalence is high among women working in the construction industry and in construction and extraction occupations, as shown in a study done by Mazurek et al. in the United States [76]. Also, the higher bladder cancer mortality in the asbestos workers sub-cohort can probably be attributed to the increased smoking habit rates in this subgroup compared to the general population, as shown in a recent study by Frost et al. [77].

Regarding the major influence of tobacco smoke in the etiopathogenesis of bladder cancer, results from a recent meta-analysis [78] delineate a pooled relative risk of bladder cancer disease-specific mortality of 1.47 (95% CI: 1.24–1.75) for all smokers. In line with this result, a combined analysis of 11 case-control studies from Europe shows that the proportion of bladder cancer cases among women attributable to ever smoking was 0.30 [79].

Furthermore, for other organs there is no evidence of a difference between men and women in the development of asbestos-related diseases. This datum could suggest that the excess in bladder cancer risk for women only is not likely to represent a causal association.

This systematic review and meta-analysis suffers from some limitations. Information about smoking habits of workers, which is the main risk factor for bladder cancer and a potential confounder, is lacking from the studies we reviewed. Thus, the pooled SMR and SIR could be overestimated.

Another limitation is the lack of quantitative data on asbestos exposure and duration of the employment in most cohorts. To analyze the dose-response effect we considered the duration of the employment as a proxy for the dose, but only 12 out 60 studies provided this datum. However, the meta-regression resulted in an absence of a dose-response effect.

Conclusions

Our meta-analysis provides evidence that workers with occupational asbestos exposure have a bladder cancer incidence and mortality rate similar to the general population.

Excesses in bladder cancer risk in selected groups of workers, and in particular women, are not likely to represent causal associations; however, further studies are needed to evaluate whether female workers are more likely to develop bladder cancer when occupationally exposed to asbestos.

Due to the limited data on the duration of the exposure, it is not clear whether length of employment has a significant role in bladder cancer.