Lung cancer risk and occupational pulmonary fibrosis: systematic review and meta-analysis

Abstract

Background:

Molecular pathways found to be important in pulmonary fibrosis are also involved in cancer pathogenesis, suggesting common pathways in the development of pulmonary fibrosis and lung cancer.

Research question:

Is pulmonary fibrosis from exposure to occupational carcinogens an independent risk factor for lung cancer?

Study design and methods:

A comprehensive search of PubMed, Embase, Web of Science and Cochrane databases with over 100 search terms regarding occupational hazards causing pulmonary fibrosis was conducted. After screening and extraction, quality of evidence and eligibility criteria for meta-analysis were assessed. Meta-analysis was performed using a random-effects model.

Results:

52 studies were identified for systematic review. Meta-analysis of subgroups identified silicosis as a risk factor for lung cancer when investigating odds ratios for silicosis in autopsy studies (OR 1.47, 95% CI 1.13–1.90) and for lung cancer mortality in patients with silicosis (OR 3.21, 95% CI 2.67–3.87). Only considering studies with an adjustment for smoking as a confounder identified a significant increase in lung cancer risk (OR 1.58, 95% CI 1.34–1.87). However, due to a lack of studies including cumulative exposure, no adjustments could be included. In a qualitative review, no definitive conclusion could be reached for asbestosis and silicosis as independent risk factors for lung cancer, partly because the studies did not take cumulative exposure into account.

Interpretation:

This systematic review confirms the current knowledge regarding asbestosis and silicosis, indicating a higher risk of lung cancer in exposed individuals compared to exposed workers without fibrosis. These individuals should be monitored for lung cancer, especially when asbestosis or silicosis is present.

Shareable abstract

Asbestosis and silicosis can indicate a higher risk of lung cancer in exposed individuals, but they are not obligatory precursors for lung cancer. Individuals should be monitored for lung cancer especially when asbestosis or silicosis is present. https://bit.ly/47u9pWa

Introduction

A recent meta-analysis identified idiopathic pulmonary fibrosis (IPF) as an independent risk factor for lung cancer [1]. Molecular pathways found to be important in pulmonary fibrosis, such as growth factor cascade, proliferative and pro-angiogenic pathways, are also involved in cancer pathways [2, 3], suggesting common pathways in the development of pulmonary fibrosis and lung cancer.

Chronic lung inflammation due to persistent inhaled particles or other foreign bodies has been described for various occupational exposures, such as asbestos, silica dust or welding fume particles [4–6]. For those exposures, the development of lung fibrosis as well as lung cancer has been reported [7, 8], both originating from chronic inflammation, e.g. via NOD-like receptor family, pyrin domain-containing 3 (NLRP3) inflammasome [9, 10]. In this context, pulmonary fibrosis caused by inhalation of asbestos is called asbestosis and that due to silica inhalation is defined as silicosis. In the German Ordinance on Occupational Diseases, a documented case of silicosis is a mandatory criterion for obtaining occupational disease compensation for diagnosed lung cancer [11].

The assumption that a pre-existing pulmonary fibrosis due to occupational exposure increases the risk for a consecutive occurrence of lung cancer is obvious. Especially since pulmonary fibrosis due to occupational exposure usually indicates a rather high cumulative dose, which also indicates an increased lung cancer risk. However, only a few studies have addressed the question of whether the existence of a pulmonary fibrosis is an independent risk factor for lung cancer independent from cumulative dose. Studies and meta-analyses have found an increased risk of lung cancer for silicotics; however, some studies have also observed an increased risk for nonsilicotics [12], while others found no risk increase [13–15]. Some studies describe an increasing risk after asbestos exposure with pulmonary fibrosis [16, 17] and a consecutive degree of severity [18]. Independent of the pre-existence of pulmonary fibrosis, the International Agency for Research on Cancer (IARC) has declared crystalline silica and asbestos as human carcinogens [7, 8].

This systematic review aims to address the question of whether pulmonary fibrosis due to occupational exposure to chemical, physical or biological hazards, irrespective of its carcinogenic potential, poses a risk factor for lung cancer in individuals exposed to asbestos, silica or other fibrogenic agents. This could be used to identify patients at increased risk for lung cancer due to occupational exposure. Additionally, it may have medicolegal implications, influencing the recognition of and compensation for occupational diseases.

Material and methods

This review adheres to the guidelines set by the Office of Health Assessment and Translation (OHAT) of the National Toxicology Program [19, 20], as well as the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [21]. The search strategy, inclusion and exclusion criteria, and data to be extracted from included articles were specified in a protocol at the beginning of the project. The systematic study protocol was registered with the international prospective register of systematic reviews (PROSPERO) beforehand. The review protocol is available online (PROSPERO-ID: CRD42021255643).

Eligibility criteria

The eligibility criteria were defined using the population, exposure, comparator, outcome, study design concept (PECOS) [19]. Journal articles were included in this review when they reported investigations of adults (>17 years old) with exposures to chemical, physical or biological hazards with potential to cause a pulmonary fibrosis (population). Studies were only eligible if they included groups with and without pulmonary fibrosis (exposure and comparator). Outcome reports were restricted to diagnosis of lung cancer. Information about exposure intensity (dose) was taken into account but was not a prerequisite.

Information sources and search strategy

Peer-reviewed journal articles written in English or German were included in this review if they described randomised controlled trials and observational studies such as cohort, case–control and cross-sectional studies, as well as autopsy studies (study design). The year of publication was not restricted. Reviews, comments, nonpeer-reviewed studies and case reports were excluded. Relevant journal articles published through May 2021 were identified using electronic database searches (all fields) in PubMed (US National Library of Medicine, National Institutes of Health, https://pubmed.ncbi.nlm.nih.gov), Embase (www.embase.com), Cochrane Central (www.cochranelibrary.com) and Web of Science (www.webofknowledge.com). In addition to databases, reference lists from review articles and the results of free searches were screened to identify further studies in July 2021.

During protocol development, a list of about 100 search terms for substances, occupations and superordinate search terms was compiled to include the highest possible number of different exposures leading to work-related pulmonary fibroses. The list can be found in supplementary table 1. The search strategy for the systematic search in the databases was as follows: (“pulmonary fibrosis” OR “lung fibrosis”) AND (“lung cancer” OR “bronchial carcinoma” OR “lung carcinoma”) AND X. X represents the insertion of the aforementioned search terms.

Selection and data collection process

Screening for eligibility of all potentially relevant articles was conducted in two stages. First, the titles and abstracts of the identified articles were screened by two authors. In the second stage, the full text was retrieved for those publications that met the inclusion criteria and the articles were independently reviewed by the same two authors. The authors jointly made a final decision about the inclusion or exclusion of the reviewed articles.

Extracted data included study identification (author, title and year), funding and conflicts of interest, description of study population (e.g. demography, geography, number of participants, recruiting strategy and inclusion and exclusion criteria), methods (e.g. study design, follow-up, outcome and measurement of exposure) and results (e.g. qualitative and statistic results, statistical power, and dose–response). Missing effect parameters were calculated if the underlying data were available [16, 17, 22–57]. All data were extracted and summarised by one reviewer, while a second reviewer performed a thorough quality check for completeness and accuracy of the data extraction.

Meta-analysis

Meta-analysis was conducted if both of the following two eligibility criteria were met to avoid low power [58]: 1) heterogeneity is acceptable (I²<50%) and 2) at least seven studies can be included. In meta-analyses that did not meet the criteria for homogeneity, we conducted subgroup analyses in an effort to account for heterogeneity due to observed effect parameters or a lack of confounder adjustment. Odds ratios (ORs) were studied as the effect parameters. Subgroups were assigned by the ORs determining lung cancer rates obtained in living patients (OR_alive) or dead patients, e.g. in autopsies (OR_dead), and the OR determining the risk of dying from lung cancer (OR_mortality).

Meta-analysis was performed with RevMan 5 (version 5.4.1) in nonCochrane mode (Cochrane, London, UK) using the generic inverse-variance method with a random-effects model.

Study risk of bias assessment

The approach recommended by the OHAT of the National Toxicology Program [19, 20] was used to assess the internal validity and the quality of the included studies. The questions included for assessment can be found in supplementary table 2. The OHAT criteria were independently assessed by two authors for all included studies. The following ratings were used: “++” definitely low risk of bias, “+” probably low risk of bias, “−” probably high risk of bias and “− −” definitely high risk of bias. Disagreements in the assessment were discussed between the authors and resolved by consensus. According to the OHAT approach, studies were categorised into three quality tiers based on their risk-of-bias ratings, as follows: first tier, high confidence in the reported results; second tier, moderate confidence in the reported results; and third tier, low confidence in the reported results.

Certainty assessment

In accordance with the OHAT guidelines, the available studies were initially grouped by key study design features and each group of studies was given an initial confidence rating based on those features. This initial rating was then assessed for downgrading factors that decreased confidence in the results or upgrading factors. Following determination of the confidence in the body of evidence, the groups were assigned one of the following four levels of confidence: “high” (the true effect is highly likely to be reflected in the apparent relationship), “moderate”, “low” and “very low” (the true effect is highly likely to be different from the apparent relationship).

Finally, the confidence in the body of evidence was translated into a descriptor of the evidence for health effects using the confidence ratings and direction of the effect (“health effect” or “no health effect”) choosing between five levels of evidence (“high”, “moderate”, “low”, “evidence of no health effect” and “inadequate evidence”).

Results

A total of 1813 articles was identified by systematic searches, as follows: 463 in PubMed, 1008 in Embase, 333 in Web of Science and nine in Cochrane Central. After removal of duplicates, 1298 articles underwent screening based on title and abstract. Among these, 1167 studies were excluded because they did not match the eligibility criteria. The full texts of the remaining 131 articles were obtained in order to check for their eligibility for inclusion in the current analysis. An additional free search in PubMed and exploration of the reference lists of 176 reviews resulted in the inclusion of another 102 articles eligible for full-text screening. A total of 233 articles entered full-text screening stage. Finally, 52 articles fulfilled the eligibility criteria and were included in this systematic review (figure 1). The screening process was conducted with moderate interrater reliability for the title and abstract screening phase (Cohen's kappa 0.744, 95% CI 0.684–0.805) and high interrater reliability for the full-text screening stage (Cohen's kappa 0.884, 95% CI 0.814–0.954) [59].

FIGURE 1.

Flow diagram of literature search, eligibility, and inclusion process according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.

Studies were categorised into three groups based on the cause of pulmonary fibrosis, as follows: asbestos (21 studies), silica (25 studies) and other exposures, including mixed and coal dusts (six studies). Studies included effect parameters determining lung cancer rates obtained in living patients (OR_alive) and in dead patients, e.g. in autopsies (OR_dead), and determining the risk of dying from lung cancer (OR_mortality). The characteristics of the studies included in the three groups can be found in table 1.

TABLE 1.

Characteristics of studies included in this review: asbestos, silica dust and mining

Study (year)	Design	Study population	Number #	Effect parameter OR (95% CI)
Asbestos
Case and Dufresne [26] (1997)	Case–control study, autopsy study	Chrysotile miners and millers	111	ORdead 2.41 (0.92–6.34) (from raw data)
Cheng and Kong [27] (1992)	Cohort study, retrospective mortality study	Workers employed in asbestos textiles, friction material and asbestos cement manufacturing	1172	ORdead 2.60 (1.01–6.32)
Churg and Vedal [22] (1994)	Cross-sectional study	Shipyard workers and insulators	144	ORdead 2.16 (0.82–5.67) (from raw data)
Coutts et al. [28] (1987)	Cohort study, prospective mortality study	Male cases of asbestosis	58	ORdead 0.63 (0.11–3.40)
de Klerk et al. [18] (1997)	Case–control study	Former crocidolite workers	896	ORalive 3.11 (1.79–5.42)
Doll [29] (1955)	Case–control study, autopsy study	Workers employed at a large asbestos works	105	ORdead 2.25 (0.6–8.42) (from raw data)
Finkelstein [30] (1997)	Cohort study, prospective mortality study	Asbestos cement workers	117	ORmortality 2.00 (0.83–4.80) (from raw data) ORdead 0.86 (0.34–2.19) (from raw data)
Green et al. [23] (1997)	Case–control study, autopsy study	Former workers in a chrysotile asbestos textile plant	55	ORdead 19.93 (2.30–172.84) (from raw data)
Hillerdal [31] (1994)	Cohort study	Men with pleural plaques from the general population	1596	ORalive 1.98 (0.95–4.16) (from raw data)
Hughes and Weill [24] (1991)	Cohort study, prospective mortality study	Workers employed in the manufacture of asbestos cement products	642	ORmortality 4.54 (1.93–10.67) (from raw data) ORdead 2.25 (0.87–5.80)
Jacob and Anspach [32] (1965)	Cohort study, retrospective mortality study	Workers exposed to asbestos	2645	ORdead 0.55 (0.19–1.63) (from raw data) ORalive 13.88 (5.75–33.49) (from raw data) ORmortality 1.98 (0.79–4.95) (from raw data)
Koskinen et al. [17] (2002)	Cohort study	Workers employed in building construction	16 696	ORalive 4.93 (3.36–7.22) (from raw data)
Liddell and McDonald [33] (1980)	Case–control study	Two cohorts of chrysotile miners and millers	1427	ORdead 1.06 (0.07–1.62) (from raw data)
Markowitz et al. [16] (2013)	Cohort study, prospective mortality study	Insulators	2377	ORmortality 2.79 (2.11–3.69) (from raw data, smokers) ORmortality 1.89 (0.72–4.98) (nonsmokers)
Pesch et al. [34] (2010)	Cohort study	High-exposed asbestos workers	576	ORmortality 3.81 (0.90–16.1) (from raw data) ORdead 3.74 (0.85–16.47) (from raw data)
Reid et al. [35] (2005)	Case–control study	Former workers and residents with known amounts of asbestos exposure	1988	ORalive 4.41 (2.54–7.67) (from raw data)
Sluis-Cremer and Bezuidenhout [36] (1989)	Case–control study, autopsy study	Amphibole asbestos miners	399	ORdead 8.70 (4.07–18.57) (from raw data)
Wagner et al. [37] (1986)	Cross-sectional study, autopsy study	Dockyard workers	189	ORdead 0.56 (0.21–1.52) (from raw data)
Wilkinson et al. [60] (1995)	Case–control study	Lung cancer patients	949	ORalive 1.44 (1.04–2.00) (from raw data)
Yotsumoto et al. [25] (2018)	Case–control study	Workers with asbestos exposure	120	ORalive 6.0 (0.65–55.40) (from raw data)
Zhong et al. [38] (2008)	Cohort study, retrospective mortality study	Workers from an asbestos plant	572	ORmortality 4.81 (2.32–9.95) (from raw data)
Silica dust
Amandus and Costello [61] (1991)	Cohort study	Metal miners	9912	ORmortality 3.15 (1.79–5.54)
Amandus et al. [39] (1995)	Cohort study	Workers in dusty trades	2689	Relative risk 3.9 (2.4–6.4) (adjusted for smoking and age) No OR
Checkoway et al. [62] (1999)	Cohort study	Workers in a diatomaceous earth facility	1879	ORmortality 1.89 (0.64–5.39)
Chen et al. [63] (1992)	Cohort study	Workers in metal mines and pottery factories	Unclear	Relative risk 1.22 (0.9–1.6) No OR
Chen and Chen [41] (2002)	Case–control study	Tin miners	757	ORdead 1.92 (1.31–2.83) (from raw data)
Chen et al. [64] (2006)	Cohort study	Tin miners	4629	ORmortality 3.29 (2.33–4.64)
Cocco et al. [65] (2001)	Case–control study	Workers in tungsten, copper, iron, tin and clay mines and in pottery factories	1024	ORmortality 1.32 (0.96–1.80)
Dong et al. [42] (1995)	Cohort study	Silica and clay brick workers	6003	ORmortality 2.7 (1.65–4.41) (from raw data) ORdead 0.94 (0.57–1.56) (from raw data)
Finkelstein [66] (1995)	Case–control study	Workers exposed to silica dust	2091	ORalive 2.87 (1.49–5.51) ORmortality 3.31 (1.5–7.29) ORdead 2.07 (0.85–5.04)
Finkelstein [67] (1995)	Case–control study	Miners	1298	ORalive 2.86 (1.40–5.85)
Finkelstein [43] (1998)	Case–control study	Workers with silicosis	168	ORalive 4.36 (1.75–10.88)
Forastiere et al. (1986) [44]	Case–control study	Workers in ceramic industry	128	ORdead 2.65 (1.15–6.09) (from raw data)
Hessel et al. [45] (1990)	Case–control study, autopsy study	Workers in gold mining industry	549	ORdead 1.10 (0.78–1.55) (from raw data)
Hessel et al. [46] (1986)	Case–control study	Workers in gold mining industry	399	ORdead 1.49 (0.94–2.34)
Hnizdo and Sluis-Cremer [68] (1991)	Cohort study	Gold miners	Unclear	ORmortality 0.9 (0.5–1.6) (adjusted for age at death, particle years and pack-year equivalents)
Hnizdo et al. [47] (1997)	Case–control study, autopsy study	Gold miners	416	ORdead 2.45 (1.25–4.79) (from raw data)
Hua et al. [69] (1994)	Case–control study	Workers in tin mines	267	ORdead+alive 2.08 (1.21–3.57) (from raw data)
Liu et al. [48] (2013)	Cohort study	Workers exposed to silica dust without exposure to carcinogenic confounders	23 628	ORalive 1.39 (1.12–1.72) (from raw data)
Mastrangelo et al. [49] (1988)	Case–control study	Lung cancer patients	261	ORalive 1.84 (1.07–3.15) (from raw data)
McLaughlin et al. [70] (1992)	Case–control study	Men with lung cancer as cause of death	1678	ORdead 1.2 (0.9–1.6) (from raw data)
Meijers et al. [50] (1996)	Cohort study	Workers in ceramic industries	160	ORdead 2.92 (1.19–7.2) ORmortality 7.24 (3.31–15.83)
Qiao et al. [71] (1997)	Cohort study	Radon- and arsenic-exposed tin miners	7867	Unclear
Sherson et al. [51] (1991)	Cohort study	Foundry workers	6054	ORalive 3.18 (1.68–6.00) (from raw data)
Tsuda et al. [52] (2002)	Case–control study	Brick workers	126	ORdead 2.05 (0.98–4.27)
Westerholm et al. [53] (1986)	Cohort study	Workers in mining, tunnelling, quarrying and iron and steel foundries	1522	ORalive 1.22 (0.59–2.55) ORmortality 2.45 (1.05–5.72) ORdead 2.34 (0.97–5.67)
Mixed dusts/coal dusts
Dalal et al. [54] (1991)	Cross-sectional study, autopsy study	Coal miners	92	ORdead 3.21 (1.05–9.84) (from raw data)
Honma et al. [72] (1993)	Cross-sectional study, autopsy study	Autopsy cases with nonasbestos pneumoconiosis	unclear	ORdead 1.15 (0.57–2.33) (from raw data)
Honma et al. [73] (1997)	Cross-sectional study, autopsy study	Autopsy cases with nonasbestos pneumoconiosis	759	ORdead 0.64 (0.38–1.08)
Katabami et al. [55] (2000)	Cross-sectional study, autopsy study	Patients with nonasbestos pneumoconiosis	563	ORdead 6.15 (3.44–11.00) (from raw data)
Rooke et al. [56] (1979)	Cross-sectional study, autopsy study	Miners and ex-miners	1003	ORdead 0.73 (0.49–1.08) (from raw data)
Vallyathan et al. [57] (1987)	Cross-sectional study, autopsy study	Miners with lung cancer	277	ORdead 3.05 (1.59–5.85) (from raw data)

Results highlighted in bold indicate a significant increase in odds ratio (confidence interval does not include 1).^#: indicates the number of patients with information about lung cancer and pulmonary fibrosis. OR_alive: obtained in living patients; OR_dead: obtained in dead patients; OR_mortality: death from lung cancer.

Regarding risk of bias for asbestos, six studies (29%) were assigned to the first tier, 13 (57%) to the second tier and two (14%) to the third tier (figure 2). For silica dust, two studies (8%) were assigned to the first tier, 21 (84%) to the second tier and two (8%) to the third tier (figure 2). Finally, for other exposures, no studies were assigned to the first tier, five (83%) to the second tier and one (17%) to third tier (figure 2).

FIGURE 2.

Risk-of-bias ratings for studies reporting on a) asbestos (n=21), b) silica (n=25) and c) mining (n=6). Criteria ratings served as the basis for the assignment of individual studies to one out of three study quality categories. ++: Definitely low risk of bias; +: probably low risk of bias; −: probably high risk of bias; − −: definitely high risk of bias; NR: not reported.

Narrative synthesis

Asbestos

Hughes and Weill [24] conducted a study of asbestos cement factory workers assessing lung cancer risk compared to lung fibrosis diagnosed from radiographs. Without fibrosis, no increased lung cancer risk could be detected compared to age-adjusted rates from the general Louisiana population; while, with asbestosis, a significant increase was detected. However, with no adjustments for smoking or cumulative exposure, it is unclear if asbestos is an independent risk factor for lung cancer or just indicates higher cumulative exposure in this study. For 16 696 male workers included in a Finnish asbestos screening campaign, Koskinen et al. [17] found no overall increased lung cancer risk in the cohort. For workers with asbestosis, an increased risk could be found, even when adjusted for smoking, age, pleural plaques, asbestos exposure index and occupation (relative risk 1.9, 95% CI 1.3–2.7). Vice versa, an association of asbestos exposure index with lung cancer risk was also detected when adjusting for asbestosis, smoking, age, pleural plaques and occupation, indicating asbestosis as a risk factor for lung cancer independent from a high cumulative exposure. Markowitz et al. [16] conducted a study comparing lung cancer mortality for asbestos-exposed and nonexposed workers. Here, combined supra-additive effects for asbestos and smoking could be identified. For this review, the increase of lung cancer risk for workers with asbestosis was compared with the risk in workers without asbestosis. A significant increase could be observed (OR_mortality 2.79, 95% CI 2.11–3.63). However, due to the small number of nonsmokers in this study, only a nonsignificant risk increase remains after only considering nonsmokers (OR_mortality 1.89, 95% CI 0.72–4.98). Therefore, asbestosis indicated an increased lung cancer risk, but it could not be verified if this association was independent of smoking or cumulative exposure.

Asbestosis as independent risk factor for lung cancer was also identified in a study by Reid et al. [35]. After adjustment for smoking, age, sex, residency and cumulative exposure, an increased risk for lung cancer was associated with asbestosis (OR_alive 1.94, 95% CI 1.02–1.42). Without asbestosis, an increased cumulative exposure was associated with an increased lung cancer risk; again indicating that asbestosis is an independent risk factor for lung cancer.

Studies indicating no increased lung cancer risk associated with asbestosis were also included in this review [25, 28, 30, 31, 33]. This could be explained at least partially by small numbers of participants or a low statistical power.

In summary, studies with adjustments for smoking and cumulative exposure indicated that asbestosis could be an independent risk factor.

Silica dust

Many of the included studies regarding silica dust exposure were conducted in overlapping or identical study populations. Multiple studies report a cohort of workers from mines and plants in China [41, 48, 63, 64, 65, 70]. Chen et al. [63] observed no significant increase in lung cancer risk for silicotics in this cohort (relative risk 1.22; 95% CI 0.9–1.6). In a case–control study of this population by McLaughlin et al. [70], increased ORs for silicotics compared to nonsilicotics could only be observed for workers in iron–copper and tin mines, and not in pottery factories or tungsten mines. In tin mines, increased lung cancer risk was also associated with exposure to arsenic and polycyclic aromatic hydrocarbons. Similarly, in a study by Cocco et al. [65] using the same cohort, an increased lung cancer risk for silicotics adjusted for age, smoking and cumulative dust exposure (OR 1.5, 95% CI 1.0–2.1) was only described if workers were exposed to carcinogenic cadmium or polycyclic aromatic hydrocarbons. Focusing on tin mines, increases in lung cancer risk for silicotics were observed in two studies, but confounding co-exposures with arsenic and smoking must be suspected [41, 64]. One study was published as an update of this cohort in 2013 [48]. Here, workers exposed to silica dust without relevant co-exposure were included. A significant increase in lung cancer risk was observed for silicotics with an adjustment for smoking (OR 1.61, 95% CI 1.29–2.03).

Another study population of workers from Canada was assessed in multiple studies [43, 66, 67]. Confounding co-exposure with radon was present for workers. However, even after adjustment for radon exposure, silicosis continued to exhibit a significant association with an increased risk of lung cancer [66, 67].

In a study involving a population employed in the ceramic industry in the Netherlands, Meijers et al. [50] found an increased risk of lung cancer for silicotics but not for nonsilicotics compared to the general population. In a direct comparison, exposed silicotics had a significantly higher lung cancer risk than exposed nonsilicotics (OR_mortality 7.24; 95% CI 3.31–15.83; OR_dead 2.92; 95%-CI 1.19–7.2).

Additional studies consisted of small numbers of participants and low statistical power [40, 51, 53].

Regarding cumulative exposure, some studies indicated a correlation between silicosis and an increased risk of lung cancer dependent on cumulative exposure [45, 46, 68, 69]. Whereas other studies reported a marginally significant (95% CI starts with 1.0) risk increase independent of cumulative exposure [41, 47, 65].

In summary, there is a moderate level of evidence for silicosis as an independent risk factor for lung cancer. It cannot be excluded that this risk association is dependent on co-exposure and cumulative dose.

Mining (mixed and coal dusts)

For mining including mixed dusts, only studies with a very low level of confidence were included in this review. The heterogeneity and lack of characterisation of exposure was especially problematic in this group. Multiple articles reported on overlapping study populations with nonasbestos pneumoconiosis, which included silicosis as well as mixed-dust pneumoconioses and coal worker's pneumoconiosis (CWP) [55, 73, 74]. Exposure characterisation in these studies was solely based on job descriptions. Two studies found no increase in lung cancer risk for patients with pulmonary fibrosis compared to nonfibrotics. However, in cases with diffuse interstitial fibrosis, lung cancer was primarily located in fibrosis sites [73, 74]. Katabami et al. [55] found a positive relationship between pneumoconioses and squamous cell carcinoma of the lung. Incidence of lung cancer was significantly higher in fibrotics compared to nonfibrotics (53% versus 15%). Due to the considerable heterogeneity of studies and exposure, no definitive conclusion can be drawn from these studies regarding lung cancer risk.

Studies reporting outcomes from coal workers usually focus on health effects related to coal dust; however, co-exposure with silica dust is common. Accordingly, in a study by Dalal et al. [54], histological findings typical of silicosis are reported. Analysis of data encompassing progressive massive fibrosis (PMF), macular CWP and silicosis did not reveal an increased risk of lung cancer (OR_dead 1.05, 95% CI 0.45–2.44). In the case of coal workers, Rooke et al. [56] reported an increased lung cancer risk specifically for workers without pneumoconiosis, but not with fibrosis solely determined via macroscopic assessment, clearly restricting the study's quality. Vallyathan et al. [57] found an increased risk for lung cancer for CWP without specific characterisation.

In summary, for mixed-dust exposure and CWP as risk factors for lung cancers, only a very low level of evidence and a general lack in quality of studies exists, rendering a statement about lung cancer risk impossible.

In terms of other exposures known to cause pulmonary fibrosis and lung cancer, e.g. hard metal dust fibrosis, several articles were included in the initial search, but none of them were included after the full-text screening. One study reported increased respiratory cancer rates in a metal fabrication facility [75]; however, those cases also included mesothelioma and laryngeal cancer. A systematic review found an increased lung cancer risk in silicon carbide workers [76]. As these workers were also exposed to silica dust, no clear attribution to hard metals could be concluded. Therefore, there is some indication in the literature that hard metals can induce pulmonary fibrosis and an increased lung cancer risk. However, the available literature is insufficient to allow for conclusions regarding their interaction.

Meta-analysis

Following the OHAT handbook [19], studies with the same effect parameters from overlapping populations were identified and one study was chosen for analysis based on publication year, number of participants and relevance before conducting the meta-analysis.

Asbestos

A total of 17 studies met the eligibility criteria and were included in the meta-analysis for asbestos. Four were excluded due to a high risk of bias (third tier) [29, 32], an overlap of study populations [18] and a mix of exposed and unexposed patients in the group determining the effect parameters [60]. With a heterogeneity of I²=73%, meta-analysis could not be conducted. This was also true for analysis of the subgroups separated by effect parameters: five studies and I²=37% for OR_alive, 12 studies and I²=74% for OR_dead and five studies and I²=0% for OR_mortality. Even when restricting the analysis to studies with adjustments for smoking as a confounder, the criteria were not met (seven studies, I²=61%).

Silica dust

Out of 25 studies on silica dust, 14 were eligible for meta-analysis. 11 were excluded due to a high risk of bias (third tier) [51, 52] and an overlap of study populations [39, 63–70]. With a heterogeneity of I²=75%, meta-analysis could not be conducted for the overall group. Analysis of subgroups separated by effect parameters was not possible for OR_alive (five studies and I²=49%). However, for OR_dead (nine studies, I²=47%) and OR_mortality (eight studies and I²=0%), the analysis criteria were met. In this subgroup analysis, studies were included that were excluded from the overall analysis if there were no overlaps in the study populations in the subgroup. Regarding lung cancer rates in dead patients, an increase in the risk for lung cancer could be found (OR_dead 1.47, 95% CI 1.13–1.90) for individuals with silicosis compared to individuals without silicosis (figure 3a). Additionally, a considerable increase could be observed in terms of lung cancer mortality (OR_mortality 3.21, 95% CI 2.67–3.87) for silicotics (figure 3b).

FIGURE 3.

Meta-analysis with forest plots for the association of silicosis and lung cancer taken from studies reporting a) lung cancer rates in dead patients, e.g. in autopsies (OR_dead), and b) risk rates to die from lung cancer (OR_mortality). c) Meta-analysis with forest plot for the association of silicosis and lung cancer taken from studies with risk adjustment for smoking. df: degree of freedom; IV: inverse variance.

Finally, when only considering studies with an adjustment for at least smoking as a confounder and including all effect parameters (OR_alive, OR_dead and OR_mortality), eight studies with I²=14% could be identified. Here, a significant increase in lung cancer risk (OR 1.58, 95% CI 1.34–1.87) could be observed for individuals with silicosis compared with individuals without silicosis (figure 3c).

Mining and other exposures

A mix of different exposures was included in this group, rendering a meta-analysis unreasonable. Overlapping study populations with silicosis, mixed-dust pneumoconioses and CWP [55, 73, 74] were included.

Evidence synthesis

Confidence of the body of evidence

Evidence synthesis began with an assessment of initial confidence based on key features of the study designs, following the guidelines of the OHAT. Since the studies in this review did not include controlled exposure, they could not reach the highest level of initial confidence (++++). Grouped by exposure, the study groups received initial confidence ratings between moderate (+++) and very low (+) (supplementary table 3).

After assessment for factors that could either decrease or increase confidence, the confidence rating for the body of evidence was established. Moderate confidence was assigned to the asbestos studies, while studies on silica received ratings of both moderate and low confidence (supplementary tables 4 and 5). In the asbestos studies with a low confidence rating, the initial confidence level was reduced due to an unexplained consistency in the data (supplementary table 4). Here, the confidence intervals in the studies did not overlap and the considerable heterogeneity could not be explained by different study designs. In the studies with mixed dust exposure (mining), the initial low confidence had to be decreased further due to an unexplained consistency and the risk of bias (supplementary table 6).

Evidence of health effects

The conclusions of the qualitative analysis for health-related effects are summarised in table 2 and below.

TABLE 2.

Conclusion of qualitative analysis for health-related effects

Exposure	Level of evidence	Conclusion
Asbestos	Moderate	Asbestosis as independent risk factor for lung cancer
Silica dust	Moderate	Silicosis as risk factor for lung cancer, confounding possible
Mining	Very low	No statement due to heterogeneity

Asbestos

Studies regarding asbestos included in this review reached levels of very low to moderate confidence. In the narrative synthesis of the included studies, asbestosis could be identified as a potential independent risk factor for asbestos-related lung cancer. Due to heterogeneity, meta-analysis could not be conducted.

Silica

Studies regarding silica included in this review reached levels of low to moderate confidence. In the narrative synthesis of the included studies, silicosis could be identified as a risk factor for lung cancer, although confounding and cumulative exposure could not be sufficiently evaluated. In the subgroup meta-analysis concerning lung cancer in dead patients (OR_dead), an increased risk for lung cancer could be found, as well as a considerable increase in lung cancer mortality. Finally, when only considering studies with an adjustment for at least smoking as a confounder and including all effect parameters, a significant increase in lung cancer risk (OR 1.58, 95% CI 1.34–1.87) could also be observed.

Mining (mixed and coal dust)

In summary, when considering mixed-dust exposure and CWP as risk factors for lung cancers, there is only a very low level of evidence and a general lack in quality of studies. As a result, making a conclusive statement about the risk of lung cancer is impossible.

Discussion

In this systematic review, asbestosis and silicosis could be identified as risk factors for lung cancer. For asbestos, the qualitative review of the literature found evidence supporting asbestosis as a risk factor independent of smoking and cumulative exposure. For silicosis, meta-analysis of subgroups and studies adjusted for smoking found a significant risk increase for silicosis and lung cancer. Qualitative analysis identified relevant co-exposure in the included studies, thereby limiting the actual significance of silicosis as a risk factor for lung cancer.

An increased risk of lung cancer has also been described for other interstitial lung diseases outside of occupational medicine. For example, IPF represents an independent risk factor for lung cancer [1]. It is frequently found in elderly smoking men [77], who are also a typical risk group for occupational lung cancer, e.g. in asbestosis [78]. IPF and lung cancer share certain pathways, such as increased programmed cell death-ligand 1 expression, transforming growth factor-β release and durotaxis [79]. Thus, it seems obvious that pulmonary fibrosis due to occupational exposure would present similar risk constellations.

Relation to previous studies

General associations between asbestos and silica exposure and the development of lung fibrosis and/or lung cancer have been reported [7, 8]. In the German Ordinance on Occupational Diseases, a documented silicosis is a mandatory requirement for approval of compensation for lung cancer as an occupational disease [11]. However, at present, it is not clear if this relationship is a direct one or if the existence of pulmonary fibrosis is merely a surrogate parameter for a high cumulative dose, since asbestosis and silicosis occur more frequently with a higher cumulative exposure [80, 81].

Asbestosis

Several studies have addressed the question of whether the presence of asbestosis should be a prerequisite for attributing lung cancer to asbestos exposure. Although investigation of that hypothesis seems similar to the task undertaken in this review, there are differences in terms of the point of view and study selection, as well as differing populations. The currently available scientific evidence indicates that the lung cancer risk is determined by cumulative asbestos exposure with no requirement for asbestosis [82].

Research reporting on asbestosis may be limited by the method used to detect pulmonary fibrosis. In their review, Hessel et al. [83] discussed histologic evidence of fibrosis being more sensitive than high-resolution computed tomography (CT) or radiographs. Thus, fibrosis detection method could influence detection rates. Hillerdal and Henderson [84] elucidated another problem when investigating asbestosis and asbestos-related lung cancer. In many epidemiological studies, asbestos exposure is so high that the number of workers without any asbestosis is rather low, especially in older studies. In fact, in this review, some studies had only a small number of patients without asbestosis.

In an exposure–response estimate, Courtice et al. [81] reported a nearly sixfold increase at the highest exposure levels for lung cancer and a threefold increase for abestosis. This suggests that the lung cancer risk linked to cumulative exposure does not solely consist of the corresponding risk for asbestosis. Weiss [85] reported a high correlation between asbestosis and lung cancer rates in 38 cohorts. However, cumulative exposure data and lung cancer relative risks in eight cohorts did not correlate to the same degree, indicating asbestosis as an independent risk factor for asbestos-related lung cancer.

In summary, this review reports an increased risk for lung cancer in asbestosis patients in accordance with previous studies. A few studies indicate asbestosis as independent risk factor for asbestos-related lung cancer; however, research with reliable adjustments for smoking and cumulative exposure is scarce.

Silicosis

In terms of the association between silicosis and lung cancer, multiple reviews and meta-analyses have been published. Lacasse et al. [86] found an association between silicosis and lung cancer even for nonsmokers. The risk increased with an increasing degree of small and large opacities. However, silica exposure was not quantified or adjusted for. Additionally, Lacasse et al. [87] published a dose–response meta-analysis regarding silica and lung cancer reporting a positive association. However, in most of the included studies, patients with silicosis were not excluded and no adjustment for silicosis was made in the analysis.

Poinen-Rughooputh et al. [12] reported a positive dose–response relationship in terms of silica exposure and lung cancer. Studies with silicotics showed a higher risk for lung cancer compared with nonsilicotics; however, the studies were not adjusted for cumulative exposure, again indicating a positive association between silicosis and lung cancer.

A meta-analysis by Pelucchi et al. [15] identified a positive association between silica exposure and lung cancer. Risk increases were observed for silicotics, but not for nonsilicotics; although only a single study of nonsilicotics was included in the meta-analysis of cohort and case–control studies, respectively. Again, no reliable conclusion could be drawn regarding the dependency of silicosis and silica exposure in association with lung cancer.

In summary, this review reporting an increased risk for lung cancer with silicosis is in accordance with previous studies regarding silicosis as a risk factor. However, the lack of studies researching the dependency of silicosis and silica exposure in association with lung cancer limits conclusions about the association between silicosis and lung cancer.

Mining (mixed and coal dusts)

A review study found no increase in lung cancer risk for CWP, with most studies including adjustments for smoking [88]. Tomášková et al. [89] reported an increased lung cancer death risk for coal miners with CWP and especially for severe CWP and PMF, but it was not adjusted for smoking or cumulative exposure. In a study that included part of the cohort of the aforementioned study, a higher risk of lung cancer was again found in miners with CWP and that risk correlated with the severity of CWP [90].

Limitations

Since only observational studies were included in this review, initial confidence at the highest level (++++) could not be achieved. Furthermore, studies with a low number of patients, limited pulmonary fibrosis cases, low statistical power or no adjustments for confounding factors were included, thereby limiting the overall conclusions of this review.

Additionally, inconsistent terms for lung cancer, e.g. pulmonary cancer, lung neoplasm and especially pulmonary fibrosis (e.g. lung fibrosis, silicosis, asbestosis and exposures) (supplementary table 1), complicated the systematic search. Thus, an additional free search including the reference lists of reviews was conducted. No registers were found that documented exposures potentially causing pulmonary fibrosis. Search terms were identified by a thorough literature search and consultations with experienced practitioners.

Overall, aside studies investigating asbestosis and silicosis, the amount of literature is rather scarce. There was a higher number of case reports, which were not included in this review per se. To reach a preliminary conclusion regarding the association between pulmonary fibrosis and lung cancer aside silicosis and asbestosis, case reports and grey literature should be considered.

As Hessel et al. [83] discussed in their review, the pulmonary fibrosis detection method used and its sensitivity directly influences detection rates. Histologic findings are more sensitive than CT or radiographs, and they represent the only method that can detect asbestosis grade I–a fibrosis limited to the wall of a respiratory bronchiole [91]. Accordingly, the studies included in this review were evaluated in terms of the detection methods used for pulmonary fibrosis and lung cancer, and that evaluation was included in the risk of bias assessment (see category detection bias: outcome). Definite low bias was only attributed if the fibrosis and lung cancer diagnoses were both secured by histology. Studies that included radiographic and CT findings as well as lung cancer information from death certificates and medical history were categorised as probably low bias. However, most studies included fibrosis detection via CT or radiographs, introducing uncertainty regarding the detection of early or mild pulmonary fibroses.

The heterogeneity between studies hindered a meta-analysis for asbestosis and only allowed for the analysis of subgroups in the case of silicosis. This limits the significance of the conclusions from this review. However, the conservative threshold for conducting a meta-analysis of at least seven studies was applied. Given the lack of studies accounting for cumulative exposure, no definitive conclusion can be reached on the independence of asbestosis/silicosis as risk factors for lung cancer. This systematic review depicts the current state and a general overview regarding lung cancer risk in individuals with pulmonary fibrosis due to occupational exposure. Future reviews should focus on particular hazards and include animal studies to further elucidate the connection between fibrosis and lung cancer.

Clinical implications and conclusion

This systematic review confirms the current status of literature describing asbestosis and silicosis, indicating a higher risk of lung cancer in exposed individuals. Nevertheless, it cannot be concluded that exposed individuals without fibrosis do not have a risk for lung cancer or that asbestosis and silicosis are obligatory precursors for lung cancer. However, individuals with exposure to asbestos and/or silica should be monitored for lung cancer, especially when asbestosis or silicosis is present.

A large number of workers are, or have been, potentially exposed to asbestos, silica or other hazards with fibrogenic potential. Accounting for the long latency between exposure and occurrence of fibrosis and lung cancer, studies and systematic reviews including long-term follow-ups are needed. To investigate the association between exposure, pulmonary fibrosis, and lung cancer risk in the future, data on cumulative exposure and smoking status are needed.

Points for clinical practice

• Asbestosis and silicosis indicate a higher risk of lung cancer in exposed individuals.

• Individuals without asbestosis or silicosis are also at risk of lung cancer and asbestosis and silicosis are not obligatory precursors.

• Individuals with exposure to asbestos and/or silica should be monitored for lung cancer, especially when asbestosis or silicosis is present.

Questions for future research

• Accounting for the long latency between exposure and occurrence of fibrosis and lung cancer, studies and systematic reviews including long-term follow-ups are needed; this should specifically include data on cumulative exposure and smoking status.

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material ERR-0224-2023.SUPPLEMENT ^{(409.4KB, pdf)}